GappinganddeterminersharinginSpanish - Luis … · GappinganddeterminersharinginSpanish ... that...
Transcript of GappinganddeterminersharinginSpanish - Luis … · GappinganddeterminersharinginSpanish ... that...
Gapping and determiner sharing in Spanish
Naiara Centeno
University of the Basque Country
November 22, 2011
Abstract
This dissertation is a study of gapping and determiner sharing in Spanish. In gapping,
the verb is missing in the final conjunct. In determiner sharing, the missing elements are
both the verb and a determiner in the final conjunct but both are still interpreted in both
parts of coordination. I make three major claims concerning the analysis of gapping and
determiner sharing. First of all, I propose that both small and large conjuncts are nec-
essary to account for these phenomena; this proposal challenges an implicit assumption
in previous works, namely, that all subtypes of gapping and determiner sharing could be
accounted for by either a small or a large conjunct analysis alone. Second, I propose that
both movement and deletion are necessary to account for all subtypes of gapping and
determiner sharing. Again, this reverses previous assumptions that either movement or
deletion alone should be enough to account for all subtypes of gapping and determiner
sharing. Third, I propose an additional condition to determiner sharing which is not
listed in the previous literature: for determiner sharing to be present in the sentence,
there needs to be an independently triggered movement that affects the DPs that enter
the sharing relationship. The dissertation is organized as follows.
In chapter 2, I focus on gapping in Spanish. I divide gapping phenomena in five differ-
ent types: ‘Basic Gapping’ where the only lexical verb is missing, ‘Gapping of Auxiliary’
where only the auxiliary is missing, ‘Gapping of Modal’ where a modal verb is missing,
‘Complex Gapping’ which can be ‘for Objects’ where both the verb and an argument of
it are missing or ‘for Verbal Complexes’ where whole verbal complexes are missing, and
‘Gapping in NEG-nor configurations’. The analysis of all these types of gapping shows
that their structures are different. Various tests developed in previous studies shows that
these types have each a different syntax. For instance, the scopal interactions of coordi-
nation with negation and modals shows that differences exist among types of gapping,
and the same happens with sloppy readings of pronouns. The structures of the first three
types can be analyzed following the proposals of Johnson (2000) and Lin (2002): vPs are
coordinated and the gap is a consequence of ATB movement of the verb. All complex
gapping types and the gapping type within NEG-nor contexts are dealt via deletion by
phase at PF. Remnants in these cases escape the deletion domain by movements to the
3
4
left of the conjuncts and the size of conjuncts in these cases vary from small to large.
In chapter 3, I turn to the determiner sharing construction and its interaction with the
gapping structures discussed in chapter 2. I show that determiner sharing can be divided
into the same five subtypes that gapping can, as discussed in chapter 1, and I analyze
determiner sharing in both subject and object position. I show that though determiner
sharing is always possible in subject position, it is possible in object position only for some
cases, namely, in cases where gapping is derived by deletion, rather than by movement.
My proposal is to account for missing elements by adapting Lin’s (2002) theory of DETs
to the needs of the structures in Spanish, which basically comprise small conjuncts and
much larger conjuncts, and movement and deletion. I show that the interaction between
the grammaticality contrasts and my proposal –conjunct sizes, movements, and deletion–
presents a scenario where determiner sharing is only possible when the DPs that enter
the sharing relationship move for independent reasons.
Finally, in chapter 4, I study a specific case of determiner sharing, namely wh-
determiner sharing, extending the analyses in Arregi and Centeno (2005) and Centeno
and Vicente (2009). The central feature of the data in this chapter is that sharing can also
affect wh-determiners. It is shown that DET positions need to be above phases and that
the analysis goes hand in hand with what is previously presented. Hence, a concluding
aspect in this chapter is that it provides additional support for some of the proposals in
the previous chapters.
Resumen
Esta tesis es un estudio de “gapping” y “determiner sharing” en espanol. En “gapping” se
omite el verbo en la parte de la coordinacion no inicial pero se interpreta como si estuviera
presente. En “determiner sharing” los elementos omitidos son el verbo y un Determinante
en la parte de la coordinacion no inicial pero tambien se interpretan como si estuvieran
presentes. Mi propuesta sobre el analisis de “gapping” y “determiner sharing” se compone
de tres postulados principales. En primer lugar, propongo que es necesario que la coordi-
nacion se de tanto al nivel de la oracion como a niveles inferiores para dar cuenta de estas
estructuras. Este analisis se opone a la asuncion implıcita que aparece en trabajos previos
de que todos los subtipos de “gapping” o “determiner sharing” se tienen que dar cuenta
o con coordinacion a nivel oracional unicamente o con coordinacion a niveles mas bajos
unicamente. En segundo lugar, propongo que los dos procesos, tanto el de movimiento
como el de borrado, son necesarios en el analisis de estas estructuras. Esto tambien se
opone a asunciones previas que conllevan que basta uno u otro proceso por sı solo para
dar cuenta de todos los tipos diferentes de “gapping” o “determiner sharing”. En tercer
lugar, propongo una condicion adicional al fenomeno de “determiner sharing” que no se
ha propuesto en la literatura previa: para que “determiner sharing” pueda ocurrir en la
frase tiene que haber un movimiento propiciado independientemente que afecte al DP que
entra en la relacion de “sharing”. La tesis esta organizada de la siguiente manera.
En el capıtulo 2 me centro en “gapping” en espanol. Divido los ejemplos de “gapping”
en cinco tipos diferentes: ‘Basic Gapping’ donde solo se omite el verbo lexico, ‘Gapping
of Auxiliary’ donde solo se omite el auxiliar, ‘Gapping of Modal’ donde se omite el verbo
modal, ‘Complex Gapping for Objects or for Verbal Complexes’ donde se omiten el verbo
y un argumento de este o toda la forma verbal, y ‘Gapping in NEG-nor configurations’
en donde gapping esta inmerso en una construccion del tipo ‘ni. . . ni. . . ’. El analisis de
todos estos tipos de “gapping” demuestra que sus estructuras son diferentes. La liter-
atura previa ofrece tests por los cuales el comportamiento de los tipos de gapping es
heterogeneo. Las estructuras de los tres primeros tipos se pueden analizar siguiendo las
propuestas de Johnson (2000) y Lin (2002): la coordinacion ocurre al nivel de vP y el
5
6
“gap” es consecuencia de un movimiento ATB del verbo. Los otros tipos de “gapping”
se analizan mediante “deletion by phase” en PF. Los “remnants” en este caso evitan ser
borrados porque se mueven al extremo izquierdo de las partes coordinadas. En estos casos
el tamano de las partes de la coordinacion varıa de caso a caso.
En el capıtulo 3, continuo con “determiner sharing” y su interaccion con las estructuras
presentadas en el capıtulo 2. Situo el fenomeno de “determiner sharing” en todos los tipos
de “gapping” que se listan arriba y analizo “determiner sharing” en posicion de sujeto y
de objeto. Explico que aunque “determiner sharing” es posible siempre en posicion de
sujeto, no lo es siempre en posicion de objeto. Los casos en los que “determiner sharing”
es posible son aquellos en los que “gapping” se deriva por borrado, en vez de movimiento.
Mi propuesta es dar cuenta de los elementos que se omiten adaptando la teorıa de Lin
(2002) sobre DETs a las necesidades que estas estructuras presentan en espanol. Sobre
todo, se necesitan coordinaciones a nivel oracional y mas abajo, y los procesos de borrado
y movimiento. Demuestro que la interaccion entre los contrastes de gramaticalidad y esto
ultimo –tamano de las partes coordinadas, borrado y movimiento– presenta un escenario
donde “determiner sharing” solo es posible cuando los DPs que entran en la relacion de
“sharing” se mueven por razones independientes.
Por ultimo, en el capıtulo 4, estudio un caso especıfico de “determiner sharing”, ‘wh-
determiner sharing’, extendiendo los analisis de Arregi and Centeno (2005) y Centeno and
Vicente (2009). La caracterıstica principal de los datos de este capıtulo es que “determiner
sharing” tambien puede afectar a Determinantes-wh. Muestro que las posiciones DET
tienen que estar encima de fases y que el analisis funciona perfectamente con lo que se ha
presentado antes. Uno de los aspectos concluyentes de este capıtulo es que proporciona
argumentos adicionales en favor de algunas de las propuestas expuestas en los capıtulos
anteriores.
Acknowledgements
A PhD is certainly not a matter of work only. And it is not only a matter of simpleknowledge. And it might not even be about some courses and some writing only. A PhDcould become a matter of everything: a matter of love, of caring, of helping, of fighting,of working, of learning, of crying, of smiling, of laughing, of jumping obstacles, of askingfor help, of persisting, of hoping, of family, of friends, of desiring, of patience, of irony, ofwaiting, of values, of ideals, of evaluating, of assessing, of re-evaluating, of changing, ofstaying the same, of business, of drama, of seeing the world, of seeing beautiful things,of seeing ugly things, of traveling, of airplanes, trains, buses, cars, of culture shocks, ofshocks in general, of spending times in libraries, bookshops, cafes, bars, shops, cities,hospitals (. . . ), of opening your eyes, of closing them, of playing around, of playing a role,of weeks, months, years, of memories, of dreams, of nightmares, of passion, of dullness. . .
What if it makes sense after all? It is the people with the names listed first those whoreally may have a sense of what this PhD has been. That is why they are the first oneshere. These people know my feelings pretty well and the only way in which I have notsaid thank you to them is publicly. So, here I solve it. To all of you a public “Thankyou!!” Words make no justice to these people but anyways. . . The general order that Ifollow is: family, directors, friends, academia.
Isabel Royo, my mum, and Juan Manuel Centeno, my dad. They have been there allthroughout this process, since day one. Always there since I was born, always magic,always incredible, always fun and full of love. . .Agustin Rius, my partner. Huge heart, huge support, always helping, heaps of love. Aworld of daily surprises, the scientist “counterpart” right at home. . . The smell of mate.Eider Centeno, my sister. Another fundamental pillar within all this. So healthy to listento your opinion. . . So much help and love. My sister, irreplaceable. Isabel Hernaez, myniece. So tiny. . . So powerful. . . And Carlos Hernaez, my brother in law. Who would havethought that day in the Old Part?I am also thankful to more family: those grandmas, grandpas, aunties, aunts, cousins andtheir families, who have been near me or have shown interest in these PhD.
Thanks to those big sentences that have always been there: “Con la constancia se llega altriunfo.” Se lo dijo Moreno a Pujol. “Siempre nos quedara Paris.” “Cuando se estudia seestudia, cuando se come se come, cuando se juega se juega. . . ” “El Tour es muy duro.”“Cuando estas en la cresta de la ola, el aire o el viento te mueven, ası, el flequillo. . . ”“Mas vale una vez colorada que ciento morada. . . ” “Redondea, vos redondea.”
7
8
Olga Porres and her family. Olga, you have also been there all along and in many waysyou should be up here in the family section. You have listened for so many hours. Wehave talked about so many things. . . From nowadays, back to those years, back to Bilbo,back to Aldapeta. . . It is a long way: thank you so very much!
Luis Vicente, director. I have been admiring your knowledge on syntax since our firstlinguistics classes. It was back then when we did our first linguistics work together. Thatwent well. Nowadays, this has worked well, too. That old admiration is still here andthat knowledge of yours about syntax has helped me write this that follows. Thank you,Luis, for all your help. It has been a pleasure to laugh doing syntax again. Thanks foreverything.Alazne Landa, director. Thank you so very much for being there and for providing mewith your help. Thank you for reading this dissertation, for your comments, for help withthe administrative side of this. Thank you for everything.
Jon Franco, the King of shadows. Certainly, thank you for being behind every email.Thank you for all meetings, for all coffees, for all conversations, for all your interest, forall your help. Thank you for everything. Those sun glasses, silent travellers; that whiterope. . .Eduardo Smith, Language House professor. You helped me with every resource you had.You answered all my questions. You read my mind pretty well in so many occasions Iwill always remember how much I have learned because of you, both as a student and asa teacher. You have been of great influence in my formation and I am extremely proudof that.
Xabi and Eva. What can I say! Thank you so very much for everything, everything!Xabi, always there with your views and the views of the whole Team. . . Thank you, thankthem! It is very difficult to describe not only the help that you offer us but also the feelingof gratitude that we feel in written words. You have done a lot for this PhD and youmight probably be among the people that understand it best. . . Thanks a million.
Nora Ibargoyen, Arrate Insausti. We have been and we will be always there. Thankyou so much. Sebastian Arriola –Seba–, Jose Herrera –Josesito–, Gaby, Hedie, Marianne,Roberto and the Blacksburg-Cellar gang in Virginia. . . You made me laugh so much in suchhard times. . . You understood so very well and got interested quite rapidly. . . Weekendswere incredible with you people. I certainly miss you. Emanuel and Ita, Soly, thank youvery much for everything and for your company.Alex Pelham Waerea, Michael Walmsley, Laura Harrison and the Waikato people thatcome to my office to have fun while working and to make of it a magical kiwi place(among them, some of “my” first 390s: Ashley Steel, Ming Xia, Royce Dodd, BrendanSheridan, and again Alex and Laura). To those Waikato students who helped me makemy first LA literature course (. . . ) so incredible, thank you. Alex, Laura, and Michael:thank you for proof-reading parts of this dissertation.
Also in the University of Waikato: thanks to Dan Zirker, Alistair Swale, Athena Cham-bers, the international languages staff members, the School of Arts.Clara, Edgardo Beacon and the Beacon family, in Hamilton, New Zealand, and beyond:you have been of great help and you have been extremely welcoming. For that and for
9
everything, thank you very much. Also, in Hamilton: Alvaro, Cecilia, Cristi, Alfredo,Pablo, Mindy, and families, thank you. Thank you also Carlos, Fabian, Isabelle.
In the University of Illinois at Urbana-Champaign: thank you to those professors whotaught me syntax and morpho-syntax courses: Luis Lopez, Karlos Arregi, James Yoon,Elabbas Benmamoun. To my past committee there and to all those who shared with mesimilar/other linguistic approaches and provided me with the possibility of having a widerpicture of the linguistics field, thank you also. Thank you also to those Spanish coordina-tors who helped me with my duties as Teaching Assistant and with all other pedagogicalmatters of teaching a foreign language. To those department mates who helped me atdifferent points, thank you very much.
In Virginia Tech: thanks a lot to Robin Panneton and the Infant Perception Lab, the peo-ple in the Psychology Department and Datapalooza meetings. Jean Hanigan and MarkHanigan: thank you for that welcoming, for opening your home doors once and again, forthose driving lessons, for letting me help at the library, for that conversation we had atyour house which really helped. . . For your interest and your help, thank you so very much.
John Frampton, Luis Seidel, thank you for technical help in the process of writing thisdocument. Thanks to the audiences in LSRL, LSA, NELS, BIDE, EHU, eggschool: thankyou for your questions, for your answers, for your comments, for your help, for your con-versation, for listening.In the University of the Basque Country: thank you to those with whom I shared a partof my work and my research, and to those who accepted me in many talks I attended sinceI was studying my undergrad studies in Bilbao. I have already mentioned Alazne Landaabove, thank you very much. Thank you very much to the linguistics research teams.
In the University of Deusto and CIDE: thanks a lot to Inmaculada Ballano, Itziar Munoz,Leyre Goitia, Jon Franco again, Margarita Otaegui, and the rest of members in the CIDEteam. Jon Ortiz de Urbina, thank you also for those judgments; to all my English Philol-ogy Professors, thank you very much. Some people in the University of Deusto have givenme a lot and it has been a university that has seen me throughout all these years. I havestudied there and worked there, both in the Bilbao and the San Sebastian campuses. Itis a big part of my life, all in all.
To everyone and the reader, I have worked on what follows and I have written about italways with the same idea in my mind: keep on learning. Echoing my first statementsof purpose, I truly hope this piece of writing gives you a lot of enlightening moments.Thank you all and enjoy.
Agradecimientos
Un doctorado no es ciertamente un asunto de trabajo unicamente. Y no es solo un asuntode simple conocimiento. E incluso puede que no sea sobre unos cuantos cursos y un pocode escritura. Un doctorado puede convertirse en un asunto que lo abarca todo: un asuntode amor, de cuidado, de ayuda, de lucha, de trabajo, de aprendizaje, de llorar, de sonreır,de reırse, de saltar obstaculos, de pedir ayuda, de persistir, de tener esperanza, de familia,de amigos, de desear, de paciencia, de ironıa, de esperar, de valores, de ideales, de evaluar,de calificar, de re-evaluar, de cambiar, de quedarse igual, de “business”, de drama, de verel mundo, de ver cosas bellas, de ver cosas feas, de viajar, de aviones, trenes, autobuses,coches, de shocks culturales, de shocks en general, de pasar tiempo en bibliotecas, entiendas de libros, cafeterıas, bares, tiendas, ciudades, hospitales (. . . ), de abrir los ojos,de cerrarlos, de jugar, de jugar con los papeles, de semanas, meses, anos, de memorias,de suenos, de pesadillas, de pasion, de aburrimiento. . .
¿Y si tiene sentido despues de todo? Son las personas cuyos nombres estan listadosprimero quienes realmente pueden tener una idea de lo que ha sido este doctorado. Esa esla razon por la cual son los primeros. Estas personas conocen mis sentimientos bastantebien y la unica manera en que no les he dado las gracias es de manera publica. Ası queaquı lo soluciono. A todos vosotros, un “¡Gracias!” en publico. Las palabras no hacenjusticia hacia estas personas pero bueno. El orden general que sigo es: familia, directores,amigos, academia.
Isabel Royo, mi ama, y Juan Manuel Centeno, mi aita. Han estado ahı a lo largo de todoel proceso, desde el primer dıa. Siempre ahı desde que nacı. Siempre magicos, siempreincreıbles, siempre divertidos y llenos de amor. . .Agustin Rius, mi pareja. Un corazon enorme, un apoyo enorme, siempre ayudando, canti-dad de amor. Un mundo de sorpresas de a diario, la “contraparte” cientıfica en casa. . . Elolor a mate.Eider Centeno, mi hermana. Otro pilar fundamental en todo esto. Escuchar tu opiniones tan sano. . . Tanta ayuda y tanto amor. Mi hermana, irremplazable. Isabel Hernaez,mi sobrina. Tan txiki. . . Tan poderosa. . . Y Carlos Hernaez, mi cunado. ¿Quien lo iba adecir aquel dıa en la parte vieja?Tambien estoy agradecida a mas familia: a esas abuelas, abuelos, tıas, tıos, primos yfamilias, quienes han estado cerca mıo o han mostrado interes en este doctorado.
Gracias a esas grandes frases que siempre han estado ahı: “Con la constancia se llega altriunfo.” Se lo dijo Moreno a Pujol. “Siempre nos quedara Paris.” “El corazon de unamujer es un oceano lleno de secretos.” “Cuando se estudia se estudia, cuando se come secome, cuando se juega se juega. . . ” “El Tour es muy duro.” “Cuando estas en la cresta
11
12
de la ola, el aire o el viento te mueven, ası, el flequillo. . . ” “Mas vale una vez coloradaque ciento morada. . . ” “Redondea, vos redondea.”
Olga Porres y familia. Olga, tu tambien has estado ahı desde el principio y tienes que estaraquı arriba en la seccion de la familia por muchas razones. . . Partiendo del hoy hacia atras,hasta aquellos anos, hasta Bilbo, hasta Aldapeta. . . Es un largo camino: ¡muchısimas gra-cias!
Luis Vicente, director. He admirado tu conocimiento sobre sintaxis desde nuestras primerasclases de linguıstica. Fue entonces cuando hicimos nuestro primer trabajo de linguısticajuntos. Aquello funciono bien. Hoy en dıa, esto tambien ha funcionado bien. Esa viejaadmiracion sigue todavıa presente y ese tu conocimiento sobre sintaxis me ha ayudadoa escribir lo que sigue. Gracias, Luis, por toda tu ayuda. Ha sido un verdadero placervolver a reırme otra vez haciendo sintaxis. Muchas gracias por todo.Alazne Landa, directora. Muchısimas gracias por estar ahı y por darme tu ayuda. Graciaspor leer esta tesis, por tus comentarios, por tu ayuda en el lado administrativo de todoesto. Muchas gracias por todo.
Jon Franco, el Rey de las sombras. Muchısimas gracias por estar detras de cada email.Muchas gracias por todas las reuniones, por todos los cafes, por todas las conversaciones,por todo tu interes, por toda la ayuda. Muchas gracias por todo. Those sun glasses, silenttravellers; that white rope. . .Eduardo Smith, profesor en Language House. Me ayudaste con todos los recursos quetenıas. Me respondiste a todas mis preguntas. Pudiste leer mi mente bastante bien entantas ocasiones. . . Voy a recordar siempre todo lo que he aprendido gracias a ti, no solocomo estudiante pero tambien como profesora. Has sido de gran influencia en mi for-macion y estoy bien orgullosa de ello.
Xabi y Eva. ¡Que puedo decir! ¡Muchısimas gracias por todo, por todo! Xabi, siempre ahıcon tus puntos de vista y los puntos de vista de todo el Grupo. . . ¡Gracias a ti, gracias atodos! Es muy difıcil poder describir con palabras escritas tanto la ayuda que nos ofrecescomo el sentimiento de agradecimiento que sentimos. Has hecho mucho mucho por estedoctorado y probablemente seas de las personas que mejor lo entienden. . . Un millon degracias.
Nora Ibargoyen, Arrate Insausti. Siempre hemos estado y estaremos ahı. Muchısimasgracias. Sebastian Arriola –Seba–, Jose Herrera –Josesito–, Gaby, Hedie, Marianne,Roberto y el grupo Blacksburg-Cellar en Virginia. . . Me hicisteis reır tanto en tiempostan difıciles. . . Entendisteis tan absolutamente bien y os interesasteis tan rapido. . . Losfines de semana fueron increıbles con vosotros. Os echo de menos, de verdad. Emanuel eIta, Soly, muchısimas gracias por todo y por vuestra companıa.
Alex Pelham Waerea, Michael Walmsley, Laura Harrison y la gente de Waikato que vienea mi oficina a divertirse mientras trabajamos, y a hacer de ella un lugar magico-kiwi (en-tre ellos, algunos de “mis” primeros 390s: Ashley Steel, Ming Xia, Royce Dodd, BrendanSheridan, y Alex y Laura, otra vez). A aquellos estudiantes de Waikato que me ayu-daron a hacer tan increıble mi primer curso de literatura LA (. . . ), gracias. Alex, Laura
13
y Michael, gracias por repasar partes de esta tesis.
En University of Waikato tambien: muchısimas gracias a Dan Zirker, Alistair Swale,Athena Chambers, al profesorado de lenguas internacionales, al School of Arts.Clara, Edgardo Beacon y la familia Beacon, en Hamilton, Nueva Zelanda, y mas alla:habeis sido de gran ayuda y habeis sido extremadamente hospitalarios. Por eso y portodo, muchısimas gracias. Tambien, en Hamilton: Alvaro, Cecilia, Cristi, Alfredo, Pablo,Mindy, y familias, gracias. Gracias tambien a Carlos, Fabian, Isabelle.
En University of Illinois at Urbana-Champaign: gracias a aquellos profesores que im-partieron mis cursos de sintaxis y morfo-sintaxis: Luis Lopez, Karlos Arregi, James Yoon,Elabbas Benmamoun. A mi comite anterior y a todos aquellos que compartieron con-migo puntos de vista liguıstico-teoricos ya fueran similares o diferentes y que me dieronla posibilidad de obtener un dibujo mayor del campo de la linguıstica, muchas gracias.Muchas gracias tambien a aquellos coordinadores de espanol que me ayudaron con misobligaciones como “Teaching Assistant” y con otros asuntos pedagogicos relacionados conla ensenanza de lenguas extranjeras. A aquellos companeros del departamento que meayudaron en diferentes momentos, muchısimas gracias.
En Virginia Tech: muchısimas gracias a Robin Panneton y el Laboratorio de PercepcionInfantil, las personas en el Departamento de Psicologıa, y las reuniones Datapalooza.Jean Hanigan y Mark Hanigan: gracias por aquella bienvenida, por abrirnos las puertasde vuestra casa una y otra vez, por la conversacion que tuvimos en vuestra casa que tantoayudo. . . Por vuestro interes y vuestra ayuda, muchısimas gracias.
John Frampton, Luis Seidel, gracias por vuestra ayuda tecnica durante el proceso deescribir este documento. Gracias a las audiencias en LSRL, LSA, NELS, BIDE, EHU,eggschool: muchas gracias por las preguntas, por las respuestas, por los comentarios, porvuestra ayuda, por la conversacion, por escuchar.En la Universidad del Paıs Vasco, Euskal Herriko Unibertsitatea: gracias a aquellos conquienes he compartido parte de mi trabajo e investigacion, y gracias a quienes me acep-taron en multitud de charlas a las que he asistido desde que estudiaba mi carrera enBilbao. Ya he mencionado a Alazne Landa, muchısimas gracias. Muchas gracias a losgrupos de investigacion de liguıstica.
En la Universidad de Deusto y CIDE: muchas gracias a Inmaculada Ballano, Itziar Munoz,Leyre Goitia, Jon Franco otra vez, Margarita Otaegui, y el resto de los miembros del grupoCIDE. Jon Ortiz de Urbina, muchas gracias tambien por aquellos juicios; a mis profesoresde Filologıa Inglesa, muchısimas gracias. Algunas personas en la Universidad de Deustome han dado mucho, y ha sido, esta, una universidad que me ha estado viendo a lo largode todos estos anos. He estudiado y trabajado aquı, en los campus tanto de Bilbao comode San Sebastian. En pocas palabras, es una parte importante de mi vida.
A todos y al lector, he trabajado en lo que sigue y he escrito sobre ello siempre con lamisma idea dentro de mı: seguir aprendiendo. Haciendo eco de mis primeros “statementsof purpose”, de verdad espero que este escrito les ofrezca muchos momentos esclarece-dores. Gracias a todos. Disfruten.
Contents
1 Introduction 21
1.1 Aim of the dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.2 Outline of the dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.2.1 Chapter 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.2.2 Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.2.3 Chapter 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2 Gapping 31
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2 Gapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.2 General properties of gapping . . . . . . . . . . . . . . . . . . . . . 33
2.2.2.1 Directionality of gapping . . . . . . . . . . . . . . . . . . . 33
2.2.2.2 Gapping in verbal complexes . . . . . . . . . . . . . . . . 33
2.2.2.3 Gapping and the features of finite verbs . . . . . . . . . . 34
2.2.2.4 Complex Gapping . . . . . . . . . . . . . . . . . . . . . . 34
2.2.2.5 Asymmetries between Gapping and Pseudogapping/VP
ellipsis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.2.2.5.1 Licensing Environments: . . . . . . . . . . . . . . 34
2.2.2.5.2 The Categories Affected Differ: . . . . . . . . . . 36
2.2.2.5.3 The Strings Affected Differ: . . . . . . . . . . . . 38
2.2.2.5.4 The Identity Conditions on Antecedence Differ: . 38
2.2.2.5.5 Connectors . . . . . . . . . . . . . . . . . . . . . 39
2.2.3 Gapping analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.2.3.1 Large conjunct analyses . . . . . . . . . . . . . . . . . . . 40
2.2.3.2 Small conjunct analyses . . . . . . . . . . . . . . . . . . . 41
2.2.3.3 Small conjuncts, deletion and movement . . . . . . . . . . 43
2.2.4 Evidence for two kinds of coordination . . . . . . . . . . . . . . . . 46
2.2.4.1 Wide scope elements . . . . . . . . . . . . . . . . . . . . . 46
15
16 CONTENTS
2.2.4.1.1 Negation can take wide scope. . . . . . . . . . . . 46
2.2.4.1.2 Wide scope of modals . . . . . . . . . . . . . . . 47
2.2.4.1.3 Variable binding across conjuncts . . . . . . . . . 48
2.2.4.2 Repp’s contribution: negation and gapping . . . . . . . . . 48
2.2.5 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3 Gapping in Spanish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.2 Properties of Spanish gapping . . . . . . . . . . . . . . . . . . . . . 52
2.3.2.1 Gapping affects the Verb in final conjuncts in Spanish co-
ordinate contexts. . . . . . . . . . . . . . . . . . . . . . . 52
2.3.2.2 Verbal Complexes . . . . . . . . . . . . . . . . . . . . . . 52
2.3.2.3 Gapping in Spanish and the features of finite verbs . . . . 53
2.3.2.4 Complex Gapping . . . . . . . . . . . . . . . . . . . . . . 53
2.3.2.5 Additional Properties of gapping in Spanish . . . . . . . . 53
2.3.2.5.1 Licensing environments for Spanish gapping . . . 54
2.3.2.5.2 Affected categories . . . . . . . . . . . . . . . . . 54
2.3.2.5.3 Affected strings . . . . . . . . . . . . . . . . . . . 55
2.3.2.5.4 Identity Conditions . . . . . . . . . . . . . . . . . 55
2.3.3 Wide scope elements . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.3.3.0.5 The interaction between negation and gapping . . 56
2.3.3.0.6 Cross-conjunct variable binding . . . . . . . . . . 56
2.3.3.0.7 Wide scope of modals . . . . . . . . . . . . . . . 57
2.3.4 Negation, conjunction, and gapping: implications of scope of nega-
tion for the analysis of gapping . . . . . . . . . . . . . . . . . . . . 58
2.3.4.1 Wide and narrow scopes of negation . . . . . . . . . . . . 61
2.3.4.2 Distributed scope of negation . . . . . . . . . . . . . . . . 62
2.3.5 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.4 Gaps and complex gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.4.2 Multiple sluicing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.4.2.1 Preference for PPs Test . . . . . . . . . . . . . . . . . . . 67
2.4.2.2 DP Heaviness Test . . . . . . . . . . . . . . . . . . . . . . 67
2.4.2.3 Not the Object of a P Test . . . . . . . . . . . . . . . . . 68
2.4.2.4 Strict Locality Test . . . . . . . . . . . . . . . . . . . . . . 69
2.4.3 Gapping is not multiple sluicing . . . . . . . . . . . . . . . . . . . . 70
2.4.4 Gapping, move and delete separately . . . . . . . . . . . . . . . . . 74
2.4.4.1 Understanding gapping properties better . . . . . . . . . . 75
CONTENTS 17
2.4.4.2 Basic gapping . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.4.4.3 Gapping of auxiliary . . . . . . . . . . . . . . . . . . . . . 79
2.4.4.4 Gapping of modal . . . . . . . . . . . . . . . . . . . . . . 80
2.4.4.5 Complex gapping . . . . . . . . . . . . . . . . . . . . . . . 84
2.4.4.5.1 ‘Complex Gapping for Objects’: . . . . . . . . . . 84
2.4.4.5.2 ‘Complex Gapping for Verbal Complexes’: . . . . 89
2.4.4.6 Gapping in NEG-nor . . . . . . . . . . . . . . . . . . . . . 97
2.4.5 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3 Determiner Sharing 101
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.2 Determiner sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.2.2 Properties of determiner sharing . . . . . . . . . . . . . . . . . . . . 102
3.2.2.1 Determiner sharing and the missing Verb . . . . . . . . . . 102
3.2.2.2 Determiner sharing and the missing T . . . . . . . . . . . 103
3.2.2.3 Conjunct initial . . . . . . . . . . . . . . . . . . . . . . . . 103
3.2.2.4 Shared constituent in Determiner position . . . . . . . . . 103
3.2.2.5 Connectors, negation, determiner sharing . . . . . . . . . . 104
3.2.2.6 Interim Summary . . . . . . . . . . . . . . . . . . . . . . . 104
3.2.3 Interim conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.3 Determiner sharing in Spanish . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.3.2 Properties of determiner sharing in Spanish . . . . . . . . . . . . . 105
3.3.2.1 Determiner sharing and the missing Verb . . . . . . . . . . 105
3.3.2.2 Determiner sharing and the missing T . . . . . . . . . . . 106
3.3.2.3 Conjunct initial in Spanish? . . . . . . . . . . . . . . . . . 107
3.3.2.4 Shared constituent in Determiner position. . . . . . . . . . 107
3.3.2.5 Connectors, negation, determiner sharing . . . . . . . . . . 108
3.3.3 Determiner sharing: Spanish data . . . . . . . . . . . . . . . . . . . 109
3.3.3.1 Spanish determiner sharing in subject position . . . . . . . 110
3.3.3.1.1 Determiner sharing in Basic Gapping: . . . . . . 110
3.3.3.1.2 Determiner sharing in Gapping of Auxiliary: . . . 112
3.3.3.1.3 Determiner sharing in Gapping of Modal: . . . . 113
3.3.3.1.4 Determiner sharing in Complex Gapping for Ob-
jects: . . . . . . . . . . . . . . . . . . . . . . . . . 113
18 CONTENTS
3.3.3.1.5 Determiner sharing in Complex Gapping for Ver-
bal Complexes: . . . . . . . . . . . . . . . . . . . 115
3.3.3.1.6 Determiner sharing in Gapping in NEG-nor : . . . 116
3.3.3.2 Spanish determiner sharing in object position . . . . . . . 117
3.3.3.2.1 Determiner sharing in Basic Gapping: . . . . . . 117
3.3.3.2.2 Determiner sharing in Gapping of Auxiliary: . . . 117
3.3.3.2.3 Determiner sharing in Gapping of Modal: . . . . 118
3.3.3.2.4 To sum up so far. . . . . . . . . . . . . . . . . . . 118
3.3.3.2.5 Determiner sharing in Complex Gapping for Ob-
jects: . . . . . . . . . . . . . . . . . . . . . . . . . 118
3.3.3.2.6 Determiner sharing in Complex Gapping for Ver-
bal Complexes: . . . . . . . . . . . . . . . . . . . 119
3.3.3.2.7 Determiner sharing in Gapping in NEG-nor : . . . 120
3.3.4 Interim conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.4 Previous determiner sharing analyses . . . . . . . . . . . . . . . . . . . . . 121
3.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.4.2 Ackema and Szendroi (2002) . . . . . . . . . . . . . . . . . . . . . . 122
3.4.2.1 Summary of the analysis . . . . . . . . . . . . . . . . . . . 122
3.4.2.2 Problems of application to Spanish . . . . . . . . . . . . . 124
3.4.3 Citko (2007) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
3.4.3.1 Summary of the analysis . . . . . . . . . . . . . . . . . . . 125
3.4.3.2 Problems of application to Spanish . . . . . . . . . . . . . 126
3.4.4 Johnson (2000), Lin (2002) via Sportiche’s DP-Split . . . . . . . . . 127
3.4.5 Interim conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.5 Spanish determiner sharing analysis . . . . . . . . . . . . . . . . . . . . . . 132
3.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.5.2 Proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
3.5.3 Analysis: subject position . . . . . . . . . . . . . . . . . . . . . . . 141
3.5.4 Analysis: object position . . . . . . . . . . . . . . . . . . . . . . . . 156
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
4 Wh-Determiner Sharing 175
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
4.2 Wh-determiner sharing in Spanish . . . . . . . . . . . . . . . . . . . . . . . 175
4.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
4.2.2 General properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
4.2.2.1 The missing V: not a condition in Wh-questions . . . . . . 176
4.2.2.2 The missing T: not a condition in Wh-questions . . . . . . 177
CONTENTS 19
4.2.2.3 Conjunct initial? . . . . . . . . . . . . . . . . . . . . . . . 178
4.2.2.4 Connectors, negation, and determiner sharing . . . . . . . 178
4.2.3 Wh-determiner sharing: data and classification . . . . . . . . . . . 178
4.2.3.1 Wh-determiner sharing with subjects . . . . . . . . . . . . 178
4.2.3.1.1 Wh-Determiner Sharing in Basic Gapping: . . . . 179
4.2.3.1.2 Wh-Determiner Sharing in Gapping of Auxiliary: 179
4.2.3.1.3 Wh-Determiner Sharing in Gapping of Modal: . . 180
4.2.3.1.4 Wh-determiner sharing in Complex Gapping for
Objects: . . . . . . . . . . . . . . . . . . . . . . . 181
4.2.3.1.5 Wh-Determiner Sharing in Complex Gapping for
Verbal Complexes: . . . . . . . . . . . . . . . . . 181
4.2.3.1.6 Wh-Determiner Sharing in NEG-nor : . . . . . . . 182
4.2.3.2 Wh-determiner sharing with objects . . . . . . . . . . . . 183
4.2.3.2.1 Wh-Determiner Sharing in Basic Gapping: . . . . 183
4.2.3.2.2 Wh-Determiner Sharing in Gapping of Auxiliary: 183
4.2.3.2.3 Wh-Determiner Sharing in Gapping of Modal: . . 184
4.2.3.2.4 Wh-Determiner Sharing in Complex Gapping for
Objects: . . . . . . . . . . . . . . . . . . . . . . . 185
4.2.3.2.5 Wh-Determiner Sharing in Complex Gapping for
Verbal Complexes: . . . . . . . . . . . . . . . . . 185
4.2.3.2.6 Wh-Determiner Sharing in NEG-nor : . . . . . . . 186
4.2.4 Interim conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
4.3 Wh-determiner sharing: analysis . . . . . . . . . . . . . . . . . . . . . . . . 187
4.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
4.3.2 Proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
4.3.3 Specific Wh-determiner sharing structures . . . . . . . . . . . . . . 188
4.3.4 Interim conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
4.4 Cyclicity in Wh-determiner sharing . . . . . . . . . . . . . . . . . . . . . . 196
4.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
4.4.2 Arregi and Centeno (2005) . . . . . . . . . . . . . . . . . . . . . . . 197
4.4.3 More evidence and phases . . . . . . . . . . . . . . . . . . . . . . . 199
4.4.4 Centeno and Vicente (2009) . . . . . . . . . . . . . . . . . . . . . . 213
4.4.5 Interim conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
5 Conclusion 219
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
5.2 Cross-linguistic comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
20 CONTENTS
5.3 Grammaticality judgments . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5.4 Further types of gapping and determiner sharing . . . . . . . . . . . . . . . 222
5.5 The syntax of Determiner Phrases . . . . . . . . . . . . . . . . . . . . . . . 224
Chapter 1
Introduction
1.1 Aim of the dissertation
The main goal in this dissertation is to understand silent material at the level of sentence
analysis. Merchant (2001) explains the significance of the study of silent elements.
The primary goal of contemporary theoretical linguistics is to develop a theory
of the correspondence between sound (or gesture) and meaning. Nowhere does
this sound-meaning correspondence break down more spectacularly than in
the case of ellipsis. And yet various forms of ellipsis are pervasive in natural
language –words and phrases which by rights should be in the linguistic signal
go missing. How is this possible?
Merchant (2001) investigates how that is possible focusing on sluicing. In this dissertation,
I target that same goal focusing on different forms of ellipsis: gapping and determiner
sharing sentences in both declarative contexts and questions. Gapping sentences are
characterized by the omission of the main verb in coordinate contexts. In languages such
as English or Spanish, the omission of the Verb occurs in non-initial conjuncts. I illustrate
this with examples in English (1) and Spanish (2).
(1) Juan bought an ice-cream and Ana [ ] an apple-pie.
(2) JuanJuan
fuewent
alto.the
cinemovies
yand
AnaAna
[ ][ ]
alto.the
teatro.theatre
‘Juan went to the movies and Ana went to the theater.’
In the English example (1), the verb ‘bought’ is omitted in the final conjunct. However,
its meaning is present in both parts of coordination. Likewise, in the Spanish example
(2), though the Verb fue ‘went’ is omitted in the second conjunct, its meaning is present
in both conjuncts. Determiner sharing is a structure very similar to gapping. It occurs in
coordinate contexts if there is omission of the main Verb, as in gapping, and also omission
21
22 CHAPTER 1. INTRODUCTION
of a Determiner, the Determiner that is said to be shared. I provide examples in English
(3) and (4).
(3) Few boys bought an ice-cream and [ ]D girls [ ]V an apple-pie.
(4) Muchosmany
chicosboys
fueronwent
alto.the
cinemovies
yand
[ ]D[ ]D
chicasgirls
[ ]V[ ]V
alto.the
teatro.theatro
‘Many boys went to the movies and many girls went to the theater.’
In the English example (3), the Verb ‘bought’ and the Determiner ‘few’ are missing from
the non initial conjunct. Still, their meaning is present in both conjuncts. The same
applies to Spanish (4). The Verb fueron ‘went’ and the Determiner muchas ‘many’ are
omitted in the non-initial conjunct. However, their meaning is still present there.
Throughout the years, the study of gapping has brought proposals which are all mainly
related to two aspects: (i) accounting for missing elements via deletion or movement; and
because of the close relation of gapping with coordination, (ii) the size of the conjuncts
upon which gapping is present. At the beginning gapping was proposed to be exactly
the same as ellipsis (Hankamer (1971, 1972), Hankamer and Sag (1976)). Afterwards,
Johnson (2004, 2009) proposed that gapping was different to other ellipsis forms and
claimed that gapping was mainly the result of movement. With respect to the size of the
conjuncts, the first proposals were that conjuncts are large, always sentential (Chomsky
(1957), Gleitman (1965)). However, this changed later with Siegel (1984, 1987), McCaw-
ley (1993), who proposed to have smaller conjuncts in contexts where propositions were
conjoined. This proposal is importantly reinforced, in more general grounds, by the work
on phrasal coordination by Lakoff and Peters (1969).
Most gapping theories make a correlation between the large conjunct proposal and the
deletion proposal on the one hand, and the small conjunct proposal and the movement
proposal on the other hand. This seems to follow naturally. On the one hand, if conjunc-
tion is of large conjuncts, ellipsis accounts for missing elements. And on the other hand,
if conjuncts can be smaller, missing elements may be missing because of movements, like
for instance, ATB movement (For more detailed explanation, see for instance, chapter 2
in Hartmann (2000), Johnson (2009), or chapter 2 in this dissertation.). However, this
correlation is not always this straightforward because deletion, for example, does not en-
tail the need of large conjuncts in the structure. Considering the latter idea, I propose an
eclectic theory for gapping and its related structure, determiner sharing, and propose to
break this correlation explaining how and why. I base my proposal on my investigations
of Spanish data. I show that both approaches, in relation to deletion/movement and the
sizes of conjuncts, are needed in accounting for the different types of gapping and deter-
1.2. OUTLINE OF THE DISSERTATION 23
miner sharing that arise.
Additionally, it is important to understand, among other aspects, that various ele-
ments, apart from gapping or determiner sharing, also play an important role in deciding
upon the structure that is assigned to each sentence. In recent literature, negation has
been considered in more detail in the analysis of gapping than has been done previously.
Repp (2009) shows in her dissertation that negation in coordinate/gapping contexts intro-
duces three different readings: a wide scope reading of negation, where the latter scopes
above conjunction; a narrow scope reading, where negation scopes below negation and
only one part of the coordination is affected by negation; and a distributed scope reading,
where negation is present in both parts of coordination but where conjuncts are negated
separately. Before Repp’s work, gapping was analyzed in relation to negation considering
only the wide and narrow scope readings. The properties of negation in gapping con-
texts in Spanish are similar to those of English. Also, as far as the two languages can
be compared, the general properties of gapping in Spanish are similar to those of English
gapping. The analysis that is proposed for the gapping phenomena in Spanish responds to
the properties that gapping displays in relation to negation and to the general properties
of gapping.
The investigation of silence in this dissertation is framed within the standard theory
of the Minimalist Program as is defined in Chomsky (1995, 2000). Part of the proposals
in chapter 2, 3, and 4, which are centered both on gapping and determiner sharing rely on
the more specific theory of phases in its standard version, as defined in Chomsky (2000,
2001, 2004). It is important to notice that through the study of silence, the scope of this
dissertation reaches other general aspects of syntax. I contribute to the analysis of the
syntax of DPs in chapter 3 more specifically, and the syntax of wh-movement or of phases
in chapter 4. In general terms, I provide evidence that the standard theory of phases
(Chomsky (2000, 2001, 2004)) is correct for both verbal and nominal categories: CPs,
vPs, DPs.
1.2 Outline of the dissertation
This dissertation is organized according to my main goals in this work: understanding
gapping and determiner sharing in Spanish. First, I investigate the gapping structure in
Spanish in chapter 2. Second, I focus on the analysis of determiner sharing sentences in
Spanish in chapter 3. Third, I further investigate determiner sharing in Spanish but this
24 CHAPTER 1. INTRODUCTION
time when it is embedded in wh-questions in chapter 4. Finally, I provide a conclusion in
chapter 5.
1.2.1 Chapter 2
The investigation of Gapping in Spanish starts with an introduction of the properties
that this structure displays in English. Gapping appears in coordinate contexts when
the main Verb is missing from part of the conjuncts. In English, the omitted Verbs hap-
pen in non-initial conjuncts. Previous literature has also remarked that gapping is more
restricted than other deletion phenomena. The intimate relationship between gapping
and coordination is reflected in the diversity of analyses that gapping receives. There
are two mainstreams in the literature: the large conjunct approach and the small con-
junct approach. In broad terms, the large conjunct approach proposes to analyze gapping
sentences as coordination of sentential conjuncts. When elements are missing in these
kinds of structures, a process like deletion applies. In most cases, in the large conjunct
approach, PF deletion is the process that is proposed. To the contrary, the small conjunct
approach allows coordination of conjuncts which are smaller than sentences. The avail-
ability of small conjuncts entails that the structure of gapping has a shared part that sits
outside conjunction and that dominates the conjoined elements. In this kind of analysis,
the main ways by which missing elements are accounted for are either by movement or
because some elements generate in the shared part of the structure.
In order to provide an analysis, I classify gapping phenomena in different types ac-
cording to missing elements or to negation. With the different types of gapping in hand,
I analyze each type one by one. I label the first type ‘Basic Gapping’ and it refers to
cases in which only the lexical Verb is missing from non initial conjuncts (5). Tests show
that the structure needs small conjuncts and movement in order to account for this type
of gapping. The next two types are ‘Gapping of Auxiliary’ (6) and ‘Gapping of Modal’
(7). These two types refer to cases where the only missing element is either the Auxiliary
or the Modal Verb respectively. The structure is similar to that of ‘Basic Gapping’, with
small conjuncts and movement when needed. Small differences arise among these struc-
tures according to nuances related to auxiliaries or modality in syntax. In the gapping
types that follow, major changes appear in structures and processes affecting structures.
In ‘Complex Gapping’ there are two subtypes: ‘Complex Gapping for Objects’ (8) and
‘Complex Gapping for Verbal Complexes’ (9, 10). The first one refers to cases in which
the missing elements are the Verb and one argument, either a Direct Object or an In-
direct Object of the missing Verb. For this type, tests show that the structure needs
1.2. OUTLINE OF THE DISSERTATION 25
small conjuncts on the one hand and a PF deletion process on the other hand. The same
structural features, small conjuncts and PF deletion, need to apply in ‘Complex Gapping
for Verbal Complexes’. In this type of gapping, there are cases where the whole verbal
complex is omitted: either both auxiliary and lexical Verb, or modal and lexical Verb.
The last division in gapping types is related to negation. In the last type, gapping is
embedded in a NEG-nor configuration. I label it ‘Gapping in NEG-nor ’ (11). Structure
wise, conjuncts in this type of gapping need to be larger than normally considered in small
conjunct approaches and the process that accounts for omitted elements is PF deletion.
(5) JuanJuan
comeeats
arrozrice
yand
PedroPedro
[ ][ ]
alubias.beans
‘Juan eats rice and Pedro eats beans.’
(6) JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
[ ][ ]
fregadowashed
elthe
suelofloor
‘Jorge has cleaned the bathroom and Pedro has washed the floor.’
(7) JuanJuan
deberıashould
arreglarfix
elthe
cochecar
yand
yoI
[ ][ ]
prepararprepare
lasthe
maletas.luggage
‘Juan should fix the car and I should prepare the luggage.’
(8) JorgeJorge
leCL
diogave
unaa
florflower
ato
MarıaMarıa
yand
PedroPedro
[ ][ ]
ato
ElenaElena
‘Jorge gave a flower to Marıa and Pedro gave a flower to Elena.’
(9) JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
[ ][ ]
elthe
suelo.floor
‘Jorge has cleaned the bathroom and Pedro has cleaned the floor.’
(10) JuanJuan
deberıashould
arreglarfix
elthe
cochecar
yand
yoI
[ ][ ]
lathe
bicicleta.bicycle
‘Juan should fix the car and I should fix the bicycle.’
(11) Marıamaria
nonot
estuvowas
enfermasick
ninor
JuanJuan
[ ][ ]
acatarrado.with.a.cold
‘Maria was not sick nor Juan was with a cold.’
1.2.2 Chapter 3
This chapter investigates determiner sharing in Spanish. It starts with a detailed exposi-
tion of the properties that are normally listed in the literature for this construction in the
English language. After that, I include a parallel list of properties for determiner sharing
in Spanish. Sentences of this type are similar in both English and Spanish; however, it
is not exactly the same in both languages. The small differences that arise are because
of general differences between the two languages or because of the wider set of Spanish
determiner sharing types that I am considering. I include a classification of determiner
26 CHAPTER 1. INTRODUCTION
sharing in Spanish, which is constructed according to the classification of gapping types
that I include in the previous chapter. Assuming this classification for determiner sharing
makes sense because the gapping of the Verb is of the same kind in determiner sharing
and in gapping. There is, additionally, a major division that is needed in the determiner
sharing classification. I provide all types for determiner sharing in subject position on
the one hand and for object position on the other hand. The grammaticality condition
of these data is important. Subject determiner sharing is grammatical in every case. Ob-
ject determiner sharing is grammatical only in some of the cases. The division between
grammatical and ungrammatical object determiner sharing examples coincides with one
of the divisions that take place in the gapping classification. Interestingly, object deter-
miner sharing is only grammatical when this is embedded in the gapping types that are
proposed to be analyzed by means of the PF-deletion process. In the analysis section I
explain the reason for such a phenomenon.
Part of my analysis is based on previous work on determiner sharing. I show that the
common features to the works by Johnson (2000), and Lin (2002), via Sportiche’s DP-
Split, are assumable in the analysis of the Spanish data that I provide. These analyses
make use of small conjuncts and propose that the Determiner that enters the sharing rela-
tionship is composed of a Determiner related position on one projection and the rest of the
Determiner on another projection. Lin makes reference to Sportiche’s DP-Split in doing
this. With this in hand, Johnson and Lin account for determiner sharing via the proposal
that the Determiner related position is shared because it is part of the shared part of
the structure that dominates the conjoined elements. With respect to the choice between
movement and deletion, Johnson accounts via movement for the missing elements, and
Lin does the same for almost every example. For just a specific case of determiner shar-
ing, Lin proposes to have deletion in a ‘salvation by deletion’ fashion. My proposal is
to adopt Sportiche’s DP-Split and to adopt the Determiner related positions to account
for determiner sharing. However, I propose some changes to the theory. First, the De-
terminer related position, DET, projects a whole projection on its own. And in order
for the licensing conditions between DET and D to be met, the DP moves to the Spec,
DETP position. Secondly, I take into account both determiner sharing sentences and
non determiner sharing contexts in my DETs proposal –this way, avoiding a construction
specific proposal–. After careful consideration of the data, the proposal is that DETPs
occur in unfixed positions in the structure, only limited by the following: DETP need to
be located at the edge of phases, DPs, vPs, and CPs, all of which have been proposed to
be phases in the literature.
1.2. OUTLINE OF THE DISSERTATION 27
My analysis of Spanish determiner sharing comprises the just explained specific pro-
posal focused on the sharing of DET and the gapping proposal: in Spanish, both small
and large conjuncts are needed, and both movement and deletion are also needed as pro-
cesses that account for omitted elements other than D. Same as with gapping, I provide
an analysis for each type of determiner sharing. In subject determiner sharing, when this
construction is embedded in ‘Basic Gapping’ (12), ‘Gapping of Auxiliary or of Modal’
(13, 14), the sentence is analyzed via small conjuncts and movement –like in the gapping
counterparts– and via the DETP projections, situated at the edge of the corresponding
phases. The rest of the types parallel their gapping counterparts too. Subject determiner
sharing embedded in ‘Complex Gapping for Objects (15) and for Verbal Complexes’ (16,
17) and ‘Gapping in NEG-nor ’ (18) is analyzed by means of the DETPs proposal and
the PF-deletion process to account for other missing elements. Within object determiner
sharing, one of the goals is to explain the grammatical contrasts among object deter-
miner sharing sentences. Only the deletion types are grammatical and are accounted
for as the subject determiner sharing cases, via deletion and the DETPs proposal. In
the cases where movement is the process that accounts for missing elements, determiner
sharing is not possible in object position. The reason is that there is no extra trigger
for these objects to move. Because these objects do not move, they cannot make a step
to Spec, DETP, and the licensing conditions for the spelling out of the Determiners are
not respected. This is what makes these sentences ungrammatical. Notice that using my
analysis of gapping as a basis, the grammaticality contrasts among determiner sharing
cases can also be explained. This highlights the validity of the analysis both for gapping
and determiner sharing.
(12) Muchosmany
chicosboys
comeneat
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]V[ ]V
heladoice-cream
deof
vainilla.vanilla‘Many boys eat apple pie and many girls eat vanilla ice-cream.’
(13) Muchosmany
chicosboys
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]T[ ]T
degustadotasted
heladoice-cream
deof
vainilla.vanilla
‘Many boys have eaten apple pie and many girls have tasted vanilla ice-cream.’
(14) Muchasmany
chicasgirls
deberıanshould
arreglarfix
sutheir
cochecar
yand
[ ]D[ ]D
chicosboys
[ ]T[ ]T
prepararprepare
sustheir
maletas.luggage
‘Many girls should fix their car and many boys should prepare their luggage.’
28 CHAPTER 1. INTRODUCTION
(15) Demasiadostoo.many
chicosboys
leCL
dierongave
unaa
florflower
ato
MarıaMarıa
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
ato
Jorge.Jorge‘Too many boys gave a flower to Marıa and too many girls gave a flower to Jorge.’
(16) Muchosmany
chicosboys
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
heladoice-cream
deof
vainilla.vanilla
‘Many boys have eaten apple pie and many girls have eaten vanilla ice-cream.’
(17) Muchasmany
chicasgirls
deberıanshould
arreglarfix
sutheir
cochecar
yand
[ ]D[ ]D
chicosboys
[ ]Cplx
[ ]Cplx
sutheir
bicicleta.bike‘Many girls should fix their car and boys their bike.’
(18) Nineither
demasiadostoo.many
ninosboys
estuvieronwere
enfermossick
ninor
[ ]D[ ]D
ninasgirls
[ ]V[ ]V
acatarradas.with.a.cold‘Neither too many boys were sick nor too many girls were with a cold.’
1.2.3 Chapter 4
In this chapter, I analyze another form of determiner sharing. This is wh-determiner
sharing, a label that refers to determiner sharing inside wh-questions. In this type of
determiner sharing, the Determiners that enter the sharing relationship are wh-words.
Because of this, wh-movement and some related aspects are an important part of the
chapter. First, I detail the properties that wh-determiner sharing display. I explain that
wh-determiner sharing is indeed another form of determiner sharing. Although there are
differences that arise between determiner sharing in declarative contexts and determiner
sharing in wh-questions, these differences are related to the nature of wh-questions. In
this chapter too, I include a complete classification of wh-determiner sharing sentences:
both for cases where the wh-phrases refer to elements that occupy the subject position
and for cases where the wh-phrases refer to elements in object position. The analysis of
these sentences relates to prior analyses in the same way. I consider the same assumptions
as before and add what is needed to analyze the new aspects of these sentences. That
is, the analysis of determiner sharing is done in the same way as in the prior chapter but
with additions that are related to wh-movements.
The analysis of Spanish wh-determiner sharing is illustrated by the analyses of four
sentences which represent the analyses of all the different types of determiner sharing that
1.2. OUTLINE OF THE DISSERTATION 29
are illustrated throughout the dissertation. First, I provide an analysis of wh-determiner
sharing embedded in a gapping type that is accounted for by movement and whose wh-
phrases refer to elements that occupy subject positions. Features of this analysis include:
small conjuncts, the needed movements to account for missing elements, the specific fea-
tures associated with wh-movement, and those associated with the DETPs proposal. The
same applies to the wh-determiner sharing counterparts whose wh-phrases refer to ele-
ments in object positions. Illustration of wh-determiner sharing embedded in gapping
types that are accounted for via PF-deletion is also done with two examples: one for
subjects and one for objects. These examples are analyzed attending to PF-deletion, the
features of the analysis that are related to the DETPs proposal, and those related to wh-
movement. The analyses of these examples illustrate the way in which determiner sharing
is analyzed by means of my proposal when embedded in wh-movement contexts. However,
wh-determiner sharing brings with it interesting evidence in favor of theories that have
been proposed in the literature in relation wh-movement. I consider these in what follows.
With respect to wh-movement specifically, I draw special attention to two aspects:
(i) the cyclic nature of wh-movement and (ii) its ability to diagnose phases via landing
sites. The basis for my analysis of these two aspects is found in two previous works on wh-
determiner sharing: Arregi and Centeno (2005) and Centeno and Vicente (2009). The first
is focused on the fact that wh-determiner sharing brings evidence of cyclicity as defined
in Chomsky (2000), Fox (2000), or Nissenbaum (2000). The second is focused on the fact
that wh-determiner sharing brings evidence that vPs corresponding not only to transitive
Verbs but also to passive, unaccusative, and rasing predicates project phase boundaries.
This evidence related to phases is relevant for the determiner sharing analysis that I
provide in this dissertation because of the latter’s reliance on the phases notion. With re-
spect to cyclicity, I also provide further evidence for such phenomenon in my present work.
Chapter 2
Gapping
2.1 Introduction
This chapter focuses on gapping constructions in Spanish. Gapping sentences can be
broadly described as coordinate sentences where the Verb in one of the conjuncts is miss-
ing. I hold that gapping in Spanish is a construction that behaves similarly to English
with respect to the properties that are normally mentioned in the literature. I enumerate
some of them for illustration at this point. The part of Verbal Complexes that need to
be missing is the same in both English and Spanish. Complex gapping, where both the
Verb and the Object are missing, is also possible in both languages. And the way that
gapping behaves in English and Spanish is similar if the following are considered: licensing
environments, categories affected, strings affected, or identity conditions on antecedence.
In order to investigate the properties of gapping adequately, first I briefly review the
debate over large conjunct and small conjunct approaches. One of the main points of
interest in such a discussion is the existence of wide scope elements, that is elements that
take scope over negation. I provide a close look at the behavior of these elements in Span-
ish gapping arriving to the conclusion that small conjuncts are needed in Spanish gapping.
Of special interest to this topic is Repp’s (2009) contribution. Before her work on the
interaction between negation and gapping, the knowledge that we had about negation’s
readings in gapping was rather incomplete. Taking into account Repp’s contributions, I
analyze negation in Spanish gapping closely. One of the conclusions is that negation not
only brings the wide and narrow scopes of negation that were previously known but also
that there is a distributed reading that gets to be important in the analysis.
This distributed reading is connected to NEG-nor configurations as in Wurmbrand
(2008). The adoption of such an account and the evidence supporting it makes coordi-
31
32 CHAPTER 2. GAPPING
nation of CPs a necessary part of the syntax of Spanish gapping. Gapping being present
in these NEG-nor configurations, there needs to be an account for the missing elements
–either the Verb on its own or the Verb and further missing material–: deletion. I assume
a PF deletion by phase approach that accounts not only for missing elements in NEG-nor
sentences but also for other types of Spanish gapping constructions. It cannot, however,
account for missing elements in all Spanish gapping types. I show that some Spanish
gapping types cannot be accounted for via deletion. Tests which appear in the gapping
and ellipsis literature show that movement needs to also be a process that accounts for
missing elements in Spanish gapping.
The typology of Spanish gapping sentences that I account for in this chapter are:
‘Basic Gapping’, ‘Gapping of Auxiliary’, ‘Gapping of Modal’, ‘Complex Gapping’, and
‘Gapping in NEG-nor ’. See the Introduction to this dissertation for examples of each
type. Final conclusions are (i) that both small conjuncts and large conjuncts are needed
in the analysis of Spanish gapping and (ii) that the deletion and movement processes are
also needed in the gapping structures that are analyzed here.
In section 2, I review the relevant literature and single out the aspects of English
gapping that I use as the basis for the study of Spanish gapping. In section 3, I investigate
Spanish gapping: enumerating its properties and comparing them to English, and showing
that wide scope elements and negation behave in the same way with as with English
gapping. Hence, I reach the interim conclusion that small conjuncts are needed in gapping
and that deletion needs to occur in at least one of the types of gapping I analyze. In
section 4, analyses for Spanish gapping are provided. Here, I first provide evidence that
this construction cannot be compared to pseudogapping (Lasnik (2006)) in Spanish. Such
evidence brings with it the additional conclusion that elements that remain not elided in
these constructions undergo movement inside their conjuncts. At this point, I proceed to
both justify and provide an analysis of each Spanish gapping type based on the possibility
of having small and large conjuncts and movement and deletion. Section 5 concludes the
chapter.
2.2 Gapping
2.2.1 Introduction
In this section, I describe gapping and I generally explain the different analyses that
have been proposed to account for gapping phenomena. The main goal of this section
2.2. GAPPING 33
is to be the basis for the discussion in the next section where I analyze gapping in Spanish.
In section 2.2.2, I explain the basic characteristics of gapping and explain some of its
main restrictions. Afterwards, I divide the proposed analyses for gapping in two major
groups: the large conjunct analyses and the small conjunct analyses. I explain these in
general terms in section 2.2.3. And as a final step, I provide evidence for two types of
coordination. This kind of evidence comes from wide scope elements, in section 2.2.4.1,
and from the analysis of the interaction between negation and gapping, in section 2.2.4.2.
In section 2.2.5, I present partial conclusions up to this point in the discussion.
2.2.2 General properties of gapping
2.2.2.1 Directionality of gapping
This phenomenon is characterized by a missing finite verb in either initial or non-initial
conjuncts depending on the type of language that is considered: in head-initial languages,
the finite verb is gapped in non-initial conjuncts; in head-final languages, the finite verb
is gapped in initial conjuncts (Kim (1997)). Consider examples (19) and (20) below.
(19) John gave a flower to Mary and Sue [ ] a book to Patrick.
(20) John-iJohn-NOM
Mary-eykeyMary-DAT
Kloch-ulflower-ACC
[ ][ ]
kulikoand
Bill-iBill-NOM
Sue-eykeySue-DAT
Chaik-ulbook-ACC
cwessta.gave
‘John gave a flower to Mary and Bill gave a book to Sue’. (Korean)
Sentence (19) is an example of a head-initial language, English. In English, the gap that
corresponds to ‘gave’ is in the final conjunct. Sentence (20) is a Korean example, a head-
final language. In this case, the gap that corresponds to cwessta‘gave’ is in the first part
of the coordination, before the connector kuliko ‘and’.
2.2.2.2 Gapping in verbal complexes
Siegel (1984, 1987), McCawley (1993), Lin (2002), and Repp (2009) show that either the
finite part of the verb is missing (21), or both the finite part and the non-finite verb are
missing (22). The non-finite part cannot be gapped while the finite part is kept (23).
(21) John has bought a flower and Mike [ ] stolen a book.
(22) John has bought a flower and Mike [ ] a book.
(23) *John has bought a flower and Mike has [ ] a book.
34 CHAPTER 2. GAPPING
2.2.2.3 Gapping and the features of finite verbs
Features like aspect, tense or mood need to be identical in both conjuncts in order for
gapping to affect one of them. However, features such as person, number, or gender may
differ. Consider examples (24) and (25) to see the difference.
(24) *The boy played with a toy car yesterday and the girl [ ] with a doll tomorrow.
(25) Erhe
trinktdrink
Weinwine
undand
diethe
anderenothers
[ ][ ]
Bier.beer
‘He is drinking wine and the others are drinking beer.’ (German)
In the English example (24), the tense in each conjunct is different as can be inferred from
the use of the two distinct adverbs ‘yesterday’ and ‘tomorrow’. This Tense difference and
the application of gapping onto one of the verbs results in ungrammaticality. However,
sentence (25) from Repp (2009, 7) is grammatical. Gapping has affected a verb ‘trinken’
that only differs in number with respect to its overt counterpart in the first conjunct.
2.2.2.4 Complex Gapping
In addition, other material like Direct Objects or Indirect Objects can also be omitted.
Consider (26).
(26) John gave Mary a flower and Bill [ ] Sue [ ].
In sentence (26), both the verb ‘gave’ and the Direct Object ‘a flower’ are missing from
the final conjunct and the sentence is still grammatical. Johnson (2004, 2009) labels the
examples of this type ‘complex gapping’.
2.2.2.5 Asymmetries between Gapping and Pseudogapping/VP ellipsis
Johnson (2004) analyzes in detail some of the restrictions that operate in gapping in com-
parison to other ellipsis processes. Comparing gapping to VP ellipsis and pseudogapping,
Johnson finds differences between the two types of ellipsis processes. These differences
are related to licensing environments, to the categories that get affected, to the strings
that get affected, and to identity conditions on antecedence.
2.2.2.5.1 Licensing Environments: Johnson (2004) shows that gapping is more
limited in its available contexts than other ellipsis processes such as VP ellipsis or pseu-
dogapping. The crucial context is coordination because Johnson shows that gapping only
works properly in coordinate sentences. Consider the English sentences in (27, 28, 29)
which correspond to examples (54a, d, f) in Johnson (2004).
(27) Betsy likes cats and Liz [ ] dogs.
2.2. GAPPING 35
(28) ?*Vivek likes Chinese action films but Wishi [ ] sci-fi movies.
(29) *Some ate natto today because others [ ] yesterday.
Example (27) is grammatical. It is a coordinate sentence where the gap is fine. However
on a different context like subordination (29), gapping is not possible and the sentence
is not grammatical. The case of coordination with ‘but’ in (28) is less clear as it brings
variability in judgments. The reason for this is that the use of ‘but’ is subject to condi-
tions on Information Structure that may interfere with those of gapping.1 Still, gapping
can only happen in coordination as subordination contexts do not work properly with it.
There are further ways in which the behavior of gapping contrasts with other ellipsis
processes which do not need to be in coordinate contexts to result in grammatical sen-
tences. Comparison is made here between gapping and pseudogapping, which is generally
assumed to be ellipsis (Jayaseelan (1990, 2001), Lasnik (1995, 1999)). In the English
examples, (30)-(32), ellipsis is fine independent of the context. These are pseudogap-
ping examples (55a, c, d) in Johnson (2004), where Tense is maintained but the verb
disappears.
(30) Betsy could like cats and Liz might [ ] dogs.
(31) Vivek might like Chinese action films, but Nishi doesn’t [ ] sci-fi movies.
(32) Some will eat Natto TODAY, because other had [ ] YESTERDAY.
There is no contrast in grammaticality between (30) and (32). This means that ellipsis
is fine in both coordination and subordination contexts. The different patterns that can
be seen in (30)-(32) and (27)-(29) show that gapping is much more restricted than other
types of ellipsis. Gapping is restricted to coordination only.
Johnson takes a further step to delimit the contexts where gapping appears. He
checks whether there is a further restriction within coordination that could possibly shape
gapping in a more specific way so that the difference between gapping and other ellipsis
processes becomes more clear. In order to do this, both processes, gapping and ellipsis,
are compared in a sentence where a coordinate structure is placed inside another one.
Compare the English ellipsis sentence in (33) with the gapping sentence in (34). These
are examples (56) and (57) in Johnson (2004).
1Notice that examples with ‘but’ improve with the simultaneous use of ‘only’ in the sentence. Considerexample (1) as illustration.
(1) Magic Johnson won five NBA titles, but Larry Bird [ ] only three.
36 CHAPTER 2. GAPPING
(33) ?Sam usually drinks martinis, or it’s an AA meeting and no one can [ ] any
alcohol.
(34) *Sam usually drinks martinis, or it’s an AA meeting and no one [ ] any alcohol.
In order to better understand these examples I present the structure (35) of example (33)
that only specifies the hierarchy of conjuncts.
(35) [ Sam usually drinks martinis] or [ [ it’s an AA meeting] and [ no one can [ ]
any alcohol] ]
There is a grammaticality contrast between gapping (34) and ellipsis (33). The structure
in (35) shows that there is a coordination inside the second conjunct of the main coordi-
nate structure. Both examples show that there are missing words inside the last conjunct
of the inner coordination. The difference between the examples is that in the grammati-
cal one, example (33), missing words undergo ellipsis. On the other hand, missing words
undergo gapping on the ungrammatical example (34). These examples show that gapping
is possible in certain coordinate contexts. Gapping is possible when the antecedent of the
gap is in the conjunct that immediately precedes the gapped conjunct. In other words,
since gapping is a result of movement, gapping is possible when the CSC is not violated.
This is evidence for the structure that Johnson (2004) and Lin (2002) propose for
gapping.2 In this type of structure, there is a part that is shared by both conjuncts
which is formed, among others, by movements of elements that come from the conjuncts.
This structure is possible because the elements that move from the second conjunct also
move from the first conjunct (ATB movement) or because there are apparent asymmetric
movements from the first conjunct. In examples with three conjuncts like (33), a gap
in the second conjunct of the embedded coordination without a gap in the first conjunct
would violate the CSC. Hence, there is no way to form a gapping structure like that in (34)
with a gap in the third conjunct and no gap in the second one. This existing difference
between gapping and ellipsis cannot be captured by the large conjunct approach because
the only device of such an approach is ellipsis.
2.2.2.5.2 The Categories Affected Differ: Johnson (2004) shows that gapping and
ellipsis affect different categories. Ellipsis can affect TPs in sluicing cases. (36) is a sluicing
English example which is example (70) in Johnson (2004):
(36) I know you talked to someone, but I can’t remember who [ ].
Ellipsis can affect not only TPs but also VPs and NPs. An example of VP ellipsis is
presented in (37) and an example of NP ellipsis is presented in (38):
2A more in-depth discussion os Johnson’s and Lin’s analysis will take place in section 2.2.3.2.
2.2. GAPPING 37
(37) John talked to someone, and Mark did [ ] too.
(38) John talked to Mark’s mother, and Bill talked to Peter’s [ ].
However, ellipsis cannot affect predicate APs. This is illustrated by the ungrammatical
examples in (39) and (40) that result from eliding the whole phrase. Eliding part of the
phrase as in pseudogapping is also ungrammatical, as can be seen in (41). Examples (39),
(40), and (41) correspond to Johnson’s examples (72c, 72b, and 74b) respectively:
(39) *I consider Betsy pretty, while you consider Sam [AP ] .
(40) *Will seems happy today, while Nishi seemed [AP ] yesterday.
(41) *Will seems happy with Vivek, while Nishi seemed [AP ] with Missy-Sue.
Another difference between gapping and ellipsis appears here because gapping can affect
As. Examples (42) and (43) (77a, b in Johnson (2004)) show that As can be gapped.
(42) I consider Liz fond of chocolates and Sam [AP ] of pies.
(43) I made Sam angry at Beaver and Betsy [AP ] at Perseus.
In order to see that this is a gapping example and not an ellipsis one, a diagnostic
test can be used. This test is related to those elements that can have wide scope over
both conjuncts. The only way an element in the first conjunct can have wide scope over
both conjuncts is when the gapping process is present. This is explained in more detail
in section 2.2.4.1 ‘Wide Scope Elements’. Hence, if an example without an adjective
is possible and still an element in the first conjunct can have wide scope, the process
responsible for the omission of that adjective can only be gapping. The following two
examples (44) and (45) ((78b) and (78a) in Johnson (2004)) show that it is possible
to have a wide scope element (namely, negation) in the first conjunct plus an omitted
adjective in the later conjunct.
(44) This won’t make every participanti angry at the judges or heri child [ ] at the
sponsors.
“It is not the case that either this will make every participant angry at the judges
or this will make every participant’s child angry at the sponsors.”
(45) I found no fatheri at the concert fond of the violent rap tunes or hisi child [ ] of
the medieval ballads.
“It is not the case that either I found each father at the concert fond of the violent
rap tunes or I found each father’s child fond of medieval ballads.”
The fact that ellipsis cannot affect APs but gapping can affect As is another difference
that arises between the two processes. This different behavior between the two processes
that arises in relation to categories shows that gapping and ellipsis cannot be subsumed
38 CHAPTER 2. GAPPING
under the same label and that they are different processes. Because of that, the analysis
of at least some types of gapping should be different to that of ellipsis.
2.2.2.5.3 The Strings Affected Differ: Gapping and pseudogapping behave differ-
ently also with respect to the strings of words that they can affect (Johnson (2004)). This
further reinforces the idea that gapping is a process different from pseudogapping or VP
ellipsis and that as such it should receive an analysis different to that of ellipsis. In certain
small clauses, the subject can be gapped together with the higher verb. However, this is
not possible in pseudogapping. In example (46) below, which is Johnson’s (2004) (79b),
gapping affects the higher verb together with the subject of the small clause. The sen-
tence is grammatical. The corresponding English pseudogapping example in (47) ((80b)
in Johnson (2004)) is ungrammatical:
(46) Some considered Sandy handsome and others [ ] plain.
(47) *Even though some considered Sandy handsome, others did [ ] plain.
Example (46) shows that when gapping affects both the subject of the small clause ‘Sandy’
and the higher verb ‘considered’ the result is fine. However, when pseudogapping applies
(47) and both the subject of the small clause ‘Sandy’ and the higher verb ‘considered’ are
affected the result is ungrammatical. Both processes are shown to be different again.
2.2.2.5.4 The Identity Conditions on Antecedence Differ: Elided material has
strict requirements in order to identify itself with the previous material that can function
as an antecedent. In spite of the strict requirements, the elided material and its antecedent
might not be exactly the same. (48) shows that an elided active VP can be identified
with a passive VP. This corresponds to example (85) in Johnson (2004).
(48) Botanist: ‘That can all be explained.’
Mr. Spock: ‘Please, do [VP ] .’
The elided VP in this example is ‘explain all that’ which is active and can be identified
because of the passive predicate in the previous sentence.
In addition, according to Johnson (2004) or Elbourne (2008), a VP can also be elided
by identification with material from two previous conjuncts at the same time, as in (49).
An elided VP can also find its antecedent within a previous DP, as in (50). These are
Johnson’s (2004) examples (86) and (87):
(49) Wendy is eager to sail around the world and Bruce is eager to climb Killimanjaro,
but neither of them can [ ] because money is tight.
(50) ?Sal is a talented forger, but Holly can’t [ ] at all.
2.2. GAPPING 39
The missing string after ‘can’ in (49) refers to the two previous conjuncts. In (50), what
is missing is ‘forge’ which can be identified because of the DP ‘a talented forger’.
Hence, although the identity conditions are strong in the ellipsis cases, sometimes
there can be some looseness. In contrast, the requirements in the gapping process are
much stricter. The looseness that is found in the ellipsis examples cannot be found in
the gapping cases. Consider (51) and (52) below, which are examples (88b) and (89b) in
Johnson (2004).
(51) *The budget cuts might be defended publicly by the chancellor, and the president
[ ] her labor policies.
(52) *Sal may be a forger of passports and Holly [ ] of paintings.
The gap in (51) would refer to the passive VP in the first conjunct to mean ‘may defend’
if the sentence was grammatical. Earlier, we saw that this kind of identification is possible
in ellipsis. The same happens in (52). The gap would be understood as ‘forge’ but it is
not possible in this example with gapping. This evidence also shows that gapping behaves
differently from ellipsis.
There is a reason why identity in gapping needs to be so strict. Identity conditions are
strict in gapping because this process results from movement. That is, the gap corresponds
to a moved element that has moved to the shared part of the structure. The moved element
and the gap are the same and identity is strict between them because they are the same.
This can only be explained straightforwardly by this approach because it is based on
movement. The ellipsis approaches cannot explain it as easily.
2.2.2.5.5 Connectors are also relevant in this context because the behavior of gap-
ping is not the same with all types of connectors. Connectors ‘and’ and ‘or’ are fine with
gapping. The connector ‘nor’ is also acceptable with gapping as illustrated in (53).
(53) Mary hasn’t a car nor Sally a garage.
2.2.3 Gapping analyses
The two major approaches to gapping proposed in the literature have been labeled as
‘large conjunct approach’ and ‘small conjunct approach’. I explain each of them and I
star with the former one.
40 CHAPTER 2. GAPPING
2.2.3.1 Large conjunct analyses
Analyses inside this mainstream are characterized by stating that the parts of the sentence
that are conjoined, the conjuncts, can only be of sentence size. Representative works
inside this mainstream are: Chomsky (1957), Gleitman (1965), Ross (1970), Hankamer
(1971), Hankamer (1972), Hankamer and Sag (1976), Neijt (1979), Brucart (1987), Wilder
(1994), Wilder (1997), Hartmann (2000), Depiante (2000). In order to account for missing
elements, these analyses need different types of mechanisms upon which further principles
apply. These principles are of different nature and of different number depending on the
authors but I will ignore these differences and present what I consider to be the common
core of the large conjunct analysis. A typical basic structure for an example like (54)
inside these analyses could be (55):
(54) John went to the concert and Dave [ ] to the movies.
(55) CoP
TP Co’
CoTP
Dave (went)
to the movies
John went
to the concert
The most commonly used mechanism to account for gaps or missing elements is deletion;
principles or restrictions upon deletion or deleted elements depend on different aspects.
Consider the following for illustration.
For Hartmann (2000), the restrictions upon missing elements come from intonation.
In this approach, elements that are omitted in ellipsis cases are those that are not fo-
calized. Wilder’s (1994, 1997) basis to account for missing elements is not intonation.
Wilder proposes to include a number of constraints that depend on content, context, or
parallelism, among others. I mention here the restrictions that he proposes and I refer
the interested reader to his work for further detail. In Wilder (1997), different restrictions
apply according to the types of ellipsis. In Forward deletion, conditions are of two types:
(i) conditions on the antecedent-ellipsis relation and (ii) conditions on ellipsis independent
of the antecedent. Inside the first group, restrictions are related to (a) content identity
with the antecedent, (b) context identity with the antecedent, and are also related to
(c) locality. The second group of restrictions, which are not related to the antecedent, is
related to aspects of formal licensing. In Backward deletion, there are constraints on (i)
2.2. GAPPING 41
peripherality, (ii) content-identity, and (iii) parallelism.
Although the core feature of these approaches is that conjuncts need to be sentential,
some of these propose small modifications to this idea. Wilder (1997) for instance differ-
entiates two types of coordination. On the one hand, Wilder proposes that DPs may be
coordinated as phrases, not sentences, basing this idea on Lakoff and Peters’s (1969) type
of evidence. On the other hand, Wilder subsumes any other type of coordination under
sentential coordination. On a different line, small conjunct analyses extend the possibility
of having phrasal coordination for DPs to every phrase. I explain it in what follows.
2.2.3.2 Small conjunct analyses
The central idea that differentiates these analyses from the ones above is the idea that
not only sentences but phrases of all types can be conjuncts. That is, conjuncts can
be smaller than sentences. Representative works inside this mainstream are: Lakoff and
Peters (1969), Siegel (1984), Siegel (1987), McCawley (1993), Johnson (2004), Johnson
(2009), Coppock (2001), Lin (2000), Lin (2002). A very basic representation of what
these analyses propose follows. However, because I cannot illustrate all of them here in a
summarized way, I choose a movement version as representative of such proposals in (56).
42 CHAPTER 2. GAPPING
(56) TP
Johni T’
went CoP
vP Co’
Co
and
vP
Dave v’
tV VP
tV PP
to the
movies
ti v’
tV VP
tV PP
to the
concert
Example (56) illustrates two points: the use of small conjuncts and one of the mechanisms
that can be used in order to account for missing elements: movement –as illustrated by
the representation of traces–. It does not illustrate, however, other possibilities that small
conjunct analyses provide: among them, deletion.
The proposal that I provide in the following sections for Spanish relies on the multiple
tools that this kind of approach makes available. These tools become explicit when
thinking about the small conjunct approaches in detail. Consider the following items:
• To begin with, the fact that small conjuncts are needed in the syntax does not entail
the exclusion of large sentential conjuncts in the system.
• Furthermore, the existence of small conjuncts does not impose that movement is
the only process that accounts for elided elements as illustrated in recent accounts.
• Finally, gapping and small conjunct analyses face sentences where elements, such
as negation, may shape the structure of the whole sentence. In such cases, items A
and B provide instruments to account for such phenomena.
2.2. GAPPING 43
I take this reasoning as part of the basic reasoning to my analysis where, among other
aspects, I show that both movement and deletion are necessary in the analysis of Spanish
gapping sentences. It is important, however, to know the shape that the discussion on
movement and deletion has taken in the literature so far.
2.2.3.3 Small conjuncts, deletion and movement
A this point, I focus on the choice between deletion and movement in small conjunct
approaches. The explanation about these two ways of accounting for missing elements,
deletion and movement, will be the starting point towards the analysis of Spanish gapping
sentences. I review aspects of relevance that appear in the cited works: Johnson (2004),
Lin (2002), and Coppock (2001).
To begin with, let us think of a regular basic gapping example like the one in (57)
below:
(57) John drove the car and Peter [ ] the motorbike.
The final conjunct in (57) is missing the verb ‘drove’. The main question here is: is the
verb missing because it has moved or because it has been affected by deletion?
A way to explain this verb gapping is by means of ATB movement. The Verb moves
Across The Board to the part of the structure that c-commands both conjuncts. In this
kind of situation, both Verbs move at the same time from each conjunct to the shared
part of the structure. This, however, is subject to whether or not the Verb moves in the
language in question (or maybe, subject to the assumptions that are made with respect
to Verb movement in each language). For instance, in Spanish, a simple analysis may be
the one provided in (58) because overt movement of the verb is normally assumed.
(58) Lathe
chicagirl
fuewent
alto.the
parquepark
yand
elthe
chicoboy
[ ][ ]
alto.the
cinemovies
‘The girl went to the park and the boy went to the movies.’
[TP la chica fuei [vP ti al parque] y [vP el chico ti al cine] ]
‘Went’ ATB moves from each conjunct to the c-commanding part of the structure as rep-
resented by the traces in each conjunct.3
In English, where the Verb is normally assumed to overtly stay in its source position,
this ATB movement of the Verb would be normally taken as impossible. However, Johnson
(2004, 2009) provides a different view of the behavior of the Verb in English basing it on
3Further analysis of gapping in Spanish is provided in the next section.
44 CHAPTER 2. GAPPING
evidence that he provides in Johnson (1991). Consider (59) for illustration of Johnson’s
(2004, 2009) analysis.
(59) TP
Johni T’
T PredP
Pred’
drove CoP
vP Co’
Co
and
vP
Peter v’
tV VP
tV the
motorbike
ti v’
tV VP
tV the
car
In structure (59), the Verbs move ATB to a PredP that is situated right above vP. The
subject in the first conjunct moves also to the shared part, according to Johnson not
violating the CSC because he takes the latter to apply only to A’-movements. Though
Johnson’s main proposals –(i) Verbs move ATB to PredP; (ii) only movement accounts
for missing elements– introduce more complexity for other examples, I consider that the
representation in (59) suffices for the discussion here.
A different possibility that is contemplated in the literature to account for English
missing Verbs in gapping constructions is introduced by examples where not only the
lexical Verb is missing but also the Auxiliary as in (60):
(60) John has bought a car and Mark [ ] a bicycle.
Although Lin (2002) contemplates English Verb movement, when facing examples of this
type, she proposes to make deletion affect the Verb as a last resort operation. Lin (2002)
makes use of Lasnik’s Salvation by Deletion tool to account for such phenomena. This
2.2. GAPPING 45
introduces the deletion process in small conjunct structures while also having movement
at the same time. The two strategies become useful in dealing with missing elements
–movement and deletion–. I represent Lin’s strategy in (61), which is the structure for
sentence (60).
(61) [TP Johni has [vP [vP ti bought a car] and [vP Mark bought a bicycle] ] ]
In Lin’s account in (61), deletion affects the Verb in the second conjunct in a Salvation
by Deletion operation.
Coppock’s position is at the opposite side of the continuum. Coppock proposes a
complete PF deletion account in order to explain the omission of elements. A broad
representation of Coppock’s analysis is in (62). With it, I return to the simpler gapping
sentence in (57).
(62) TP
Johni T’
T CoP
VP Co’
Co
and
VP
Peterj VP
the
motorbikek
VP
tj V’
(drove) tk
ti V’
drove the car
The structure in (57) basically represents what Coppock specifies. She proposes (i) ad-
junction of the Verb’s arguments to VP alluding to Shortest Move and (ii) deletion of VP.
No further specifications with respect to other phrases, such as vP, are made in Coppock’s
analysis.
Although Johnson, Lin and Coppock provide all small conjunct approaches, they differ
from each other in their choice between movement and deletion to account for missing
elements. On the one side, Johnson only makes use of movement. On the opposite side,
46 CHAPTER 2. GAPPING
Coppock only uses deletion to account for missing elements. And in between Johnson
and Coppock, Lin uses both deletion and movement for the same purpose.
To sum up, within these approaches, gapping is analyzed by means of small conjuncts
and two other processes, deletion and movement. Part of my proposal points towards the
need to use these two in the analysis of Spanish gapping. This is included in the next
section 2.2.4. Before that, I develop the small-large conjunct debate further.
2.2.4 Evidence for two kinds of coordination
Among the arguments which proponents in both the large and the small conjunct approach
use there are some arguments that demonstrate the necessity of a part in the structure
that c-commands the whole coordination. The main idea is that there are elements that
scope above coordination. This type of argument is evidence that small conjuncts are
needed in the syntax and, with this argument, the possibility of large conjuncts does
not get excluded. The following are arguments in favor of two types of coordination,
coordination of large conjuncts and of small conjuncts.
2.2.4.1 Wide scope elements
2.2.4.1.1 Negation can take wide scope. Although I provide more details below,
in the next subsection, consider these data as illustration. According to many small
conjunct approaches, the scope of negation depends on whether the ellipsis process is
gapping or not. Consider (63-65) where negation is overt only in the first part of the
coordination and where there is no gapping. The scope of the negation stops in its same
conjunct, as in (63), when all the elements are present in both conjuncts. The same
happens when there is VP ellipsis (64) or pseudogapping (65): we can only find narrow
scope in examples of this type. Examples are Johnson’s (2004): (63a), (63b), (63c).
(63) a. Kim didn’t eat natto and Sandy ate rice.
b. [TP . . . NOT eat natto] AND [TP Sandy eat rice]
(64) Kim didn’t eat natto and Sandy did [ ].
(65) Kim didn’t eat natto and Sandy did [ ] rice.
The negation in (63) gets to negate the first conjunct. Hence, the second conjunct re-
mains affirmative. The same happens in the VP ellipsis example in (64) where the second
conjunct without ellipsis would be ‘ate natto’ (no negative meaning present). The pseudo-
gapping (65) example behaves in exactly the same way. Its second conjunct is affirmative
and negation has narrow scope.
2.2. GAPPING 47
The same does not happen in gapping contexts. Siegel (1987) and Oehrle (1987) dis-
covered that when there is a negation in the first conjunct and gapping in non initial
conjuncts at the same time, the scope of negation becomes wide. That is, the meaning
of negation reaches the second conjunct. Example (66) is a gapping coordinate structure
where negation is present only in the left conjunct. This is example (64) in Johnson which
was inspired by Oehrle’s (1987) work.
(66) a. Kim didn’t play bingo and Sandy [ ] sit at home all evening.
b. [Shared . . . NOT [Conj−vP play bingo AND Sandy sit at home all evening] ]
Still the reading of this sentence is: ‘it is not the case that both Kim played bingo and
Sandy sat at home all evening’. The meaning of this sentence shows that the scope of
negation is wider in this case because the second conjunct can be false in order for the
sentence to be true. These data (63)-(66) show that the behavior of negation in gapping
contexts is different from that of negation in other coordinate contexts. I analyze the
interaction between negation and gapping in a separate section that sheds light both on
scope of negation and on the analysis of the corresponding sentences. I continue now with
further wide scope phenomena.
2.2.4.1.2 Wide scope of modals The scope of the modal verb also changes de-
pending on whether there is gapping or not in non initial conjuncts. Compare these two
examples (67, 68) (Johnson (2004): 65a, 65b).
(67) a. One man must get the majority of votes and the other must win the election.
b. [TP . . . MUST get the m. of votes] AND [TP the other MUST win the election]
(68) a. One man must get the majority of votes and the other [ ] win the election.
b. [Shared . . . MUST [Conj−vP get the m. of votes AND the other win the election]
The sentence in (67) has a reading where there is a requirement that one man should obtain
the majority of the votes and the other requirement that one should win the election. This
reading is different to the reading in the sentence in (68) because it describes a ‘perverse
anti-democracy’ (Johnson’s words) where one man gets the majority of the votes and a
different man wins the election. The reading of the gapping sentence (68) is different to the
reading where all elements are present in the conjuncts (67). The reason is that the modal
can scope over both conjuncts in the gapping sentence. The behavior of modal verbs in
coordinate sentences where gapping is present shows that the structure of sentences of
this type should allow a part of the structure to be above both conjuncts. This way, those
elements that would be situated in that part of the structure would be able to scope over
both conjuncts.
48 CHAPTER 2. GAPPING
2.2.4.1.3 Variable binding across conjuncts McCawley (1993) discovered cross-
conjunct binding. In examples of this type, a variable in the second conjunct can be
bound by an element that precedes ‘and’. In the following example (69) (Johnson (2004):
66a)), the pronoun ‘her’ is bound by the quantifier ‘not every girl’.
(69) Not every girli ate a green banana and heri mother [ ] a ripe one.
This can only happen if there is a part in the structure of the whole sentence that is above
both conjuncts and that hence c-commands the material in both conjuncts. This reading
with the bound pronoun is only possible when the verb is gapped. Once all the elements
are included in the sentence, the pronoun cannot be bound by the quantifier in the first
conjunct. This is shown by the ungrammaticality of the reading represented in example
(70).
(70) *Not every girli ate a green banana and heri mother ate a ripe one.
In this example (70) where no element is omitted, ‘her’ in the second conjunct cannot
be c-commanded by ‘not every girl’ in the first one. This means that there is no part of
the structure that c-commands all the conjoined material when all elements are present
in the sentence. It is only gapping that forces the presence of a c-commanding part in
the structure which is shared by the conjuncts. There are more elements that can scope
widely over the whole coordination but these, which have been presented here, illustrate
the phenomenon.
To sum up this subsection, there are two types of coordination that are needed: senten-
tial coordination and small conjunct coordination. The existence of wide scope elements
in coordinate contexts constitutes evidence that small conjuncts are needed. And those
coordinate contexts where elements do not scope widely are sentential coordination cases.
With respect to the readings of wide scope elements, it is important to have a closer look
at negation because it adds to this division in coordination types. Repp (2009) provides
a closer analysis of negation and its interaction with gapping that sheds light on this
discussion. Repp’s closer look at the readings that negation displays with respect to co-
ordination and gapping completes the picture that I use as the basis for my analysis on
gapping in Spanish.
2.2.4.2 Repp’s contribution: negation and gapping
An important point that Repp (2009) mentions in her work is that negation has been
taken into account in small conjunct approaches in a limited way. In other words, small
conjunct approaches have been able to account for wide scope readings of negation but
these approaches have not deepened enough in the nature of the interaction of negation
2.2. GAPPING 49
and gapping. Repp (2009) provides data that the readings of negation are three: readings
where negation has (i) wide scope, (ii) narrow scope, and (iii) distributed meaning. Small
conjunct approaches so far have not been able to deal with these three readings. I provide
details about Repp’s work here. My final goal is to apply this onto Spanish in the next
section and to propose a small conjunct analysis as complete as possible, also in light of
Repp’s contribution.
In the analysis of the three meanings of negation that Repp finds in gapping, Repp
uses data where only the first part of the coordination contains a negative marker. The
three meanings of negation in these cases (71)-(73) depend on whether or not, and how
the negation meaning reaches the second conjunct. Repp provides the following examples
(71)-(73).
(71) a. Pete hasn’t got a video and John [ ] a DVD.
b. Distributed scope reading of negation in coordination: [¬ p] ∧ [¬ q]
(72) a. Pete didn’t clean the whole flat and John [ ] laze around all afternoon.
b. Wide scope reading of negation in coordination: ¬ [[ p ] ∧ [ q ]]
(73) a. Pete wasn’t called by Vanessa and John [ ] only by Jessie.
b. Narrow scope reading of negation in coordination: [¬ p] ∧ [ q ]
In all three examples, there is a negative marker in the first part of the coordination and
the second conjunct does not contain any.
In the example in (71), although negation is not present in the second conjunct, this
final conjunct is interpreted as negative like the first one. This is the distributed reading
of negation according to Repp’s terminology.4 The example in (72) illustrates the wide
scope reading of negation. Although negation reaches the second conjunct in this case, it
does it in a different way. In this example (72), conjuncts are not negated separately like
in (71). What is negated instead is a situation where ‘Pete cleans the whole flat while
John lazes around all afternoon’. Finally, the example in (73) illustrates the narrow scope
cases, where negation only affects the first conjunct. That is, in (73), John was called by
Jessie and negation only affects the calling by Vanessa.
Intonation in (72) requires special attention: the sentence must be read in a single
intonational phrase, without any pause between the conjoined parts, and the auxiliary
needs to be stressed. The intonation of the distributed scope reading (71) is different. In
this case, there are two individual intonational phrases, the Verbs are deaccented, and
4From now on, I follow this terminology in order to explain phenomena that are related to negation.
50 CHAPTER 2. GAPPING
contrast pairs are highlighted by accents. Repp mentions that there is an ambiguity in
example (71). Apart from having a distributed reading, negation can have wide scope in
this example provided that intonation features change to those of the wide scope reading.
Repp’s work is extensive on the characterization of the intonational conditions that are
required for sentences of this type. For more details, the reader is referred to her work.
Other empirical data which Repp provides comes from German. A similar pattern to that
of English can also be found in that language. Consider examples in (74)-(76):
(74) a. MaxMax
hathad
dasthe
Buchbook
nichtnot
gelesenread
undand
MarthaMartha
??(nicht)??(not)
diethe
Zeitschriftmagazine
‘Max didn’t read the book and Martha the magazine.’ (P. 41 (2b), German)
b. Distributed Scope: [¬ p] ∧ [¬ q]
(75) a. LeonLeon
kanncan
nichtnot
Kaviarcaviar
esseneat
undand
AnnaAnna
*(gleichzeitig)*(at.the.same.time)
Bohnenbeans
‘Leon can’t eat caviar and at the same time, Anna beans.’ (P. 189 (2),
German)
b. Wide Scope: ¬ [[ p ] ∧ [ q ]]
(76) a. HANSHans
hathad
dasthe
BUCHbook
nichtnot
gekauftbought
undand
MaRIaMaria
diethe
ZEItungmagazine
vergessenforgot
‘Hans didn’t buy the book and Mary forgot the magazine.’ (P. 230 (2d),
German)
b. Narrow Scope: [¬ p] ∧ [ q ]
The descriptive and analytical picture that Repp provides in reference to these examples
is much more exhaustive than what I explain here. Details are provided in that work. I
would like to focus on the following. The three readings arise in German though there is
a difference. The readings of negation in these German examples in (74)-(76) follow the
same order as in the English examples (71)-(73). Sentence (74) corresponds to the dis-
tributed reading where the two conjuncts are separately interpreted as negative. Sentence
(75) corresponds to the wide scope reading of negation. And the last sentence (76) corre-
sponds to the narrow scope reading where only the first conjunct is interpreted as negative.
The difference between English and German lies on the distributed reading example
(74). In German the negative marker nicht ‘not’ is also needed in the second conjunct,
however, this is not the case in English. As illustrated in (71), the negative marker in
English may only be present in the first part of the coordination and still be interpreted
in both conjuncts. Repp explains this difference attributing it to the categorial status of
the negation marker in German.
2.3. GAPPING IN SPANISH 51
Cross-linguistic differences arise between English and Spanish also. Because of that, I
leave the discussion here and I retake it in section 2.3 in reference to gapping in Spanish,
and the interaction of gapping and negation in Spanish plus the existing differences with
respect to English. So far, I have provided a part of the paradigm that Repp offers in her
work, the part of the paradigm that is of use for the purposes of this research. In the next
section, when dealing with negation, I reproduce this paradigm and provide an analysis
for the Spanish cases presented there.
2.2.5 Interim summary
Among the extensive work that has been done on gapping, this section has focused on
describing the gapping phenomenon and on explaining the way in which the literature
has shown the need for two types of coordination: coordination at the CP level and
coordination at lower levels. Wide scope elements play an important role in this matter
and Repp’s closer look at one of those elements, negation, brings light on small conjunct
approaches to coordination. As a whole, this section aims to be the basis for the next
section where an analysis for gapping in Spanish is searched for. In the rest of the chapter,
gapping in Spanish is described; two types of coordination are proven to be also needed
in Spanish via a close look at wide scope elements in Spanish; and an analysis for Spanish
gaps and complex gaps is provided, all this done in light of the literature that was reviewed
in this section.
2.3 Gapping in Spanish
2.3.1 Introduction
In this section, I begin with the analysis of gapping in Spanish. Although the analysis
on Spanish gapping continues in the rest of the chapter, one goal in this section is to
show that the behavior of gapping in Spanish is comparable to that of English. Another
goal is to start providing an analysis for gapping as complete as possible in light of recent
literature. Structure wise, in this section, I put attention on the size of conjuncts, leaving
the analysis of gaps for section 2.4.
In subsection 2.3.2, I explain the properties of gapping in Spanish. In subsection
2.3.3, I start to prove the necessity of two types of conjuncts in Spanish coordination
via wide scope elements: coordination at the CP level and conjunction at lower levels. I
follow the same line of thought in subsection 2.3.4 and I take Repp’s work as reference. I
analyze negation’s behavior closely in order to get a better picture of gapping phenomena
52 CHAPTER 2. GAPPING
in Spanish than what is normally exposed in either large or small conjunct approaches to
coordination cross linguistically. I provide interim conclusions in subsection 2.3.5.
2.3.2 Properties of Spanish gapping
The properties of gapping in Spanish are similar to the properties of gapping in English.
2.3.2.1 Gapping affects the Verb in final conjuncts in Spanish coordinate
contexts.
This is the same that occurs in English and any other head-initial languages. Consider
(77). The ungrammatical example (78) shows that gapping cannot affect initial conjuncts
in Spanish.
(77) JorgeJorge
leCL
diogave
unaa
florflower
ato
MarıaMarıa
yand
PedroPedro
[ ][ ]
una
librobook
ato
ElenaElena
‘Jorge gave a flower to Marıa and Pedro gave a book to Elena.’
(78) *JorgeJorge
[ ][ ]
unaa
florflower
ato
MarıaMarıa
yand
PedroPedro
leCL
diogave
una
librobook
ato
ElenaElena
‘Jorge gave a flower to Marıa and Pedro gave a book to Elena.’
The gap that corresponds to dio ‘gave’ in (77) is in the final conjunct as in other head-
initial languages. If the gap is in the first conjunct, ungrammaticality arises. Spanish
gapping behaves like other head-initial languages in this respect.
2.3.2.2 Verbal Complexes
The same paradigm as in English or German (Siegel (1984, 1987), Lin (2002), Repp
(2009)) is also present in Spanish with respect to the examples in (79)-(81).
(79) JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
[ ][ ]
fregadowashed
elthe
suelofloor
‘Jorge has cleaned the bathroom and Pedro has washed the floor.’
(80) JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
[ ][ ]
elthe
suelofloor
‘Jorge has cleaned the bathroom and Pedro has cleaned the floor.’
(81) *JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
hahas
[ ][ ]
elthe
suelofloor
‘Jorge has cleaned the bathroom and Pedro has cleaned the floor.’
Example (79) illustrates that in Spanish gapping, it may be the case that only the finite
verb is missing, the auxiliary ha ‘has’. The next example (80) illustrates that it can also
be the case that the finite verb and the lexical verb, ha limpiado ‘has cleaned’, are missing.
2.3. GAPPING IN SPANISH 53
And finally, the last example in the paradigm, (81), illustrates that leaving the finite part
of the verb and omitting only the lexical verb is not possible in Spanish gapping.
2.3.2.3 Gapping in Spanish and the features of finite verbs
In order for a verb to be gapped in a coordinate context, Tense, Aspect, and Mood features
of both verbs need to be the same but features like person, number, or gender may differ.
Consider (82) and (83) below.
(82) ??Losthe
chicosboys
ayeryesterday
fueronwent
alto.the
cinemovies
yand
lasthe
chicasgirls
mananatomorrow
[ ][ ]
alto.the
teatrotheater‘The boys went to the movies yesterday and the girls will go to the theater tomor-
row.’
(83) Ellosthey
fueronwent
alto.the
cinemovies
yand
nosotraswe
[ ][ ]
alto.the
teatrotheater
‘They went to the movies and we went to the theater.’
The verbs in each conjunct in example (82) have different Tense features: the verb in the
first conjunct is in the past tense and the verb in the second conjunct needs to be in the
future tense as the adverb manana ‘tomorrow’ indicates. The two question marks indicate
that the example is hard to accept: gapping of the second verb is not fine because of this
semantic difference between the two verbs. Example (83) is different. The combination
of different person features yields a fine result.
2.3.2.4 Complex Gapping
Another feature that is similar to English is that in Spanish, when the verb is gapped,
the Direct Object or the Indirect Object can also be elided. Consider (84).
(84) JorgeJorge
leCL
diogave
unaa
florflower
ato
MarıaMarıa
yand
PedroPedro
[ ][ ]
ato
ElenaElena
‘Jorge gave a flower to Marıa and Pedro gave a flower to Elena.’
In this example (84), both the main verb dio ‘gave’ and the Direct Object una flor ‘a
flower’ are missing from the second conjunct. The fact that the sentence is grammatical
indicates that Spanish behaves similarly to English in this respect too.
2.3.2.5 Additional Properties of gapping in Spanish
The properties of gapping that Johnson’s work enumerates are also repeated in Spanish.
However, research on these properties cannot be done in the same way as Johnson (2004)
does. The reason for this is that ellipsis phenomena in Spanish are much more limited
54 CHAPTER 2. GAPPING
than in English (Brucart (1987), Depiante (2000)). VP ellipsis or pseudogapping are not
possible in Spanish while gapping, sluicing, or bare argument ellipsis (BAE) are possible.
That is why I do not compare Spanish gapping cases with ellipsis processes. According
to Johnson (2004), comparison should be done between gapping and pseudogapping/VP
ellipsis because of the similarities between them. Because these ellipsis processes are not
present in Spanish, that comparison cannot be done. In any case, the possibility still
exists to show that the behavior of Spanish gapping is parallel to that of English. This is
done in what follows showing that gapping behaves similarly in both languages.
2.3.2.5.1 Licensing environments for Spanish gapping These environments are
similar to those for gapping in English: gapping in Spanish is only possible in coordination.
Notice the grammaticality of example (85) which is a coordination context and its contrast
with the use of subordinate clauses in (86) and (87):
(85) JuanJuan
comeeats
arrozrice
yand
PedroPedro
[ ][ ]
alubias.beans
(86) ?Juanjuan
comeeats
arrozrice
perobut
PedroPedro
[ ][ ]
alubias.beans
(87) *Juanjuan
comeeats
arrozrice
ain
pesarspite
deof
quethat
PedroPedro
[ ][ ]
alubias.beans
‘Juan eats rice although Pedro beans’
Restrictions on gapping in Spanish are similar to restrictions on English gapping:
gapping is subject both to coordination contexts and to configurations that obey rules
on movement. In the examples that Johnson creates where there are three clauses like in
(88), the CSC needs to be obeyed. In the Spanish example in (88), the CSC is violated
and hence the result is ungrammatical, the same as in English:
(88) *Juanjuan
siemprealways
bebedrinks
martinismartinis
oor
esis
unaa
reunionmeeting
importanteimportant
yand
nadienoone
[ ][ ]
nadaany
deof
alcohol.alcohol
2.3.2.5.2 Affected categories Although ellipsis normally cannot affect Adjectival
Phrases, gapping can. Spanish example (89) shows that in Spanish, gapping can also
affect APs.
(89) Dejeleft.1SG
ato
mimy
madremother
enfadadaangry
conat
mimy
padrefather
yand
tuyou
ato
tuyour
hermanasister
[ ][ ]
conat
suher
novio.boyfriend
‘I left my mother angry at my father and you left your sister angry at her
boyfriend.’
2.3. GAPPING IN SPANISH 55
To show that this Spanish example (89) is a gapping example, we can use the same
test as in English. Elements that can scope above coordination can be placed in the first
part of the sentence (90, 91). This shows that gapping is present since, otherwise, wide
scope of such elements would be impossible.
(90) Dejeleft.1SG
ato
cadaeach
periodistai
journalistenfadadoangry
contigoat.you
yand
ato
sui
hisjefeboss
[ ][ ]
conat
elthe
chivato.informer‘I left each journalist angry at you and his boss at the informer.’
(91) Nonot
encontrefound.1SG
ato
ningunany
padreifather
enin
elthe
conciertoconcert
contentohappy
conwith
elthe
raprap
violentoviolent
oor
ato
sui
hishijoson
[ ][ ]
conwith
lathe
lıricalyrics
medieval.medieval
‘I didn’t find any father in the concert happy at violent rap or his son at medieval
lyrics.’
2.3.2.5.3 Affected strings Example (92) illustrates that in Spanish gapping, the
same as in English, when small clauses are present, the subject of a small clause and the
verb in the main clause can be affected. It has been shown above that ellipsis cannot
behave in such a way.
(92) Algunossome
habrıanhave
podidocould
considerarconsider
ato
JuanJuan
tontostupid
yand
otrosothers
[ ][ ]
inteligente.intelligent
‘Some could have considered Juan stupid and others intelligent.’
2.3.2.5.4 Identity Conditions The examples in (93) and in (94) show that gapping
in Spanish needs strict identity conditions the same as in English. A gap in a conjunct
where voice is active cannot be understood by reference to a passive voice antecedent (93).
Similarly, a verbal gap cannot be understood as making reference to previous DPs (94).
(93) *Losthe
recortescuts
deof
presupuestobudget
puedencan
serbe
defendidosdefended
publicamentepublicly
porby
elthe
vicepresidentevicepresident
yand
elthe
presidentepresident
[ ][ ]
sushis
polıticaspolitics
deof
trabajo.work
‘The cuts of budget can be defended publicly by the vicepresident and the president
can defend publicly his work politics.’
(94) *PedroPedro
puedecan
serbe
una
falsificadorforger
deof
pasaportespassports
yand
HollyHolly
[ ][ ]
pinturas.paintings
Pedro can be a forger of passports and Holly [ ] paintings.’
Gapping in Spanish has been shown to be similar to English gapping through the
analysis of all the properties that are listed in this section. Previously when discussing
56 CHAPTER 2. GAPPING
English gapping, I have provided evidence that two types of coordination are needed:
small conjunct coordination and sentential coordination. I continue now with this related
aspect to gapping, checking whether or not the same happens in Spanish. I start with it
in the next section, which deals with wide scope elements.
2.3.3 Wide scope elements
It has been shown in section 2.2.4 that wide scope elements bring evidence that two types
of coordination are needed. I provide Spanish data here with the same purpose of showing
that two types of coordination are needed: small conjunct coordination and sentential co-
ordination. In Spanish, there are also elements that can scope widely above coordination.
Negation, modal verbs, and cross-conjunct binding constructions are examples of such
cases.
2.3.3.0.5 The interaction between negation and gapping Although subsection
2.3.4 is entirely focused on negation, here, I roughly illustrate that at least two readings
for negation arise depending on the presence or the absence of gapping. Consider (95)
and (96) so that the difference in readings from one example to another is illustrated. The
scope of negation in (95) is narrow where all the elements in the coordination are present.
Wide scope negation is illustrated in (96) where the auxiliary habıa ‘had’ is gapped.
(95) PedroPedro
nonot
comioate
alubiasbeans
yand
JuanJuan
comioate
arroz.rice
‘Pedro didn’t eat beans and Juan ate rice.’
(96) JuanJuan
nonot
habıahad
comidoeaten
alubiasbeans
yand
MarıaMaria
[ ][ ]
bebidodrunk
champan.champagne
‘Juan hadn’t eaten beans and Maria drunk champagne.’
Both the narrow scope and the wide scope of negation are evidence that two types of
conjuncts are needed: large or sentential for narrow scope, and small or phrasal for wide
scope. Further evidence supports the same conclusion.
2.3.3.0.6 Cross-conjunct variable binding Cross-conjunct binding is further evi-
dence that there needs to be a shared part of the structure that c-commands all conjoined
material. This is further reason to claim that coordination needs to be phrasal in certain
contexts. Cross-conjunct binding is possible in the next Spanish gapping example (97)
the same as in its English counterpart.
(97) Ningunno
estudianteistudent
leyoread
‘Brooklyn‘Brooklyn
Follies’Follies’
yand
sui
hismadremother
[ ][ ]
‘El‘The
GuardianCatcher
entrein
elthe
Centeno’.Rye’
2.3. GAPPING IN SPANISH 57
‘No student read ‘Brooklyn Follies’ and his mother ‘The Catcher in the Rye’.’
When gapping is not present (98), binding is impossible.
(98) *Ningunno
estudianteistudent
leyoread
‘Brooklyn‘Brooklyn
Follies’Follies’
yand
sui
hismadremother
leyoread
‘El‘The
Guardian’.Catcher’‘No student read ‘Brooklyn Follies’ and his mother read ‘The Catcher’.’
2.3.3.0.7 Wide scope of modals Modal Verbs scope widely with respect to coor-
dination when there is gapping in the sentence. Modal verbs are high enough in the
structure to analyze scope with respect to coordination in gapping examples. Notice the
different readings in (99) and (100). The second example (100) is a gapping coordinate
sentence where the modal verb scopes widely above conjoined material like in Johnson’s
examples.
(99) JuanJuan
deberıashould
arreglarfix
elthe
cochecar
yand
yoI
deberıashould
prepararprepare
lasthe
maletas.luggage
‘Juan should fix the car and I should prepare the luggage.’
(100) JuanJuan
deberıashould
arreglarfix
elthe
cochecar
yand
yoI
[ ][ ]
prepararprepare
lasthe
maletas.luggage
‘Juan should fix the car and I prepare the luggage.’
In order to understand the wide scope reading of this modal verb deberıa ‘should’ and
to see how it differs from the narrow scope reading, it helps to imagine different situ-
ations for each example. In the narrow scope reading (99), the two requirements that
are conjoined are understood as separate requirements, like for instance, in a situation
where Juan and I are preparing to move to a different city. In such a situation, there are
many aspects that we have to complete, among which: one is the fixing of the car which
is Juan’s job, and another is preparing the luggage which is my job. On the wide scope
reading (100) the two conjoined requirements are understood as one. In this case, a situ-
ation that would fit is one where Juan and I are preparing a sudden trip. The day prior
to departure we should fix the car and prepare the luggage, both things understood as one.
This is further evidence that two different analyses are needed for each type of sen-
tence: coordinate sentences where all elements are present and coordinate sentences where
the verb is gapped. Before I start explaining the analysis of gapping that I propose in
detail, I investigate the interaction of Spanish gapping with negation following Repp’s
(?) contribution. This makes me able to also take into consideration the details about
negation’s behavior in my analysis.
58 CHAPTER 2. GAPPING
2.3.4 Negation, conjunction, and gapping: implications of scope
of negation for the analysis of gapping
Among the elements that can have both wide and narrow scope with respect to coordi-
nation, negation proves to be a useful element in the analysis of Spanish gapping. Recent
literature has put attention on it providing a detailed analysis on its interaction with
gapping that brings light on the discussion here about types of coordination. Specifically,
Repp (2009) analyzes negation and gapping in German and English. Repp pays special
attention to the scope of negation with respect to coordination, gapping, different types of
coordinating connectors, focus particles like ‘only’ or ‘even’, among others. One of Repp’s
crucial goals is to prove that small conjunct approaches that rely on movements to the
shared part cannot account for all the phenomena which are related to negation. I do
tackle these phenomena in order to show that it is possible to account for what negation
brings to gapping configurations with my analysis.
Here, I analyze the interaction of negation with coordinate structures and gapping
in Spanish. The data that I provide for Spanish is similar to that provided by Repp for
German or English. In accounting for the resulting paradigms, I reach the conclusion that
small conjuncts are needed, which becomes another argument in favor of the claim that
two types of coordination are needed. Furthermore, I also claim that connectors play an
important role in understanding the scope of negation in coordinate structures and I show
that the need for either big or small conjuncts is also subject to the type of connector
that is used in each case.
In the coordinate sentences in (101)-(103), negation is always near the beginning of
the sentence. This negation interacts with three aspects that become relevant in its
interpretation. For now, I focus on two of them: (i) the type of connector and (ii) the
presence or absence of gapping in the sentence, leaving the third one, intonation, for later.
Consider the three sentences in (101)-(103).
(101) Marıamaria
nonot
estuvowas
enfermasick
yand
JuanJuan
estuvowas
acatarrado.with.a.cold
‘Maria was not sick and Juan was with a cold’.
(102) Marıamaria
nonot
estuvowas
enfermasick
yand
JuanJuan
[ ][ ]
acatarrado.with.a.cold
(103) Marıamaria
nonot
estuvowas
enfermasick
ninor
JuanJuan
[ ][ ]
acatarrado.with.a.cold
The scope of negation in these three sentences changes from one another. I claim that the
differences that exist among sentences, that is, the presence or absence of every element
2.3. GAPPING IN SPANISH 59
in the sentence and the type of connector that is used in each case, correlate with different
structures. And it is this structural distinctiveness that account for the different scopes.
I first, describe the meaning nuances and the scope properties and I account syntactically
for such distinctive structures later.
In sentence (101), the negation is in the first part of the coordination and every other
element is overt in both conjuncts. The meaning of this first sentence (101) is such that
the proposition that is negated is the one in the first part of the coordination only. In
sentence (101), Maria was not sick but Juan was. I label this reading of negation in
coordination ‘narrow scope reading’ following Repp (2009).
The second sentence (102) is similar to (101) in that the negation is positioned in the
same way and in that the connector is the same: y ‘and’. The difference comes with
the missing element estuvo ‘was’ in the second conjunct. The gapped verb affects the
meaning of this sentence (102), which changes with respect to (101) in the following way.
Negation, in (102), affects the propositions of both conjuncts as a unity. That is, you
negate the fact that Maria was sick while Juan was with that cold. I refer to the reading
in this example as ‘wide scope reading’ of negation as it is customary in the literature.
The last sentence (103) has similarities with both, the two prior sentences, (101) and
(102). It is similar to (102) in that the verb is gapped in the second conjunct. It is similar
to (101) in that propositions of conjuncts are not interpreted as one. But it is different
to (101) and (102) in the usage of the connector ni ‘nor’. This, in (103), negation affects
both conjuncts but separately. The meaning of the sentence is such that negation affects
two propositions: Maria was not sick; Juan was not with a cold. Following Repp (2009),
I label this type of reading in coordinate contexts ‘distributed scope reading’ of negation.
The third aspect that needs to be taken into consideration is intonation. Its role, how-
ever, is more important in English. The distributed scope example changes in its reading
according to intonation in this language. If the English distributed scope example is read
with the wide scope intonation, its reading changes to wide scope. This ambiguity in
the distributed scope example does not appear in Spanish. In Spanish, the wide scope
reading can only appear in the configuration in (102). The distributed scope reading
can only appear in the configuration in (103). The narrow scope reading also appears in
contexts whose configuration is like in (101). Recall that intonational features of these
two readings are quite different in both languages. The wide scope reading is pronounced
with a single intonational phrase in both Spanish and English, without any pause between
60 CHAPTER 2. GAPPING
the conjoined parts, and with an accent on the auxiliary. The distributed scope reading
is pronounced with two individual intonational phrases, with deaccented Verbs, and with
accents in contrast pairs to highlight them.
Consider the schemas in (104), (105) and (106) which correspond to (101), (102) and
(103) respectively.
(104) Narrow scope reading of negation in coordination: [¬ p] ∧ [ q ]
(105) Wide scope reading of negation in coordination: ¬ [[ p ] ∧ [ q ]]
(106) Distributed scope reading of negation in coordination: [¬ p] ∧ [¬ q]
Below I provide additional examples illustrating this phenomenon. When all elements
are present and the connector that is used is y ‘and’, negation scopes narrowly with respect
to coordination. When there is gapping in the non initial part of coordination and the
connector is y ‘and’, the scope of negation is wide with respect to coordination. And
finally, when there is negation in the first part of the coordination and the connector is
ni ‘nor’, negation displays a distributed reading across conjuncts. Consider the following
examples (107-109, 110-112) and the way negation is interpreted in such cases in the
schemas in (b). These sentences (107-109, 110-112) are parallel to (101, 102, 103).
(107) a. MarıaMaria
nonot
quierewants
irgo
alto.the
cinemovies
yand
PedroPedro
quierewants
irgo
alto.the
teatrotheatre
‘Maria does not want to go to the movies and Pedro wants to go to the theatre’.
b. Narrow Scope: [¬ p] ∧ [ q ]
(108) a. MarıaMaria
nonot
quierewants
irgo
alto.the
cinemovies
yand
PedroPedro
[ ][ ]
alto.the
teatrotheatre
b. Wide Scope: ¬ [[ p ] ∧ [ q ]]
(109) a. MarıaMaria
nonot
quierewants
irgo
alto.the
cinemovies
ninor
PedroPedro
[ ][ ]
alto.the
teatrotheatre
b. Distributed Scope: [¬ p] ∧ [¬ q]
(110) a. Elthe
anestesistaanesthetist
nonot
tienehas
queto
dormirmake.sleep
alto.the
pacientepatient
yand
elthe
medicodoctor
tienehas
queto
extirparleremove
elthe
apendiceappendix
‘The anesthetist does not have to make the patient sleep and the doctor has
to remove the appendix to him’
b. Narrow Scope: [¬ p] ∧ [ q ]
(111) a. Elthe
anestesistaanesthetist
nonot
tienehas
queto
dormirmake.sleep
alto.the
pacientepatient
yand
elthe
medicodoctor
[ ][ ]
extirparleremove
elthe
apendiceappendix
2.3. GAPPING IN SPANISH 61
b. Wide Scope: ¬ [[ p ] ∧ [ q ]]
(112) a. Elthe
anestesistaanesthetist
nonot
tienehas
queto
dormirmake.sleep
alto.the
pacientepatient
ninor
elthe
medicodoctor
[ ][ ]
extirparleremove
elthe
apendiceappendix
b. Distributed Scope: [¬ p] ∧ [¬ q]
I account for these scope differences in the interaction between negation and coordination
syntactically. I claim that different structures account for the different scopes that nega-
tion displays. In doing so, I follow Siegel (1984, 1987), Johnson (2000, 2004, 2009), Lin
(2002), and Wurmbrand (2008). I provide details about my proposal focusing first on the
contrast of two types of readings and I move on to the third reading later.
2.3.4.1 Wide and narrow scopes of negation
I start with the types of readings in (101) and (102), which correspond to schemas (104)
and (105) respectively. In (113) below the contrast between narrow and wide scope
readings is presented:
(113) a. The narrow scope reading of negation: [¬ p] ∧ [ q ]
b. The wide scope reading of negation: ¬ [[ p] ∧ [ q ]]
I assume Siegel (1984, 1987), McCawley (1993), Johnson (2000, 2004, 2009), Lin (2002)
who account for these differences in scope via conjuncts of different sizes in coordinate
structures. The presence of small conjuncts in the structure allows for some part of the
structure to be shared. This is relevant in that this kind of structure leaves space for
elements that scope above all the conjoined material. The analysis for the wide scope
reading of negation in (113b, 102) is broadly illustrated in (114) below. Negation is
located in the shared part of the structure and the conjoined material is lower in the
structure.
(114) [TP Maria not was [vP [vP tSbj tV sick] and [vP Juan with.a.cold ] ] ]
In (114), because negation sits outside the conjuncts, it can scope above all conjoined
material. The wide scope reading is properly accounted for by (114).
The contrast in (113) is completed by a narrow scope reading of negation. This reading
in (113a) can easily be accounted for with a structure where negation sits in its conjunct,
not outside of it. This follows naturally if we take into account that, in example (101),
conjoined elements are complete in the sense that every element is overt. I take this to
mean that conjuncts in examples of this type are sentential following accounts in small
62 CHAPTER 2. GAPPING
conjunct approaches to coordination; and that negation sits inside one of these sentences
that are conjoined, in this case, the first conjunct. Example (115) illustrates the idea.
(115) [TP [TP Maria not was sick] and [TP Juan was with.a.cold] ]
So far, I have provided analyses for the wide scope reading and the narrow scope
reading of negation. The difference between both structures comes from the size of the
conjuncts. I continue now with the distributed reading of negation previously explained
in (103) and (106). I need to focus on both the size of its conjuncts and what brings the
difference between this example and the other two, the type of connector that is present
in this sentence.
2.3.4.2 Distributed scope of negation
In order to provide an account for this distributed scope reading of negation, I contrast
the wide scope reading with the distributed scope reading (116).
(116) a. The wide scope reading of negation in coordination: ¬ [[ p ] ∧ [ q ]]
b. The distributed scope reading of negation in coordination: [¬ p] ∧ [¬ q]
These two readings (116a and b) correspond to the types of sentences in (102, 103) that
I repeat here as (117, 118) for convenience.
(117) MarıaMaria
nonot
estuvowas
enfermasick
yand
JuanJuan
[ ][ ]
acatarrado.with.a.cold
(118) MarıaMaria
nonot
estuvowas
enfermasick
ninor
JuanJuan
[ ][ ]
acatarrado.with.a.cold
The first sentence (117) corresponds to the reading of negation where it scopes widely
with respect to coordination. Sentence (118) corresponds to the distributed reading of
negation. Crucially, the only element that brings the difference between these two ex-
amples (117, 118) is the connector ni ‘nor’ in the second sentence (118). I claim that
this connector and its necessary presence in the structure is what brings the difference
in interpretation between (117) and (118). I follow Wurmbrand (2008) in my analysis of
(118). I show that such analysis enables an account for the distributed reading sentence.
Assuming it, I can explain why the interpretation of negation in this sentence (118) is
different to the other two. The difference comes from its syntax.
Wurmbrand (2008) focuses on connectors of the type NEG-nor and takes the side
that constructions with such connectors are coordinations. The possible two sides to take
are to describe these constructions as disjunctions or conjunctions. This is due to the
logical equivalence of ¬ [p ∨ q] and [¬ p] ∧ [¬ q]. Wurmbrand argues that NEG-nor
2.3. GAPPING IN SPANISH 63
constructions can only be constructed when the logical equivalence does not hold favoring
a coordinate structure that is detailed below. Her arguments come from English and
German sentences and they are based on an observation by Lechner (2000).
Lechner observes that in German the negative element in the first part of the coordi-
nation can be low in the conjunct. When a Negative Polarity Item is introduced higher in
the structure, this does not get licensed. This means that negation cannot scope above it
which is what happens in the disjunction part of the logical equivalence: ¬ [p ∨ q] and [¬
p] ∧ [¬ q]. Another argument comes from quantifier scope. Instances where an existential
quantifier in the first conjunct can only scope above negation in both regular sentences
and NEG-nor constructions also show that a configuration where negation scopes above
coordination is not possible.
These arguments lead Wurmbrand (2008) to propose an analysis with three main
points: (i) the nor connector is composed of ‘and’ plus negation; (ii) in NEG-nor sen-
tences, Wurmbrand shows that NEG is in Spec, CP in the German and English cases that
she accounts for (argumentation for this point is based on Lechner’s (2000) observations
that are specific to German); and (iii) in these types of NEG-nor coordinations, conjuncts
are CPs. These three points are broadly illustrated in (119) below.
(119) a. [CP [CP NEG [C′ . . . ] ] andb [CP notb . . . ] ]
b. ‘and’ - ‘not’: nor ; both merged at PF and spell out as single item ‘nor ’.
The same claim about the distributivity of NEG-nor constructions is put forward
for the Spanish equivalent ni-ni ‘neither-nor’ in Bosque and Demonte (1999), who make
reference to Jimenez-Julia (1984) and Franchini (1986). The idea is that Spanish ni-ni
connectors provide an “excluding” interpretation. In other words, negation, in this con-
struction, never affects both conjuncts as a unity. Instead, negation in ni-ni constructions
affects the conjoined parts separately. Evidence from Franchini (1986) is provided in favor
of this idea. Conjuncts which are coordinated by ni ‘nor’ cannot be modified by collec-
tive predicates. Consider sentences (120) and (121) which are taken from Bosque and
Demonte (1999) (with judgments as quoted).
(120) *Nineither
JuanJuan
ninor
PedroPedro
hanhave
idogone
juntostogether
alto.the
cinemovies
‘Neither Juan nor Pedro have gone to the movies together’.
(121) */#Nonot
hehave.1SG
reunidogot.together
nineither
ato
JuanJuan
ninor
ato
PedroPedro
‘I have not got neither Juan nor Pedro together’.
64 CHAPTER 2. GAPPING
The sentence in (120) is ungrammatical because the collective meaning of the predicate
clashes with the distributive meaning of the connectors ni-ni. The example in (121) is
ungrammatical in the collective reading where Juan and Pedro are set to be together
in the same meeting. Notice that the only reading which could lead to a grammatical
result5 is one where there are two meetings organized: one for Juan to meet with someone
else, and another one for Pedro to meet with other people. This reading arises, to my
understanding, because of the distributive meaning content of the NEG-nor construction
we are dealing with.
My proposal is to adopt Wurmbrand’s (2008) structure to analyze the distributed
reading of negation (118). This analysis provides an account for why the only presence
of the connector ni ‘nor’ brings the distributed reading to the sentence. In Wurmbrand’s
(2008) analysis, the negation content in the connector ‘nor’ is in the second part of the
coordination. With both the NEG in the first conjunct and the negation content of nor
in the second conjunct, the distributed reading of negation is already accounted for. This
is the reason why it is advantageous to analyze example (118) as in (119). The structure
of (118) in (122) represents that negation is present in the structure of both conjuncts
and this only happens because of the presence of ni ‘nor’ in the sentence.
(122) [CP [CP Maria NOT was sick] andnor [CP notnor Juan with.a.cold] ]
With the structure in (122), which accounts for the distributed scope reading of negation,
and the other two structures for wide and narrow scope readings of negation (114, 115), I
have explained the reasons why the three readings arise and I have accounted syntactically
for the differences. Syntactically, two types of coordination are needed: coordination at
the level of CP and coordination at lower levels.
2.3.5 Interim summary
In this section, gapping in Spanish has been shown to behave in a similar way to gapping
in English. This is summarized in table 2.1 below. It has also been proven that two types
of coordination are needed the same as in English: coordination at the top sentential level
and coordination at lower levels. Coordination at the CP level takes place in cases where
either all elements are present in the coordinate sentence or where configurations Neg-nor
appear. Coordination at lower levels is necessary when elements, such as negation or
modals for instance, scope widely above all conjuncts. So far, descriptively, the analysis
of gapping in Spanish is complete. Theoretically, analyses of gapping Spanish sentences
are still general and only specify the size of the conjuncts that is needed in each case. In
5This is not acknowledged in Bosque and Demonte (1999).
2.4. GAPS AND COMPLEX GAPS 65
the next section, I complete the analysis accounting for gaps, discussing on whether it is
deletion or movement that accounts for them. In that section, I also compare these gaps
to ‘complex gaps’ providing an account for both the former and the latter.
Table 2.1: Properties of Gapping in Spanish and English
Gapping . . . Eng. Span.1 occurs in Coordinate Contexts + +2 ’s Gap is in Final Conjuncts in Head-initial languages + +3 ’s Verbal Features: T, Asp, Mod: =/Nr, Gr: may not = + +4 permits DO/IO to also be omitted + +
In Johnson’s View. . . . . . . . .
5 Licensing environments + +6 Coordinate Embeddedness + +7 Categories Affected Differ + +8 Strings Affected Differ + +9 Identity Conditions on Antecedence Differ + +10 Array of Connectors + +
And the Scope of Elements. . . . . . . . .
11 Wide Scope Elements: Neg, Modals, Cross-conjunct Binding + +12 Negation: Wide, Narrow and Distributed Scopes + +
2.4 Gaps and complex gaps
2.4.1 Introduction
In this section, I analyze the types of gapping sentences that have been mentioned in the
chapter taking into account the different aspects that have been proven.
The types of gapping sentences are:
1. Basic gapping: the only lexical Verb is missing.
2. Gapping of Auxiliary: only the Auxiliary is gapped.
3. Gapping of Modal: only the Modal Verb is gapped.
4. Complex gapping: the lexical Verb and one object are missing; or the lexical Verb
and the Auxiliary/Modal are missing.
5. Gapping in NEG-nor : there is gapping inside a NEG-nor configuration.
In order to successfully analyze all of these subtypes, the following factors need to be
taken into account:
66 CHAPTER 2. GAPPING
• The necessity of large and small conjuncts
• The necessity of movement and ellipsis.
In order to analyze these sentences, I investigate different options that the literature
provides. I first investigate whether these gapping sentences can be analyzed following
Lasnik (2006), a multiple sluicing analysis where the final remnant moves to the right.
I show that although this is applicable to multiple sluicing in Spanish, the analysis is
inconclusive with the gapping sentences in hand. I hence depart from this kind of analysis
and show that it is appropriate to consider a hybrid small conjunct approach where both
movement and deletion need to be present. Tests from previous literature show that both
processes are needed in the analysis.
2.4.2 Multiple sluicing
One external feature that gapping possesses is that final conjuncts in English and Spanish
gapping maintain a surface configuration similar to that of multiple sluicing. Taking this
as a starting point, it is natural to hypothesize that gapping sentences could be analyzed
as multiple sluicing ones. In order to check whether Lasnik’s (2006) multiple sluicing
analysis is applicable to Spanish gapping sentences, first I check whether Spanish multi-
ple sluicing behaves in a similar way to multiple sluicing in English. If this is the case, it
may provide a possible basis for applying such an analysis to the types of sentences that
are the target of this dissertation.
Lasnik (2006) departs from Richards’s (1997) or Merchant’s (2001) sluicing accounts
in that the former does not assume that multiple sluicing is a case of multiple wh-fronting.
Lasnik shows that in a sentence like (123), the second element that remains in the part
of the sentence affected by deletion is subject to rightwards focus movement:
(123) I know that in each instance one of the girls got something from one of the boys.
But they didn’t tell me which from which.
PF deletion affects IP in the second sentence in (123) after the elements that remain
move from inside the IP. Elements like ‘from which’ in this sentence are shown to display
properties similar to those that rightwards focus movement normally displays. Lasnik
(2006) shows it for English.
My goal here is to show that those properties that phrases like ‘from which’ in (123)
display are similar to those that parallel Spanish phrases display. First, I start by exem-
plifying a case of multiple sluicing in Spanish (124):
2.4. GAPS AND COMPLEX GAPS 67
(124) Una
estudiantestudent
hahas
habladotalked
conwith
una
profesor,professor
perobut
nonot
seI.know
quewhich
estudiantestudent
conwith
quewhich
profesor.professor
‘A student has talked with a professor but I don’t know which student with which
professor.’
The second part of the sentence in (124), no se que estudiante con que profesor ‘I not know
which student with which professor’, is similar to the second part of the English multiple
sluicing sentence in (123). I now make usage of the same tests which Lasnik (2006) uses in
order to show that the behavior of the second remaining element in Spanish is similar to
that of the English one. In what follows, for ease of exposition all throughout the section
I assign a label to each test.
2.4.2.1 Preference for PPs Test
Both in examples of rightwards movements and multiple sluicing in English, there is a
preference for having PPs over DPs. In the case of multiple sluicing, this preference hap-
pens with the second wh-remnant. I show here that multiple sluicing in English behaves
similarly to multiple sluicing in Spanish. This is illustrated by the parallel contrasts in
(125)-(126) and (127)-(128):
(125) ?Someone talked about something, but I cant remember who about what.
(126) ?*Someone saw something, but I cant remember who what.
(127) Alguiensomeone
hablotalked
sobreabout
algosomething
perobut
nonot
recuerdoI.recall
quienwho
sobreabout
quewhat
(128) *Alguiensomeone
viosaw
algo,something
perobut
nonot
recuerdoI.recall
quienwho
quewhat
In both cases (125, 127) where the second wh-phrase is a PP sobre que ‘about what’
the example becomes preferable. This behavior is empirically similar to what happens in
rightwards movement cases.
2.4.2.2 DP Heaviness Test
Another characteristic that has been observed in rightwards focus movement is that DP
heaviness improves the acceptability of the sentence where the DP moves rightwards. In
what follows, I provide multiple sluicing examples in English and Spanish (129, 130, 131,
132), where heavy DPs move to the right.
(129) Which linguist criticized yesterday which paper about sluicing?
68 CHAPTER 2. GAPPING
(130) ?Some linguist criticized (yesterday) some paper about sluicing, but I don’t know
which linguist which paper about sluicing.
(131) ¿Quewhich
linguistalinguist
criticocriticized
ayeryesterday
quewhich
artıculosarticles
sobreabout
gapping?gapping
(132) Algunsome
linguistalinguist
criticocriticized
ayeryesterday
algunossome
artıculosarticles
sobreabout
gappinggapping
perobut
nonot
recuerdoI.recall
quewhich
linguistalinguist
quewhich
artıculosarticles
sobreabout
gappinggapping
These examples (129-132) are better than those in (126) and (128) where DPs are not as
heavy as these ones.
2.4.2.3 Not the Object of a P Test
Rightwards DP movement does not affect the object of a preposition. The same happens
with the kind of movement that affects the second wh-phrase in English multiple sluicing
as Lasnik (2006) shows. And the same happens in Spanish too. In this case, I compare
both languages both for rightwards focus movement and for multiple sluicing. Examples
(133)-(135) illustrate what happens in English with respect to the movement of objects
of prepositions. Examples (136)-(138) illustrate the behavior of Spanish in this respect:
(133) *A linguist spoke about yesterday a paper on sluicing.
(134) A linguist criticized yesterday a paper on sluicing.
(135) Some linguist spoke about some paper on sluicing, but I don’t know which linguist
?*(about) which paper on sluicing.
(136) *Una
linguistalinguist
hablotalked
sobreabout
ayeryesterday
unan
artıculoarticle
sobreabout
gapping.gapping
(137) Una
linguistalinguist
criticocriticized
ayeryesterday
unan
artıculoarticle
sobreabout
gapping.gapping
(138) *Algunsome
linguistalinguist
hablotalked
sobreabout
algunsome
artıculoarticle
deabout
gappinggapping
perobut
nonot
seI.know
quewhich
linguistalinguist
(–)(–)
quewhich
artıculoarticle
deabout
gapping.gapping
The paradigm illustrated in these examples (133)-(138) clearly shows the parallel behav-
ior of English and Spanish. In both languages, movement cannot affect the object of
a preposition. This is typical of rightwards focus movement, shown by ungrammatical
examples (133) and (136), and does also happen in multiple sluicing both in English and
Spanish (135, 138). This suggests that the type of movement that affects the second wh-
2.4. GAPS AND COMPLEX GAPS 69
phrase in multiple sluicing is similar to the rightwards movements that are illustrated here.
2.4.2.4 Strict Locality Test
This last test refers to the strict locality that rightwards movement displays. The same
as in the tests above, movement by the second wh-phrase in multiple sluicing behaves
similarly. Although a Prepositional Phrase can freely extrapose to the end of its clause,
it cannot move to a higher clause. This restriction is known as the Right Roof Constraint
(Ross (1967)). There are also exceptions to this generalization, exemplified here by control
clauses. The strict locality of rightward movement is illustrated in (139) and (140); and
the exemption brought by the control verb ‘want’ is in (141) below. The parallel behavior
between rightwards movements and the movements that affect the second wh-phrase in
multiple sluicing is shown in (142, 143). Parallel Spanish examples are (144)-(147):
(139) Some students spoke yesterday to some professors.
(140) *Some students said that Mary will speak yesterday to some professors.
(141) ?Mary wanted to go until yesterday to the public lecture.
(142) *Some students wanted John to go to some lectures, but I’m not sure which to
which.
(143) ?Some of the students wanted to go to some of the lectures, but I’m not sure which
to which.
(144) Algunossome
estudiantesstudents
hablarontalked
ayeryesterday
conwith
algunossome
profesores.professors
(145) *Algunossome
estudiantesstudents
dijeronsaid
quethat
MarıaMarıa
hablarawill.talk
ayeryesterday
conwith
algunossome
profesores.professors
(146) MarıaMarıa
prometiopromised
irgo
hastauntil
ayeryesterday
ato
lathe
charla.talk
(147) *Una
estudiantestudent
hahas
dichosaid
quethat
JuanJuan
hahas
habladotalked
conwith
una
profesor,professor
perobut
nonot
seI.know
quewhich
estudiantestudent
conwith
quewhich
profesor.professor
(148) Algunossome
estudiantesstudents
prometieronpromised
irgo
ato
algunassome
charlastalks
perobut
nonot
seI.know
quieneswho
ato
quewhich
charlas.talks
70 CHAPTER 2. GAPPING
The constraint on locality in rightwards movement is shown by the grammaticality con-
trasts in (139)-(140) for English and in (144)-(145) for Spanish. The exception to this
constraint on rightwards movements that is introduced by control verbs is illustrated by
sentences (141) and (146) for English and Spanish. The same locality constraint affects
the wh-phrase in multiple sluicing in English and Spanish. This can be inferred from
ungrammatical examples (142) and (147) where extraction of the second wh-phrase to
higher clauses is impossible. The same exemption to the constraint that appears in right-
wards movement is paralleled in multiple sluicing both in English (143) and Spanish (148).
All these four tests show that multiple sluicing behaves similarly both in English and
Spanish. This is important because it means that the same kind of sluicing that Lasnik
(2006) proposes for English can be applied to Spanish, and hence, that this kind of process
exists in the latter language. Now that we know that this process exists in Spanish, the
next question is whether this analysis can account for the gapping examples that are the
target of this work. I focus on it in subsection 2.4.3.
2.4.3 Gapping is not multiple sluicing
I explain in this section that gapping is not multiple sluicing, and I show it by looking at
Lasnik’s (2006) tests more closely in their application to Spanish. I conclude that two of
the tests are not indicative of rightwards movement in Spanish and that the other two are
not conclusive because grammaticality contrasts are not significant. I start with the latter.
PP and DP heaviness Consider examples in (149)-(151) which are all equally accept-
able.
(149) RobinRobin
hablaspeaks
francesFrench
yand
LeslieLeslie
aleman.German
‘Robin speaks French and Leslie speaks German.’
(150) RobinRobin
hablaspeaks
enin
francesFrench
yand
LeslieLeslie
enin
aleman.Germasn
‘Robin speaks in French and Leslie speaks in German.’
(151) RobinRobin
hablaspeaks
sobreabout
animalesanimals
salvajeswild
yand
LeslieLeslie
sobreabout
especiesspecies
enin
peligrorisk
deof
extincion.extinction‘Robin speaks about wild animals and Leslie speaks about endangered species.’
In sentences where the verb and an auxiliary are gapped, preference for PPs or DP
heaviness is not observable. Examples (152)-(154) below illustrate this point:
2.4. GAPS AND COMPLEX GAPS 71
(152) RobinRobin
hahas
habladospoken
francesFrench
yand
LeslieLeslie
aleman.German
‘Robin has spoken French and Leslie has spoken German.’
(153) RobinRobin
hahas
habladospoken
enin
francesFrench
yand
LeslieLeslie
enin
aleman.German
‘Robin has spoken in French and Leslie has spoken in German.’
(154) RobinRobin
hahas
habladospoken
sobreabout
animalesanimals
salvajeswild
yand
LeslieLeslie
sobreabout
especiesspecies
enin
peligrorisk
deof
extincion.extinction
‘Robin has spoken about wild animals and Leslie has spoken about endangered
animals.’
In gapping sentences where a modal verb is also gapped, PPs or heavy DPs are not
preferable either. The sentences that illustrate this are (155)-(157):
(155) RobinRobin
quierewants
hablarspeak
francesFrench
yand
LeslieLeslie
aleman.German
‘Robin wants to speak French and Leslie wants to speak German.’
(156) RobinRobin
quierewants
hablarspeak
enin
francesFrench
yand
LeslieLeslie
enin
aleman.German
‘Robin wants to speak in French and Leslie wants to speak in German.’
(157) RobinRobin
quierewants
hablarspeak
sobreabout
elthe
librobook
deon
logicalogic
yand
LeslieLeslie
sobreabout
elthe
artıculoarticle
deon
matematicas.maths‘Robin wants to speak about the logic book and Leslie wants to speak about the
maths article.’
Gapping sentences within NEG-nor configurations display a too small preference for PPs
or DP heaviness. Hence, I conclude that this small preference is not indicative of Lasnik’s
(2006) multiple sluicing proposal with rightwards movement. Consider (159) and (160)
below in comparison with (158):
(158) ?Nineither
JuanJuan
comprobought
aguawater
ninor
PedroPedro
pan.bread
(159) Nineither
JuanJuan
hablotalked
conwith
MarıaMarıa
ninor
PedroPedro
conwith
Susana.Susana
(160) Nineither
JuanJuan
comprobought
elthe
librobook
sobreabout
gappinggapping
ninor
PedroPedro
elthe
artıculoarticle
sobreabout
elipsis.ellipsis
72 CHAPTER 2. GAPPING
Gapping where the verb and an object are gapped display the same behavior as gapping
within NEG-nor configurations. Consider (161)-(162) and (163)-(164) where the gram-
maticality contrasts are too small to adopt the proposal in Lasnik (2006) for the Spanish
data:
(161) ?JorgeJorge
leCL
diogave
aguawater
ato
MarıaMarıa
yand
PedroPedro
pan.bread
‘Jorge gave water to Maria and Pedro gave bread to Maria.’
(162) JorgeJorge
leCL
diogave
aguawater
ato
MarıaMarıa
yand
PedroPedro
ato
Elena.Elena
‘Jorge gave water to Maria and Pedro gave water to Elena.
(163) ?JuanJuan
regalogave.as.present
ato
MarıaMarıa
librosbooks
yand
PedroPedro
musica.music
(164) JuanJuan
regalogave.as.present
ato
MarıaMarıa
libroslibros
sobreabout
linguısticalinguistics
yand
PedroPedro
musicamusic
clasicaclassic
deby
Wagner.Wagner
The Preference for PPs Test and the DP Heaviness Test show that gapping’s behavior
is not the same as multiple sluicing’s behavior. The two other tests in Lasnik (2006),
which he uses as indicative of rightwards movement, are applicable to Spanish only from
a movement-no movement perspective. The Not the Object of a P Test in Spanish will
always yield the same result: sentences that are not acceptable. The reason for this is
that in Spanish preposition stranding is not possible. The Strict Locality Test is only
indicative of movement in Spanish. To sum up, two tests are inconclusive, one is not
applicable, and the last one, as I show in what follows, is not uniform.
In any case, I apply the two remaining tests onto the different types of gapping to
complete the paradigm. I show that the group of gapping types that I am analyzing still
proves to be heterogenous.
Locality Example (165) shows that PPs cannot be separated in Spanish. The NP cannot
be away from the Preposition.
(165) ??RobinRobin
hablaspeaks
sobreabout
animalesanimals
salvajeswild
yand
LeslieLeslie
especiesspecies
enin
peligrorisk
deof
extincion.extinction‘Robin speaks about wild animals and Leslie speaks about endangered species.’
(166) RobinRobin
dicesays
quethat
elthe
artıculoarticle
estabawas
enin
francesFrench
yand
LeslieLeslie
enin
aleman.German
2.4. GAPS AND COMPLEX GAPS 73
‘Robin says that the article was in French and Leslie says that the article was in
German.’
The exact same results can be observed with gapping sentences where both the verb and
an auxiliary are gapped. Consider (167) and (168) below:
(167) ??RobinRobin
hahas
habladospoken
sobreabout
animalesanimals
salvajeswild
yand
LeslieLeslie
especiesspecies
enin
peligrorisk
deof
extincion.extinction‘Robin has spoken about wild animals and Leslie has spoken about endangered
species.’
(168) RobinRobin
hahas
dichosaid
quethat
elthe
artıculoarticle
estabawas
enin
francesFrench
yand
LeslieLeslie
enin
aleman.German
‘Robin has said that the article was in French and Leslie has said that the article
was in German.’
The behavior of gapping sentences where the modal verb is also gapped is the same.
Consider (170) and (169) below:
(169) ??RobinRobin
quierewants
hablarspeak
sobreabout
logicalogics
yand
LeslieLeslie
matematicas.mathematics
‘Robin wants to speak about logic and Leslie wants to speak about mathematics.’
(170) RobinRobin
quierewants
decirsay
quethat
elthe
artıculoarticle
estabawas
enin
francesFrench
yand
LeslieLeslie
enin
aleman.German
‘Robin wants to say that the article was in French and Leslie wants to say that
the article was in German.’
Gapping in NEG-nor behaves differently. In this case both tests yield unacceptable
results due to the movement of the final remnant. Consider sentence (172) for the Not the
Object of a P Test and sentence (173) for the Strict Locality Test; both of them contrast
grammatically with (171):
(171) Nineither
JuanJuan
hahas
habladotalked
conwith
MarıaMarıa
ninor
PedroPedro
conwith
Susana.Susana
(172) ??Nineither
JuanJuan
hablotalked
sobreabout
unan
artıculoarticle
deabout
gappinggapping
ninor
PedroPedro
una
librobook
deon
elipsis.ellipsis
(173) ?*Nineither
JuanJuan
hahas
dichosaid
quethat
PedroPedro
hahas
habladotalked
conwith
Susana,Susana
ninor
AlfonsoAlfonso
conwith
Marıa.Marıa
74 CHAPTER 2. GAPPING
Gapping where both the verb and an object are gapped behave like gapping in NEG-
nor configurations. Movement tests show that final remnants move. Consider (174) and
(175):
(174) *RobinRobin
estais
leyendoreading
una
librobook
deby
JohnJohn
UpdikeUpdike
yand
LeslieLeslie
AnnAnn
Tyler.Tyler
‘Robin is reading a book written by John Updike and Leslie Ann Tyler.’
(175) *Juan dijo que David le contaba mentiras a Pedro y Marıa a Ana.
Juan said that David CL told lies to Pedro and Marıa to Ana
‘Juan said that David told lies to Pedro and Marıa said that David told lies to
Ana.’
Both (173) and (175) are ungrammatical. Their ungrammaticality is explained because
of the movement of the final remnant. The phenomenon in each example is the same as
in (176) and (177) respectively which are ungrammatical because there is movement of
the phrases con Marıa ‘with Marıa and a Ana ‘to Ana’. Consider (176) and (177) below.
(176) *?Con Marıa, Alfonso no ha dicho que Pedro ha hablado.
with Marıa, Alfonso not has said that Pedro has talked
‘With Marıa, Alfonso has not said that Pedro has talked.’
(177) *?A Ana, Marıa dijo que David le contaba mentiras.
to Ana, Marıa said that David CL told lies
‘To Ana, Marıa said that David told lies.’
In this subsection, I have shown that the types of gapping which are examined here
do not respond to tests like multiple sluicing. Because of this I head onto a different
analysis in the next subsection. It is important to bare in mind for the analyses that
come next that two types of gapping only, Complex Gapping and Gapping in NEG-nor,
present remnants which undergo movement inside their conjunct.
2.4.4 Gapping, move and delete separately
In this subsection, I provide an analysis for the five types of gapping that I mention: Basic
Gapping, Gapping of Auxiliary, Gapping of Modal, Complex Gapping, and Gapping in
NEG-nor. In order to provide their analyses, I take into account arguments that have
been proposed already or that I supply gradually for each case. Taking these as potential
tests for diagnosing gapping instances of one type of structure or the other may help to
distinguish one type of gapping from the other. This is explained below.
The arguments in Johnson’s work that gapping is different from other deletion pro-
cesses underlie different aspects of his analysis. Some arguments suggest that conjuncts
2.4. GAPS AND COMPLEX GAPS 75
need to be smaller than sentences. Some others suggest actual movement of some element
to a part of the structure which is higher than coordination. In section 2.4.4.1, I explain
the way in which arguments or tests have the potential to distinguish among the different
types of gapping.
Afterwards, I consider each type of gapping separately because each one responds
differently to tests. There are two consequences that are important to this fact: (i) all
sentences that are considered gapping form a heterogeneous group and, hence, analyses
for each sentence should be different in light of test results; (ii) the analysis that I propose
needs to be eclectic enough to be capable of reaching all differences. Both small and large
conjuncts are allowed while both movement and deletion play a part on the syntax of
gapping depending on what tests dictate.
2.4.4.1 Understanding gapping properties better
It is important to understand the properties of gapping better so that the analysis of
gapping sentences is properly assigned. I show that gapping sentences are of different
types and hence have different structures depending on the way these types respond to
properties. In order to do so, I specifically focus on those properties of gapping that are
of relevance to the syntactic analysis of sentences of this type. Before I discuss how to get
to the differences, I explain what it is that unifies all these types under the label ‘gapping’
as opposed to other deletion phenomena.
Johnson (2004) discusses the uniqueness of coordination as relevant to gapping. I
extend this discussion here in a series of statements and generalizations that lead the
reader to the point that I take as the basis for my analysis of gapping.
1. Johnson (2004) states that the property that makes coordination unique is ‘sharing’,
which he describes as satisfying “parallel grammatical requirements in each of the
conjuncts” (page 35).
2. I show in subsequent subsections that all types of gapping that are analyzed in this
work either have a shared part in their structure or a correlative connector.
3. If statement (2) is carefully thought of, the following two aspects are also true:
• If any type of gapping has a shared part in their structure, no doubt, that
shared part satisfies parallel requirements in each conjunct. (This is part of
the definition.)
76 CHAPTER 2. GAPPING
• If any type of gapping is embedded in a correlative conjunction, there is a part
in this structure that satisfies parallel requirements in each conjunct in the
sense that each part of the correlation serves as a connector for each conjunct.
4. Overall, the conclusion is that sharing is a property that occurs in all gapping cases,
as made clear in the statements in (3). Since sharing is a unique property of coor-
dination, gapping –through sharing– necessarily gets to be unique to coordination
(independently of the process that accounts for missing elements). Hence, it is shar-
ing, the satisfaction of parallel grammatical requirements, that unifies all gapping
types and that differentiates them from the rest of ellipsis phenomena.
This conclusion justifies that the analysis of all types of gapping is different from the
rest of ellipsis phenomena. My proposal makes syntactically explicit the difference be-
tween gapping and other types of ellipsis. Although explained below in detail, gapping is
analyzed either via ATB movement or a type of PF deletion which deviates from other
PF deletion processes in an aspect that is explained below. See section 2.4.4.5.
With the justification in hand that all types of gapping can be dubbed under such a
label, I proceed to explain what makes the difference among gapping types. Taking gap-
ping properties one by one, it is noticeable that not all of them prove the same about the
structure of gapping. Step by step, I move along relevant properties and their meanings
towards the structure of gapping.
In terms of structure, one first step in distinguishing between types of gapping is know-
ing the size of conjuncts that is needed in each type. Types of gapping can be tested with
respect to the size of conjuncts by placing gapping in contexts where wide scope elements
are present. In these cases, if the scope of these elements can be wide, conjuncts need
to be small as has been explained before (see sections 2.2.4.1, 2.3.3, 2.3.4.1). If not, con-
juncts need to be large. So, for instance, from previous sections, we know that ‘Gapping
in NEG-nor ’ needs coordination of CPs because Wurmbrand’s (2008) analysis suggests
so. A test with wide scope elements suggests the structure of this type of sentences to be
a coordination of CPs.
As a second step, after the size of conjuncts which is needed is known, the process that
accounts for missing elements needs to be identified. A test that makes a difference here
comes from coordinate examples in Johnson (2004) where a coordination is embedded in
a higher coordination: [ . . . & [ . . . & . . . ]]. In cases of this kind, Johnson places the gap
in the very final embedded conjunct and it corresponds to an overt counterpart that is in
2.4. GAPS AND COMPLEX GAPS 77
the first conjunct. This leaves a conjunct in the middle where there is neither a gap nor
its corresponding pronounced counterpart. Ungrammatical instances prove that elements
in the gap undergo movement, the reason being that the CSC gets to be violated, which
is what produces the ungrammaticality. If the contrary happens, movement is not present
and deletion needs to be the process that accounts for missing elements.
Another step which is based on the ellipsis literature rather than on the gapping lit-
erature is related to sloppy readings of pronouns (Sag (1976)). These sloppy readings of
pronouns can only appear in deletion contexts, hence, if they are present in the gapping
sentences that I provide here, this could only mean that the sentence in question is a
gapping by deletion example. Consider section 2.4.4.5 for examples.
At this point, it is important to remember that other properties that are listed in the
characterization of Spanish and English are based on the comparison, in English, between
gapping and ellipsis processes, VP-ellipsis and pseudogapping, that do not exist in Span-
ish. This has also been stated before. Hence, some of those properties cannot be used as
tests for the purposes in this subsection because their use would need impossible compar-
isons between gapping and nonexistent ellipsis processes in Spanish. Overall, considering
previous gapping literature, the tests that I have mentioned in the first and second steps
are those that make a difference among gapping types in Spanish. However, they cannot
all be applied in every case. For instance, those gapping tests that elide pronouns can
only be applicable to types of gapping that also omit objects. Also, the difficulty brought
by judgements is also relevant in that I try to base myself on judgements that are clear
cut. These kinds of cases do not arise at all times for all gapping types. Hence, these
tests need to be understood as those that have ‘potential’ to differentiate gapping types
apart.
Summarizing, relevant gapping properties or tests that can distinguish gapping from
any other types of ellipsis and which can also distinguish gapping types from each other
are:
1. The property of sharing distinguishes gapping from other ellipsis phenomena. As
such, it does also unify all gapping types together under the same label.
Gapping: + sharing
Elsewhere ellipsis: - sharing
2. The property of wide scope elements distinguishes gapping types with small con-
juncts from gapping types with large conjuncts.
78 CHAPTER 2. GAPPING
Gapping and wide scope elements: small conjuncts
Gapping and narrow scope elements: large conjuncts
3. The property of embedded coordinate structures distinguishes gapping with move-
ment from gapping without movement.
Impossible gapping in embedded coordinate structures: movement
Possible gapping in embedded coordinate structures: deletion
4. The property of sloppy readings distinguishes gapping via deletion from gapping via
a different process.
Sloppy readings possible in gapping: deletion
Sloppy readings impossible in gapping: movement
Application of these follow for the different types of gapping: ‘Basic Gapping’, ‘Gapping
of Auxiliary’, ‘Gapping of Modal’, ‘Complex Gapping’ and ‘Gapping in NEG-nor.
2.4.4.2 Basic gapping
According to the small conjunct literature (Siegel (1984, 1987), McCawley (1993), Lin
(2002), Johnson (2004, 2009)) and to the cross-linguistic support that Spanish provides,
sentences like (85), repeated below as (178) for convenience, the existence of cross conjunct
binding configurations suggests the need for small conjuncts. The cross conjunct binding
phenomenon which proves the need for small conjuncts is illustrated by examples (97) and
(98) which are reproduced here as (179) and (180) respectively. The presence of gapping
in (179) makes binding possible.
(178) JuanJuan
comeeats
arrozrice
yand
PedroPedro
[ ][ ]
alubias.beans
(179) Ninguni
noestudiantestudent
leyoread
‘Brooklyn‘Brooklyn
Follies’Follies’
yand
sui
hismadremother
[ ][ ]
‘El‘The
Guardian’.Catcher’
‘No student read ‘Brooklyn Follies’ and his mother ‘The Catcher’.’
(180) *Ninguni
noestudiantestudent
leyoread
‘Brooklyn‘Brooklyn
Follies’Follies’
yand
sui
hismadremother
leyoread
‘El‘The
Guardian’.Catcher’‘No student read ‘Brooklyn Follies’ and his mother read ‘The Catcher’.’
In addition to small conjuncts, movement is also needed in this kind of construction
as example (88) from above illustrates, where gapping is tested under the coordinate
embeddedness test. Example (88) is repeated here as (181).
2.4. GAPS AND COMPLEX GAPS 79
(181) *Juanjuan
siemprealways
bebedrinks
martinismartinis
oor
esis
unaa
reunionmeeting
importanteimportant
yand
nadienoone
[ ][ ]
nadaany
deof
alcohol.alcohol
The analysis of sentence (178) that I propose is easy to understand if some features
in the small conjunct literature are taken into account (Johnson (2004)). Because small
conjuncts are needed, I propose to have coordination of vPs. I propose for the Verb in each
conjunct to ATB move to T to check features in T because Verb movement in Spanish
is generally assumed and because of example (181). I also propose to have movement of
the subject in the first conjunct to Spec, TP to satisfy the EPP. (See Johnson (2004)
for a discussion on the latter) Among other considerations, Johnson assumes that the
CSC applies to A’-movements and not to A-movements.6 This way, the structures that I
propose do not violate the CSC. The specific structure for ’Basic Gapping’ is schematized
in (182).
(182) TP
Juani T’
come CoP
vP Co’
Co
y
vP
Pedro v’
tV VP
tV alubias
ti v’
tV VP
tV arroz
2.4.4.3 Gapping of auxiliary
I have labeled ‘Gapping of Auxiliary’ the group of gapping sentences where only the
Auxiliary is gapped. This type of gapping is illustrated in (79) which is repeated here as
(183):
6For discussion on this topic, the reader is referred to Johnson (2004). Also, for discussion on Caseassignment of both subjects in the first and second conjuncts.
80 CHAPTER 2. GAPPING
(183) JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
[ ][ ]
fregadowashed
elthe
suelofloor
‘Jorge has cleaned the bathroom and Pedro has washed the floor.’
If we consider wide scope elements within sentences of this kind, the same as before
happens. Consider (96), repeated here as (184), where negation scopes widely above
coordination:
(184) JuanJuan
nonot
habıahad
comidoeaten
alubiasbeans
yand
MarıaMaria
[ ][ ]
bebidodrunk
champan.champagne
‘Juan hadn’t eaten beans and Maria drunk champagne.’
The analysis of sentence (183) is similar to the analysis which I have proposed so far:
vP coordination, movement of the first conjunct subject to Spec, TP. The difference comes
with the presence of the Auxiliary in T, which makes no further movement of the Verbs
away from vP necessary. Verbs simply stay in vP. Consider (185) which is the structure
of sentence (183).
(185) TP
Jorgei T’
ha CoP
vP Co’
Co
y
vP
Pedro v’
fregado VP
tV el suelo
ti v’
limpiado VP
tV el bano
2.4.4.4 Gapping of modal
I focus here on gapping sentences where only the Modal Verb is gapped. An example of
this is (100), which I repeat here as (186):
(186) JuanJuan
deberıashould
arreglarfix
elthe
cochecar
yand
yoI
[ ][ ]
prepararprepare
lasthe
maletas.luggage
‘Juan should fix the car and I should pack.’
2.4. GAPS AND COMPLEX GAPS 81
Examples like (186) in which a Modal Verb is missing from the final conjunct bring with
them evidence for small conjuncts. That is, Modal Verbs per se constitute one of the
types of wide scope elements which are employed in the literature in order to prove the
existence of small conjuncts. Because of this, the mere absence of the Modal in (186)
produces a different reading from the reading in a sentence like (99), repeated here as
(187), where the Modal Verb is overt in both conjuncts7:
(187) JuanJuan
deberıashould
arreglarfix
elthe
cochecar
yand
yoI
deberıashould
prepararprepare
lasthe
maletas.luggage
‘Juan should fix the car and I should pack.’
The analysis of sentence (186) does not have many differences with respect to previous
analyses. The differences, in fact, are related to the use of the Modal Verb rather than to
the coordination or gapping facets of the sentence. The steps to follow are (i) to determine
what the structure is in Spanish when Modal Verbs are present and (ii) to provide the
relevant gapping analysis.
First, complements of modal verbs are TPs because temporal features in the matrix
clause can be different to those temporal features in the complement. Different time
adverbials can modify each part of the sentence, the matrix part on the one hand and the
complement on the other hand. Consider examples (188)-(191) with all types of modals
in which adverbial modification takes place in both matrix and complement parts:
(188) Ayeryesterday
tenıahad.I
queto
presentarpresent
elthe
trabajopaper
lahe
semanaweek
quethat
vienecomes
yand
ahoranow
esis
parafor
dentrowithin
deof
dostwo
semanas.weeks
‘Yesterday, I had to present the paper for next week, and now, the deadline is in
two weeks.’
(189) Ayer,yesterday
JuanJuan
querıawanted
irgo
alto.the
cinemovies
mananatomorrow
mismoright
yand
hoytoday
dicesays
quethat
lathe
pelıculamovie
esis
mala.bad
‘Yesterday, Juan wanted to go to the movies right tomorrow and now, he is saying
that the movie is bad’.
(190) Hehave.I
trabajadoworked
muchoa.lot
conwith
buenagood
ayudahelp
yand
ahoranow
puedocan.I
terminarfinish
lathe
tesisdissertation
enin
tresthree
meses.months
‘I have worked a lot with good help and now I can finish the dissertation in three
months.’
7The semantic difference has been explained in section 2.3.3
82 CHAPTER 2. GAPPING
(191) Haceago
unaone
semanaweek
elhe
debıamust
deof
tenerhave
unaa
reunionmeeting
conwith
lathe
mafiamafia
elthe
1first
deof
agostoAugust
perobut
lasthe
nuevasnew
pruebasproofs
indicanshow
quethat
lathe
reunionmeeting
serıawould.be
unone
mesmonth
despues.after‘One week ago, he would’ve had a meeting with the mafia the first of August but
new proofs show that the meeting would be one month later’.
In all these examples, (188)-(191), adverbials contrast from the matrix parts of sentences
to the complements of the modal verbs.
Second, the use of evidential adverbials which express the speaker’s attitude towards
the sentence that is uttered shows that the complement of the Modal Verb needs to be
smaller than a CP. Adverbs of this type can only be adjoined to whole CPs. Consider
sentence (192) below that illustrates its use in a sentence with a Modal Verb:
(192) Poraccording
loCL
quethat
pareceit.looks
ser,to.be,
JuanJuan
debeshould
leerread
esethat
libro.book
‘It is apparent that it is necessary that Juan reads this book.’
If we try to make this adverbial modify the complement of the Modal Verb, either the
result is not that acceptable or the meaning is such that what gets to be modified is
the whole verbal complex with the Modal Verb included. Consider (193) below which is
parallel to (192) but which presents the adverbial structurally lower than the Modal Verb.
I construct the example in this way in order to try to force the explained meaning.
(193) ?JuanJuan
debe,should,
poraccording
loCL
quethat
pareceit.looks
ser,to.be,
leerread
esethat
libro.book
‘It is necessary that it is apparent that Juan reads this book.’
The only possible meaning of sentence (193) is parallel to the one in sentence (192). The
fact that evidential adverbials cannot modify the lower verb in these kinds of situations
syntactically means that the complement of the Modal Verb is not a CP because adver-
bials of this type can only be adjoined to such phrases.
The conclusion is that Modal Verbs’ complements need to be TPs. It can be deduced
from examples (188)-(191) that Tense needs to be present in the complement of Modal
Verbs. In addition, it can be concluded from example (193) that complements are not
CPs. The structure I propose for a regular declarative sentence containing a Modal Verb
such as (194) is represented in (195) in which, as illustrated, the complement of the Modal
Verb is a TP and the Modal Verb behaves like a raising Verb:
2.4. GAPS AND COMPLEX GAPS 83
(194) JuanJuan
debeshould
leerread
esethat
libro.book
‘Juan should read that book.’
(195) TP
Juani T’
deberıa CoP
vP Co’
Co
y
vP
yoj v’
tVMod VP
tVMod TP
tj T’
preparar vP
tj v’
tV VP
tV las
maletas
ti v’
tVMod VP
tVMod TP
ti T’
arreglar vP
ti v’
tV VP
tV el
coche
In (195), conjuncts are small because they are vPs. The Modal Verb is in the shared
part of the structure because it has moved ATB from the conjuncts as dashed arrows
indicate. Finally, as in other sharing structures, the subject in the first conjunct moves
to the shared TP to satisfy the EPP. This is done after it has raised from below within
its conjunct. The subject raising movement is parallel in both conjuncts as traces ‘ti’ and
‘ tj ’ indicate.
84 CHAPTER 2. GAPPING
2.4.4.5 Complex gapping
Johnson (2004, 2009) groups the prior gapping types together with ‘Complex Gapping’
and proposes to have the same analysis for all these in English. In this dissertation, I
extend the ‘Complex Gapping’ label to more types of gapping sentences than Johnson and
I make the following distinction: ‘Complex Gapping for Objects’ and ‘Complex Gapping
for Verbal Complexes’. I show that both types of ‘Complex Gapping’ do not behave in
the same way as other gapping types with respect to tests.
2.4.4.5.1 ‘Complex Gapping for Objects’: To begin with, sentence (84) which I
repeat below as (196) illustrates what the label ‘Complex Gapping for Objects’ refers to
coordinate cases where both the Verb and an object are gapped:
(196) JorgeJorge
leCL
diogave
unaa
florflower
ato
MarıaMarıa
yand
Pedro.Pedro
[ ][ ]
ato
ElenaElena
‘Jorge gave a flower to Marıa and Pedro gave a flower to Elena.’
In this sentence, the Verb le dio ‘gave (her)’ and the object una flor ‘a flower’ are gapped
in the final conjunct.
The fact that the object is also gapped is a good reason to think about the analysis
in more detail. In order to provide an analysis for this type of sentence, I focus on two
aspects: (i) the size of conjuncts that is needed: large or small; and (ii) the process that
accounts for missing elements: movement or deletion. I start with (i) the size of conjuncts
in ‘Complex Gapping for Objects’. I refer to tests that are related to wide scope elements.
Consider cross conjunct binding in (197)-(198).
(197) Cadaeach
chicoi
boyleCL
diogave
unaa
florflower
ato
MarıaMarıa
yand
sui
hispadredad
[ ][ ]
ato
Elena.Elena
‘Each boyi gave a flower to Marıa and hisi dad gave a flower to Elena.’
(198) *Cadaeach
chicoi
boyleCL
diogave
unaa
florflower
ato
MarıaMarıa
yand
sui
hispadredad
leCL
diogave
unaa
florflower
ato
Elena.Elena‘Each boyi gave a flower to Marıa and hisi dad gave a flower to Elena.’
It can be observed in (197)-(198) that the cross conjunct binding phenomenon applies
similarly in ‘Complex Gapping for Objects’. That is, when the Verb and the object are
gapped, binding from the top left onto the final conjunct is possible as in (197). When
the Verb and the object are overt in both conjuncts, binding is not possible as seen in
(198). This observation makes the first conclusion possible: ‘Complex Gapping for Ob-
jects’ needs to also be analyzed through small conjuncts. I move onto the second step now.
2.4. GAPS AND COMPLEX GAPS 85
As for (ii), choosing between movement or deletion to account for missing elements,
notice that ‘Complex Gapping’ sentences behave like deletion sentences in contexts where
strict and sloppy readings of pronouns arise. Consider the ‘Complex Gapping’ sentences
in (199) and (200) in which a referential pronoun that is used in the first conjunct is
omitted in the second one:
(199) Mimy
amigofriend
celebracelebrates
suhis
cumpleanosbirthday
enin
agostoAugust
yand
yoI
[ ][ ]
enin
febrero.February
‘My friend celebrates his birthday in August and I celebrate my/his birthday in
February.’
(200) Mimy
jefeboss
despidiodismissed
ato
suhis
empleadoemployee
aat
lasthe
tresthree
yand
tuyou
[ ][ ]
aat
lasthe
cuatro.four
‘My boss dismissed his employee at three and you dismissed your/his employee at
four.’
Like in deletion contexts, the omitted mi/su cumpleanos ‘my/his birthday’ in (199) can
refer to ‘My friend’, ‘My friend’s birthday’, in the strict reading and to ‘me’, ‘my birth-
day’, in the sloppy reading of the referential pronoun. Equally, in example (200), the
pronoun can refer both to the subject in the first conjunct and the subject in the second
conjunct. In the strict reading the omitted tu/su empleado ‘your/his employee’ refers to
‘my boss’s employee’ and in the sloppy reading it refers to ‘your employee’. The presence
of sloppy readings in these sentences (199)-(200) is indicative of deletion as the process
that accounts for missing elements.
So far, I have shown that gapping sentences do not behave all in the same way. The
three prior cases of gapping –‘Basic Gapping, Gapping of Auxiliary, or Gapping of Modal’–
are accounted for both with small conjuncts and movement of the Verb. The latter case,
‘Complex Gapping for Objects’, however, although its analysis requires small conjuncts,
it also requires deletion. At this moment of the small conjunct literature development
this is not surprising since small conjunct theory is not exclusively tight to movement any
more. The main goal now is to know how deletion works in these examples.
I propose a deletion by phase approach which is inspired in the work by Gallego (2009)
and which assumes the basic idea there but that slightly changes some detail. Gallego
(2009) is based on Chomsky (2000) and considers that ellipsis cases undergo a deletion
process that targets the complement domain of phase heads. It is desirable to have an
analysis with such a device because this analysis would leave the edge of phases as a
domain for pronounceable material that is available for movements of remnants that get
pronounced. Recall that in applying Lasnik’s (2006) tests to this type of gapping it be-
86 CHAPTER 2. GAPPING
comes clear that final remnants undergo movement to the left. I explain my proposal for
this deletion process in four brief steps: (i) I make more explicit what this deleted domain
is; (ii) I assume a way to license deletion; (iii) I explain in detail the movements of the
remnants; (iv) I provide the analysis of the whole sentence.
(i) Deleted Domain:
Gallego (2009) holds that deletion targets the complement domain of the phase head.
Considering (201) below as a basis, in his view, deletion affects the underlined part in
(202):
(201) a. [PhaseP XP [PhaseP Spec [Phase′ Phase-Head Complement] ] ]
b. Edge of Phase is {XP, Spec}
Phase is {PhaseP}
Complement Domain of Phase Head is {Complement}
(202) [PhaseP XP [PhaseP Spec [Phase′ Phase-Head Complement] ] ]
My proposal slightly changes here. I would like to propose that, in ‘Complex Gapping for
Objects’, deletion affects the whole phase except for the edge, that is, both the Comple-
ment Domain –like Gallego (2009)– and the Phase Head. This only leaves the elements
that are located at the edge to be pronounced. I represent this idea in (203), where
underlined elements are affected by deletion:
(203) [PhaseP XP [PhaseP Spec [Phase′ Phase-Head Complement] ] ]
(ii) Licensing Deletion:
Merchant (2001) makes usage of the E feature to license deletion basing himself on the
general intuition in Lobeck (1999) but deviating from it. I assume the idea in Merchant
(2001) but I also deviate from it in some details. Merchant proposes an E feature on
I that gets checked with [+Wh, +Q] on the C head via I to C movement. Deletion of
the IP at PF is triggered because “E issues an instruction to the PF system to skip its
complement for purposes of parsing and production” (Merchant (2001, 60)).
I would also like to propose that the licensing of deletion is done through the E fea-
ture, assuming Merchant (2001), in a feature-feature relation. Merchant instantiates the
feature-feature relation in a head-head relation. Instead, I propose to instantiate the
feature-feature relation in a Spec-head relation. This of course means that the location
of features need to be different too. Maintaining also the link between E and [+Wh]
features, I propose that Foc is the location of the E feature which is checked with the
2.4. GAPS AND COMPLEX GAPS 87
[+Wh] in remnants that move to Spec, FocP. I provide details about the FocP projection
I am assuming in step (iii).
(iii) Movements of Remnants and Checking of E:
We know from the reasoning presented so far in section 2.4 that the final remnant in ‘Com-
plex Gapping for Objects’ moves to the left. I analyze this movement following Butler
(2004). One of the main points in this work is that right above each phase there is place
which is dedicated to information structure. This means that this part of the structure is
available for elements that move from lower positions to check features at projections like
FocusP or TopicP. Assuming such a structure provides space for the E feature that I lo-
cate in Foc in step (ii) and that gets to be checked by the phrase that moves to Spec, FocP.
Considering also step (i), a clarification needs to be made. It is said there that deletion
affects the whole phase except for the edge in this proposal. My understanding of the
edge of phases8 is that it is the locus of elements that may keep on moving further up and
that may be there at a specific point of the derivation because they have moved here from
below. Bearing this in mind, I take Spec, PhaseP and the projections that compose the
information structure above each phase to be the edge of PhaseP. Basically, the reason
for this is that these projections are the locations to which and from which elements may
keep on moving in subsequent steps of the derivation. Notice that this is not a major
change in the conceptualization of edges since what is normally considered to be the edge
of phases is Spec, PhaseP and some further phrase that may be adjoined to the Phase.
(iv) Analysis including steps (i), (ii), and (iii):
The analysis of ‘Complex Gapping for Objects’ that I propose is (204) which corresponds
to (196), the first ‘Complex Gapping’ sentence in the present subsection:
8I leave technicalities apart at this point for ease of exposition.
88 CHAPTER 2. GAPPING
(204) TP
Jorgei T’
le dio CoP
TopP Co’
Co
y
TopP
Pedroj Top’
Top FocP
a Elenak Foc’
Foc vP
tj v’
le dio VP
tV V’
una
flor
V’
tV tk
Top’
Top FocP
Foc’
Foc vP
ti v’
tV VP
tV V’
una
flor
V’
tV a
Marıa
In (204)9, the structure is a coordinate structure with small conjuncts. In this case,
instead of having vP coordination the information structure projections are above vP, the
phase boundary in each conjunct. The remnant in the last conjunct a Elena ‘to Elena’
moves to Spec, FocP hereby checking the E feature in Foc and making the deletion process
possible. The subject in the last conjunct still has space for its landing site in Spec, TopP.
In the first conjunct, elements behave as usual in small conjunct structures. The subject
moves to Spec, TP to satisfy EPP. Verb movement from the vP area to IP only takes
9I represent ‘PF-deletion’ and ‘movement’ in this dissertation as is represented in this structure.Movement is represented with traces and arrows and PF-deletion is represented with grey letters. I onlyrepresent movement and deletion in structures, not in sentences or examples, where gaps are representedby [ ]. The processes that affect each example are sometimes determined by means of additional dataand longer explanations. This is why, the specific representation of such processes in each example couldbe confusing and I avoid it.
2.4. GAPS AND COMPLEX GAPS 89
place in the first part of coordination. I claim that the movement of the second conjunct
gets affected by the PF deletion process and it is suppressed.10
The deletion process makes these types of structures have a considerable resemblance
with what is normally proposed in pseudogapping phenomena. Because of this, I base
the explanation about the behavior of the Verb on the work by Boeckx and Stjepanovic
(2001) in which they account for Verbs’ behavior in pseudogapping via a PF treatment of
this type of movement. They basically state that if Verb movement is a PF phenomenon,
at that same PF level when deletion applies, both PF operations compete, resulting in the
suppression of Verb movement. The treatment of head movement as a PF phenomenon
is already discussed in Chomsky(1995, 2000) and targets the different behavior between
phrasal movement and head movement. Support for this PF treatment of head move-
ment comes both from theoretical and psychological works: Brody (2000), Fredin (1999),
Grodzinsky and Finkel (1998). And Boeckx and Stjepanovic (2001) adjoin to this line of
thought. Their proposal can be summarized in three points. (i) Head movement happens
after spell out. Because it happens in the phonological component, it follows all syntactic
movements and it can be ‘superseded’ by deletion. (ii) The trigger of this PF verbal
movement is not featural. The trigger is either morphological, prosodical, or a mixture
of both. (iii) Ellipsis considered to be a PF operation and Verb movement too, both
operations compete and the result is that the Verb either moves or is deleted.
2.4.4.5.2 ‘Complex Gapping for Verbal Complexes’: Under this label, I place
gapping sentences where both the Verb and either an Auxiliary or a Modal are missing.
I start with the former. Gapping sentences where both the Verb and the Auxiliary are
missing are illustrated in (80), repeated here as (205) below:
(205) JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
[ ][ ]
elthe
suelo.floor
‘Jorge has cleaned the bathroom and Pedro has cleaned the floor.’
The same process needs to be followed. I first determine whether it is small conjuncts or
large conjuncts that are needed and afterwards whether it is deletion or movement that
accounts for missing elements.
Cross conjunct binding proves the need for small conjuncts in Gapping with Auxiliary.
This is illustrated in (206). It is only when gapping is present that binding is possible.
10In Johnson (2009), the analysis is different with respect to Verb movement sentences of this type.See footnote 11 in this chapter; and the explanation upon example (357) in section 3.5.4 in the nextchapter for discussion on this aspect. I include the major discussion of this topic in the following chapteron determiner sharing because I need to make reference to a determiner sharing example in order to fullyillustrate my point.
90 CHAPTER 2. GAPPING
Example (207) illustrates that when all elements are present cross-conjunct binding is
impossible.
(206) Ningun chicoi ha limpiado el bano y sui padre [ ] el suelo.
no boy has cleaned the bathroom and his dad [ ] the floor
‘No boyi has cleaned the bathroom and hisi dad has cleaned the floor.’
(207) *Ningun chicoi ha limpiado el bano y sui padre ha limpiado el suelo.
no boy has cleaned the bathroom and his dad has washed the floor
‘No boyi has cleaned the bathroom and hisi dad has washed the floor.’
In relation to the question whether it is deletion or movement that accounts for missing
elements, it is not new that sentences of this kind bring some complexity to small con-
junct analyses. I focus on explaining the phenomenon in Spanish so that cross-linguistic
differences do not distort the discussion here. Details on English data and on a different
analysis to the one provided here can be found in Lin (2002). In order to provide an
answer to whether it is deletion or movement that accounts for missing elements in this
kind of gapping, I concentrate on two aspects: what we know so far about its structure
(i) in relation to conjunction and (ii) in relation to the sentence per se. I explain the logic
according to these aspects.
We know that we have parallel examples like (183), that I repeat here as (208), which
also contains a participle in the second conjunct:
(208) JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
[ ][ ]
fregadowashed
elthe
suelo.floor
‘Jorge has cleaned the bathroom and Pedro has washed the floor.’
As discussed before, because wide scope elements can be present in sentences where the
participle is also present in the second conjunct, coordinations need to conjoin vPs. In
these cases, we know that the subject in the second conjunct –‘Pedro’ in (208)– is in the
Spec, vP. And also, because the participle is behind the subject, this participle needs to be
inside vP. If we take this conclusion seriously, this means that participles need to always
be inside the vP; and because of it, participles need to always be inside the conjunction.
Hence, because these participles cannot move from the conjoined part, the only way that
they do not get to be pronounced in gapping cases is via PF deletion.11
11I make reference to this lack of movement in participles in the next chapter, section 3.5.4, whendiscussing example (357), in order to explain the reason why I do not assume Johnson (2009). I explainthat ’Complex Gapping for Objects’ may be analyzed following the strategy in Johnson (2009) –ATBmovement plus deletion simultaneously– and that I leave this debate open. However, I explain that ifa unified analysis to all deletion sentences covered in this dissertation is desired, the necessary lack ofmovement of the Verbs in some of them –participles in this case– is to be taken into account. Johnson’sanalysis does not comprehend a solution for the Spanish sentences in which participles do not move. There,
2.4. GAPS AND COMPLEX GAPS 91
The PF deletion account for sentence (205) is parallel to the prior PF deletion analyses
I have proposed. Remnants move to the left, to the ‘edge’ of the conjunct, and the
subsequent PF deletion process accounts for missing elements. Consider (209).
(209) TP
Jorgex T’
ha CoP
TopP Co’
Co
y
TopP
Pedroi Top’
Top FocP
el sueloj Foc’
Foc vP
ti v’
limpiado vP
tV tj
Top’
Top FocP
Foc’
Foc vP
tx v’
limpiado VP
tV el bano
In (209), in the first conjunct, the subject moves up to Spec, TP satisfying the EPP.
I follow Johnson in proposing so. Because the Auxiliary is already present in T head, the
Verbs do not move from their conjuncts. In the first conjunct, the Verb moves to v head
and stays there. In the second conjunct, Verb movement, as a PF process, is suppressed
in the competition with PF deletion (Boeckx and Stjepanovic (2001)). The Verb is PF
deleted this way. The PF deletion process affects the phase as stated above. This is the
location where remnants move: specifically, to Focus and Topic projections above vP. PF
I also make reference to a different type of determiner sharing sentence which cannot be accounted forvia the strategy in Johnson (2009). See also footnote 10 in this chapter.
92 CHAPTER 2. GAPPING
deletion, as proposed here, accounts for this type of sentence.
One of the advantages of this approach is that the close analysis of gapping properties
makes available their use as tests which prove that both PF deletion and movement are
needed processes in gapping. The existence of these two processes makes natural the
account for ‘Complex Gapping for Verbal Complexes’. Lin (2002), for instance, makes
use of Lasnik’s Salvation by Deletion tool which is something exceptional in her analysis
since everything else is accounted via movement. I have shown that deletion need not be
exceptional in gapping since it is needed in other types of gapping as well, like ‘Complex
Gapping for Objects’ or ‘Gapping in NEG-nor ’ in subsection 2.4.4.6.
The other possibility within ‘Complex Gapping for Verbal Complexes’ is to have gap-
ping of both the lexical Verb and a Modal Verb. This is illustrated in (210):
(210) JuanJuan
deberıashould
arreglarfix
elthe
cochecar
yand
yoI
[ ][ ]
lathe
bicicleta.bicycle
‘Juan should fix the car and I should fix the bicycle.’
Notice furthermore that cross conjunct binding also takes place in sentences of this type
(211)-(214):
(211) Cadaeach
chicoi
boyhahas
arregladofixed
elthe
portatillaptop
yand
sui
hismadremom
[ ][ ]
lathe
baterıa.battery
‘Each boyi has fixed the laptop and hisi mom has fixed the battery.’
(212) *Cadaeach
chicoi
boyhahas
arregladofixed
elthe
portatillaptop
yand
sui
hismadremom
hahas
arregladofixed
lathe
baterıa.battery
‘Each boyi has fixed the laptop and hisi mom has fixed the battery.’
(213) Cadaeach
amigoi
friendhahas
compradobought
una
comiccomic
yand
sui
hishermanobrother
[ ][ ]
lathe
coleccioncollection
entera.entire
‘Each friendi has bought a comic and hisi brother has bought the entire collection.’
(214) *Cadaeach
amigoi
friendhahas
compradobought
una
comiccomic
yand
sui
hishermanobrother
hahas
compradobought
lathe
coleccioncollection
entera.entire
‘Each friendi has bought a comic and hisi brother has bought the entire collection.’
Hence, according to the contrasts between (211) and (212) on the one hand, and (213)
and (214) on the other hand, small conjuncts are also needed in this type of gapping.
Basing myself on the same reasoning that the Verb cannot escape the conjuncts via
movement, I propose the same PF deletion mechanism as before. Consider the structure
in (215) below:
94 CHAPTER 2. GAPPING
(215) TP
Juani T’
deberıa CoP
TopP Co’
Co
y
TopP
yoj Top’
Top FocP
la
bicicletak
Foc’
Foc vP
tj v’
tVMod VP
tVMod TP
tj T’
arreglar TopP
tj Top’
Top FocP
tk Foc’
Foc vP
tj v’
tV VP
tV tk
Top’
Top FocP
Foc’
Foc vP
ti v’
tVMod VP
tVMod TP
ti T’
arreglar TopP
Top’
Top FocP
Foc’
Foc vP
ti v’
tV VP
tV el coche
2.4. GAPS AND COMPLEX GAPS 95
In (215), the modal Verbs move in their conjuncts from their source position to the left so
that they keep on moving ATB to T in the shared part. Consider the ‘tModalV ’ instances.
In the second conjunct, remnants move to the edge –TopP and FocP– so that, when PF
deletion affects the whole phase except for the edge, they are not deleted. The lexical
Verb in the first conjunct stays in that conjunct moving up to T its TP. In the second
conjunct, the movement of the lexical Verb gets affected by the PF process and is deleted
(Boeckx and Stjepanovic (2001)).
In structures like (215), there are two possible phases that could be considered to be
the ones affected by deletion: the vP phase corresponding to the modal Verb and the vP
phase corresponding the lexical Verb. The coordinate structure which is affected by sub-
sequent deletion could then be considered to be at those two levels. However, structural
consequences are different for each one and it is important to make clear the reason why
the appropriate structure is (215), where PF deletion applies.
In comparing (215) with the other possibility in (216) in which there is conjunction
of the lower vPs, it can be seen that in (215) deletion applies but that in (216) the Verb
could escape the conjuncts in an ATB fashion. I represent the aspect in question very
schematically in (216) for the sake of clarity.
96 CHAPTER 2. GAPPING
(216) TP
T’
Modal V vP
v’
tModalV VP
V’
tModalV TP
T’
Lexical V CoP
vP Co’
Co vP
v’
tLexicalV VP
tLexicalV
v’
tLexicalV VP
tLexicalV
Though (216) could be a possible derivation according to the tools we have so far, I
show below that it is not possible. I base myself on evidence that is related to negation.
Consider example (217) below:
(217) Elthe
RealReal
MadridMadrid
podrıacould
nonot
ficharhire
alto.the
porterogoalkeeper
delof.the
AthleticAthletic
yand
elthe
BarcelonaBarcelona
[ ][ ]
alto.the
delanteroforward
deof
lathe
Real.Real
‘Real Madrid could not hire Athletic’s goalkeeper and Barcelona could hire Real’s
forward.’
In (217), negation is placed inside a ‘Complex Gapping’ example where the second con-
junct is missing the modal Verb and the lexical Verb. Negation is placed in the first part
of the coordination and the judgment for this example is that negation’s meaning only
reaches that conjunct. That is, the scope of negation in this example is narrow because
2.4. GAPS AND COMPLEX GAPS 97
the second conjunct does not get to be negated: [¬ p ] ∧ [ q ]. The only way this can be
explained is if conjunction is situated above negation in the structure so that the scope of
negation finds the limits of its conjunct (218). Since in example (217), negation is above
the lower vP phase and coordination needs to be above negation, conjunction needs to be
located at the height of the next vP phase. This means that conjunction is at the level
of the vP of the modal Verb, which is what we have in (215). Consider (218):
(218) TP
T’
Modal V CoP
vP(Modal V) Co’
Co vP(Modal V)
v’
tModalV VP
tModalV TP
. . .
Lexical V
v’
tModalV VP
tModalV TP
NegP
Neg’
Neg . . .
Lexical V
Notice that this adds to the evidence that PF deletion is needed in sentences of this
type. Recall that in this kind of structure, PF deletion needs to apply because the
Verb cannot escape its conjuncts. Since I have shown that a structure which is derived
by movement (216) is not possible basing myself on evidence from negation, deletion is
the process that needs to account for missing elements in ‘Complex Gapping for Verbal
Complexes’.
2.4.4.6 Gapping in NEG-nor
The last gapping type I am covering is ‘Gapping in NEG-nor ’ which is where gapping
appears in NEG-nor configurations. Consider example (103) repeated here as (219):
(219) Marıamaria
nonot
estuvowas
enfermasick
ninor
JuanJuan
[ ][ ]
acatarrado.with.a.cold
98 CHAPTER 2. GAPPING
In the previous section, I have assumed Wurmbrand (2008) to account for such configu-
rations. In doing so, I have assumed that both conjuncts are CPs. This means that this
type of gapping brings another difference with respect to the four prior cases because it
needs to be analyzed with large conjuncts. Consequently, missing elements need to be
affected by deletion, which is what explains their absence. Taking the latter into con-
sideration, notice that the different types of gapping sentences differ from each other in
the two following ways. With respect to the size of conjuncts, this is the only gapping
type that responds to large conjunct coordination. And with respect to PF deletion or
movement, movement plays a major role in the three first types: ‘Basic Gapping’, ‘Gap-
ping of Auxiliary’, and ‘Gapping of Modal’. However, PF deletion plays a major role in
‘Complex Gapping’ and ‘Gapping in NEG-nor ’.
Before I proceed to provide the analysis, I will show that Wurmbrand’s proposal that
these sentences are conjunctions of CPs is further supported by the following tests. When
testing this type of sentence with respect to elements that can scope widely, the result is
that these elements cannot scope widely. Consider sentence (220) in which a Modal Verb
can only scope narrowly. Consider also the impossibility of binding in the cross conjuncts
binding configurations in NEG-nor sentences (221)-(223):
(220) Nineither
JuanJuan
puedecan
comereat
alubiasbeans
ninor
PedroPedro
[ ][ ]
gambitas.shrimp
‘Neither Juan can eat beans nor Pedro can eat shrimp.’
(221) *Cadaeach
chicoi
boynonot
hahas
tocadoplayed
lathe
guitarraguitar
ninor
sui
hispadrefather
[ ][ ]
elthe
trombon.trombone
‘Each boy has not played the guitar nor his father has played the trombone.’
(222) ??Nineither
cadaeach
chicoi
boyhahas
tocadoplayed
lathe
guitarraguitar
ninor
sui
hispadrefather
[ ][ ]
elthe
trombon.trombone
‘Neither each boy has played the guitar nor his father has played the trombone.’
(223) ??Ningunno
chicoi
boyhahas
tocadoplayed
lathe
guitarraguitar
ninor
sui
hispadrefather
[ ][ ]
elthe
trombon.trombone
‘No boy has played the guitar nor his father has played the trombone.’
The four sentences in (220)-(223) in which typical wide scope elements cannot scope
widely prove that NEG-nor configurations cannot be accounted for via small conjuncts.
In these cases, conjunction should be CP coordination. Because conjuncts are CPs, the
only possible way to account for missing elements is PF deletion.
I propose a similar approach to deletion as the one for ‘Complex Gapping’. In order
to illustrate it, I represent the structure of sentence (219) in (224) below.
(224) a. [CP [CP Marıa no estuvo enferma] . . .
2.5. CONCLUSION 99
b. . . . [CP [C+NEG ni] [TopP Juan [FocP acatarrado [CP . . . ] ] ] ]
There is CP coordination in structure (224). In the second conjunct, the remnants move
from below and the checking of the F feature in Foc triggers the deletion of the Phase
Phrase except for the edge as proposed earlier.12
2.4.5 Interim summary
The goal of this section has been to provide an analysis of Spanish gapping basing it on
the findings in previous sections and attending to arguments which are normally part of
the gapping literature. I have hypothesized of the possibility of these sentences to be
multiple sluicing examples due to their similarity with these constructions. Although the
investigation has shown the hypothesis to be not apropriate for gapping, Lasnik’s (2006)
tests have revealed details of interest for the analysis, such as, the needed movements of
the remnants.
The solution has proven to be a hybrid analysis that can account for the heterogeneous
nature of the gapping types here discussed. A small conjunct approach which naturally
leaves space for large conjuncts and which considers both movement and PF deletion in
the analysis of missing elements is shown to be needed. It may be the case that the
fact that gapping in Spanish has been shown to be that heterogeneous opens up further
possibilities for gapping analyses in other languages.
2.5 Conclusion
The focus of this chapter has been to analyze gapping constructions in Spanish. In or-
der to do this, first, an overview of gapping constructions in other languages has been
provided with emphasis on aspects that have been shown central in the analysis, like
NEG-nor configurations, for instance. Afterwards, by placing Spanish side to side with
English, I have shown that what is normally considered gapping in the literature is also
present in Spanish in a similar way to English.
12In chapter 3, section 3.5, when analyzing NEG-nor configurations in determiner sharing I consider adifferent possibility for the analysis which is, however, similar to the analysis here provided for gapping.With a closer look at evidence in (220)-(223), there I conclude that coordination can be a bit lower inthe structure. It cannot be far lower because the modal Verbs need to be kept inside conjunction. Still,deletion needs to apply in these cases because the modal Verb, which is inside coordination, can still begapped. Notice that Wurmbrand (2008) considers both options –CP coordination and lower coordination–and only rejects one based on data specific to German.
100 CHAPTER 2. GAPPING
Among the interim conclusions, two of the most important ones in this analysis have
been that both small conjuncts and large conjuncts are needed in the syntax of Spanish
gapping, and that also two processes are needed in accounting for missing elements: move-
ment and deletion depending on the structures. The last step of the chapter has been to
put everything together and to analyze five types of Spanish gapping sentences that arise
in the close investigation of such a structure in this language. The main conclusion is that
the analysis of gapping in Spanish needs to be hybrid enough to cover all phenomenology
that can be observed in the different tests that the literature provides as instruments of
diagnosis. Gapping, then, in the surface looks the same in a variety of sentences. But,
when getting closer to the structure the picture starts getting different.
‘Basic Gapping’ sentences are claimed to be small conjunct structures with ATB move-
ment of the verb. ‘Gapping of Auxiliary and Modal’ are also small conjunct structures
in which movement is also present. PF deletion plays a major role both in ‘Complex
Gapping’ and ‘Gapping in NEG-nor ’. However, ‘Complex Gapping’ is also analyzed via
small conjuncts whereas ‘Gapping in NEG-nor ’ is analyzed via large conjuncts. This is,
in a nutshell, the subclassification of gapping in Spanish.
Chapter 3
Determiner Sharing
3.1 Introduction
In this chapter, I investigate determiner sharing sentences with a main focus on Spanish.
Determiner sharing is a phenomenon by which, in coordination contexts, the Verb and a
Determiner are missing in, for instance, non initial conjuncts in English and Spanish. In
order to start with a basis for the analysis of such structure, I review the properties of
determiner sharing which are normally listed in previous related literature and previous
analyses. I review these properties with respect to Spanish determiner sharing and check
which of the existing approaches can be applied to this language. Same as in chapter 2,
the adaptation will require flexibility both for small and large conjuncts, and also both
for deletion and movement as processes that account for missing elements. I organize this
chapter in the following way.
In section 2, I review the general properties of determiner sharing in English and,
in section 3, the determiner sharing properties in Spanish. In section 3 I also include a
complete paradigm of Spanish determiner sharing in both subject and object position.
This paradigm is divided according to the different gapping types that are listed in chapter
2 due to the existing similarities between gapping and determiner sharing. Afterwards, I
continue with the previous analyses in section 4 in order to find the one with the potential
to be adapted into Spanish. Finally, section 5 is dedicated to the analysis of determiner
sharing. I explain my proposal and analyze Spanish determiner sharing in subject position.
I also explain both grammatical and ungrammatical sentences where object determiner
sharing is present.
101
102 CHAPTER 3. DETERMINER SHARING
3.2 Determiner sharing
3.2.1 Introduction
Determiner sharing has mainly been investigated in English. This section is dedicated to
describe the phenomenon in order to set the picture towards my analysis of determiner
sharing in Spanish. I only list the main properties that determiner sharing displays in
English and I leave all aspects related to the existing analyses for the following sections
where I investigate the phenomenon in Spanish.
3.2.2 Properties of determiner sharing
The general properties of determiner sharing are the following.
3.2.2.1 Determiner sharing and the missing Verb
Determiner sharing is a phenomenon that happens in coordinate contexts and that is
closely related to gapping sentences. Broadly, in these kinds of constructions, a determiner
and a Verb are missing in one of the conjuncts. McCawley (1993) was the first to provide
a detailed description of determiner sharing constructions. I provide examples of the
phenomenon first and provide the features which McCawley lists afterwards. Consider
(225).
(225) Few cats eat fried fish and [ ]D dogs [ ]V roast beef.
In this example (225), the final conjunct is missing a determiner and the Verb. The de-
terminer is the one that corresponds to the DP subject position and it is understood as
‘few’ from the subject in the first conjunct. The Verb ‘eat’ is also gapped in this final
conjunct.
The close relation between determiner sharing and gapping has been established since
the earliest work published on the topic. McCawley (1993) provides empirical evidence
that determiner sharing is not possible if the Verb is not missing. Consider (226)-(227)
from McCawley (1993: 245).
(226) Too many Irish setters are named Kelly, [ ]D German shepherds [ ]V Fritz, and
[ ]D huskies [ ]V Nanook.
(227) *Too many Irish setters are named Kelly, [ ]D German shepherds are named
Fritz, and [ ]D huskies are named Nanook.
In example (226), determiner sharing of ‘too many’ is possible because the Verb is missing:
the verbal complex ‘are named’ is not overt in non initial conjuncts. On the contrary,
3.2. DETERMINER SHARING 103
example (227) is ungrammatical because the presence of the Verb in non initial conjuncts
makes determiner sharing impossible.
3.2.2.2 Determiner sharing and the missing T
Lin (2002), in this respect, further specifies that, in order for determiner sharing to be
grammatical, the missing Verbal element needs to be any part of the verbal complex
containing Tense, either a tensed lexical Verb or an Auxiliary. In other words, the depen-
dency is upon the omission of Tense. The latter idea goes back to Siegel (1984), Siegel
(1987) and Oehrle (1987) who establish that the dependency of shared elements is specifi-
cally on Tense. Consider, for instance, determiner sharing examples (228)-(230) from Lin
(2002:88).
(228) The girls will drink whiskey, and [ ]D boys [ ]T drink wine.
(229) * The girls will drink whiskey and [ ]D boys will drink wine.
(230) *The girls will drink whiskey, and [ ]D boys will [ ]V wine.
The only grammatical example among these sentences (228)-(230) is number (228). Al-
though the determiner ‘the’ is missing in non-initial conjuncts in all three examples (228)-
(230), it is only grammatical (228) when the other missing element is the Auxiliary ‘will’
which contains Tense, not the lexical Verb (230). When all elements are overt except for
the determiner, the sentence is also ungrammatical (229).
3.2.2.3 Conjunct initial
One property of determiner sharing that is stated by McCawley and that is normally
mentioned is that the shared determiner has to be initial in all the conjuncts. Examples
(231, 232) are from McCawley (1993:246).
(231) Too many films are reviewed by Ebert and [ ]D concerts [ ]V by von Rhein.
(232) *Ebert reviews too many films and von Rhein [ ]V [ ]D concerts.
The contrast between the two examples is that the shared determiner in (231) is at the
initial position in the conjuncts while in (232) the case is different: the shared determiner
is not initial. When it is not initial, the sentence is ungrammatical.
3.2.2.4 Shared constituent in Determiner position
McCawley proves that the shared initial constituent needs to occupy the determiner po-
sition. He compares, for instance, constructions like ‘Italy’s red wines’ and ‘Italian red
wines’. The adjectival counterpart cannot be shared while the other can. Consider (233,
234), from (McCawley1993:246).
104 CHAPTER 3. DETERMINER SHARING
(233) Italy’s red wines are outstanding and [ ] white wines excellent.
(234) *Italian red wines are outstanding and [ ]A whites wines excellent.
Example (234) is ungrammatical. The constituent that is shared here is an Adjective.
In sentence (233), where the shared constituent is in determiner position, the result is
grammatical.
3.2.2.5 Connectors, negation, determiner sharing
McCawley further observes that in the comparison between determiner sharing cases and
their counterparts where there is no determiner sharing, there is preference for differ-
ent connectors. In the determiner sharing cases (235), the ‘or’ conjunction is preferred,
whereas, in their counterparts (236), ‘and’ conjunction is preferred. The contrasts in each
pair of sentences, (235) and (236) which are in McCawley (1993: 247), illustrate the point.
(235) a. Not enough linguists study Russian, [ ]D literary scholars [ ]V French, or
[ ]D engineers [ ]V Japanese.
b. ?? Not enough linguists study Russian, [ ]D literary scholars [ ]V French,
and [ ]D engineers [ ]V Japanese.
(236) a. *Not enough linguists study Russian, not enough literary scholars [ ]V French,
or not enough engineers [ ]V Japanese.
b. Not enough linguists study Russian, not enough literary scholars [ ]V French,
and not enough engineers [ ]V Japanese.
3.2.2.6 Interim Summary
Among this list of properties, some of them have produced more discussion than others
in the literature: specifically the first two properties, (i) the dependency of determiner
sharing upon T gapping and (ii) the necessity of the shared determiner to be initial in
its conjunct. (i) McCawley himself notices some exceptions to the rule that in order to
have determiner sharing, Tense needs to also be missing. Examples that illustrate this
exception are in McCawley (1993: 245) which I repeat here as (237).
(237) . . . I began to wonder how many paintings will never be seen, songs will never
be heard, and books will never be read because of wars yet to come. (Funky
Winkerbean, 19 May 91)
In example (237), ‘how many’ phrases share this determiner though verbal complexes ‘will
be’ are fully overt in all the conjuncts. Ackema and Szendroi (2002) acknowledge examples
of this kind too, for English, and Arregi and Centeno (2005) and Centeno (2007b), Centeno
(2007a)) account for them in Spanish. I develop this further in the next chapter. (ii) With
3.3. DETERMINER SHARING IN SPANISH 105
respect to the second property, the necessity of the shared determiner to be initial in its
conjunct, there is a noteworthy aspect. Subsequent approaches to determiner sharing
manage to maintain this generalization across types of sentences when cross linguistic
data challenge this rule. Discussion on this topic appears in sections 3.3.2.3 and 3.5.4.
3.2.3 Interim conclusion
These are the properties that characterize determiner sharing in English. I continue with
Spanish in the next section, where I provide an analysis for determiner sharing in this
language. First, I explain the properties of determiner sharing in Spanish and provide
determiner sharing data. Afterwards, I continue with their syntactic analyses.
3.3 Determiner sharing in Spanish
3.3.1 Introduction
This section is dedicated to determiner sharing in Spanish. First, I explain the main
properties of this phenomenon in Spanish in comparison to English. I follow existing
discussions on English determiner sharing, mainly considering McCawley (1993), as was
carried out in the previous section. Afterwards, I introduce determiner sharing data in
Spanish in more detail. I base the Spanish determiner sharing paradigm which I provide
on the classification of Spanish gapping which I propose in this work. Hence, I add
determiner sharing to all gapping types and test the results. I conclude afterwards with
the main generalizations so that I may proceed with the syntactic analyses later.
3.3.2 Properties of determiner sharing in Spanish
Determiner sharing displays similar properties in Spanish and in English. However, these
are not completely equal. The general properties of Spanish determiner sharing follow.
3.3.2.1 Determiner sharing and the missing Verb
First, example (238) illustrates determiner sharing in Spanish: both the Determiner and
the Verb are missing from non-initial conjuncts.
(238) Pocosfew
gatoscats
comeneat
pescaditofish.DIM
fritofried
yand
[ ]D[ ]D
perrosdogs
[ ]V[ ]V
solomillosirloin
aat
lathe
plancha.grill‘Few cats eat fried fish and few dogs eat grilled sirloin.’
106 CHAPTER 3. DETERMINER SHARING
Determiner sharing in Spanish (238) occurs in coordinate contexts and the non-initial
conjuncts are missing the Determiner, pocos ‘few’ in this case, and the Verb, comen ‘eat’
here, but still both are understood in all conjuncts.
In Spanish, same as in English, there is the dependency which McCawley (1993) noted
that determiner sharing is dependent upon gapping of the Verb. Consider the following
contrast to illustrate the point (239)-(240).
(239) Muchosmany
hombresmen
ricosrich
llevanwear
chaquetasjackets
conwith
agujerosholes
yand
[ ]D[ ]D
hombresmen
pobrespoor
[ ]V[ ]V
corbatasties
deof
sedasilk
‘Many rich men wear jackets with holes and many poor men wear silk ties.
(240) *Muchosmany
hombresmen
ricosrich
llevanwear
chaquetasjackets
conwith
agujerosholes
yand
[ ]D[ ]D
hombresmen
pobrespoor
llevanwear
corbatasties
deof
sedasilk
‘Many rich men wear jackets with holes and many poor men wear silk ties.
The grammaticality contrast in (239)-(240) is of the same type that can be found in
English. In example (239), sharing of the determiner muchos ‘many’ is possible because
there is also gapping of the Verb llevan ‘wear’. When the Verb is not missing, like in
(240), the example is ungrammatical.
3.3.2.2 Determiner sharing and the missing T
With respect to this property (Siegel (1984, 1987), Oehrle (1987), and Lin (2002)) Spanish
behaves in the same way as English: the dependency is upon the omission of Tense. In
order to have determiner sharing the Tense bearing element in the Verbal complex is what
needs to be missing. Consider (241)-(243).
(241) Muchosmany
chicosboys
hanhave
cocinadocooked
carnemeat
yand
[ ]D[ ]D
chicasgirls
[ ]T[ ]T
cocinadocooked
paellapaella
‘Many boys have cooked meat and many girls have cooked paella.’
(242) *Muchosmany
chicosboys
hanhave
cocinadocooked
carnemeat
yand
[ ]D[ ]D
chicasgirls
hanhave
cocinadocooked
paellapaella
‘Many boys have cooked meat and many girls have cooked paella.’
(243) *Muchosmany
chicosboys
hanhave
cocinadocooked
carnemeat
yand
[ ]D[ ]D
chicasgirls
hanhave
[ ]V[ ]V
paellapaella
‘Many boys have cooked meat and many girls have cooked paella.’
This paradigm (241, 242, 243) shows that determiner sharing is dependent upon T gapping
(241). In the first example (241), the Determiner muchos ‘many’ and the Tense bearing
3.3. DETERMINER SHARING IN SPANISH 107
element han ‘have’ are missing and the sentence is grammatical. In the next example
(242), only the Determiner is missing and the sentence is not correct. And finally, in the
last example (243), the Determiner and the lexical Verb cocinado ‘cooked’ are missing
but the Tense bearing element han ‘have’ is maintained: the result is that the sentence is
not grammatical.
3.3.2.3 Conjunct initial in Spanish?
One of the properties that McCawley (1993) lists for English determiner sharing does not
hold for Spanish. In Spanish, the shared Determiner does not need to be initial in all
the conjuncts. (In English the case is the contrary.) Consider the following two examples
(244)-(245), among which there is no grammatical contrast.
(244) Muchosmany
chicosboys
comeneat
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]V[ ]V
heladoice-cream
deof
vainilla.vanilla‘Many boys eat apple pie and many girls eat vanilla ice-cream.’
(245) PedroPedro
hahas
comidoeaten
demasiadostoo.many
pastelescakes
yand
JuanJuan
[ ]V[ ]V
[ ]D[ ]D
heladosice-creams
deof
vainilla.vanilla‘Pedro has eaten too many cakes and Juan has eaten too many vanilla ice-creams.’
The word order in the second conjunct of example (245) shows that the subject ‘Juan’
has to always precede the shared Determiner demasiados ‘too many’, represented by the
bracketed gap for D in this example. These two grammatical examples (244, 245) are
also relevant in this other respect. In both examples, coordination is conjoining elements
that are composed of a subject and a predicate. Within these circumstances, determiner
sharing in object position, which is what is grammatical in (245), is not possible in
English.1 This is another difference between English and Spanish. Determiner sharing in
subject position, illustrated by (244) at this point, is possible in both languages.
3.3.2.4 Shared constituent in Determiner position. . .
This property that is listed in McCawley (1993) for English cannot be applied to Spanish
for different reasons that I explain. The corresponding English examples (375) and (376)
cannot be reproduced in Spanish. The first reason for it is that prenominal Adjectives in
Spanish are more limited than in English. Prenominal Adjectives in Spanish can only be
1Notice that not all cases of determiner sharing in object position are grammatical. I explain in thenext section which cases of determiner sharing in object position are grammatical and which cases arenot.
108 CHAPTER 3. DETERMINER SHARING
used with an appositive reading and this type of reading is difficult to get with Adjectives
that specify the origin of a country or region. Consider the oddness of example (387).
(246) ???losthe
italianosItalian
vinoswines
‘the Italian wines’
The second reason is that restrictions on bare NPs are much more in Spanish than in
English. The following example (388) illustrates the fact that in Spanish bare NPs are
not possible in preverbal subject position.
(247) *Vinoswines
italianosItalian
sonare
excelentesexcelent
‘Italian wines are excellent.’
The bare NP vinos italianos ‘Italian wines’ is not possible in the illustrated context, in
preverbal subject position. Due to these reasons then, the property that McCawley lists
as characteristic of determiner sharing in English cannot be replicated for Spanish.
3.3.2.5 Connectors, negation, determiner sharing
McCawley (1993) also notices that the preference for connectors changes from gapping
contexts alone to determiner sharing contexts when some specific Determiners are used.
Examples in (248, 249) illustrate this property. Notice that there is a grammaticality
contrast in both (248) and (249).
(248) a. Nono
muchosmany
linguistaslinguists
estudianstudy
rusoRussian
oor
[ ]D[ ]D
literatosmen.of.letters
[ ]V[ ]V
francesFrench
‘Not many linguists study Russian or not many men of letters study French.’
b. ??Nono
muchosmany
linguistaslinguists
estudianstudy
rusoRussian
yand
[ ]D[ ]D
literatosmen.of.letters
[ ]V[ ]V
francesFrench‘Not many linguists study Russian and not many men of letters study French.’
(249) a. ??Nonot
muchosmany
linguistaslinguists
estudianstudy
rusoRussian
oor
nonot
muchosmany
literatosmen.of.letters
[ ]V[ ]V
francesFrench‘Not many linguist study Russian or not many men of letters study French.’
b. Nonot
muchosmany
linguistaslinguists
estudianstudy
rusoRussian
yand
nonot
muchosmany
literatosmen.of.letters
[ ]V[ ]V
francesFrench‘Not many linguist study Russian and not many men of letters study French.’
3.3. DETERMINER SHARING IN SPANISH 109
The paradigm of Spanish sentences (248, 249) parallels that of English. With Determin-
ers of this type, no muchos ‘not many’, when determiner sharing is present (248), the
preferred connector is o ‘or’ as the (248 a) sentence illustrates. On the contrary, when
only gapping is present (249), the preferred connector is y ‘and’ as the (249 b) sentence
illustrates.
I explained in the discussion on English determiner sharing that some of these prop-
erties are more extensively discussed in the literature than others, among others, because
of some existing exceptions. Exceptions are normally reported in relation to the first
properties listed here. Like in English, Spanish has exceptions to the rule that determiner
sharing is only possible if Tense is also gapped. Although I explain the related aspects
in the next chapter, it is important to note that the exceptions are the same in both
languages. Consider (250) for illustration. Example (250a) is a Spanish sentence and
example (250b) is an English one taken from Ackema and Szendroi (2002, 210).
(250) a. ¿Cuantoshow.many
librosbooks
leyoread
JuanJuan
yand
[ ]D[ ]D
revistasmagazines
revisoreviewed
Pedro?Pedro?
‘How many books did Juan read and how many magazines did Pedro review?’
b. I began to wonder how many paintings will never be seen, [ ]D songs will
never be heard, and [ ]D books will never be read because of wars yet to
come.
In sentence (250a), the sharing of the Wh-Determiner cuantos ‘how many’ is possible even
though Tense bearing elements, leyo ‘read’ and reviso ‘reviewed’, are overtly pronounced
in both conjuncts. Hence, wh-determiner sharing is possible under such circumstances.
The same happens in the English example (250b), where the Wh-Determiner can be
shared while the Tense bearing element ‘will’ is overt in all the conjuncts.
Overall, I have illustrated that determiner sharing is a phenomenon that is present
in Spanish and I have investigated determiner sharing properties in Spanish making a
comparison between this language and English. In the following, I provide a determiner
sharing paradigm for Spanish that comprises a set of examples ordered according to the
findings that I present in the gapping chapter. I also explain that existing analyses do
not account for the Spanish determiner sharing pattern presented there.
3.3.3 Determiner sharing: Spanish data
The determiner sharing analyses that exist in the literature account for such phenomenon
either by means of large conjuncts or by means of small conjuncts. This labeling of ap-
110 CHAPTER 3. DETERMINER SHARING
proaches is basically the same for approaches on gapping because of the dependence of
determiner sharing upon gapping. This dependence is such that determiner sharing could
be considered a sort of gapping. Notice, however, that one of the major conclusions in
the chapter where gapping in Spanish is analyzed is that both large and small conjuncts
are needed.
Due to the same dependence in Spanish (of determiner sharing upon gapping), I ob-
viously take into account the classification of gapping types that I make in the gapping
chapter in order to find out the type of analysis that Spanish determiner sharing needs.
In more specific terms, I add determiner sharing to all gapping types that are proposed
in the gapping chapter. The first idea that is worth noting in relation to all of the dif-
ferent types of determiner sharing is that judgments for determiner sharing structures do
also differ from type to type. This makes the classification of determiner sharing types
reasonable.
I start the classification by dividing determiner sharing examples between those where
the Determiner is shared in subject position and those where the sharing occurs in object
position. Recall from the previous section that object determiner sharing is possible in
Spanish. I begin with subject position and continue with object position.
3.3.3.1 Spanish determiner sharing in subject position
Determiner sharing in subject position in Spanish is possible in all the gapping types
that I have provided in the previous chapter. Judgments are the same both (i) for the
sentences that exemplify the interaction between each type of gapping and determiner
sharing and (ii) for the sentences which are related to the tests that provide guidance
towards the analysis. I will consider each gapping type one by one showing that the
behavior of determiner sharing in subject position is the same as what has been found for
gapping.
3.3.3.1.1 Determiner sharing in Basic Gapping: Consider example (251) which
shows that determiner sharing with basic gapping –the only lexical Verb is missing– is
possible.
(251) Muchosmany
chicosboys
comeneat
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]V[ ]V
heladoice-cream
deof
vainilla.vanilla‘Many boys eat apple pie and many girls eat vanilla ice-cream.’
3.3. DETERMINER SHARING IN SPANISH 111
In (251), the determiner muchos ‘many’ and the Verb comen ‘eat’ are missing from the
non-initial conjunct. As represented in (251), the example is grammatical and the inter-
pretation of each word is also present in the non-initial conjunct. In conclusion, determiner
sharing in this case is possible.
Furthermore, the contrasts (252)-(253) and (254)-(255) show that determiner sharing,
like gapping, needs small conjuncts because elements with wide scope, such as negation,
can scope widely only when gapping affects the sentence.
(252) Muchosmany
chicosboys
nonot
comeneat
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]V[ ]V
heladoice-cream
deof
vainilla.vanilla
‘It is not the case that many boys eat apple pie and girls vanilla ice-cream.’
(253) Muchosmany
chicosboys
nonot
comeneat
pastelpie
deof
manzanaapple
yand
muchasmany
chicasgirls
comeneat
heladoice-cream
deof
vainilla.vanilla
‘Many boys do not eat apple pie and many girls eat vanilla ice-cream.’
(254) Muchosmany
pacientespatients
nonot
duermensleep
enat
elthe
pasillocorridor
yand
[ ]D[ ]D
medicosdoctors
[ ]V[ ]V
enin
lasthe
habitaciones.rooms‘It is not the case that many patients sleep at the corridor and doctors in the
rooms.’
(255) Muchosmany
pacientespatients
nonot
duermensleep
enat
elthe
pasillocorridor
yand
muchosmany
medicosdoctors
duermensleep
enin
lasthe
habitaciones.rooms.
‘Many patients do not sleep at the corridor and many doctors sleep in the rooms.’
Semantically, (252) and (253) contrast with respect to the scope of negation. In (253),
where no word is missing, negation only negates the first conjunct. When determiner
sharing affects the sentence (252), negation scopes above coordination and the meaning
of the sentence can be expressed: ‘It cannot be the case that many boys eat apple pie and
many girls eat vanilla ice-cream’. The judgment for the wide scope reading of negation
may be simpler in example (254), whose meaning is: ‘It is not the case that many patients
sleep at the corridors while many doctors sleep at the rooms’. The latter contrasts with
the scope of negation in sentence (255), which is narrow because there are no missing
elements in the sentence. When there is only coordination with no missing element and
negation is only on one part of the coordination, negation only affects its conjunct. Hence,
112 CHAPTER 3. DETERMINER SHARING
in (255), the second conjunct is affirmative and negation’s scope stops at the coordination
level.
3.3.3.1.2 Determiner sharing in Gapping of Auxiliary: The same correspon-
dence between determiner sharing and gapping is observed with ‘Gapping of Auxiliary’.
Determiner sharing is possible when the other missing element is the Auxiliary (256).
(256) Muchosmany
chicosboys
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]T[ ]T
degustadotasted
heladoice-cream
deof
vainilla.vanilla
‘Many boys have eaten apple pie and many girls have tasted vanilla ice-cream.’
In sentence (256), the Determiner muchos ‘many’ and the Auxiliary han ‘have’ are missing
from the second conjunct; the sentence is grammatical; and they get to be interpreted in
the final conjunct. The behavior of wide scope elements is also the same: consider the
semantic contrasts between sentences (257) and (258), and (259) and (260).
(257) Demasiadostoo.many
chicosboys
nonot
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]T[ ]T
degustadotasted
heladoice-cream
deof
vainilla.vanilla
‘It is not the case that too many boys have eaten apple pie and girls tasted vanilla
ice-cream.’
(258) Demasiadostoo.many
chicosboys
nohave
hannot
comidoeaten
pastelpie
deof
manzanaapple
yand
demasiadastoo.many
chicasgirls
hanhave
degustadotasted
heladoice-cream
deof
vainilla.vanilla
‘Too many boy have not eaten apple pie and too many gilrs have tasted vanilla
ice-cream.’
(259) Muchosmany
pacientespatients
nonot
hanhave
dormidoslept
enat
elthe
pasillocorridor
yand
[ ]D[ ]D
medicosdoctors
[ ]T[ ]T
ocupadooccupied
lasthe
habitaciones.rooms
‘It is not the case that many patients have slept at the corridor and doctors
occupied the rooms.’
(260) Muchosmany
pacientespatients
nonot
hanhave
dormidoslept
enat
elthe
pasillocorridor
yand
muchosmany
medicosdoctors
hanhave
ocupadooccupied
lasthe
habitaciones.rooms
‘Many patients have not slept at the corridor and many doctors have occupied the
rooms.’
3.3. DETERMINER SHARING IN SPANISH 113
In examples (258) and (260), negation only affects the initial conjuncts. However, when
there is determiner sharing and gapping, like in (257) and (259), the scope of negation is
wide. Sentence (257) for instance can be paraphrased: ‘It is not that too many boys have
eaten apple pie while too many girls have tasted vanilla ice-cream.
3.3.3.1.3 Determiner sharing in Gapping of Modal: When determiner sharing
happens in ‘Gapping of Modal’ cases, similar conclusions can be reached again. Sentence
(261) exemplifies the fact that determiner sharing, of muchas ‘many’ in this case, is
possible when the gapped element is a modal Verb, deberıan ‘should’ in this sentence.
(261) Muchasmany
chicasgirls
deberıanshould
arreglarfix
sutheir
cochecar
yand
[ ]D[ ]D
chicosboys
[ ]T[ ]T
prepararprepare
sustheir
maletas.luggage
‘Many girls should fix their car and many boys should prepare their luggage.’
(262) Muchasmany
chicasgirls
deberıanshould
arreglarfix
sutheir
cochecar
yand
muchosmany
chicosboys
deberıanshould
prepararprepare
sustheir
maletas.luggage
‘Many girls should fix their car and many boys should prepare their luggage.’
In determiner sharing within ‘Gapping of Modal’ cases, the scope of the wide scope
elements behaves similarly. In sentence (261), the modal Verb deberıan ‘should’ refers
to one requirement that has two parts: ‘it should be the case that many girls fix their
cars and many boys prepare their luggage’. The modal Verb scopes widely because of the
presence of gapping. In (262), because every word is overt in both conjuncts, the modal
Verbs in each conjunct express two separate requirements: on the one side, the fixing of
the car, and on the other side, the preparing of the luggage.
3.3.3.1.4 Determiner sharing in Complex Gapping for Objects: Determiner
sharing with ‘Complex Gapping’ yields similar results. Determiner sharing is also possible
in ‘Complex Gapping for Objects’ examples (263).
(263) Demasiadostoo.many
chicosboys
leCL
dierongave
unaa
florflower
ato
MarıaMarıa
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
ato
Jorge.Jorge‘Too many boys gave a flower to Marıa and too many girls gave a flower to Jorge.’
In sentence (263), the final conjunct is missing the Determiner demasiados ‘too many’,
the Verb dieron ‘gave’ and the Direct Object una flor ‘a flower’ though all of them are
interpreted there.
114 CHAPTER 3. DETERMINER SHARING
In sentences of this kind, elements like negation can scope widely above coordination.
The meaning of (264) and (266) contrasts with the meaning of their counterparts (265)
and (267) where negation only affects the first conjunct.
(264) Demasiadostoo.many
chicosboys
nonot
leCL
dierongave
unaa
florflower
ato
MarıaMarıa
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
ato
Jorge.Jorge
‘It is not the case that too many boys gave a flower to Marıa and girls to Jorge.’
(265) Demasiadostoo.many
chicosboys
nonot
leCL
dierongave
unaa
florflower
ato
MarıaMarıa
yand
demasiadastoo.many
chicasgirls
leCL
dierongave
unaa
florflower
ato
Jorge.Jorge
‘Too many boys did not give a flower to Marıa and too many girls gave a flower
to Jorge.’
(266) Demasiadostoo.many
chicosboys
nonot
pudieroncould
darlegive.CL
unaa
florflower
ato
MarıaMarıa
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
ato
Jorge.Jorge
‘It is not the case that too many boys were able to give a flower to Marıa and girls
to Jorge.’
(267) Demasiadostoo.many
chicosboys
nonot
pudieroncould
darlegive.CL
unaa
florflower
ato
MarıaMarıa
yand
demasiadastoo.many
chicasgirls
pudieroncould
darlegive.CL
unaa
florflower
ato
Jorge.Jorge
‘Too many boys were not able to give a flower to Marıa and too many girls were
able to give a flower to Jorge.’
A further detail in the parallelism between determiner sharing and gapping types
concerns the usage of pronouns. This can be seen in (268).
(268) Muchosmany
chicosboys
leCL
dierongave
sutheir
numeronumber
deof
telefonotelephone
ato
MarıaMarıa
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
ato
Jorge.Jorge
‘Many boys gave their telephone number to Marıa and girls (their telephone num-
ber) to Jorge.’
One of the readings of the second conjunct in (268) corresponds to the sloppy reading
of the pronoun where ‘many girls’ telephone numbers were given to Jorge’. This specific
fact also happens in the corresponding ‘Complex Gapping’ cases and means that deletion
accounts for missing elements in these cases. The behavior of determiner sharing in subject
position and the behavior of gapping are parallel for all types of contexts investigated here.
The following subsections provide additional evidence for this conclusion.
3.3. DETERMINER SHARING IN SPANISH 115
3.3.3.1.5 Determiner sharing in Complex Gapping for Verbal Complexes:
Determiner sharing in ‘Complex Gapping for Verbal Complexes’ behaves like its gapping
counterpart.
(269) Muchosmany
chicosboys
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
heladoice-cream
deof
vainilla.vanilla
‘Many boys have eaten apple pie and many girls have eaten vanilla ice-cream.’
In sentence (269), the Determiner muchos ‘many’, the Auxiliary han ‘have’ and the Verb
comido ‘eaten’ are missing from the final conjunct but they are still understood there.
The behavior of wide scope elements is the same. When gapping is present, negation
scopes widely (270, 272). But when every element is present, negation only affects the
first conjunct, like in (271, 273).
(270) Demasiadostoo.many
chicosboys
nonot
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
heladoice-cream
deof
vainilla.vanilla
‘It is not the case that too many boys have eaten apple pie and girls vanilla ice-
cream.’
(271) Demasiadostoo.many
chicosboys
nonot
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
demasiadastoo.many
chicasgirls
hanhave
comidoeaten
heladoice-cream
deof
vainilla.vanilla
‘Too many boys have not eaten apple pie and too many girls have eaten vanilla
ice-cream.’
(272) Muchosmany
pacientespatients
nonot
hanhave
dormidoslept
enin
elthe
pasillocorridor
yand
[ ]D[ ]D
medicosdoctors
[ ]Cplx
[ ]Cplx
enin
lasthe
habitaciones.rooms
‘It is not the case that many patients have slept in the corridor and doctors in the
rooms.’
(273) Muchosmany
pacientespatients
nonot
hanhave
dormidoslept
enin
elthe
pasillocorridor
yand
muchosmany
medicosdoctors
hanhave
dormidoslept
enin
lasthe
habitaciones.rooms
‘Many patients have not slept in the corridor and many doctors have slept in the
rooms.’
‘Complex Gapping for Verbal Complexes’ with modal Verbs is similar to the deter-
miner sharing counterpart. Consider (274): the Determiner muchos ‘many’ and the verbal
116 CHAPTER 3. DETERMINER SHARING
complex with the modal Verb deberıan arreglar ‘should fix’ are missing in the final con-
junct. Still, both ‘many’ and ‘should fix’ are interpreted in the non-initial conjunct. This
sentence (274) contrasts semantically with (275) where there is no determiner sharing.
This indicates that the scope of the modal Verb in (274) is wide.
(274) Muchasmany
chicasgirls
deberıanshould
arreglarfix
sutheir
cochecar
yand
[ ]D[ ]D
chicosboys
[ ]Cplx
[ ]Cplx
sutheir
bicicleta.bike‘Many girls should fix their car and boys their bike.’
(275) Muchasmany
chicasgirls
deberıanshould
arreglarfix
sutheir
cochecar
yand
muchosmany
chicosboys
deberıanshould
arreglarfix
sutheir
bicicleta.bike
‘Many girls should fix their car and many boys should fix their bike.’
3.3.3.1.6 Determiner sharing in Gapping in NEG-nor : In NEG-nor configu-
rations, determiner sharing behaves the same way as gapping. First, determiner sharing
is possible in such constructions. Example (276) is a grammatical sentence where the
Determiner demasiadas ‘too many’ and the Verb estuvieron ‘were’ are missing from the
non-initial conjunct, however, not from its interpretation.
(276) Nineither
demasiadostoo.many
ninosboys
estuvieronwere
enfermossick
ninor
[ ]D[ ]D
ninasgirls
[ ]V[ ]V
acatarradas.with.a.cold‘Neither too many boys were sick nor too many girls were with a cold.’
It is important to take into account sentence (277) because it reinforces the parallelism
between determiner sharing in subject position and the gapping types which are provided
and explained in the previous chapter. Like its gapping counterparts, determiner sharing
in NEG-nor configurations blocks wide scope readings.
(277) Nineither
muchosmany
profesoresprofessors
puedencan
comereat
caviarcaviar
ninor
[ ]D[ ]D
estudiantesstudents
[ ]Cplx
[ ]Cplx
arrozrice
conwith
tomate.tomato.sauce
‘Neither too many professors can eat caviar nor many students can eat rice with
tomato sauce.’
The modal Verb pueden ‘can’ in this sentence (277) does not scope widely above conjunc-
tion. The two facts that are conjoined in sentence (277) are understood as separate like
in distributed readings: ‘on the one hand, many teachers cannot eat caviar, and on the
other hand, many students cannot eat rice with tomato sauce’. The parallelism between
3.3. DETERMINER SHARING IN SPANISH 117
determiner sharing in subject position and the different gapping types that I propose is
clear. I continue now investigating the behavior of determiner sharing in object position.
3.3.3.2 Spanish determiner sharing in object position
Determiner sharing in object position is more problematic than determiner sharing in
subject position. That is why, I provide data first which only take into account the
interaction between the different gapping types and determiner sharing. For now, other
data are left to the side. I provide the complete paradigm of gapping types in determiner
sharing situations indicating whether or not the judgments are that determiner sharing is
possible in each sentence. I will now proceed to explain that determiner sharing in object
position is possible only in some cases.2
3.3.3.2.1 Determiner sharing in Basic Gapping: Determiner sharing in Basic
Gapping is not possible when it happens in object position (278). With this type of
determiner sharing, the elements that are missing are the Determiner and the only lexical
Verb. In example (278), for instance, the missing Determiner is demasiados ‘too many’
in the object position in demasiados helados de vainilla ‘too many vanilla ice-creams’ and
the missing Verb is come ‘eats’. The interpretation of the Determiner that appears in the
first part of the coordination does not reach the object in the second conjunct.
(278) #PedroPedro
comeeats
demasiadostoo.many
pastelespies
deof
manzanaapple
yand
JuanJuan
[ ]V[ ]V
[ ]D[ ]D
heladosice-creams
deof
vainilla.vanilla
‘Pedro eats too many apple pies and Juan eats too many vanilla ice-creams.’
Notice that example (278) is grammatical in the reading where helados de vainilla ‘vanilla
ice-creams’ is interpreted as a bare plural. The interpretation in such a case is ‘Pedro eats
too many apple pies and Juan eats vanilla ice-creams’. However, this is not the determiner
sharing interpretation, which is our target, and that is why the grammaticality judgement
is represented with #. This also applies to the following relevant examples.
3.3.3.2.2 Determiner sharing in Gapping of Auxiliary: This type of determiner
sharing is such that it happens in object position and the missing elements in non initial
conjuncts are the Determiner and the Auxiliary. This type of determiner sharing in object
position is not possible. As in the previous example, the interpretation of the Determiner
2This is different from what is stated in Arregi and Centeno (2005). Since the publication of thatarticle, closer analysis to sentences of this type has led me to the conclusion in this work: object determinersharing is not always possible in Spanish. For a summary of the main points in that article see the nextchapter on wh-determiner sharing.
118 CHAPTER 3. DETERMINER SHARING
in the first object DP does not reach the second object DP. Consider (279) for illustration
where the Auxiliary ha ‘has’ and the Determiner demasiados ‘too many’ in the object
of the second conjunct are missing. The interpretation of demasiados ‘too many’ in the
first object DP, demasiados pasteles de manzana ‘too many apple pies’ does not reach the
second object DP.
(279) #PedroPedro
hahas
comidoeaten
demasiadostoo.many
pastelespies
deof
manzanaapples
yand
JuanJuan
[ ]T[ ]T
degustadotasted
[ ]D[ ]D
heladosice-creams
deof
vainilla.vanilla
‘Pedro has eaten too many apple pies and Juan has tasted too many vanilla ice-
creams.
3.3.3.2.3 Determiner sharing in Gapping of Modal: Here, determiner sharing
affects objects again but sharing is not possible. In example (280), the affected object
is muchos arandanos ‘many blueberries’ and it is the Determiner muchos ‘many’ that
is missing. Such a Determiner is not interpreted in the second conjunct. The other
missing element in this type of determiner sharing is the modal Verb. In the example
that illustrates this, (280), the missing modal Verb is deberıa ‘should’ in the non initial
conjunct.
(280) #PedroPedro
deberıashould
tomardrink
muchosmany
zumosjuices
deof
naranjaorange
yand
JuanJuan
[ ]T[ ]
comereat
[ ]D[ ]
arandanos.blueberries‘Pedro should drink many glasses of orange juice and Juan should eat many blue-
berries.’
3.3.3.2.4 To sum up so far. Judgments so far indicate that determiner sharing in
object position in these types of gapping, ‘Basic Gapping’ (278), ‘Gapping of Auxiliary’
(279), ‘Gapping of Modal’ (280), is not possible. These sentences are not ungrammatical
but the interpretation of the determiner which is only pronounced in initial conjuncts
does not modify the NP that would get affected in the final conjuncts. The situation
changes from now on, however. Consider the following two examples (281, 282) under the
‘Complex Gapping for Objects’ type.
3.3.3.2.5 Determiner sharing in Complex Gapping for Objects: This type of
determiner sharing is characterized by such phenomenon per se embedded in the type of
gapping where the missing elements are the Verb and one of its arguments. In this case,
determiner sharing is possible. Hence, though the Determiner, the Verb, and one of its
3.3. DETERMINER SHARING IN SPANISH 119
arguments are missing, all of them are interpreted in the non initial conjuncts. Consider
(281, 282).
(281) Alvaro
Alvaro
leCL
diogave
ato
JuanJuan
demasiadastoo.many
responsabilidadesresponsibilities
yand
AnaAna
[ ]Cplx
[ ]Cplx
[ ]D[ ]D
disgustos.problems
‘Alvaro gave Juan too many responsibilities and Ana caused Juan too many prob-
lems.’
(282) Alvaro
Alvaro
leCL
diogave
demasiadastoo.many
responsabilidadesresponsibilities
ato
JuanJuan
yand
AnaAna
[ ]Cplx
[ ]Cplx
[ ]D[ ]D
disgustos.problems
‘Alvaro gave too many responsibilities to Juan and Ana caused too many problems
for Juan.’
In the first example (281), the interpretation of the determiner that would be shared de-
masiadas ‘too many’ does not modify the NP in the final conjunct disgustos ‘problems’.
Determiner sharing is not present in example (281). However, consider (282). The only
difference in this example with respect to (281) is that the DP that enters into the de-
terminer sharing relationship, demasiadas responsabilidades ‘too many responsibilities’ in
the first conjunct is before the Indirect Object a Juan ‘to Juan’. With this being the only
difference, determiner sharing becomes possible in example (282). In (282), the interpre-
tation of demasiadas ‘too many’ modifies responsabilidades ‘responsibilities’ in the final
conjunct.
Before I explain the implications of this grammatical contrast (281)-(282) in terms of
the analysis, I provide the rest of the examples in the paradigm. This includes deter-
miner sharing in ‘Complex Gapping for Verbal Complexes’ (283, 284) and in NEG-nor
configurations (285).
3.3.3.2.6 Determiner sharing in Complex Gapping for Verbal Complexes:
According to the logic I am following in this classification, identification of this type
of determiner sharing is not difficult. In this type of determiner sharing, the missing
elements in non initial conjuncts are the Determiner, the lexical Verb, and the Auxiliary
or the modal Verb, depending on the example. Sentences (283) and (284) illustrate both
cases respectively. The determiner demasiados ‘too many’, the Auxiliary ha ‘has’, and
the lexical Verb comido ‘eaten’ are missing in (283). Similarly, in (284), the Determiner
muchos ‘many’, the modal Verb deberıa ‘should’, and the lexical Verb tomar ‘drink’ are
120 CHAPTER 3. DETERMINER SHARING
the missing elements. Determiner sharing is possible in both cases. In other words, the
Determiner can be interpreted in both conjuncts in both sentences.
(283) PedroPedro
hahas
comidoeaten
demasiadostoo.many
pastelescakes
yand
JuanJuan
[ ]Cplx
[ ]Cplx
[ ]D[ ]D
heladosice-creams
deof
vainilla.vanilla‘Pedro has eaten too many cakes and Juan has eaten too many vanilla ice-creams.’
(284) PedroPedro
deberıashould
tomardrink
muchosmany
zumosjuices
deof
naranjaorange
yand
JuanJuan
[ ]Cplx
[ ]Cplx
[ ]D[ ]D
batidosmilkshakes
deof
arandanos.blueberries
‘Pedro should drink many glasses of orange juice and Juan should drink many
blueberry milkshakes.’
3.3.3.2.7 Determiner sharing in Gapping in NEG-nor : This type of determiner
sharing is the one where the sharing of the Determiner happens in a NEG-nor configura-
tion. Consider (285).
(285) Nineither
PedroPedro
comeeats
demasiadostoo.many
pastelespies
deof
manzanaapple
ninor
JuanJuan
[ ]V[ ]V
[ ]D[ ]D
heladosice-creams
deof
vainilla.vanilla
‘Neither Pedro eats too many apple pies nor Juan eats too many vanilla ice-
creams.’
Determiner sharing in these two last types of gapping is possible (283)-(285). The de-
terminer that is interpreted in the DPs that are in object position in initial conjuncts is
interpreted in the object DPs in final conjuncts.
3.3.4 Interim conclusion
Given the paradigm provided above, several aspects can be observed at this point which
need to condition the analysis that should be provided for determiner sharing in Spanish.
These are:
• It is clear from what can be observed in the paradigm that, as in gapping, both small
and large conjuncts are needed. Because determiner sharing is a type of gapping,
the deletion and movement processes used in the gapping analyses in chapter 2,
section 2.4.4, are also needed in determiner sharing.
• The dependence that determiner sharing has upon the gapping of Tense needs to
also be captured by the analysis, as in English.
3.4. PREVIOUS DETERMINER SHARING ANALYSES 121
• The different pattern in judgments that can be found between subject determiner
sharing and object determiner sharing in Spanish needs to be explained by the
analysis.
• The proposal needs to also capture the pattern within object determiner sharing,
which is only possible within the gapping types that are accounted for by deletion.
The first listed item is important in deciding about existing accounts for determiner
sharing. The reason for this is that the need for both sizes of conjuncts leaves little space
for a plain assumption of either one of the existing approaches. Every existing approach
so far is based on the assumption that all determiner sharing conjuncts are large or small.
They do not consider the combination of the two types of conjuncts and the two types of
processes, movement and deletion, in one single approach. In the next section, I explain
the steps I follow in accounting for Spanish determiner sharing data considering the given
scenario.
3.4 Previous determiner sharing analyses
3.4.1 Introduction
It is clear from the previous section that the Spanish determiner sharing paradigm com-
prises the complexity of the gapping paradigm plus some added device to account for
determiner sharing that I introduce in this section. Consequently, my analysis of de-
terminer sharing in Spanish needs to capture the distinctions between large and small
conjuncts and movement vs. deletion, and their interactions with the process of deter-
miner sharing.
Just as in gapping, existing determiner sharing analyses can be divided according to
their usage of large or small conjuncts. Ackema and Szendroi (2002) is a large conjunct
analysis of determiner sharing that only considers large conjuncts as possible elements
of coordination for every determiner sharing sentence. The DP Split theories so far –
Johnson (2000), Lin (2002), Arregi and Centeno (2005)– or Citko (2007) consider that
small conjuncts are the only means by which sharing can be explained. In broad terms,
the difference between both approaches is sustained on the same arguments that are used
in large and small conjunct approaches to gapping. Also similarly to the gapping scenario,
none of the existing approaches accounts for the Spanish data which has been provided
in this work. Logically, one possible next step that could be considered is to investigate
some adaptation of any of the existing theories of determiner sharing into the empirical
data which I introduce here. I present the mechanisms of these analyses and I explain
122 CHAPTER 3. DETERMINER SHARING
whether any plausible adaptation would be suitable or not for each analysis. I start with
Ackema and Szendroi (2002).
3.4.2 Ackema and Szendroi (2002)
3.4.2.1 Summary of the analysis
This is a large conjunct analysis to determiner sharing: the authors argue that conjoined
elements in determiner sharing structures are always sentential. Ackema and Szendroi
account for determiner sharing by means of a process called ‘dependent ellipsis’. I explain
details of this process and their analysis dividing it into three main assumptions which
are central to their work.
(i) Ackema and Szendroi (2002) assume Williams’s (1997) theory of coordination:
coordinate phrases are double headed. For instance, a phrase such as IP in a coordination
context has two heads I and I (286).
(286) [IP [I′ [ I] [ I] ] ]
Consider (287, 288) for further illustration. In (288), the coordinate phrase under which
two IPs are conjoined, like in (287), is labeled [I,I]P. This means that it is a phrase with
two I heads (288).
(287) I think that John will eat meat and Mary will drink wine.
(288) [I,I]P → . . . [[I,I]P [IP John . . . ] and [IP Mary . . . ] ]
(ii) On the basis of this structure, Ackema and Szendroi (2002) make a further assump-
tion in preparation for their analysis of determiner sharing. They account for gapping
sentences assuming a device labeled ‘coordinate ellipsis’, which is also found in Williams
(1997). In ‘coordinate ellipsis’ one of the heads in a coordinate phrase affected by gapping
is a 0 which is a head that has neither phonological material nor syntactic features. The
gap of sentence (289), for instance, is accounted for by the phrase in (290).
(289) I think that John will eat meat and Mary 0 drink wine.
(290) [I,0]P = IP and 0P
The phrase in (290), [I,0]P, is a double headed phrase where one of the heads is 0. This
produces coordination of an IP and a 0P, as I represent there. This 0 head accounts for
the gap created by the omission of ‘will’.
(iii) Ackema and Szendroi (2002) take the fact that determiner sharing is dependent
upon gapping as the basis for labeling the process responsible for determiner sharing
3.4. PREVIOUS DETERMINER SHARING ANALYSES 123
‘dependent ellipsis’. For the simplest cases of dependent ellipsis, they propose two 0
heads. The 0 head in coordinate ellipsis –which is the regular gapping one– allows the
head of a dependent phrase to be also 0. Consider:
• One 0 head for coordinate ellipsis in the gapped phrase.
• One 0 head for dependent ellipsis in the dependent phrase.
This dependent phrase, which has its 0 head, can also become 0P and have other overt
material. The result of the two processes put together, coordinate ellipsis and dependent
ellipsis, is that non-initial conjuncts in an English coordinate sentence may end up having
multiple 0 heads that are anaphoric on elements in initial conjuncts. Consider the analysis
of an English determiner sentence (291) in (292) for illustration.
(291) The girls will drink whiskey and boys drink wine.
(292) [T,0]P
TP 0P
DP 0’
0 VP
drink
wine
D
0
NP
boys
DP t’
T
will
VP
drink
whiskey
D
the
NP
girls
Ackema and Szendroi (2002) do not specify many details about the exact meaning of
‘dependent’ in ‘dependent ellipsis’. They only specify that phrases that are affected by
‘dependent ellipsis’ are those that are ‘directly dependent’ on the phrase that undergoes
coordinate ellipsis (Ackema and Szendroi (2002): p. 1, 7). For subject determiner sharing,
they state “0 head licenses heads of its other dependents, like its specifier to be 0” (Ackema
and Szendroi (2002): p. 7). For determiner sharing in object position, the details in
their analysis are explained in the following way: “After all, complements should be
possible targets of dependent ellipsis” (Ackema and Szendroi (2002): p. 15). Hence,
dependent ellipsis in their analysis sometimes targets the specifier of a 0P and sometimes
the complement of such 0P. It can be inferred that the needed relationship between the
initial 0P and the one affected by dependent ellipsis should happen inside the initial 0P.
However, the direction of the dependent ellipsis process is not unique. It can both affect
phrases upwards and downwards with respect to the initial 0 head.
124 CHAPTER 3. DETERMINER SHARING
3.4.2.2 Problems of application to Spanish
Once I have provided the main points in Ackema and Szendroi (2002), let us see whether
it could be applied to Spanish determiner sharing or not. A way to make an adapta-
tion of this theory would be to make their idea on dependent 0 heads applicable to an
analysis where both large and small conjuncts were present. To check if this is possi-
ble I make a rational assumption on determiner sharing sentences. Because determiner
sharing in Spanish behaves similarly to gapping according to the different gapping types
that I list, I take as basis my approach on gapping: in order to provide each determiner
sharing analysis, I base myself on the corresponding gapping analysis for such type of
‘gapping/determiner sharing’.
Following this line of thinking, in order to check whether the adaptation of Ackema and
Szendroi (2002) could work, I need to find a 0 head in every gapping structure. This would
enable me to postulate the second 0 head that accounts for determiner sharing and that
could be dependent on the first 0 head. A major problem arises. When checking one by
one each of the structures, there is at least one structure upon which this adaptation could
not be applied because there is no 0 head. This is the one that corresponds to ‘Gapping
of Auxiliary’, which is the basis for the analysis of ‘Determiner sharing in Gapping of
Auxiliary’. Consider the ‘Gapping of Auxiliary’ structure (294) that is provided in chapter
2, section 2.4.4.3, for sentence (293).
(293) JorgeJorge
hahas
limpiadocleaned
elthe
banobathroom
yand
PedroPedro
[ ][ ]
fregadowashed
elthe
suelofloor
‘Jorge has cleaned the bathroom and Pedro has washed the floor.’
(294) [TP Jorgei ha [vP [vP ti limpiado el bano] y [vP Pedro fregado el suelo] ] ]
I propose that in this structure (294), the Auxiliary is generated in T and that the lexical
Verbs occupy v respectively in their conjuncts. In this structure (294), there is no 0 head
upon which any other 0 head could be dependent.
Notice that the same applies to ‘Determiner sharing in Gapping of Modal’. I base the
analysis of ‘Determiner sharing in Gapping of Modal’ on the structure that I provide for
‘Gapping of Modal’ in chapter 2, section 2.4.4.4. The analysis of Ackema and Szendroi
(2002) could not be adapted either to such cases because, as can be seen in chapter 2,
section 2.4.4.4, the modal Verb is located in T, the same way the auxiliary does in (294)
above. This leaves no space for a 0 head upon which any other 0 head could be depen-
dent. This is another instance that accounting for determiner sharing in these sentences
following an adaptation of Ackema and Szendroi (2002) is impossible.
3.4. PREVIOUS DETERMINER SHARING ANALYSES 125
One of the major aspects in this dissertation is that in order to analyze either gapping
or determiner sharing in Spanish, conjuncts need to be both small and large. The analysis
in Ackema and Szendroi (2002) considers conjuncts that can only be of a large size. This is
another drawback for any potential adaptation into Spanish data of the analysis provided
by Ackema and Szendroi (2002) because both sizes need to be taken into consideration
when analyzing Spanish gapping and determiner sharing.
3.4.3 Citko (2007)
3.4.3.1 Summary of the analysis
Citko (2007) is a small conjunct analysis to determiner sharing applied to English and
Polish. Citko’s two main themes are the usage of small conjuncts and the proposal
that the determiner that is missing in the conjuncts that are also affected by gapping
is literally shared. Her claim is that this determiner is actually shared by NPs that
are in different conjuncts, in different planes. Citko assumes the multiplanar version
of coordinate structures where a given node in a structure can simultaneously dominate
nodes that are each in a different plane.3 The mechanism by which this determiner gets to
be shared by the NPs is a process called ‘parallel merge’ by which the determiner merges
at the same time to these NPs that are in different planes. Example (295) is a structure
that represents ‘Parallel Merge’ in determiner sharing. It allows for the determiner to
combine with α and β, the NPs that share D, and creates multi-rooted, multi-dominant
objects. Consider (295).
(295) Parallel Merge
DP DP
D βα
The proposal is that parallel merge can only take place if the shared element moves overtly
later in the derivation. For this, Citko assumes Moro (1997, 2000) dynamic approach to
antisymmetry where Kayne’s (1994) antisymmetry is assumed but where symmetric struc-
tures are allowed as long as a later overt movement breaks this symmetry. Citko uses this
in order to account for linearization in the parallel merge structure.
3Without deepening into details, conjuncts constitute the different planes in approaches of this type,while the shared part remains one plane. See Moltmann (1992a), Moltmann (1992b), Muadz (1991), orGoodall (1987), among others, for details.
126 CHAPTER 3. DETERMINER SHARING
A specific example that shows the mechanism of Parallel Merge in determiner sharing
is (296). This is the structure of the derivation at a moment by which multidominance
has affected both the DPs ‘few dogs’ and ‘few cats’ and the Verb ‘eat’.
(296) T’
T CoP
vP Co’
Co vP
DP v’
v VP
Alpo
DP v’
v VP
eat Whiskas cats
few dogs
3.4.3.2 Problems of application to Spanish
In order to explain the specifics of this structure, I focus on what is crucial to determiner
sharing structures, which is what Citko also explains. As the verb ‘eat’ and the determiner
‘few’ are shared elements by the conjuncts, they need to move following the dynamic
approach to antisymmetry so that linearization is possible. They move to T and Spec,
TP respectively in subsequent steps of the derivation, which I illustrate in (297).
(297) [TP few dogs[T′ eat [CoP [vP Whiskas] [Co′ [vP cats Alpo] ] ] ] ]
Citko’s analysis is suitable for Polish. Her main argument to propose a literally shared
Determiner is based on agreement. In Polish, in order to have determiner sharing, the
Determiner and the NPs that enter into the determiner sharing relationship need to all
have the same agreement features. Consider (298, 299) which illustrates it with gender
features.4
(298) FidoFido
zjadlate
duzomuch
AlpoAlpo
aand
WhiskersWhiskers
[ ][ ]
Whiskas.Whiskas
‘Fido ate too much Alpo and Whiskers ate too much Whiskas’. (Polish)
(299) FidoFido
zobaczy lsaw
*ta/*tegothis.FEM/MASC
kotkecat.FEM
aand
WhiskersWhiskers
psadog.MASC
4Citko does not provide any gender specifications for example (298) in the version of her article thatI am using as reference. I assume both ‘Alpo’ and ‘Whiskas’ have the same features in Polish.
3.4. PREVIOUS DETERMINER SHARING ANALYSES 127
‘Fido saw this cat and Whiskers saw this dog.’ (P. 24, Polish)
In Spanish, that is not the case. Agreement among the D and the NPs can be via same
features in every element that enters the determiner sharing relationship or it can be like
in unbalanced coordination, where features do not match (300).
(300) Nineither
muchosmany.MASC.PL
chicosboys.MASC.PL
hanhave
leıdoread
losthe
librosbooks
ninor
[ ]D[ ]D
chicasgirls.FEM.PL
[ ]T[ ]T
revisadoreviewed
losthe
artıculos.articles
‘Neither many boys have read the books nor many girls have reviewed the articles.’
Hence, Citko’s analysis specifically works for Polish, but not for Spanish. In addition,
notice that Citko has claimed that object determiner sharing is possible in Polish without
further specification. Such an analysis is not suitable for Spanish. As we know from the
previous section, object determiner sharing is not always possible in Spanish. Citko’s
analysis cannot capture these facts. There is a further aspect that needs to be considered
in relation to Citko’s analysis. One of the semantic features of the Determiner in deter-
miner sharing is that its meaning is distributive. Though the literature on determiner
sharing does not provide many details about such fact, the distributivity of Determiners
in determiner sharing contexts is acknowledged both in McCawley (1993:248-249) and in
Lin (2002:102-103). Lin expresses this as follows: “determiner sharing normally results in
a distributed interpretation; that is, missing determiners are usually interpreted as if they
were present in each conjunct”.5 It is difficult to evaluate whether Citko’s analysis can
actually derive the obligatory distributive reading of Determiners in determiner sharing
examples, without any further discussion of the semantics of multidomination structures
(Citko does not provide either such type of discussion.). I continue with other alternatives.
3.4.4 Johnson (2000), Lin (2002) via Sportiche’s DP-Split
Central pieces of work in the small conjunct approach to determiner sharing are John-
son (2000), Lin (2002) and Sportiche (1996, 2005). Johnson (2000) explains determiner
sharing phenomena by assigning two different structural parts to the determiner. The
proposal for this division is related to the fact that the determiners that he analyzes ‘few,
no’ in, for instance, the Whiskas sentence (301) or a ‘no’ sentence (302), have meanings
which can be decomposed into two parts.
(301) Few dogs eat Whiskers and [ ]D cats [ ]V Alpo.
(302) No representative voted for the proposition or [ ]D senator [ ]V against it.
5Lin clarifies in a footnote that “the interpretation of shared negative determiners across disjunctsmay be an exception to this observation”.
128 CHAPTER 3. DETERMINER SHARING
Structure wise, Johnson proposes to analyze these Determiners, ‘few’ and ‘no’, each
as the sum of two projections, an AdvP and a DP, which correspond to the two meaning
parts that he finds in these Determiners:
- ‘Few’: sentence negator ‘not’ (in AdvP) + indefinite ‘any’ (in DP)
- ‘No’: sentence negator ‘not’ (in AdvP) + indefinite ‘any’ (in DP)
Johnson proposes that the AdvP is above the DP part and ensures their relation by means
of two constraints that the two parts need to meet. Consider these constraints applied to
‘few’ (303).
(303) a. φ must be within the c-command of few at LF.
b. Few must be adjoined to a DP headed by φ by Spell Out.
In these constraints (303) φ represents the second part of the Determiner which is below
the AdvP part in the structure. These ensure that each part always appears dependent
upon the other and that “few is an adverb that has the unusual requirement of surfacing
attached to a DP”.
Lin (2002) follows Johnson’s (2000) idea of assigning two parts to the determiner in
determiner sharing constructions but finds inspiration in Sportiche’s (1996, 2005) work
in order to propose DET positions as one of the two determiner related positions. An
important part of Sportiche’s work –Sportiche (1996, 1997, 2005)– is focused on the neces-
sity of splitting the DP. He has used different labels for this: DP-Partitioning Hypothesis
(Sportiche (1996)) or Split DP structure (Sportiche (2005)) in later work. The main fea-
ture of Sportiche’s DP Split proposal is that the D part of a DP is generated separately
from the NP.
In Sportiche (1996), he provides arguments that regular DP properties associated to
Ds like referential, quantificational, or numeral properties are not selected as part of the
arguments of the verb. Because of this, he suggests that the D head, which is the locus
for all those properties, is generated above the verbal predicate and hence, that Ds and
NPs are not part of the same constituent. In Sportiche (2005), the reasons why Sportiche
proposes this configuration are within a major reasoning that concerns movement, recon-
struction, and scope phenomena of different kinds. The DP Split proposal is just one part
of the solving strategy that targets a detailed picture of such phenomena.6
The structure in (305), which is represented through steps (a-c) illustrates the parti-
tioning between D and NP. The sentence that can be used as referent is (304).
6For details about these arguments and approaches, the reader is referred to Sportiche (1996, 2005).
3.4. PREVIOUS DETERMINER SHARING ANALYSES 129
(304) The man walked many paths.
(305) a. D
the vP
NP
man
v’
v
walked
D
many VP
V
tV
NP
paths
b. D
the mani vP
NP
ti
v’
v
walked
D
many pathsj VP
V
tV
NP
tj
c. [TP the manx[T′ [D tx [vP [NP ti] walked [D many pathsj [VP tV [NP tj] ] ] ] ] ] ]
Both the subject NP ‘man’ and the object NP ‘paths’ are generated in argument posi-
tions. These NPs are generated separately from their determiners ‘the’ and ‘many’ which
are in the D positions. The determiner associated with the subject NP ‘the’ is generated
above vP and the one associated with the object NP ‘many’ is generated above VP. Later
in the derivation, ‘the man’ moves to Spec, TP.
In Lin (2002), determiner related positions that are generated separately from the
arguments are just DETs, which represent some part of the determiner whose features
are not clearly specified there. Contrary to Sportiche, the arguments of Verbs in Lin’s
130 CHAPTER 3. DETERMINER SHARING
theory are DPs. Lin proposes two DET positions in the structure: one above vP for
subjects and one above VP for objects. Locations are the same as in Sportiche, though.
This is represented in (306).
(306) DET
DET vP
DP-Sbj v’
v DET
DET VP
V DP-Obj
Following Johnson’s (2000) constraints, Lin imposes two requirements (307a, b) and a
further condition (307c) that need to be held between DETs and Ds in order to restrict
the association between the two:
(307) a. D must be within the c-command domain of DET at LF.
b. DET must be adjoined to a DP headed by D by Spell-Out.
c. D needs to be adjoined to DET to be spelled-out.
Lin’s analysis of sentence (304) is represented in (308) through steps (308a-c), where I
provide relevant details about the slight changes in this analysis with respect to Sportiche’s
one (305). In (308), requirements in (307) are followed.
(308) a. DET
DET vP
DP
the man
v’
v
walked
DET
DET VP
V
tV
DP
many paths
3.4. PREVIOUS DETERMINER SHARING ANALYSES 131
b. DET
DET
the mani
vP
DP
ti
v’
v
walked
DET
DET
many pathsj
VP
V
tV
DP
tj
c. TP
The mani T’
walked DET
DET
ti
vP
DP
ti
v’
v
tV
DET
DET
many pathsj
VP
V
tV
DP
tj
In this case, both arguments are generated as DPs in subject and object positions (308a).
The object DP ‘many paths’ is situated inside VP as complement of the verb and its cor-
respondent DET position c-commands it. The subject DP ‘the man’ gets c-commanded
at Spec, vP position by the higher DET position. In this respect, the sentence structure
follows principle (307a). Both DPs make a movement towards each DET so that the
determiner in both the subject and the object DPs gets spelled-out (308b). Principles
(307b) and (307c) are also followed. Afterwards, the further movement of the subject up
132 CHAPTER 3. DETERMINER SHARING
to Spec, TP is finally done in the derivation (308c).
Notice also that the generalization that determiner sharing is dependent upon T gap-
ping is captured by the structure proposed by Lin in the following way. As all DETs are
situated either right above vP or right above VP and as DET is already in the shared part,
T needs to be in the shared part too, because it is above DET in all the cases –higher in
the structure than vP or VP– and hence, T is shared. Thus, so far, Lin’s (2002) proposal is
the most suitable one because it accounts for determiner sharing with all Ds; and because
it can also account for the semantics of these Determiners in determiner sharing contexts.
However, this proposal lacks some specifications that are needed in the consideration not
only of Spanish determiner sharing but also of the major spectrum where other regular
sentences –coordinate and non-coordinate– are also included.
3.4.5 Interim conclusion
To summarize, the analysis in Ackema and Szendroi (2002) cannot account for all deter-
miner sharing sentences in Spanish. In order to have dependent ellipsis, an initial 0 head
is always needed in the structure and at least one case in Spanish can be found where
the initial 0 head is not present. Citko’s (2007) analysis also lacks the potential for an
adaptation onto Spanish determiner sharing because its arguments are based on Polish
phenomena that are not part of the Spanish language. Finally, in general terms, the
DP-Split proposals are the most suitable because they are more flexible with respect to
Determiners and more accurate regarding Ds’ semantics. This is why, in the next section,
I adapt the DP-Split proposal into Spanish determiner sharing. This adaptation com-
prises (i) the accounts for the different types of determiner sharing that exist in Spanish,
(ii) the necessary changes that are needed for this adaptation onto Spanish, (iii) and the
necessary changes which the determiner sharing theory needs in order for it to be correctly
set in the more general frame of both conjoined and non conjoined sentences.
3.5 Spanish determiner sharing analysis
3.5.1 Introduction
At this point, the determiner sharing paradigm has been completed and previous analyses
have also been explained and scrutinized in their relation to Spanish determiner sharing.
According to the conclusions in the previous section, I assume Lin (2002) in proposing
DET positions but I propose some changes. I also assume the gapping classification done
in the previous chapter and I analyze each type of determiner sharing attending to the
3.5. SPANISH DETERMINER SHARING ANALYSIS 133
division between subject and object determiner sharing. I start from determiner sharing
in subject position and I gradually move to object determiner sharing cases analyzing
them case by case.
3.5.2 Proposal
The theory I propose here has different parts. The two principal assumptions are: (i)
the gapping classification and its analysis as I propose in the previous chapter and (ii)
a theory with DET positions. I make further specifications in what follows particularly
related to assumption (ii).
A theory with DET positions is preferred as stated before and this is why I assume it.
However, this theory so far lacks some specifications necessary to integrate it with the
wider framework of syntactic theory. Integrating it with the wider framework is useful
in order to consider not only conjoined sentences but also non conjoined sentences, and
not only determiner sharing sentences but also non determiner sharing sentences. I will
expose these necessary specifications in the following order. First, I introduce the discus-
sion on the conditions that need to hold between DETs and DPs in page 133. Within
that discussion I explain the need for the phrase projected by DET, DETP, in page 133.
Second, I move onto explaining how many DET positions I propose exist in the structure
and their locations. This starts on page 134. Third, on page 137, I continue with how I
account for the optionality of determiner sharing, which is a discussion that is very much
related to the location of DET positions.
Lin’s conditions upon the relation between DET and DPs are based on Johnson’s (2000).
Lin’s conditions are specified in (307) but I repeat them here for easy reference in (309).
(309) a. D must be within the c-command domain of DET at LF.
b. DET must be adjoined to a DP headed by D by Spell-Out.
c. D needs to be adjoined to DET to be spelled-out.
They maintain the shape in Johnson’s work conditions and include Sportiche’s DP-Split
proposal. One of the goals of these conditions is to describe how movements should be
from D to DET, however, these conditions are vague in this description. I revisit them sub-
sequently in order to make them more specific. First, I propose that the DET-Determiner
relationship is a Spec, Head relationship. This gets reflected in the type of movement
that the DP undergoes and the type of projection which hosts the DET position. The
projection is DET Phrase and the DP moves to Spec, DETP.
134 CHAPTER 3. DETERMINER SHARING
Additionally, a closer look at these conditions leads me to conclude that it is desirable
to simplify them. Consider the interaction of two of the conditions (309a, 309b) which Lin
(2002) proposes. Condition (309a) states that D must be within the c-command domain
of DET at LF and condition (309b) states that DET must be adjoined to a DP headed
by D by Spell-Out. If by spell-out, the DP needs to move to DET by condition (309b), it
is not possible that, at LF, this DP remains in a position that is c-commanded by DET,
which is what condition (309a) states. The reason for this is that by LF it has already
moved to this same DET by the (309b) condition.
Condition (309a) is thus incompatible with the other conditions and unnecessary. DET
is above D simply because determiner sharing sentences are structured this way. Hence,
I propose to have conditions (309b) and (309c) as references.
These are the conditions that I propose in (310) based on the aspects that I have just
mentioned.
(310) a. The Spec in a DET projection must be occupied by a DP headed by D by
Spell-Out.
b. D needs to move to the Spec in a DET projection to be spelled-out.
Before I illustrate the way these conditions are implemented with specific determiner
sharing examples, revision of other factors needs to be carried out. Lin (2002) does not
elaborate on the optional condition of determiner sharing in either conjoined or regular
sentences. Consider for instance these two examples (311, 312).
(311) Muchosmany
chicosboys
comeneat
pastelpie
deof
manzanaapple
yand
[ ][ ]
chicasgirls
[ ][ ]
heladoice-cream
deof
vainilla.vanilla‘Many boys eat apple pie and many girls eat vanilla ice-cream.’
(312) Muchosmany
chicosboys
comeneat
pastelpie
deof
manzanaapple
yand
muchasmany
chicasgirls
[ ][ ]
heladoice-cream
deof
vainilla.vanilla‘Many boys eat apple pie and many girls eat vanilla ice-cream.’
The Determiner muchos ‘many’ in the second conjunct can both be shared (311) or not
shared like in a regular gapping example (312).
In order to target such question in this work, I investigate different hypotheses that
are related to the number of DET positions which are needed and the location or locations
of such DET projections. I illustrate these hypotheses in what follows and I explain the
3.5. SPANISH DETERMINER SHARING ANALYSIS 135
reasoning behind the existence of each hypothesis. First, I introduce them in a simplified
way. Afterwards, I explain them in detail and analyze their consequences. I conclude that
Hypothesis 3 is the correct one.
• Hypothesis 1: There is one DET projection that only appears in coordination con-
texts when determiner sharing is present. The logic behind proposing such a scenario
would be: since determiner sharing can only happen in gapping contexts, and gap-
ping contexts can only happen in coordination, a DET projection, which is what
accounts for determiner sharing can only be present when coordination is present
too. Non determiner sharing contexts are accounted for without DETs.
• Hypothesis 2: There are various DET projections whose positioning is fixed (always
the same). The concept is to think that each DET projection corresponds to each
DP. Determiner sharing is accounted for because of the relative position of such pro-
jections with respect to coordination. When coordination happens below the DET
projection, there is determiner sharing. When coordination happens above coordi-
nation, there is no determiner sharing because DETs are inside their conjuncts, one
per conjunct.
• Hypothesis 3: There are various DET projections whose position is not completely
fixed. There is one DET projection per each DP but the position of the DET can
slightly vary within the limits of milestones in the structure: phases. When there is
no determiner sharing, DET projections happen at the level of DP phases. When
determiner sharing is present, the DET projection is at the level of higher phases.
Hypothesis 1 differs from Hypotheses 2 and 3 in the number of DET projections that
are proposed. So, this is the first step I take: I determine whether or not it is one DET
that is needed. If one thinks of determiner sharing and the fact that it only happens in
gapping, and hence, in conjunction, one could think that it is one DET position which
gets generated only in conjoined contexts that accounts for determiner sharing. In such a
situation, non determiner sharing contexts are accounted for without any DET. I illustrate
these two possibilities in two partial structures (313, 314).
(313) Determiner sharing: DET
DET CoP
vP Co’
Co vP
136 CHAPTER 3. DETERMINER SHARING
(314) No determiner sharing: CoP
vP Co’
Co vP
However, such a solution would be undesirable from a theoretical point of view: taking
such step would be too construction specific in that I would be claiming that a functional
projection only arises in the very specific cases where not only conjunction but also de-
terminer sharing appears.
Hypotheses 2 and 3 are considered in this next step. Hypothesis 2 differs from Hy-
pothesis 3 in that positions of DETs in Hypothesis 2 are fixed. Let us then investigate
the consequences of having fixed positioning of DETs. If DET projections are proposed
to be at the same height in the structure at all times, optionality of determiner sharing is
accounted for according to the relative height of coordination. When determiner sharing
is present in the sentence, the DET projection is generated above conjunction. When
determiner sharing is not present, conjunction is above DET projections. The two possi-
bilities in Hypothesis 2 are illustrated in the partial structures (315, 316) which consider
the potential positions Lin (2002) does.
(315) Determiner sharing: TP
. . . DET
DET CoP
vP Co’
Co vP
TP
. . . vP
. . . DET
DET CoP
VP Co’
Co VP
3.5. SPANISH DETERMINER SHARING ANALYSIS 137
(316) No determiner sharing: TP
. . . CoP
vP Co’
Co vP
DET . . .
DET . . .
TP
. . . vP
. . . CoP
VP Co’
Co VP
DET . . .
DET . . .
I illustrate in structures (315, 316) that the optional character of determiner sharing de-
pends upon the relative height of CoP with respect to DETP in this Hypothesis 2.
This way of setting elements in the structure fails to explain some features of deter-
miner sharing. I explain in the next chapter that some determiner sharing data which are
related to questions and wh-movement require that these DET positions do not occupy
fixed locations. I show there that DETs need to be right above phases –either vP or CP–
but with enough flexibility to leave enough space for the amount of information structure
needed in each phase.
Hypothesis 3 works. I propose that there are various DET positions, one per DP, and
that the positioning of these DETs must not be completely rigid. I have also explained,
making reference to the next chapter, that phases are needed in the explanation of where
DETs should be. Hence, there are three factors that need to be taken into account: (i)
both possibilities of having and not having determiner sharing in the sentence, (ii) the
non rigid character of DETs’ positioning, and (iii) the notion of phases.
The optionality of determiner sharing can be nicely captured if all types of phrases
that have been labeled phases are taken into account. In the literature, the different
types of phases are: DPs, vPs, and CPs. And considering DPs as phases here is a plus
since a logical relationship between DETs and DP phases –and hence phases in general–
is established from the beginning, from the core where Determiners are generated. The
proposal is to take into consideration DP, vP, and CP phases from small to large with
respect to DET’s location. I propose that DETPs are generated as close as possible to
the DP phase by default, the smallest of phases. Notice that the need for full DETPs
in this proposal has already been explained at the beginning of this section on page 133.
138 CHAPTER 3. DETERMINER SHARING
Consider a case where no determiner sharing is present, like (312) above, that I repeat
here as (317), which can be considered the default contexts for DPs.
(317) Muchosmany
chicosboys
comeneat
pastelpie
deof
manzanaapple
yand
muchasmany
chicasgirls
[ ][ ]
heladoice-cream
deof
vainilla.vanilla‘Many boys eat apple pie and many girls eat vanilla ice-cream.’
Because DETP needs to be right above phases, in this kind of example (317) where
determiner sharing is not present, DETP is generated above the DP phases. I propose
that the DETP is generated as close as possible to the DP phase by default. A very
simplified way of representing such a structure would be the following (318). Again,
although this could correspond to (317), I leave details for later because many other
details need to be taken into account.
(318) No determiner sharing:
TP
. . . CoP
vP Co’
CovP
DETP(Subj) v’
v VP
V DETP(Obj)
DET’
DET DP
DET’
DET DP
DETP(Subj) v’
v VP
V DETP(Obj)
DET’
DET DP
DET’
DET DP
With this kind of structure (318), the phases notion is taken into account and non deter-
miner sharing cases can be easily explained: DP-phases and their corresponding DETPs
are present in both conjuncts.
How does the proposal in Hypothesis 3 account for determiner sharing? The answer:
by means of the notion of having DETP whose positioning is not fixed. I propose DETPs
that can ‘float’ around within the limits of phases in the sense that they can either be
generated at the level of the smallest phase, DP, or they can ‘float’ up to be generated at
3.5. SPANISH DETERMINER SHARING ANALYSIS 139
the level of the next larger phase, vP. And if needed, DETP can float all the way up to
the largest of phases, CP. In a way, the type of DETP that I propose is similar to other
functional projections that are claimed exist within what is normally called ‘left periph-
ery’. Notice that the behavior of such projections in the left periphery, like TopP or FocP,
also appear in the structure when needed, like I claim happens with DETP. For instance,
in determiner sharing cases, which are the cases where the notion that DETP position-
ing is not rigid becomes useful, at the DPs affected by determiner sharing, generation of
DETP does not take place. DETP, in such cases, ‘floats up’ to be generated at the vP
phase level. It is in this sense that I claim locations of DETPs are not rigid. In addition,
DETPs also resemble the functional projections that appear in the left periphery in that
DETPs are also claimed to potentially appear at the edges of all phases (Butler (2004)).
In a determiner sharing sentence, then, DETP is generated at the vP phase level. Coordi-
nation affects vPs and DETP stays above it enabling the determiner sharing configuration.
A determiner sharing theory with various DETPs whose location is not fixed, like in
Hypothesis 3, easily accounts for determiner sharing cases and non determiner sharing
cases. It can also account for those aspects that are problematical for Hypothesis 2. I
list all these aspects next further explaining those that have not been developed, and
explaining the ways in which Hypothesis 3 accounts for the aspects that otherwise are
unexplained.
• The proposal accounts for determiner sharing cases and for non determiner sharing
cases. Hence, the optionality of such phenomenon is accounted for and explained.
• With the initial association that is made between DETPs and DP phases, the re-
lation between DETPs and phases gets to be established. Establishing this rela-
tionship justifies the proposal that DETPs generate above phases, which is useful
because it provides an explanation for the locations where DETP can appear. A
consequence of this is that within this proposal the impossibility of having DETP
above TP is explained and not stipulated –as was needed in Hypothesis 2–. DETP
cannot be situated right above TP because TP is not a phase.
• The proposal that DETPs can be at the level of all phases with the flexibility of them
appearing when needed also avoids the problem that existed within Hypothesis 3
related to the associations between DETPs and DPs. Recall that the problem was
that with a fixed DET projection, the fixed DETP needed to be above conjunction
because it was shared. This brought the drawback that with the word order that
arises in every sentence, the association between DET and the DPs entering the
determiner sharing relationship was problematical. Those problems are avoided
140 CHAPTER 3. DETERMINER SHARING
with the proposal that DETP potentially appear at every phase and that they
appear if needed. Imagine an object determiner sharing context where, like in other
examples, the word order of the second conjunct is subject-object. I schematize
such a situation within the analysis here in (319). Again, I consider such a situation
in general terms because real examples do have further complications that I need
to target later for the sake of exposition.
(319) . . . DETP
DETP
for Obj DP
CoP
vP Co’
Co vP
DETP v’
vVP
. . . Obj DP
DETP
for Sbj DPSbj
DETP v’
vVP
. . . Obj DP
DETP
for Sbj DPSbj
This proposal (319) explains the association between Ds and DETs in object deter-
miner sharing. I have explained that when Ds do not get into a determiner sharing
relationship, DETP generated at the DP-phase level. In (319), the Determiners in
the subject DPs are associated with the DETPs that are situated right at the level
of the DP-phases. These DETPs that are at the DP-phase level are the nearest to
the subject DPs. Because the latter association is done this way, the association
that is done at the level of object DPs is easy to explain. It works like regular
determiner sharing, with no relevant intervening element –because subject Ds are
already associated with their DETs–. Because object DPs enter the determiner
sharing relationship, their DET gets generated at the level of the highest phase and
not right above the DP-phase. The next higher phase is vP; conjunction is at that
level; and DET generates above conjunction. DETP is in the shared part and it is
only the object Determiner in the first conjunct that moves to Spec, DETP and the
only one that can be spelled out. The Determiner in the second conjunct object DP
is not spelled out because it cannot move to Spec, DETP.
3.5. SPANISH DETERMINER SHARING ANALYSIS 141
Another strength of this proposal is that the devices used here –in Hypothesis 3– are
also useful in accounting for wh-movement phenomena which are related to determiner
sharing and which I present, explain, and account for in the next chapter. Also, though
on a more theoretical basis, proposing the availability of this DETP projection at the
level of every phase adds to the parallelism that is frequently claimed to exist among
the different phases: DP, vP, CP. The adding factor to this parallelism comes from the
fact that a Determiner related position which is more on the nominal side of categories is
proposed to also be spread across phases, vP and CP, that are related to the verbal side
of categories. Discussion on phases is further extended in the next chapter. Now, I focus
on analyzing determiner sharing in detail. I explain all the determiner sharing types that
I have presented in declarative contexts in section 2 of this chapter.
3.5.3 Analysis: subject position
I show in this section that my proposal, which is detailed in section 3.5.2, easily accounts
for the most basic cases of determiner sharing in subject position in Spanish. Before
doing that, however, I illustrate the fact that I also account for non determiner sharing
sentences easily. I provide the analysis of sentence (320) in structure (321).
(320) Muchosmany
chicosboys
comeneat
pastelpie
deof
manzanaapple
yand
muchasmany
chicasgirls
[ ][ ]
heladoice-cream
deof
vainilla.vanilla‘Many boys eat apple pie and many girls eat vanilla ice-cream.’
142 CHAPTER 3. DETERMINER SHARING
(321) TP
Muchos
chicosi
T’
comen CoP
vP Co’
CovP
DETP v’
tV VP
tV helado de
vainilla
muchas
chicasj
DET’
DET tj
DETP v’
tV VP
tV pastel de
manzana
ti DET’
DET ti
The main aspects to understand in this structure (321) are related to the relationship
between Ds and DETPs. Because determiner sharing is not present in the structure,
DETPs generate at the level of DP phases, below coordination. This is the default
scenario. Both Determiners in the DP subjects, muchos chicos, muchas chicas ‘many
boys, many girls’, move to Spec, DETPs and hence both Ds get to be spelled-out. The
structure is a regular basic gapping example that contains small conjuncts and a shared
part on top.7
Determiner Sharing in Basic Gapping: Analysis
The first subject determiner sharing sentence is within ‘Basic Gapping’ (322). I provide
its structure in (323).
(322) Muchosmany
chicosboys
comeneat
pastelpie
deof
manzanaapple
yand
[ ][ ]
chicasgirls
[ ][ ]
heladoice-cream
deof
vainilla.vanilla‘Many boys eat apple pie and many girls eat vanilla ice-cream.’
7In the interest of conciseness, I do not provide any further derivations for the variants withoutdeterminer sharing. In any event, the analysis would always be the same, i.e., DETP would appear atthe DP level, below coordination.
3.5. SPANISH DETERMINER SHARING ANALYSIS 143
(323) TP
muchos
chicosi
T’
comen DETP
ti DET’
DET CoP
vP Co’
Co
y
vP
(-)chicas v’
tV VP
tV helado de
vainilla
ti v’
tV VP
tV pastel de
manzana
Just like in the gapping counterpart, small conjuncts are conjoined in (323). In this
case the Verb ATB-moves from the conjuncts to the shared part and the DET position
enables the determiner in the DP in the first conjunct muchos chicos ‘many boys’ to be
spelled out. This is done via movement of this DP from the specifier of vP to the spec-
ifier of DETP. Because the subject in the second conjunct cannot move to Spec, DET,
its Determiner fails to be spelled out. Afterwards, once the spelling out of the complete
subject DP in the first conjunct is guaranteed, this DP moves up to the specifier of TP
because of the EPP feature.
Determiner Sharing in Gapping of Auxiliary: Analysis
The next subject determiner sharing example is within ‘Gapping of Auxiliary’. This
again is a conjunction of small conjuncts. Consider the structure of (324) in (325).
144 CHAPTER 3. DETERMINER SHARING
(324) Muchosmany
chicosboys
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
[ ][ ]
chicasgirls
[ ][ ]
degustadotasted
heladoice-cream
deof
vainilla.vanilla
‘Many boys have eaten apple pie and many girls have tasted vanilla ice-cream.’
(325) TP
muchos
chicosi
T’
han DETP
ti DET’
DET CoP
vP Co’
Co
y
vP
(-)chicas v’
v
degustado
VP
tV helado de
vainilla
ti v’
v
comido
VP
tV pastel de
manzana
This structure (325) is quite similar to the prior one (323). In this case, the lexical
Verbs comido, degustado ‘eaten, tasted’ stay in each conjunct. In the shared part, the
Auxiliary Verb generates in the Tense position. The subject in the first conjunct, moves
to the specifier in DET in order for its Determiner to be spelled out. Afterwards, this
subject DP, muchos chicos ‘many boys’ keeps on moving up to Spec, TP. The subject in
the second conjunct presents an unpronounced Determiner because the DP cannot make
the movement to DET.
3.5. SPANISH DETERMINER SHARING ANALYSIS 145
Determiner Sharing in Gapping of Modal: Analysis
The behavior of the lexical Verbs and the Modal is similar in subject determiner
sharing within the ‘Gapping of Modal’ type. And the same applies for subjects. The
structure for the relevant example (326) is (327) with the analysis of the Modal being
based on the discussion in chapter 2, section 2.4.4.4.
(326) Muchasmany
chicasgirls
deberıanshould
arreglarfix
sutheir
cochecar
yand
[ ][ ]
chicosboys
[ ][ ]
prepararprepare
sustheir
maletas.luggage‘Many girls should fix their car and many boys should prepare their luggage.’
146 CHAPTER 3. DETERMINER SHARING
(327) TP
muchas
chicasi
T’
deberıan DETP
ti DET’
DET CoP
vP Co’
Co
y
vP
(-)chicosj v’
tVMod VP
tVMod TP
tj T’
preparar vP
tj v’
v
tV
VP
V
tV
sus
maletas
vP
ti v’
tVMod VP
tVMod TP
ti T’
arreglar vP
ti v’
v
tV
VP
V
tV
su coche
In this structure (327) the lexical Verbs arreglar, preparar ‘fix, prepare’ stay in their
conjuncts, however, the modal Verbs ATB-move from the conjuncts to T in the shared
part. The subjects in each conjunct move up step by step up to the vPs that pertain to
Modal Verbs and once there the subject DP of the first conjunct moves to the specifier
in DETP so that the Determiner can be spelled out. This muchas chicas ‘many girls’
moves to Spec, TP afterwards as a final step. Again, this movement cannot be done by
3.5. SPANISH DETERMINER SHARING ANALYSIS 147
the second subject and, hence, its Determiner is not spelled out.
Determiner Sharing in Complex Gapping for Objects: Analysis
The next set of examples and structures are also subject determiner sharing examples but
the related gapping types are accounted for by deletion. I start with subject determiner
sharing in ‘Complex Gapping for Objects’. The sentence is (328) and the structure is
(329).
(328) Demasiadostoo.many
chicosboys
leCL
dierongave
unaa
florflower
ato
MarıaMarıa
yand
[ ][ ]
chicasgirls
[ ][ ]
ato
Jorge.Jorge
‘Too many boys gave a flower to Marıa and too many girls gave a flower to Jorge.’
148 CHAPTER 3. DETERMINER SHARING
(329) TP
demasiados
chicosi
T’
le dieron DETP
ti DET’
DET CoP
TopP Co’
Co
y
TopP
(-)chicasj Top’
Top FocP
a Jorgek Foc’
Foc vP
tj v’
le
dieron
VP
tV V’
una flor V’
tV tk
ti Top’
Top FocP
Foc’
Foc vP
ti v’
tV VP
tV V’
una flor V’
tV a Marıa
Because of evidence that is found in gapping sentences of this type, the remnants in (329)
move to the left of the final conjunct. Both chicas ‘girls’ and a Jorge ‘to Jorge’ move up
to the edge and deletion affects the rest of the vP phase. The subject in the first conjunct
moves towards the specifier in DETP. This is done in order for the Determiner in DP to
be spelled out. A further movement up to Spec, TP completes the structure. The DP in
3.5. SPANISH DETERMINER SHARING ANALYSIS 149
the second conjunct cannot move to DETP and because of this its D cannot be spelled
out. Regarding Verbs, I have represented Verb movement to T in (329) as moving only
from the first conjunct. For justification of asymmetric Verb movement, see chapter 2,
section 2.4.4.5, and the references cited there.
Determiner Sharing in Complex Gapping for Verbal Complexes: Analysis
Subject determiner sharing is also grammatical within ‘Complex Gapping for Verbal Com-
plexes’. Consider the example where it is the Auxiliary that is missing in conjunction with
the Verb and the Determiner (330). Its structure is in (331).
(330) Muchosmany
chicosboys
hanhave
comidoeaten
pastelpie
deof
manzanaapple
yand
[ ][ ]
chicasgirls
[ ][ ]
heladoice-cream
deof
vainilla.vanilla‘Many boys have eaten apple pie and many girls have eaten vanilla ice-cream.’
150 CHAPTER 3. DETERMINER SHARING
(331) TP
muchos
chicosi
T’
han DETP
ti DET’
DET CoP
TopP Co’
Co
y
TopP
(-)chicasj Top’
Top FocP
helado
de vainillak
Foc’
Foc vP
tj v’
v
comido
VP
tV tk
Top’
Top FocP
Foc’
Foc vP
ti v’
v
comido
VP
tV pastel
de manzana
In (331), the Verbs stay in their conjuncts and T is occupied by the Auxiliary han ‘have’.
The remnants in the second conjunct move to the left of the conjunct to the Specifier
positions in TopP and FocP. The subject in the first conjunct muchos chicos ‘many boys’
moves onto Spec, DET and finally to Spec, TP. With these last two steps, the Determiner
in this DP can be spelled out and the EPP feature gets checked/valuated. Again, the
subject in the second conjunct finishes up without a spelled out Determiner because of
its impossibility to move to Spec, DET. The rest of the unpronounced elements in the
second conjunct are accounted for via the deletion by phase process that affects the vP
phase. Notice that in this case (331), the Verb is also affected by deletion. Assuming
Boeckx and Stjepanovic (2001), any movement which this Verb could undergo in the PF
3.5. SPANISH DETERMINER SHARING ANALYSIS 151
component is suppressed by PF deletion.
Subject determiner sharing in ‘Complex Gapping for Verbal Complexes’ where it is
the modal Verb that is also missing is illustrated in (332) and the structure (333).
(332) Muchasmany
chicasgirls
deberıanshould
arreglarfix
sutheir
cochecar
yand
[ ][ ]
chicosboys
[ ][ ]
sutheir
bicicleta.bike
‘Many girls should fix their car and boys their bike.’
152 CHAPTER 3. DETERMINER SHARING
(333) TP
much. chicasi T’
deberıan DETP
ti DET’
DET CoP
TopP
Co’
Co
y
TopP
(-)chic.j Top’
Top FocP
su bici-
cletax
Foc’
Foc vP
tj v’
deberıan VP
tVMod TP
tj T’
arreglar TopP
tj Top’
Top FocP
tx Foc’
Foc vP
tj v’
tV VP
tV tx
Top’
Top FocP
Foc’
Foc vP
ti v’
tVMod VP
tVMod TP
ti T’
arreglar TopP
Top’
Top FocP
Foc’
Foc vP
ti v’
tV VP
tV su
coche
3.5. SPANISH DETERMINER SHARING ANALYSIS 153
In this case (333), where the modal Verb is present, I adopt a structure with two TPs
for each verbal complex composed of modal Verb and lexical Verb. See chapter 2, section
2.4.4.4, for justification. The subject in each conjunct raises from the lexical Verb part of
the structure to the modal Verb part of the structure. In the final conjunct, the object
also raises to the left of the conjunct, the Spec, FocP. This is done via intermediate steps.
The lexical Verbs stay in the conjuncts and the modal Verb in the first conjunct moves
to finish up in T. I propose that the modal Verb in the second conjunct does not move
assuming Boeckx and Stjepanovic (2001). The shared part of the structure is also formed
by the movement of muchas chicas ‘many girls’ to Spec, DET where the Determiner gets
to be spelled out. The Determiner in the second subject fails to be spelled out because
the whole DP cannot move to Spec, DET. A further movement by the first subject to
Spec, TP follows the prior completion of DETP. In this sentence (333), deletion by phase
affects the vP phase associated with the modal Verb.
Determiner Sharing in Gapping in NEG-nor : Analysis
Subject determiner sharing is also present in NEG-nor constructions. One example is
(276) which is repeated here as (334). Sentences of this type need to be analyzed as
coordination of large conjuncts because of the evidence that I provide in (277) in the
previous section.
(334) Nineither
demasiadostoo.many
ninosboys
estuvieronwere
enfermossick
ninor
[ ]D[ ]D
ninasgirls
[ ]V[ ]V
acatarradas.with.a.cold‘Neither too many boys were sick nor too many girls had a cold.’
The question which arises is whether or not a theory that accounts for determiner sharing
via DETs has to be applied to large conjunct coordination (334).
In order to answer this question, I look into the real size of conjuncts attending to
gapping sentences first. Consider the following examples (335, 336), which are examples
(222, 223) in the previous chapter, in comparison with (337).
(335) ??Nineither
cadaeach
chicoi
boyhahas
tocadoplayed
lathe
guitarraguitar
ninor
sui
hispadrefather
[ ][ ]
elthe
trombontrombone
‘Non of the boys has played the guitar nor has his father played the trombone.’
(336) ?Ningunno
chicoi
boyhahas
tocadoplayed
lathe
guitarraguitar
ninor
sui
hispadrefather
[ ][ ]
elthe
trombontrombone
‘No boy has played the guitar nor has his father played the trombone.’
(337) Nineither
JuanJuan
puedecan
comereat
alubiasbeans
ninor
PedroPedro
[ ][ ]
caviarcaviar
154 CHAPTER 3. DETERMINER SHARING
‘Neither Juan can eat beans nor Pedro can eat caviar.’
Though examples (335, 336) may sound a bit strange, these are reasonably acceptable.
This means that the conclusion can be reached that actually the Quantifier in the subject
position in the first part of the coordination can bind the pronoun in the second conjunct.
We know that this means the subject has to be above coordination for the binding to be
possible. It is interesting to compare this with that which the reading of example (337)
implies. The modal Verb in this example puede ‘can’ appears only once in the sentence.
The reading of the whole sentence expresses two different conditions –not only one con-
dition like in the cases where the modal Verb scopes widely–. The two conditions are:
(a) that Juan cannot eat beans and (ii) that Pedro cannot eat caviar. This separation
between conditions translates into a distributed meaning of the modal Verb and into the
fact that the modal Verbs need to be inside conjunction. Hence, the comparison of read-
ings in examples (335)-(337) shows that conjunction should be above the Modal Verb but
below a shared position where the subject can bind the material in the second conjunct.
To sum up so far, subject determiner sharing in NEG-nor cases comprises coordi-
nation of conjuncts that are not sentential but which are nonetheless larger than vPs.
Remember that the gapping structures are assumed for all types in the classification of
gapping/determiner sharing sentences. That is, conjuncts are big because modal Verbs
need to be inside but they are not sentential because the first subject needs to be out-
side the conjoined phrase. In (338, 339, 340), I provide the structures for (335, 336,337)
respectively.
(338) [TP AdvNEG-ni [TP Subj-cada chicoi [CoP [T′ T-ha ti tocado la guitarra]
[Co′ Co+NEG-ni [T′ su padre el trombon]
(339) [TP NEGDP-ningun chicoi [CoP [T′ T-ha ti tocado la guitarra]
[Co′ Co+NEG-ni [T′ su padre el trombon]
3.5. SPANISH DETERMINER SHARING ANALYSIS 155
(340) TP
AdvNEG
ni
TP
Subj
Juan
CoP
T’ Co’
CO+NEG
ni
T’
Pedro caviar (. . . )
puede comer alubias
Structure (338) corresponds to example (335). Following Johannessen (1998) I assume
that the ’initial conjunction’ ni ‘neither’ is an Adverb. In order for the subject in the first
part of the coordination to bind the second conjunct, it ends up in Spec, TP position.
Because in NEG-nor configuration the modal Verb does not scope widely, its structural
position needs to be inside the conjuncts. That is why I propose T’ conjunction for these
examples.
In structure (339), which corresponds to (336), the negation in the first conjunct is
in the n-word ningun ‘no’ in ningun chico ‘no boy’. This DP is the subject in the first
conjunct. Again, because it binds the second conjunct, it is located outside conjunction,
in Spec, TP after it has moved from the vP internal position. And conjuncts are also T’
because the distributed scope of modal Verbs needs to be taken into account. Finally,
the structure in (340), which corresponds to (337), shows the location of the first modal
Verb inside conjunction. The second modal Verb does not escape deletion because of
the assumption of Boeckx’s and Stjepanovic’s (2001) proposal. The ‘initial conjunction’
is analyzed as an Adverb and the second ni ‘nor’ combines Johannessen’s (1998) and
Wurmbrand’s (2008) ideas. (See chapter 2, section 2.4.4.6.)
The combination of this proposal with the syntax of determiner sharing is not complex.
The order that phrases follow in determiner sharing in NEG-nor sentences is similar to
that of gapping. There is only one difference which is that the DETP needs to be added:
previous explanations situate this projection right above CoP. Consider the analysis of
the determiner sharing sentence (334) in (341).
156 CHAPTER 3. DETERMINER SHARING
(341) TP
AdvNEG
ni
TP
demasiados
ninosi
DETP
ti DET’
DET CoP
T’ Co’
CO+NEG
ni
T’
(-) ninas acatarradas (. . . )
estuvieron ti tV enfermos
The subject in the first conjunct, demasiados ninos ‘too many boys’, which needs to move
out from the conjunction, moves first to Spec, DET so that its Determiner is spelled out
before further moving up to Spec, TP. The Determiner of the subject DP in the second
conjunct cannot move to Spec, DETP and hence its Determiner is not spelled out. The
heads of the TPs are part of the conjuncts because of the reasoning above, and the actual
conjunction head is the second ni ‘nor’. The first ni ‘neither’ is analyzed as an Adverb.
3.5.4 Analysis: object position
As described above, object determiner sharing is only possible in cases where deletion
accounts for missing elements. This is a fact which can be explained by means of the
analysis that I provide. Two parts of my analysis are of relevance here: (i) the division
among gapping types and their analyses, (ii) the requirements that DET and DPs need
to hold in order for determiner sharing to appear in the sentence. These requirements
are met only in the deletion cases and this happens because of the structures that these
cases display in my proposal. In these deletion cases, remnants overtly move to the left
making it possible for the requirements between DET and DP to be met. The structures
of the non deletion cases are set out in such a way that the requirements between DET
and DP cannot be met, and that is why determiner sharing is not possible in these cases.
I explain this in further detail for each case.
3.5. SPANISH DETERMINER SHARING ANALYSIS 157
In order to explain all object determiner sharing cases, I divide them into two groups.
Object determiner sharing sentences which are ungrammatical are those whose missing
elements are only accounted for by movement. I first explain what happens in the un-
grammatical cases and I continue afterwards with deletion cases.
158 CHAPTER 3. DETERMINER SHARING
Determiner Sharing in Basic Gapping: Analysis
The first one is the example where object determiner sharing happens within ‘Basic Gap-
ping’. Sentence (342) is analyzed in (343).
(342) #PedroPedro
comeeats
demasiadostoo.many
pastelespies
deof
manzanaapple
yand
JuanJuan
[ ][ ][ ][ ]
heladosice-creams
deof
vainilla.vanilla‘Pedro eats too many apple pies and Juan eats too many vanilla ice-creams.’
(343) # TP
Pedroi T’
come DETP
# DET’
DET CoP
vP Co’
Co
y
vP
Juan v’
tV VP
tV (-) helados
de vainilla
ti v’
tV VP
tV demasiados past.
de manzana
Example (342) is ungrammatical and this is reflected in the structure (343). The analy-
sis of ‘Basic Gapping’ is not derived by deletion. It is only movement that accounts for
missing elements in this case. Because there is no deletion, the remnant object in the
non-initial conjunct does not move. In the initial conjunct the object does not move either
because there is no specific trigger for object movement in regular declarative sentences
in Spanish. Because this object DP does not get to be in Spec, DETP, the Determiner
cannot be spelled out. This is the reason why the sentence is ungrammatical and it suf-
fices in order to account for it. However, it is of interest to detail the behavior of objects
in cases of this type. Objects move neither overtly nor covertly in 343. There is only one
3.5. SPANISH DETERMINER SHARING ANALYSIS 159
Quantifier in each conjunct and Quantifier Raising is only possible if there is more than
one Quantifier (Fox (2000)). This also applies to structures (345) and (347). These two
remaining ungrammatical examples are explained by the same kind of logic. I follow the
same order as in other sections: the next type of determiner sharing is within Gapping of
Auxiliary.
Determiner Sharing in Gapping of Auxiliary: Analysis
Object determiner sharing within the ‘Gapping of Auxiliary’ type is not possible either.
The relevant sentence is (344) and its structure is (345).
(344) #PedroPedro
hahas
comidoeaten
demasiadostoo.many
pastelespies
deof
manzanaapples
yand
JuanJuan
[ ][ ]
degustadotasted
[ ][ ]
heladosice-creams
deof
vainilla.vanilla
‘Pedro has eaten too many apple pies and Juan has tasted too many vanilla ice-
creams.
(345) # TP
Pedrox T’
ha DETP
# DET’
DET CoP
vP
Co’
Co
y
vP
Juan v’
degustado VP
tV (-) helados
de vainilla
tx v’
comido VP
tV demasiados past.
de manzana
160 CHAPTER 3. DETERMINER SHARING
In this sentence (344), its structure (345) accounts for ungrammaticality. Conjunction is
done at the level of vP. The Auxiliary is generated at T and Spec, TP is occupied by the
subject in the first conjunct that has moved there to satisfy EPP. The lexical Verbs do
not move and because basic sentence word order tells us that the objects have to follow
the Verbs, the object DP in the first conjunct cannot move to Spec, DET. There is no
way the Determiner in the first object DP can be spelled out without movement to Spec,
DET according to the theory. Ungrammaticality is accounted for on the same basis here.
Determiner Sharing in Gapping of Modal: Analysis
Determiner sharing within the ‘Gapping of Modal’ type behaves similarly. Consider (346)
and its structure (347).
(346) #PedroPedro
deberıashould
tomardrink
muchosmany
zumosjuices
deof
naranjaorange
yand
JuanJuan
[ ][ ]
comereat
[ ][ ]
arandanos.blueberries‘Pedro should drink many glasses of orange juice and Juan should eat many blue-
berries.’
3.5. SPANISH DETERMINER SHARING ANALYSIS 161
(347) # TP
Pedrox T’
deberıa DETP
# DET’
DET CoP
vP
Co’
Co
y
vP
Juani v’
tVMod VP
tVMod TP
ti T’
comer vP
ti v’
tV VP
tV (-)arandanos
tx v’
tV mod VP
tVMod TP
tx T’
tomar vP
tx v’
tV VP
tV muchos zumos
de naranja
The shared part of this structure (347) is composed of TP, to which the subject of the first
conjunct moves and the modal Verbs ATB move from both conjuncts. Because lexical
Verbs stay in their conjuncts and objects naturally follow them, there is no way the object
in the first conjunct can move to the specifier of DETP. This leaves the pronunciation of
the Determiner unaccounted for and this is why the sentence is ungrammatical.
Determiner sharing in object position in deletion cases: In sentences (342), (344),
(346), which are object determiner sharing cases in gapping contexts that are accounted
for by movement, there is no reason why the object needs to move in any case, in any
conjunct. Let us look deeper into this lack of object DP movement by comparing these
latter examples with the following paradigm, which is composed of the deletion cases.
162 CHAPTER 3. DETERMINER SHARING
The following are examples (281, 283, 284, 285) which I repeat here as (348, 349, 350,
351).
(348) Alvaro
ALvaro
leCL
diogave
ato
JuanJuan
demasiadastoo.many
responsabilidadesresponsibilities
yand
AnaAna
[ ][ ]
[ ][ ]
disgustos.problems
‘Alvaro gave Juan too many responsibilities and Ana caused Juan too many prob-
lems.’
(349) PedroPedro
hahas
comidoeaten
demasiadostoo.many
pastelescakes
yand
JuanJuan
[ ][ ]
[ ][ ]
heladosice-creams
deof
vainilla.vanilla
‘Pedro has eaten too many cakes and Juan has eaten too many vanilla ice-creams.’
(350) PedroPedro
deberıashould
tomardrink
muchosmany
zumosjuices
deof
naranjaorange
yand
JuanJuan
[ ][ ]
[ ][ ]
batidosmilk
deshakes
arandanos.of blueberries
‘Pedro should drink many glasses of orange juice and Juan should drink many
blueberry milk shakes.’
(351) Nineither
PedroPedro
comeeats
demasiadostoo.many
pastelespies
deof
manzanaapple
ninor
JuanJuan
[ ]V[ ]V
[ ]D[ ]D
heladosice-creams
deof
vainilla.vanilla
‘Neither Pedro eats too many apple pies nor does Juan eat too many vanilla ice-
creams.’
The one thing that all these sentences have in common –in contrast to the already analyzed
cases– is that they are all grammatical. Although I provide all the details of the derivations
later, I only focus on what is related to object movement here. The relevance of this
paradigm in relation to object movement resides in example (349). As I showed in chapter
2, section 2.4.4.3, this type of gapping, ‘Gapping of Auxiliary’, needs to be analyzed
by maintaining the participles inside the conjuncts. Conditions on determiner sharing
phenomena state that in order to have this type of sharing, the DP with the pronounced
Determiner needs to move to DETP, which has to be outside the conjunction so that it is
shared. However, word order in example (349) tells us that the object DP needs to follow
the participle which is inside the conjunct. So, it needs to be inside the conjunction but
it also needs to move outside to Spec, DETP. A simple way of thinking would be to claim
that the movement of object DPs to Spec, DETP in determiner sharing is covert. Notice,
however, that making this a universal claim about all object determiner sharing cases
would not work because in such a case no explanation would be available for examples
(342, 344, 346), where ungrammaticality is explained via the lack of movement of object
DPs. I hence conclude (i) and (ii) and I explain it as follows.
3.5. SPANISH DETERMINER SHARING ANALYSIS 163
• (i) Grammaticality of object determiner sharing in these cases –deletion ones–, in
opposition to those where there is no deletion, is due to the fact that the conditions
between DET and DPs are met: object DPs move to Spec, DETP.
• (ii) The reason why there is only movement of the object DPs in these examples
(348, 349, 350, 351) is related to what differentiates these examples from the un-
grammatical ones: the presence of deletion and movements of remnants.
The explanation is based on evidence from PF deletion approaches that when remnants
move, correlates should move too. Their evidence is based on scopal parallelism in ellipsis
(Chung et al. (1995), Romero (1998), Merchant (2001)). Scopal parallelism is found both
in ellipsis and deaccenting. I illustrate this with sluicing phenomena here. The basic
idea is that the scope which the wh-phrase presents in relation to its clause needs to be
the same which its correlate presents with the antecedent clause. Consider the following
examples (352, 353) that I take from Johnson (2001).
(352) She always reads a book at dinner time.
This sentence has two readings which depend upon the scope of ‘a book’ with respect to
its predicate. When ‘a book’ has narrow scope with respect to its predicate, the reading is
‘she reads a book which could be different each day’. Instead, when ‘a book’ scopes widely,
the sentence means that ‘she reads that one book everyday’. This ambiguity disappears
in a comparable sluicing context like (353).
(353) She always reads a book at dinner time. I can’t figure out which one [ ].
In this case (353), the only available reading is the one where ‘a book’ has wide scope
with respect to ‘always’. This happens because of scope parallelism between sluice and
antecedent: the wh-phrase has moved to the left and its scope with respect to its clause
is wide; the same wide scope is the only possibility then for the wh-phrase’s correlate ‘a
book’. Syntactically, for ‘a book’ to scope widely it needs to move to the left to a position
where its scope is wide.
The same happens in Spanish. Consider parallel examples (354, 355) where disam-
biguation also takes place when the sluice is present.
(354) MarıaMarıa
siemprealways
leereads
una
librobook
duranteat
lathe
cena.dinner
‘Maria always reads a book at dinner time.’
(355) MarıaMarıa
siemprealways
leereads
una
librobook
duranteat
ladinner.
cena.But
Peronever
nuncaCl.me
mehas
hatold
dichowhich.one
cual [ ].
164 CHAPTER 3. DETERMINER SHARING
Marıa always reads a book at dinner time but she has never told me which one.’
In Spanish also, the sluice which is present in example (355) makes de ambiguity in (354)
disappear. In (355), the only meaning possible is the one where ‘un libro’ scopes widely
with respect to its predicate.
For Spanish determiner sharing in object position, I propose a parallel situation: move-
ments of remnants, which make arguments escape from PF deletion in the final conjunct,
enable covert movement of the object DP in the first conjunct to Spec, DETP, in the
shared part. It is only by means of movements of remnants and scope parallelism in dele-
tion contexts that covert movement of the correlate object DP gets justified. This is the
reason why object determiner sharing is possible in deletion cases and it is not possible
in the rest of cases. I start now with the detailed analyses of derivations. I maintain the
deletion by phase process and illustrate the way object covert movement works in deter-
miner sharing derivations further for each case. Determiner sharing per se is analyzed the
same way as before in the following cases.
Determiner Sharing in Complex Gapping for Objects: Analysis
The first example in this set is an object determiner sharing sentence that is situated
within the ‘Complex Gapping for Objects’ type. Consider sentence (356) and its structure
(357).
(356) Alvaro
ALvaro
leCL
diogave
ato
JuanJuan
demasiadastoo.many
responsabilidadesresponsibilities
yand
AnaAna
[ ][ ]
[ ][ ]
disgustos.problems
‘Alvaro gave Juan too many responsibilities and Ana caused Juan too many prob-
lems.’
3.5. SPANISH DETERMINER SHARING ANALYSIS 165
(357) TP
Alvaroi T’
le dio DETP
tx DET’
DET CoP
TopP
Co’
Co
y
TopP
Anak Top’
Top FocP
(-)disgustosy Foc’
Foc vP
tk v’
le dio VP
tV V’
a Juan V’
tV ty
Top’
Top FocP
Foc’
Foc vP
ti v’
tV VP
tV V’
a Juan V’
tV demasiadas
respons.x
In this sentence (357), because determiner sharing happens in object position, the DET
Projection is associated with the DPs that are generated in object position. Because
this is a deletion example, the remnants in the second conjunct move up to the specifier
positions in FocP and TopP at the left of the conjunct. The deletion by phase process
affects the vP phase in the final conjunct.
In this final conjunct the movement of the Verb is suppressed following Boeckx and
Stjepanovic (2001) as explained above and the Verb in the first conjunct moves up to
T. This Verb movement from the first conjunct onto the shared part may be seen as an
166 CHAPTER 3. DETERMINER SHARING
asymmetric movement and as such problematic. However, I think it is unproblematic if
we think carefully of cyclicity in movements. Let us consider, for instance, Nissenbaum’s
(2000) notion of multiple spell out. The cycle in this work gets to be completed once all
operations both overt and covert have taken place. That is, each cycle is comprised by
(i) overt syntactic operations, which happen before spell out, (ii) spell out per se, and
(iii) covert operations that happen after spell out, at LF and PF levels. If this is taken
into account in the scenario in example (357), the moment in the derivation when the
Verb in the first conjunct moves is as follows. The two conjuncts have been formed and,
because they happen to be at phase level, the cycle/phase that comprises each conjunct
has been finalized too. So, I explain each cycle –one per conjunct– separately. By the time
the cycle in the second conjunct is complete, because both overt and covert operations
happen before the completion of the cycle, the suppression of the Verb movement at PF
has already happened. This is different in the cycle that comprises the first conjunct. By
the time the cycle ends, the Verb in the first conjunct is in a position that is accessible
to the rest of the derivation. When T attracts this lexical Verb dio ‘gave’, it is only this
one Verb that is accessible for such movement. Either asymmetric or not with respect to
coordination, this type of movement is what the derivation both allows –because of what
is known about cycles– and needs -because of T’s attraction of the nearest and only Verb
available at that time of the derivation–.
Because this sentence is grammatical, we need the DP in the first conjunct to move
to DETP so that the DP’s Determiner is pronounced. Following what I have explained
about scope parallelism, in this example (357), the object of the first conjunct demasiadas
responsabilidades ‘too many responsibilities’ covertly moves to Spec, DET where it gets
its Determiner spelled out. Finally, the subject in the first conjunct moves up to Spec,
TP because of EPP, and the remaining argument of the Verb a Juan ‘to Juan’ in the first
conjunct remains in its argument position.
The sentence (356), an object determiner sharing sentence within ‘Complex Gapping
for Objects’, could be analyzed claiming that deletion affects the argument of the Verb
which is missing and that the Verb is affected by ATB movement –in the lines of Johnson
(2009)–. Though I understand that this specific type of determiner sharing is also com-
patible with such analysis and I leave it open for further discussion, I would like to draw
attention to the following. If we consider that a unified analysis of all determiner sharing
cases that are accounted for via deletion is appropriate, examples (349) and (351) may be
of importance here. Example (349) is the focus of the next paragraph and it is repeated
as (358) and analyzed in (359). Example (351) is explained and analyzed at the end of
3.5. SPANISH DETERMINER SHARING ANALYSIS 167
this section. There, example (351) is repeated as (362) and analyzed in (363). Example
(358) is a sentence where the missing elements in non initial conjuncts are the determiner,
the lexical Verb and the auxiliary. The relevance of this kind of sentence comes with a
fact that is explained in the previous chapter, when ‘Complex Gapping for Verbal Com-
plexes’ is analyzed. There, I explain that in sentences of this type, the participles in the
Spanish sentence need to stay inside the conjuncts. This of course, is incompatible with
an analysis where the lexical Verbs ATB move from conjuncts to the shared part. So, if
a unified analysis to these deletion cases is desired, consider that this sentence cannot be
analyzed following the proposal that both ATB movement of the Verb and also deletion
are present in the sentence.
The same applies to sentence (362) as can be seen in its structure (363) at the end of
this section. Recall that with the approach in Johnson (2009) Verbs move, and that with
my approach the movement of the Verb in the second conjunct is suppressed. It can be
seen in the structure in (363) that with the suppression of the movement of the second
Verb, sentence (362) can be accounted for. Sentence (362) cannot be accounted for if the
Verbs in both conjuncts move. The approach in Johnson (2009) cannot account for this
because if the second Verb also moves it has to be spelled-out in the T position in the
second conjunct and that is not the result that we need to obtain. In sentence (362) the
Verb in the second conjunct is not pronounced and an approach such as Johnson’s where
the second Verb also moves is not suitable because this cannot account for the missing
Verb. The proposal in this dissertation is able to provide a unified analysis to not only
examples (349) and (351), repeated as (358) and (362) respectively, but also to examples
(348) and (350). Because of this, I favor this kind of analysis over Johnson’s which can
only account for part of the examples.
Determiner Sharing within Complex Gapping for Verbal Complexes: Analysis
Object determiner sharing in ‘Complex Gapping for Verbal Complexes’ is instantiated first
in sentence (358) and its structure is (359). These two represent object determiner sharing
in sentences where what is also missing is the auxiliary and the lexical Verb.
(358) PedroPedro
hahas
comidoeaten
demasiadostoo.many
pastelescakes
yand
JuanJuan
[ ][ ]
[ ][ ]
heladosice-creams
deof
vainilla.vanilla
‘Pedro has eaten too many cakes and Juan has eaten too many vanilla ice-creams.’
168 CHAPTER 3. DETERMINER SHARING
(359) TP
Pedroi T’
ha DETP
tj DET’
DET CoP
TopP
Co’
Co
y
TopP
Juany Top’
Top FocP
(-)helados
de vainillax
Foc’
Foc vP
ty v’
comido VP
tV tx
Top’
Top FocP
Foc’
Foc vP
ti v’
comido VP
tV demasiados
pastelesj
In structure (359), the remnants in the final conjunct move to the specifier positions
in FocP and TopP. Deletion by phase affects the vP phase in the final conjunct also. The
regular association is made between DET and the object DPs. The first Determiner is
the only one that can move to Spec, DET. Because of the evidence which I have cited
above with respect to scope parallelism, this DPS movement needs to be done and it is
done covertly in the first conjunct. Tense is occupied by the auxiliary and the Verbs stay
in their conjuncts. Notice that I assume the PF suppression of Verb movement to higher
projections in its final conjunct, which is done by deletion as in Boeckx and Stjepanovic
(2001). Finally, the subject in the first conjunct makes the usual movement to Spec, TP
to satisfy EPP.
3.5. SPANISH DETERMINER SHARING ANALYSIS 169
Object determiner sharing within ‘Gapping for Verbal Complexes’, where the Deter-
miner, the lexical Verb and the modal Verb are missing, is instantiated in sentence (360)
and analyzed in (361).
(360) PedroPedro
deberıashould
tomardrink
muchosmany
zumosjuices
deof
naranjaorange
yand
JuanJuan
[ ][ ]
[ ][ ]
batidosmilk
deshakes
arandanos.of blueberries
‘Pedro should drink many glasses of orange juice and Juan should drink many
blueberry milk shakes.’
170 CHAPTER 3. DETERMINER SHARING
(361) TP
Pedrox T’
deberıaDETP
ty DET’
DET CoP
TopP
Co’
Co
y
TopP
Juani Top’
Top FocP
(-)bat de
aranjFoc’
Foc vP
ti v’
deberıa VP
tV Mod TP
ti T’
tomar TopP
ti Top’
Top FocP
tj Foc’
Foc vP
ti v’
tV VP
tV tj
Top’
Top FocP
Foc’
Foc vP
tx v’
tV Mod VP
tV Mod TP
tx T’
tomar TopP
tx Top’
Top FocP
Foc’
Foc vP
tx v’
tV VP
tV m. zum.
de nar.y
3.5. SPANISH DETERMINER SHARING ANALYSIS 171
The mechanism in this example (361) follows same strategies as above. In the final con-
junct, remnants move to Spec, FocP and Spec, TopP to scape PF deletion. Deletion by
phase affects the deletion domain as explained in the Gapping chapter. And this PF dele-
tion operation suppresses Verb movement also in PF following Boeckx and Stjepanovic
(2001). The shared part of the structure is composed of Spec, TP, to which the subject
in the first conjunct moves satisfying EPP. The Head in TP is satisfied by movement of
the modal Verb in the first conjunct. I have also explained above that following succes-
sive cyclicity, by the time the derivation gets to this point, this is the only Verb that
is available to move to T. Finally, determiner sharing happens in this example because
the object DP in the first conjunct can move to Spec, DETP covertly and the DP in
the second conjunct cannot. I support this covert movement above by means of evidence
related to scope parallelism (Chung, Ladusaw, and McCloskey (1995), Romero (1998),
Merchant (2001)).
Determiner Sharing in Gapping in NEG-nor : Analysis
I finally cover the last of the determiner sharing cases within declarative sentences. This
is object determiner sharing within NEG-nor. Such example is (362) and its structure
(363).
(362) Nineither
PedroPedro
comeeats
demasiadostoo.many
pastelespies
deof
manzanaapple
ninor
JuanJuan
[ ]V[ ]V
[ ]D[ ]D
heladosice-creams
deof
vainilla.vanilla
‘Neither Pedro eats too many apple pies nor Juan eats too many vanilla ice-
creams.’
172 CHAPTER 3. DETERMINER SHARING
(363) TP
AdvNEG
ni
TP
Pedroi DETP
tj DET’
DET CoP
T’
Co’
Co+NEG
ni
T’
T TopP
Juanx Top’
Top FocP
(-) helados
de vainillay
Foc’
Foc vP
tx v’
v VP
come ty
come TopP
Top’
Top FocP
Foc’
Foc vP
ti v’
tV VP
tV demasiados past.
de manzanaj
As I illustrate in (363), the derivation has similar features to the structures above. There
are some differences, however, because of the need of larger conjuncts in NEG-nor con-
figurations which has been proven in the parallel subject determiner sharing case. Part of
the Tense projection gets to be inside conjunction because of the analysis done before on
the scope of elements inside this projection. Verbal operations happen inside each con-
junct. In the first conjunct, the Verb moves to T as lexical Verbs do in Spanish when no
auxiliary fills T. In the second conjunct, however, following the line of thought in Boeckx
and Stjepanovic (2001), the lexical Verb and its movement gets suppressed. This is the
result of the competition that arises among the PF operations: PF deletion and Verb
movement in PF. Notice that the analysis in Boeckx and Stjepanovic (2001) is focused
3.6. CONCLUSION 173
on pseudogapping. Hence one of the aspects that play a part in the consideration of Verb
movement as a PF phenomenon is do support. Because in Spanish, do support is not
available, I consider the fact that T in the second conjunct is empty unproblematic. Also,
because motivations for head movement are claimed to be either phonological or mor-
phological, I consider the emptiness of T a consequence of the competition in PF of both
PF processes, deletion and Verb movement; and as such, still unproblematic. Determiner
sharing is explained as above. The DP in the first conjunct can move to DETP and the
DP in the second conjunct cannot. This is why, the first DP’s Determiner gets to be
pronounced. Covert movement of this DP is explained because of the scopal parallelism
facts that happen in deletion cases. The structure finishes its derivation with the shared
part which is comprised of the DETP, the subject of the first conjunct that moves to Spec,
TP asymmetrically, and the ‘neither’ part of the connector which I analyze as a negated
Adverb following Johannessen (1998).
3.6 Conclusion
The divisions among gapping types defined in chapter 2 have proven themselves useful for
the analysis of determiner sharing: contrasts between grammatical sentences and ungram-
matical sentences happen in accordance to the different gapping types. Instantiation of
such a fact can be found on the grammaticality contrasts among object determiner sharing
cases. Analysis of determiner sharing in Spanish is done via a theory with DETs which
presents differences with respect to previous literature. In order to account for determiner
sharing and the missing determiner in non initial conjuncts in Spanish, I propose that
DET projections occupy non fixed positions right above phases –DPs, vPs, and CPs– in
the structure. With this, I am able to account for sentences where determiner sharing is
either present or not in both conjoined and non conjoined sentences. The compatibility
with the phenomena presented in the next chapter, which is focused on wh-determiner
sharing, is also crucial in such proposal. In accounting for the determiner sharing sen-
tences, this theory leaves space for small and large conjuncts, and space for not only
movement but also deletion in the derivations.
Chapter 4
Wh-Determiner Sharing
4.1 Introduction
This chapter is dedicated to the last step that I take in the analysis of gapping and
determiner sharing in this dissertation. I investigate the determiner sharing structure
embedded into wh-questions in Spanish. This structure, which I label wh-determiner
sharing, is characterized by the sharing of the wh-word by the conjuncts in coordinate
questions. First, in section 4.2 I compare the properties of determiner sharing in wh-
questions and in declarative sentences. I also provide wh-determiner sharing data for all
types of determiner sharing that I have introduced in previous chapters.
Next, in section 4.3, I analyze the structures of these sentences. I briefly explain what
the proposal for wh-determiner sharing in Spanish consists of. It mainly brings together
the proposal in the previous chapters with the features of wh-movement. I analyze four
sentences types to illustrate this. Finally, I focus on one of the aspects of the analysis of
wh-determiner sharing, which is of interest for the general theory of syntax: the effects of
cyclicity and phases in wh-determiner sharing and their interaction between each other.
Wh-determiner sharing and the analysis here provided can be used as evidence in favor of
cyclicity in wh-movement and as diagnostic for categories which project phase boundaries.
4.2 Wh-determiner sharing in Spanish
4.2.1 Introduction
In this section, I introduce the wh-determiner sharing phenomenon. This is a subtype of
determiner sharing that arises in wh-questions. I explain its behavior with respect to the
general properties of determiner sharing discussed in the previous chapters. I also provide
175
176 CHAPTER 4. WH-DETERMINER SHARING
a complete paradigm where wh-determiner sharing sentences are divided according to
the type of gapping in which these are embedded. In this respect, the classification is
identical to the ones developed in previous chapters of this work. I start with the behavior
that wh-determiner sharing displays with respect to the general properties of determiner
sharing.
4.2.2 General properties
Wh-determiner sharing is a subtype of determiner sharing which takes place in wh-
questions. In this type of construction, the determiner that is shared is the wh-word.
In general terms, its main feature is the same: the wh-Determiner gets to be shared in
the sense that it is only pronounced in one part of the coordination but it gets to be
interpreted in all conjunction parts. In comparison to regular determiner sharing struc-
tures, some of the general properties of wh-determiner sharing appear to be possible and
optional, rather than compulsory conditions. Though I explain this later, I claim that
this fact is related to the nature of wh-questions.
4.2.2.1 The missing V: not a condition in Wh-questions
To begin with, the condition which is mentioned in McCawley (1993) that determiner
sharing is dependent upon Verb gapping is only a possibility in wh-determiner sharing.1
That is, this condition does not restrict wh-determiner sharing because this type of deter-
miner sharing is grammatical either way. Consider (364) where Verb gapping is present
and (365) where Verb gapping is absent.
(364) ¿Cuantoshow.many
estudiantesstudents
acudengo
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]V[ ]V
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students do go to the syntax class and how many professors do go to
the monthly meeting?’
(365) ¿Cuantoshow.many
estudiantesstudents
acudengo
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
faltanmiss
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students go to the syntax class and how many professors miss the
monthly meeting?’
In example (364), wh-determiner sharing is present in the sentence at the same time as
Verb gapping. Both the wh-determiner cuantos ‘how many’ and the Verb acuden ‘go’ are
1See section 3.2.2 in the previous chapter to compare the behavior of wh-determiner sharing withdeclarative determiner sharing.
4.2. WH-DETERMINER SHARING IN SPANISH 177
missing in the final conjunct. However, they both get interpreted here. Notice, though,
that wh-determiner sharing is different from determiner sharing in that Verb gapping is
not a compulsory condition. Example (365) is a wh-determiner sharing sentence, with
sharing of cuantos ‘how many’ where Verbs acuden ‘go’ and faltan ‘miss’ are overt in the
conjuncts.2 Still both sentences are grammatical.
4.2.2.2 The missing T: not a condition in Wh-questions
The same pattern can be observed with respect to the specification that it is T, not
simply V, that needs to be missing in determiner sharing for such sharing to happen
(Siegel (1984), Siegel (1987), Oehrle (1987), Lin (2002). In wh-determiner sharing, T
gapping is possible as example (366) illustrates, but it is not compulsory as illustrated in
sentence (367).
(366) ¿Cuantoshow.many
estudiantesstudents
hanhave
acudidogone
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]T[ ]T
faltadomissed
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students have gone to the syntax class and how many professors have
missed the monthly meeting?’
(367) ¿Cuantoshow.many
estudiantesstudents
hanhave
acudidogone
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
hanhave
faltadomissed
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students have gone to the syntax class and how many professors have
missed the monthly meeting?’
In sentence (366), wh-determiner sharing and T gapping are present at the same time.
Both the wh-determiner cuantos ‘how many’ and the auxiliary han ‘have’ are missing
from the non-initial conjunct. However, this conjunct is interpreted with the semantic
content of these two elements, the wh-determiner and the auxiliary. In sentence (367),
wh-determiner sharing is present with the absence of T gapping. Hence, in this example,
cuantos ‘how many’ can be shared without gapping of han ‘have’. This illustrates the
fact that T gapping is not compulsory for wh-determiner sharing to be present.
2Examples of this type, with no T sharing but with sharing of D, are presented for the first time inMcCawley (1993, 245), and is discussed by Ackema and Szendroi (2002, 210) for English. In Spanish,the first work in which wh-determiner sharing is considered is by Arregi and Centeno (2005). Centeno(2007) investigates some aspects of the structure with more specification, and Centeno and Vicente (2009)investigate this further in relation to the theory of phases. I develop some aspects of these works later.At this point, I present the basics that are needed before expanding on the topics presented in thoseworks.
178 CHAPTER 4. WH-DETERMINER SHARING
4.2.2.3 Conjunct initial?
In a different respect, but within the general properties which are normally listed for
determiner sharing, let us now consider McCawley’s observation that the DPs that enter
the determiner sharing relationship need to be conjunct initial. Consider (368).
(368) ¿PedroPedro
cuantoshow.many
librosbooks
hahas
leıdoread
yand
JuanJuan
[ ]D[ ]D
revistasmagazines
[ ]T[ ]T
revisado?reviewed
‘How many books has Pedro read and how many magazines has Juan reviewed?
In sentence (368), the wh-DPs that enter the determiner sharing relationship are cuantos
libros ‘how many books’ and cuantas revistas ‘how many magazines’. These DPs are not
initial in either the first conjunct or in the second, yet the sentence in example (368) is as
grammatical as sentences where the DPs that enter the sharing relationship are conjunct
initial. In wh-determiner sharing, “shared Ds” need not be conjunct initial.
4.2.2.4 Connectors, negation, and determiner sharing
A comparison between wh-determiner sharing and determiner sharing in declarative sen-
tences with respect to this property cannot be done. The reason for this is that the types
of Determiners which are used in the declarative sentences cannot be used in wh-questions.
In the declarative sentences, the Determiners that are shared to check on this property
are ‘not many’ in each conjunct. This type of D with negation in it cannot be used in
wh-questions. Because the property is intimately related to negation, without this type
of Determiner with negation in it, the property cannot apply to wh-determiner sharing.
4.2.3 Wh-determiner sharing: data and classification
This section is dedicated to present wh-determiner sharing data. The main goal is to
provide a complete paradigm according to the classification that has been developed in
the previous chapters. For each type of determiner sharing, I pay attention both to cases
which present gapping of T and cases which do not present gapping of T, because these
two possibilities are part of the wh-determiner sharing paradigm. I divide wh-determiner
sharing data according to whether wh-phrases are subjects or objects.
4.2.3.1 Wh-determiner sharing with subjects
Here, I provide wh-determiner sharing sentences where wh-phrases are associated with the
subjects in each conjunct. I divide them according to the gapping and determiner sharing
classification that has been done in previous chapters in sections 2.4.4, for gapping, and
3.3.3, for determiner sharing.
4.2. WH-DETERMINER SHARING IN SPANISH 179
4.2.3.1.1 Wh-Determiner Sharing in Basic Gapping: This kind of wh-determiner
sharing can be illustrated in example (364), which I repeat here as (369).3 This type of
wh-determiner sharing is characterized by the fact that the sharing relationship between
wh-Determiners is embedded in basic gapping, where the only lexical Verb is missing.
(369) ¿Cuantoshow.many
estudiantesstudents
acudengo
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]V[ ]V
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students go to the syntax class and how many professors go to the
monthly meeting?’
In this example (369), the wh-Determiner cuantos ‘how many’ and the only lexical Verb
acuden ‘go’ are missing from the non initial conjunct. The sharing of the wh-Determiner
is established between the two subjects cuantos estudiantes ‘how many students’ and
cuantos profesores ‘how many professors’. Because T or V gapping is not obligatory in
wh-determiner sharing, the non T sharing counterpart to (369) is (365), which I repeat
as (370) here.
(370) ¿Cuantoshow.many
estudiantesstudents
acudengo
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
faltanmiss
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students go to the syntax class and how many professors miss the
monthly meeting?’
The novel aspect in this example (370) is that there is a Verb in each conjunct: acuden
‘go’ and faltan ‘miss’. Wh-determiner sharing is possible either way.
4.2.3.1.2 Wh-Determiner Sharing in Gapping of Auxiliary: This type of wh-
determiner sharing has also appeared already in this section in example (366). I repeat
it here as (371).
(371) ¿Cuantoshow.many
estudiantesstudents
hanhave
acudidogone
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]T[ ]T
faltadomissed
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students have gone to the syntax class and how many professors have
missed the monthly meeting?’
3The wh-Determiner which I construct all of the sentences with is cuanto/s/as ‘how many’ becauseother wh-Determiners that could be used such as que ‘what’ or cual/es ‘which’ sound strange in theSpanish dialect I pertain to: the one spoken in the Northern part of the Iberian Peninsula. Though thisgeographical area is known, among others, for its different language contact situations, the dialect of theNorthern part of Spain is commonly also referred to as a whole.
180 CHAPTER 4. WH-DETERMINER SHARING
Here, same as in other ‘Gapping of Auxiliary’ sentences, the Auxiliary is missing. In
addition, in this specific case (371), the wh-Determiner is also missing. The Auxiliary
han ‘have’ and the wh-Determiner cuantos ‘how many’ are missing but interpreted in the
non initial conjunct: ‘how many professors have missed the monthly meeting’. The non
T gapping counterpart of example (371) is (367). I repeat it here as (372).
(372) ¿Cuantoshow.many
estudiantesstudents
hanhave
acudidogone
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
hanhave
faltadomissed
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students have gone to the syntax class and how many professors have
missed the monthly meeting?’
In this example (372), the whole verbal complex with lexical Verb and auxiliary is present
in both conjuncts. The whole verbal complexes are han acudido ‘have gone’ and han
faltado ‘have missed’.
4.2.3.1.3 Wh-Determiner Sharing in Gapping of Modal: In this type of deter-
miner sharing, the modal Verb is also missing from the non initial conjunct. Consider
sentence (373).
(373) ¿Cuantoshow.many
estudiantesstudents
deberıanshould
faltarmiss
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]T[ ]T
acudirgo
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students should miss the syntax class and how many professors should
go to the monthly meeting?’
In this example (373), the modal Verb deberıan ‘should’ and the wh-Determiner cuantos
‘how many’ are missing from the second conjunct. Still, both elements are interpreted
here. Notice that the wh-Determiner is also interpreted in the non initial conjunct in the
following example (374). This is a similar sentence to (373) with the only difference being
that T is present in both conjuncts.
(374) ¿Cuantoshow.many
estudiantesstudents
deberıanshould
faltarmiss
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
deberıanshould
acudirgo
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students should miss the syntax class and how many professors should
go to the monthly meeting?’
In this example (374), both modal Verbs are part of the complete Verbal complexes
deberıan faltar ‘should miss’ and deberıan acudir ‘should go’ which are part of each
conjunct.
4.2. WH-DETERMINER SHARING IN SPANISH 181
4.2.3.1.4 Wh-determiner sharing in Complex Gapping for Objects: This case
is a wh-determiner sharing example which is embedded in what I have labeled ‘Complex
Gapping for Objects’ in this work. The missing elements in non initial conjuncts in this
type of determiner sharing are: the wh-Determiner, the Verb, and an argument of this
Verb. Consider examples (375, 376) for illustration.
(375) ¿Cuantoshow.many
chicosboys
leCL
dierongave
unaa
florflower
ato
MarıaMarıa
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
ato
Jorge?Jorge‘How many boys gave a flower to Marıa and how many girls gave a flower to
Jorge?’
(376) ¿Cuantoshow.many
chicosboys
leCL
dierongave
unaa
florflower
ato
MarıaMarıa
yand
[ ]D[ ]D
chicasgirls
[ ]Cplx
[ ]Cplx
una
libro?book‘How many boys gave a flower to Marıa and how many girls gave a book to Marıa?’
In sentence (375), the wh-Determiner cuantos ‘how many’ is shared by both wh-phrases
cuantos chicos ‘how many boys’ and cuantas chicas ‘how many girls’. This wh-Determiner
is omitted in the second conjunct. In this same conjunct, there is also omission of the
Verb dieron ‘gave’ and the direct object una flor ‘a flower’. Though these are all omitted
they are still interpreted in the second conjunct whose final interpretation is the one
illustrated: ‘how many boys gave a flower to Jorge?’. Sentence (376) is similar to the
one just explained (375). The missing elements in (376) are the wh-Determiner cuantas
‘how many’, the Verb dieron ‘gave’, and the indirect object a Marıa ‘to Marıa’. Within
this type of gapping, ‘Complex Gapping for Objects’, constructing a non T or V gapping
counterpart is not relevant to this paradigm because in this case, the resulting gap would
not be complex. The only missing element, apart from the omitted Determiner, would be
an argument of the Verb.
4.2.3.1.5 Wh-Determiner Sharing in Complex Gapping for Verbal Complexes:
Determiner sharing here is characterized by the following missing elements: the wh-
Determiner and the whole verbal complex, which may be composed of auxiliary and
participle on the one hand, or modal and infinitive on the other hand, in the sentences
which I construct. The following two examples (377, 378) illustrate this point.
(377) ¿Cuantoshow.many
estudiantesstudents
hanhave
acudidogone
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]Cplx
[ ]Cplx
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
182 CHAPTER 4. WH-DETERMINER SHARING
‘How many students have gone to the syntax class and how many professors have
gone to the monthly meeting?
(378) ¿Cuantoshow.many
estudiantesstudents
deberıanshould
acudirgo
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]Cplx
[ ]Cplx
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students should go to the syntax class and how many professors should
go to the monthly meeting?
In sentence (377), the missing elements that are still interpreted in the final conjunct
are the wh-Determiner cuantos ‘how many’ and the elements which compose the verbal
complex, namely the auxiliary han ‘have’ and the participle acudido ‘gone’. Similarly,
in sentence (378) the wh-Determiner cuantos ‘how many’ and the modal and lexical
Verbs deberıan acudir ‘should go’ are missing, but are also interpreted in all parts of
coordination. In these cases, the non T gapping counterparts coincide with two examples
already provided: sentence (372) for verbal complexes with auxiliary ‘have’ (377), and
sentence (374) for verbal complexes with a modal Verb (378).
4.2.3.1.6 Wh-Determiner Sharing in NEG-nor : This type of wh-determiner shar-
ing is embedded in the NEG-nor construction. Its two main features are easily observable:
the wh-Determiner is shared, and the whole conjunction is in this NEG-nor configuration.
In the example that illustrates this (379), the lexical Verb is also missing from the final
conjunct. Consider (379).
(379) ¿Cuantoshow.many
fısicosphysicists
nonot
fueronwent
ato
lathe
claseclass
deof
matematicasmathematics
ninor
[ ]D[ ]D
linguistaslinguists
[ ]V[ ]V
ato
lathe
claseclass
deof
semantica?semantics
‘How many physicists did not go to the mathematics class and how many linguists
did not go to the semantics class?’
In this sentence (379), negation is present in the first part of coordination and in the
connector. The wh-Determiner cuantos ‘how many’ is shared and the lexical Verb fueron
‘went’ is not overtly pronounced in the final conjunct. Still, interpretation of such elements
is present in the affected conjunct. The next example is the NEG-nor sentence where T
sharing is not present (380).
(380) ¿Cuantoshow.many
fısicosphysicists
nonot
saltaronjumped
enin
lathe
claseclass
deof
matematicasmathematics
ninor
[ ]D[ ]D
linguistaslinguists
cantaronsang
enin
lathe
claseclass
deof
semantica?semantics
‘How many physicists did not jump in the mathematics class and how many lin-
guists did not sing in the semantics class?’
4.2. WH-DETERMINER SHARING IN SPANISH 183
In this example, wh-determiner sharing of cuantos ‘how many’ is possible and main Verbs
saltaron ‘jumped’ and cantaron‘sang’ are present in both conjuncts.
4.2.3.2 Wh-determiner sharing with objects
The following wh-determiner sharing sentences are divided according to the same clas-
sification. The difference between these sentences and the preceding ones is that the
wh-phrases in the following examples are associated to objects instead of subjects. Same
as before I start with wh-determiner sharing embedded in Basic Gapping.
4.2.3.2.1 Wh-Determiner Sharing in Basic Gapping: In this type of wh-determiner
sharing, the wh-phrases are the objects in the conjuncts. The wh-Determiner is shared
and, as in other ‘Basic Gapping’ cases, the only lexical Verb is missing from the final
conjunct. Sentence (381) illustrates the point.
(381) ¿Cuantashow.many
clasesclasses
aat
lathe
semanaweek
tienenhave
losthe
estudiantesstudents
yand
[ ]D[ ]D
reunionesmeetings
alat.each
mesmonth
[ ]V[ ]V
losthe
profesores?professors
‘How many weekly classes do students have and how many monthly meetings do
professors have?’
In this example (381), the wh-word cuantas ‘how many’ and the Verb tienen ‘have’ are
missing from the final conjunct. However, they are interpreted here: ‘how many monthly
meetings do professors have?’. As with examples in which wh-phrases refer to subjects,
example (381) can be expressed without V or T gapping. This is illustrated in (382).
(382) ¿Cuantashow.many
clasesclasses
aat
lathe
semanaweek
tienenhave
losthe
estudiantesstudents
yand
[ ]D[ ]D
reunionesmeetings
alat.each
mesmonth
disfrutanenjoy
losthe
profesores?professors
‘How many weekly classes do students have and how many monthly meetings do
professors enjoy?’
In example (382), both main Verbs tienen ‘have’ and disfrutan ‘enjoy’ are present, and
the wh-word is interpreted in both conjuncts.
4.2.3.2.2 Wh-Determiner Sharing in Gapping of Auxiliary: The logic is the
same in this kind of wh-determiner sharing. Determiner sharing is embedded in ‘Gapping
of Auxiliary’ this time. Missing elements in non-initial conjuncts are: the wh-Determiner
and the Auxiliary. Consider sentence (383).
184 CHAPTER 4. WH-DETERMINER SHARING
(383) ¿Cuantoshow.many
librosbooks
hanhave
leıdoread
losthe
estudiantesstudents
yand
[ ]D[ ]D
revistasmagazines
[ ]T[ ]T
revisadoreviewed
losthe
profesores?professors
‘How many books have students read and how many magazines have professors
reviewed?’
In (383), the wh-Determiner is shared by both objects cuantos libros ‘how many books’
and cuantas revistas ‘how many magazines’. The other missing element is the Auxiliary
han ‘have’ in the final conjunct. This type of determiner sharing also has the non T
gapping counterpart(384).
(384) ¿Cuantoshow.many
librosbooks
hanhave
leıdoread
losthe
estudiantesstudents
yand
[ ]D[ ]D
revistasmagazines
hanhave
revisadoreviewed
losthe
profesores?professors
‘How many books have students read and how many magazines have professors
reviewed?’
In (384), auxiliary han ‘have’ is overt in both conjuncts and wh-determiner sharing is
perfectly possible.
4.2.3.2.3 Wh-Determiner Sharing in Gapping of Modal: Within ‘Gapping of
Modal’, wh-determiner sharing is characterized by the omission of a modal Verb and a
wh-Determiner. Consider (385).
(385) ¿Cuantoshow.many
librosbooks
deberıanshould
leerread
losthe
estudiantesstudents
yand
[ ]D[ ]D
revistasmagazines
[ ]T[ ]T
revisarreview
losthe
profesores?professors
‘How many books should students read and how many magazines should professors
review?’
Here, in sentence (385), the wh-word cuantos ‘how many’ and the modal Verb deberıan
‘should’ are omitted in the second conjunct. This conjunct is interpreted as if every
element was overt. In example (386), the wh-Determiner is also shared because it is
interpreted in both conjuncts and T is present in both conjuncts: the modal Verb deberıan
‘should’ is overt in both sides of the coordination.
(386) ¿Cuantoshow.many
librosbooks
deberıanshould
leerread
losthe
estudiantesstudents
yand
[ ]D[ ]D
revistasmagazines
deberıanshould
revisarreview
losthe
profesores?professors
‘How many books should students read and how many magazines should professors
review?’
4.2. WH-DETERMINER SHARING IN SPANISH 185
4.2.3.2.4 Wh-Determiner Sharing in Complex Gapping for Objects: Wh-
determiner sharing is present in sentences of this type together with the omission of
the Verb and one argument of this Verb. Non initial conjuncts lack the wh-Determiner
that enters the sharing relationship, the Verb and one argument. Still, these elements are
interpreted there as in other determiner sharing cases. Consider sentences (387, 388) for
illustration.
(387) ¿Cuantashow.many
floresflowers
leCL
diogave
suher
sobrinonephew
favoritofavorite
ato
MarıaMarıa
yand
[ ]D[ ]D
librosbooks
[ ]Cplx
[ ]Cplx
suher
primonephew
defrom
Zaragoza?Zaragoza
‘How many flowers did her favorite nephew give to Marıa and how many books
did her nephew from Zaragoza give to Marıa?’
(388) ¿Cuantashow.many
cajasboxes
mandosent
JuanJuan
ato
AnaAna
porby
marsea
yand
[ ]D[ ]D
paquetespackages
[ ]Cplx
[ ]Cplx
PedroPedro
porby
aire?air
‘How many boxes did Juan send by sea and how many packages did Pedro send
by air?’
In sentence (388), the wh-word cuantas ‘how many’ is shared by both wh-phrases cuantas
cajas ‘how many boxes’ and cuantos paquetes ‘how many packages’. The Verb mando
‘sent’ is omitted in the second conjunct and the indirect object a Ana ‘to Ana’ is also
omitted here. All of these elements are interpreted as if they were overtly present in the
conjunct. As explained before, within the type of determiner sharing that occurs within
‘Complex Gapping for Objects’, I do not provide the non T gapping counterpart to the
types of sentences in (387) or (388) because without gapping of T or V the gap would not
be complex.
4.2.3.2.5 Wh-Determiner Sharing in Complex Gapping for Verbal Complexes:
As in other wh-determiner sharing types in this section, wh-phrases which enter into the
sharing relationship are objects. In sentences of this type, apart from the missing Deter-
miner, it is the Verbal Complex that is omitted. Either the Auxiliary and the participle
are missing or the modal Verb and the infinitive are omitted. Sentences (389) and (390)
illustrate this.
(389) ¿Cuantoshow.many
librosbooks
deof
sintaxissyntax
habramight.have
leıdoread
NoamNoam
ChomskyChomsky
yand
[ ]D[ ]D
revistasmagazines
deof
estilostyle
yand
dicciondiction
[ ]Cplx
[ ]Cplx
BarackBarack
Obama?Obama
‘How many syntax books might Noam Chomsky have read and how many style
and diction magazines might Barack Obama have read?
186 CHAPTER 4. WH-DETERMINER SHARING
(390) ¿Cuantashow.many
clasesclasses
deof
sintaxissyntax
deberıanshould
tenerhave
losthe
estudiantesstudents
yand
[ ]D[ ]D
reunionesmeetings
mensualesmonthly
[ ]Cplx
[ ]Cplx
losthe
profesores?professors
‘How many syntax classes should the students have and how many monthly meet-
ings should professors have?’
In example (389), the missing elements in the second conjunct are the wh-word cuantas
‘how many’, the Auxiliary habra ‘might have’ and the lexical Verb leıdo ‘read’. Still,
they are all interpreted in both parts of coordination. In the second example (390), the
missing verbal complex is composed of the modal Verb deberıan ‘should’ and the infinitive
tener ‘have’. Still, all elements are interpreted as if they were overt: ‘how many monthly
meetings should professors have?’. The non T gapping counterparts to this type of wh-
determiner sharing coincide with examples (384) and (386) where verbal complexes are
overt and complete in every conjunct.
4.2.3.2.6 Wh-Determiner Sharing in NEG-nor : Spanish wh-determiner sharing
with objects happens in sentences of this type within the NEG-nor configuration. Specific
features of this configuration are: that the wh-word is shared by both wh-phrases, that
the Verb is missing from the non initial conjuncts, and that conjunction is of the NEG-nor
type. Consider example (391) for illustration.
(391) ¿Cuantashow.many
clasesclasses
deof
matematicasmathematics
nonot
tuvierontook
losthe
fısicosphysicians
ninor
[ ]D[ ]D
clasesclasses
deof
semanticasemantics
[ ]Cplx
[ ]Cplx
losthe
linguistas?linguists
‘How many mathematics classes did physicians not take and how many semantics
classes did linguists not take?’
In this sentence, negation is present in the first conjunct and in the connector ni ‘nor’.
The wh-Determiner cuantas ‘how many’ is shared by both wh-phrases as in the other
determiner sharing examples in this section and the Verb tuvieron ‘took’ is missing from
the final conjunct. All missing elements cuantas ‘how many’ and tuvieron ‘took’ are
interpreted in both parts of coordination. The non T sharing counterpart example is
(392). In the latter, main Verbs boicotearon ‘boycott’ and cancelaron ‘cancel’ are present
in both conjuncts.
(392) ¿Cuantashow.many
clasesclasses
deof
matematicasmathematics
nonot
boicotearonboycotted
losthe
fısicosphysicians
ninor
[ ]D[ ]D
clasesclasses
deof
semanticasemantics
cancelaroncanceled
losthe
linguistas?linguists
‘How many mathematics classes did physicians not boycott and how many seman-
tics classes did linguists not cancel?’
4.3. WH-DETERMINER SHARING: ANALYSIS 187
4.2.4 Interim conclusion
In this section, I have examined the behavior of wh-determiner sharing with respect to:
(i) the general properties of determiner sharing and (ii) to the classification among types
of gapping and determiner sharing sentences that has been done in previous chapters.
Though the behavior of wh-determiner sharing is not exactly the same as the behavior
of regular determiner sharing (i.e., unlike in regular determiner sharing, T sharing is
only optional), wh-Determiners get to be interpreted as shared in these sentences, too.
In what follows, I claim that differences are related to the properties of wh-movement
and wh-movement constructions. Wh-determiner sharing sentences can also be classified
naturally according to the division among determiner sharing types that I have been using
throughout this work. Exposition of these basic facts leads me onto the next section.
4.3 Wh-determiner sharing: analysis
4.3.1 Introduction
The analysis of wh-determiner sharing is similar to the analysis of declarative determiner
sharing sentences that I provide in section 3.5 in chapter 3. In this section, I analyze
specific cases of wh-determiner sharing. This illustrates the way in which the analysis
that I provide in this work accounts for determiner sharing questions. I only focus on the
V or T sharing cases. The non V or T sharing sentences are analyzed in section 4.4.3
below.
4.3.2 Proposal
In this section, I provide the analysis of wh-determiner sharing. In order to do this, I
explain general aspects of these sentences and their structures. Afterwards, I analyze
some of the sentences individually and provide details about them. General aspects to
wh-determiner sharing that need to be considered are:
• The structural analysis of these types of sentences follows the same logic as examples
in prior chapters. The only additional aspect is that wh-movement features are
also included in such structures. Determiner sharing in these sentences is possible
because of the independently motivated wh-movement that affects the wh-DPs that
enter into the sharing relationship. Recall from the previous chapter (section 3.5)
that this is a generalization that applies to all types of determiner sharing.
188 CHAPTER 4. WH-DETERMINER SHARING
• Wh-determiner sharing brings two additional factors to the analysis: (i) the presence
of T or V gapping in the sentences –like in prior chapters–, or (ii) the absence of T
or V gapping in the sentences –which is only possible in wh-determiner sharing–.
• With respect to factor (i), all cases are analyzed according to the types of gap-
ping/determiner sharing in the classification. I propose the same usage of both
small and large conjuncts, and of the processes of movement and deletion. The rea-
sons why I propose both sizes of conjuncts and both types of processes are explained
in sections 2.4.4 and 3.5. In other words, I maintain the details of the analysis so
far and add the typical assumption that phrases affected by wh-movement land at
the specifiers of Focus Phrases (Rizzi (1997), Butler (2004)).
• Factor (ii) is about cases where gapping of T or V is not present in the sentences. In
this section, I analyze the sentences that correspond to factor number (i). Sentences
where there is no gapping of T are analyzed later in the chapter in section 4.4.3.
In the following section, I analyze some specific wh-determiner sharing examples with
gapping of V or T both with subjects and objects to illustrate my proposal. Further
details about the analysis are provided below in the section where the topic of cyclicity
is covered.
4.3.3 Specific Wh-determiner sharing structures
In order to explain the details of the analysis of wh-determiner sharing sentences, I pro-
vide the structures of some of the examples above. I analyze wh-determiner sharing in
basic gapping and wh-determiner sharing in complex gapping for verbal complexes both
for subject and object cases. The choice of these two cases is motivated by the fact that
they illustrate the two mechanisms that derive gapping (see chapter 2 for discussion),
namely, gapping by movement (basic gapping) and gapping by deletion (complex gapping
for verbal complexes). These structures can then be generalized to other subtypes of
gapping (gapping of auxiliary, gapping of modal . . . ).
Wh-determiner sharing in Basic Gapping with Subjects: The sentence that
illustrates this type of wh-determiner sharing is sentence (369), which I repeat here as
(393).
(393) ¿Cuantoshow.many
estudiantesstudents
acudengo
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]V[ ]V
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
4.3. WH-DETERMINER SHARING: ANALYSIS 189
‘How many students go to the syntax class and how many professors go to the
monthly meeting?’
The more salient features of the structure of this sentence (393) come from the fact that
this type of sharing is embedded in a Basic Gapping context. This means that conjuncts
are vPs and that the Verbs acuden ‘go’ are gapped because they ATB move from the
conjuncts to the shared part.4 Determiner sharing of the wh-word is possible in this
sentence because the movement of the wh-phrases is independently triggered. Consider
the structure of sentence (393) in (394).
(394) FocP
cuantos
estudiantesi
Foc’
Foc TP
ti T’
acuden DETP
ti DET’
DET CoP
vP Co’
Co
y
vP
(-)
profesores
v’
tV VP
tV PP
a la reunion
de cada mes
ti v’
tV VP
tV PP
a la clase
de sintaxis
Wh-phrases are generated at the edge of the vPs, the conjuncts. These wh-phrases cuantos
estudiantes ‘how many students’ and cuantos profesores ‘how many professors’ are the
phrases that enter the wh-determiner sharing relationship. For the first wh-Determiner
4The reasoning for these features is explained in section 2.4.4.2 in chapter 2.
190 CHAPTER 4. WH-DETERMINER SHARING
to be spelled out, the first wh-DP makes an intermediate stop at Spec, DETP, in the
path of its movement to Spec, FocP. The spelling out of the other Determiners, those in
the DPs which do not enter the sharing relationship (inside the PPs in this sentence), is
guaranteed because of the DETP at the edges of those DPs. See footnote 6 for further
clarification about both the structure and the context. The construction of the shared
part is continued with the two following steps: (i) the Verbs ATB move to T as explained
before. I follow Depiante and Vicente (2010) and the references cited there to assume
that the Verb movement stops at T. These authors explain that T to C movement does
not take place in Spanish. This is why, in the structure, I depict that the Verb is at T
without further moving to Foc. (ii) The subject keeps on moving upwards: first, to Spec,
TP, second to Spec, FocP. Details and justification for all these movements, other than
wh-movements, are offered in the sections which I have already referred to, section 2.4.4
in chapter 2 and section 3.5 in chapter 3.
Wh-determiner sharing in Complex Gapping for Verbal Complexes with
Subjects: The example which introduces this type of wh-determiner sharing is (377). I
repeat the sentence here as (395) and its structure is represented in (396).
(395) ¿Cuantoshow.many
estudiantesstudents
hanhave
acudidogone
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
[ ]Cplx
[ ]Cplx
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students have gone to the syntax class and how many professors have
gone to the monthly meeting?
4.3. WH-DETERMINER SHARING: ANALYSIS 191
(396) FocP
cuantos
est.i
Foc’
Foc TP
ti T’
han DETP
ti DET’
DET CoP
FocP Co’
Co
y
FocP
(-)
profesores
Foc’
Foc TopP
PP
a la r. de
cada mes
Top’
Top vP
tSbj v’
acudido VP
tV tPP
ti Foc’
Foc TopP
Top’
Top vP
ti v’
acudido VP
tV PP
a la clase
de sintaxis
The central aspects to this structure are the features in Complex Gapping for Verbal
Complexes, namely, PF deletion and small conjuncts.5 In this example (396), the wh-
phrases, that are the subjects of each conjunct, are at the edge of the conjoined parts.
The wh-phrase in the second conjunct cuantos profesores ‘how many professors’ moves
to the edge from its Spec, vP position. Its landing site is Spec, FocP. The subject in the
5See section 2.4.4.5 in chapter 2.
192 CHAPTER 4. WH-DETERMINER SHARING
first conjunct remains in Spec, vP, already the edge of the phase. This first wh-phrase
cuantos estudiantes ‘how many students’ moves to Spec, DETP to get its D spelled out.
Because the second wh-phrase cannot move there, the second wh-D is not spelled out. See
footnote 6 for further clarification about the spelling out of the Determiners in the other
DPs in the sentence. The remnant in the second conjunct moves to the edge in order to
escape from the PF-deletion process. The shared part is formed with the movement of
the first subject to Spec, TP and afterwards to Spec, FocP to validate the wh-features.
In this case, following the explanations in Depiante and Vicente (2010), the auxiliary han
‘have’, that generates at T, does not move upwards to Foc.
Wh-determiner sharing in Basic Gapping with Objects: Sentence (381), which
I repeat here as (397), is representative of this type of determiner sharing. This type of
determiner sharing occurs in wh-movement contexts and in Basic Gapping. Wh-words in
sentences of this type refer to the objects of the conjuncts. The analysis of this sentence
is shown in number (398).
(397) ¿Cuantashow.many
clasesclasses
aat
lathe
semanaweek
tienenhave
losthe
estudiantesstudents
yand
[ ]D[ ]D
reunionesmeetings
alat.each
mesmonth
[ ]V[ ]V
losthe
profesores?professors
‘How many weekly classes do students have and how many monthly meetings do
professors have?’
4.3. WH-DETERMINER SHARING: ANALYSIS 193
(398) FocP
cuantas clases
a la semanai
Foc’
Foc TP
T’
tienen DETP
ti DET’
DET CoP
FocP Co’
Co
y
FocP
(-)reuniones
al mesj
Foc’
Foc vP
los
profesores
v’
tV VP
tV tj
ti Foc’
Foc vP
los
estudiantes
v’
tV VP
tV ti
Because this type of determiner sharing (397)-(398) is embedded in Basic Gapping, the
omission of elements is explained because they move upwards in different types of move-
ments. The Verbs, for instance, ATB move to the shared part, to T specifically. Once in
T, the Verbs stay there because T to C movement does not take place in Spanish Depiante
and Vicente (2010)). Also, following the latter work, the authors state that the presence
of A’-movement blocks the movement of the subject to Spec, TP. In this structure (398),
the location of both los estudiantes ‘the students’ and los profesores ‘the professors’ il-
lustrates the point. The sharing of the wh-element is explained via the DET projection.
The wh-phrase in the first conjunct moves to Spec, DETP, first, in its path to Spec, FocP.
194 CHAPTER 4. WH-DETERMINER SHARING
In Spec, DETP, the wh-Determiner is spelled out. In this example, the motivation of the
movement of the wh-phrase is also independent of any condition related to determiner
sharing. As argued in the previous chapter, the sharing relationship is possible because
of this independent motivation. Like in other examples, the wh-Determiner in the second
conjunct is not spelled out because it cannot move to Spec, DETP. In this chapter, this
is the first structure where different DPs from those which enter the sharing relationship
appear as arguments without being embedded in PPs. Recall from chapter 3, section
3.5.2, that DPs that do not enter the sharing relationship get their Determiners spelled
out at the DETPs that are situated at the edge of the DP phases. I do not include these
DETPs in the structure in order to avoid cluttering the tree.6
Wh-determiner sharing in Complex Gapping for Verbal Complexes with
Objects: Example (389) illustrates this type of wh-determiner sharing. The main fea-
tures in this example and its structure are: that missing elements are affected by deletion,
that conjuncts are small in this type of sentence, and that the wh-movement trigger makes
the sharing of the wh-Determiner possible. I repeat the example here in (399) and I pro-
vide the analysis in (400).
(399) R¿Cuantoshow.many
librosbooks
deof
sintaxissyntax
habramight.have
leıdoread
NoamNoam
ChomskyChomsky
yand
[ ]D[ ]D
revistasmagazines
deof
estilostyle
yand
dicciondiction
[ ]Cplx
[ ]Cplx
BarackBarack
Obama?Obama
‘How many syntax books might Noam Chomsky have read and how many style
and diction magazines might Barack Obama have read?
6This part of the proposal applies to every structure. I normally avoid the representation of theDETPs, which are not part of the determiner sharing relationship, to avoid the mentioned cluttering andto emphasize the sharing relationship better.
4.3. WH-DETERMINER SHARING: ANALYSIS 195
(400) FocP
cuantos libros
de sintaxisj
Foc’
Foc TP
T’
habra DETP
tj DET’
DET CoP
FocP Co’
Co
y
FocP
(-)revistas
de e. y d.i
Foc’
Foc TopP
Barack
Obamax
Top’
Top vP
tx v’
leıdo VP
tV ti
tj Foc’
leıdo TopP
Top’
Top vP
Noam
Chomsky
v’
tV VP
tV tj
As can be seen in (400), wh-Determiner sharing is explained like in other examples.
Because of the independent trigger of the wh-movement, sharing of DET is possible in
the sentence. Following what has been explained about wh-movement in Spanish in the
previous examples, the auxiliary in T, habra ‘might have’, does not move to Foc, and the
movement of the subject to Spec, TP is blocked by the presence of the A’-movement. Both
wh-phrases move to Spec, FocP in the conjuncts. The wh-phrase in the first conjunct
196 CHAPTER 4. WH-DETERMINER SHARING
keeps on moving up to the shared part of the structure: its first step is in Spec, DETP
and its second step is its final landing site, Spec, FocP. The lexical Verbs are kept in the
conjuncts and the second lexical Verb is affected by the PF deletion by phase process.
Before PF deletion applies, the remnant that is not affected by wh-movement does also
move up to Spec, TopP escaping deletion.
4.3.4 Interim conclusion
In this section, I have explained the details of the analysis which I have used to account
for determiner sharing in both declarative sentences and wh-questions. Wh-determiner
sharing is possible because the phrases that enter the determiner sharing relationship
undergo movement for reasons that are independent from the sharing configuration. I
have shown that the specific analyses respond to the conditions that need to hold in
determiner sharing contexts and wh-questions. In the next section, I investigate the
behavior of wh-movement in determiner sharing sentences in more depth. I explain that
determiner sharing, via the analysis which is provided here, is further proof that wh-
movement is cyclic. I easily relate this with the idea which is provided in the determiner
sharing chapter that phases play an important role in the analysis of determiner sharing.
4.4 Cyclicity in Wh-determiner sharing
4.4.1 Introduction
This section is dedicated to the development of aspects that are related to cyclicity in de-
terminer sharing sentences that are affected by wh-movement. The structures that I have
provided so far (394, 396, 398, 400) respond to an analysis which not only accounts for
declarative sentences and questions but also brings evidence in favor of the hypothesis that
wh-movement is cyclic. An example which is easy to consider is (400). In this example,
the positions of the wh-phrases cuantos libros de sintaxis ‘how many syntax books’ and
cuantas revistas de estilo y diccion ‘how many style and diction magazines’ are crucial.
The final position of the first wh-phrase is straightforward. This position is the specifier
of the highest Focus Phrase. Notice that, before getting to this final landing site and also
before moving to Spec, DETP to get its wh-Determiner spelled out, this wh-phrase moves
to the specifier of the IP-internal Focus Projection. This latter position is an intermedi-
ate step of the wh-phrase in the first conjunct. Straight evidence for this intermediate
step is right in the second conjunct. In structure (400), overt copies of remnants, among
which the wh-phrase is present, occupy the specifiers of the IP-internal Focus and Topic
projections. In this non initial conjunct, the overt copy of the wh-movement is located
4.4. CYCLICITY IN WH-DETERMINER SHARING 197
at the intermediate landing site: the specifier of the IP-internal FocP. The same happens
in structure (398). The wh-phrase in the non initial conjunct cuantas reuniones al mes
‘how many meetings each month’ moves to the specifier of IP-internal FocP. In this case,
the word order of the sentence is such that there needs to be a landing site for a derived
position preceding the subject of the final conjunct. I explain further details in relation
to word order in what follows.
Wh-determiner sharing is firstly used as evidence for cyclicity in Arregi and Centeno
(2005). I explain the major contributions in that work in subsection 4.4.2 below. After-
wards, I continue with subsection 4.4.3 which is dedicated to providing more evidence in
favor of the cyclic nature of wh-movement and to relate this phenomenon to phases as
were covered in the previous chapter. Next, I explain the work in Centeno and Vicente
(2009) –subsection 4.4.4– because not only is it intimately related to cyclicity and phases,
but also because my analysis on determiner sharing can capture the examples presented
there. In the end, I conclude this section in subsection 4.4.5.
4.4.2 Arregi and Centeno (2005)
Their evidence in favor of cyclicity is based on word order in wh-determiner sharing
sentences. The crucial examples there are (401)-(405).
(401) ¿Cuantoshow.many
ninosboys
hanhave
leıdoread
librosbooks
yand
[ ]D[ ]D
ninasgirls
[ ]T[ ]T
revisadoreviewed
revistas?magazines
‘How many boys have read books and how many girls have reviewed magazines?’
(402) *¿Cuantoshow.many
ninosboys
hanhave
leıdoread
librosbooks
yand
[ ]T[ ]T
revisadoreviewed
[ ]D[ ]D
ninasgirls
revistas?magazines
‘How many boys have read books and how many girls have reviewed magazines?’
(403) *¿Cuantoshow.many
ninosboys
hanhave
leıdoread
librosbooks
yand
[ ]T[ ]T
revisadoreviewed
revistasmagazines
[ ]D[ ]D
ninas?girls
‘How many boys have read books and how many girls have reviewed magazines?’
(404) ¿Cuantoshow.many
librosbooks
hashave.you
leıdoread
yand
[ ]D[ ]D
revistasmagazines
[ ]T[ ]T
revisado?reviewed
‘How many books have you read and how many magazines have you reviewed?’
(405) *¿Cuantoshow.many
librosbooks
hashave.you
leıdoread
yand
[ ]T[ ]T
revisadoreviewed
[ ]D[ ]D
revistas?magazines
‘How many books have you read and how many magazines have you reviewed?’
The first three examples (401)-(403) are wh-determiner sharing sentences in which the
wh-phrases refer to subjects. In the last two examples (404)-(405), the wh-phrases refer
to objects. The contrast between grammatical word orders (401, 404) and ungrammatical
198 CHAPTER 4. WH-DETERMINER SHARING
word orders (402, 403, 405) shows that final landing sites of wh-phrases need to be in de-
rived positions preceding the Verb. Sentence (401), for example, contrasts with its other
two counterparts (402, 403) because [ ]D ninas ‘[ ]D girls’ precedes the Verb revisado
‘reviewed’. The other two possibilities are ungrammatical. Also, in sentence (404), the
wh-phrase [ ]D revistas ‘[ ]D magazines’ precedes the Verb. This is what makes the
sentence grammatical in comparison to example (405).
The other set of examples that is presented in that work in relation to wh-questions
is a group of wh-determiner sharing sentences where there is no gapping of T. I include
these sentences here (406)-(410):
(406) ¿Cuantoshow.many
ninosboys
hanhave
leıdoread
librosbooks
yand
[ ]D[ ]D
ninasgirls
hanhave
revisadoreviewed
revistas?magazines
‘How many boys have read books and how many girls have reviewed magazines?’
(407) *¿Cuantoshow.many
ninosboys
hanhave
leıdoread
librosbooks
yand
hanhave
revisadoreviewed
[ ]D[ ]D
ninasgirls
revistas?magazines
‘How many boys have read books and how many girls have reviewed magazines?’
(408) *¿Cuantoshow.many
ninosboys
hanhave
leıdoread
librosbooks
yand
hanhave
revisadoreviewed
revistasmagazines
[ ]D[ ]D
ninas?girls
‘How many boys have read books and how many girls have reviewed magazines?’
(409) ¿Cuantoshow.many
librosbooks
hashave.you
leıdoread
yand
[ ]D[ ]D
revistasmagazines
hashave.you
revisado?reviewed
‘How many books have you read and how many magazines have you reviewed?’
(410) *¿Cuantoshow.many
librosbooks
hashave.you
leıdoread
yand
hashave.you
revisadoreviewed
[ ]D[ ]D
revistas?magazines
‘How many books have you read and how many magazines have you reviewed?’
The non sharing of T can be seen in that there is an auxiliary in every conjunct. In the
first three examples, (406)-(408), the wh-phrases that are affected by the sharing rela-
tionship refer to the subjects of the conjuncts. In the other two examples (409)-(410)
in the paradigm, these wh-phrases refer to the objects. The grammaticality contrasts
in these examples only show that wh-phrases need to be in derived positions also when
T sharing is not present. Grammatical examples (406, 409) show that wh-phrases need
to be in derived positions because, in reference to word order, wh-phrases [ ]D ninas
‘[ ]D girls’ and [ ]D revistas ‘[ ]D magazines’ always precede the auxiliaries. I leave
the analysis of this set of examples (406)-(410) for the moment for expository purposes.
I focus here on the analysis of the T sharing examples.
Structure wise, the set of examples with T sharing (401)-(405) is analyzed in Arregi
and Centeno (2005) by means of proposing two fixed DET positions between VP and
4.4. CYCLICITY IN WH-DETERMINER SHARING 199
TP: one for subjects above vP and one for objects above AgroP. For instance, in the
non initial conjunct of example (404), Spec, AgroP is an intermediate position of the
object [ ]D revistas ‘[ ]D magazines’ in its cyclic movement towards Spec, CP. This
idea is schematically represented in (411) which is the structure of the second conjunct in
example (404).
(411) [AgroP (-) revistasi [Agro′ revisado [vP pro tV [VP tV ti] ] ] ]
This necessary intermediate step of the wh-phrases in AgroP in the two conjuncts, which
happens before the movement to Spec, DET, is located between TP and VP. This is evi-
dence in favor of the cyclicity proposals in Chomsky (1986), Fox (2000), and Nissenbaum
(2000).
4.4.3 More evidence and phases
The first goal here is to provide further evidence in favor of cyclicity in wh-movement
within the wh-determiner sharing frame. A second goal is to relate this phenomena to
phases taking into account what has been presented in this work about determiner sharing
and phases. The latter enables me to finally introduce the analysis of sentences which are
wh-determiner sharing structures without gapping of T.
The further evidence in favor of the cyclic condition of wh-movement is brought by a
paradigm of four wh-determiner sharing grammatical sentences (412)-(415). These four
sentences illustrate the possible intermediate sites that wh-movement may use with overt
copies. These sentences are constructed considering the possibilities of both having and
not having T sharing. The sentences are (412)-(415).
(412) ¿Cuantoshow.many
librosbooks
hashave.you
dichosaid
tuyou
quethat
hahas
leıdoread
MarıaMarıa
yand
[ ]D[ ]D
revistasmagazines
[ ]Treviewed
revisadoSusana
Susana?
‘How many books have you said that Marıa has read and how many magazines
has Susana reviewed?’
(413) ¿Cuantoshow.many
librosbooks
hahas
dichosaid
MarıaMarıa
quethat
hahas
leıdoread
yand
[ ]D[ ]D
revistasmagazines
quethat
hemoshave.we
revisado?reviewed
‘How many books has Marıa said that she has read and how many magazines that
we have reviewed?’
200 CHAPTER 4. WH-DETERMINER SHARING
(414) ¿Cuantoshow.many
librosbooks
hahas
dichosaid
MarıaMarıa
quethat
hahas
leıdoread
yand
[ ]D[ ]D
revistasmagazines
[ ]T[ ]T
comentadodeclared
SusanaSusana
quethat
hahas
revisado?reviewed
‘How many books has Marıa said that she has read and how many magazines has
Susana declared that she has reviewed?’
(415) ¿Cuantoshow.many
librosbooks
hahas
dichosaid
MarıaMarıa
quethat
hahas
leıdoread
yand
[ ]D[ ]D
revistasmagazines
hahas
comentadodeclared
SusanaSusana
quethat
hahas
revisado?reviewed
‘How many books has Marıa said that she has read and how many magazines has
Susana declared that she has reviewed?’
In all four of these sentences, (412)-(415), there is a combination of main and embed-
ded Verbs, and coordination with omitted parts of verbal complexes or entire verbal
complexes. This happens in every case within wh-determiner sharing. This paradigm
illustrates the possibilities in omission of verbal complexes gradually from the main part
of the sentence to the embedded part. I explain this idea in more detail with each example.
In the last example (415), there is no omission of any part of the verbal complexes.
In this example, there is only sharing of the wh-Determiner. In the third example (414),
there is omission of T, the auxiliary ha ‘has’, in the main part of the sentence –and the
wh-word is shared–. In the second example (413), the main verbal complex is pronounced
only once at the beginning of the sentence. Still, wh-determiner sharing is possible. And
finally, the first example in the paradigm (412) illustrates cases in which Tense in the
embedded part is omitted and the main verbal complex is overt only at the beginning of
the sentence. In this sentence, the wh-Determiner is also shared.
All four of these possibilities provide evidence for the existence of four potential sites
which are either final or intermediate steps of wh-movement. In these sentences, the wh-
phrase in the final conjuncts is overtly pronounced at these final or intermediate steps,
which constitute the evidence for the cyclic steps of wh-movement that have been proposed
in previous literature. I explain where the landing sites are in each structure example by
example of the paradigm. In order to do this, I represent the structures (416)-(419) of
the examples in the paradigm (412)-(415). For clarity, I only specify the positions of the
wh-phrases and of the elements in the verbal complexes. The first structure (416) corre-
sponds to sentence (412). The second structure (417) corresponds to sentence (413). The
third structure (418) corresponds to sentence (414). The last structure (419) corresponds
to sentence (415).
4.4. CYCLICITY IN WH-DETERMINER SHARING 201
It can be seen in the paradigm of structures (416)-(419) that the intermediate copies of
wh-phrases appear sooner in the sentences (lower in the structures) as we move from one
sentence to another analyzing their structures. These intermediate copies are underlined
here for emphasis.
202 CHAPTER 4. WH-DETERMINER SHARING
(416) FocP
c. librosi Foc’
Foc TP
T’
has FocP
ti Foc’
dicho vP
tu v’
tV VP
tV FocP
ti Foc’
Foc TP
T’
ha DETP
ti DET’
DET CoP
FocPCo’
Co
y
FocP
(-)revistasjFoc’
revisado vP
Susana v’
tV VP
tV tj
ti Foc’
leıdo vP
Marıa v’
tV VP
tV ti
4.4. CYCLICITY IN WH-DETERMINER SHARING 203
In structure (416), coordination is below the main part of the sentence and the Tense
element of the embedded part. The intermediate step of the wh-phrases need to be above
the embedded vP and below the embedded TP. Hence, in this structure, the intermediate
steps are at the embedded IP-internal Focus projection. Evidence for these intermediate
steps comes from the fact that the wh-phrase in the second conjunct displays its overt
copy at that step precisely. The next structure (417) illustrates the location for another
intermediate step.
204 CHAPTER 4. WH-DETERMINER SHARING
(417) FocP
c. librosi Foc’
Foc TP
T’
ha FocP
ti Foc’
dicho vP
Marıa v’
tV VP
tV DETP
ti DET’
DET CoP
FocPCo’
Co
y
FocP
(-)revistasjFoc’
Foc TP
T’
hemos FocP
tj Foc’
Foc vP
v’
revisado VP
tV tj
ti Foc’
Foc TP
T’
ha FocP
ti Foc’
Foc vP
v’
leıdo VP
tV ti
4.4. CYCLICITY IN WH-DETERMINER SHARING 205
In structure (417), there is a difference: coordination needs to be above the Tense element
in the embedded part. The reason for this is simple: the embedded auxiliary is pronounced
in both conjuncts. As a consequence, in this structure, the intermediate step of the wh-
phrases is above the embedded TP but below the main part. That is, in an equivalent
position to the edge of the embedded CP. I locate this intermediate step at the Focus
Projection that is generated when the whole embedded sentence is already formed. In
this example also, the overt copy of the wh-phrase in the first conjunct is straight evidence
in favor the existence of this intermediate step and cyclicity. In the next two structures
(418-419), the intermediate steps are already in the main part of the sentence.
206 CHAPTER 4. WH-DETERMINER SHARING
(418) FocP
c. librosi Foc’
Foc TP
T’
ha DETP
ti DET’
DET CoP
FocPCo’
Co
y
FocP
(-)revistasjFoc’
comentado vP
Susana v’
tV VP
tV FocP
tj Foc’
Foc TP
T’
ha FocP
tj Foc’
Foc vP
v’
revisado VP
tV tj
ti Foc’
dicho vP
Marıa v’
tV VP
tV FocP
ti Foc’
Foc TP
T’
ha FocP
ti Foc’
Foc vP
v’
leıdo VP
tV ti
4.4. CYCLICITY IN WH-DETERMINER SHARING 207
In structure (418), the only verbal element that appears above conjunction is Tense in
the main part of the sentence. Because of this, the intermediate step of wh-movement
in this example is above the main vP and below the main TP. Thus, it is located in the
main IP-internal Focus Projection. Again, the overt copy in the final conjunct is clear
evidence for this intermediate step.
208 CHAPTER 4. WH-DETERMINER SHARING
(419) DETP
c. librosiDET’
DET CoP
FocPCo’
Co
y
FocP
(-)revistasjFoc’
Foc TP
T’
ha FocP
tj Foc’
comentado vP
Susana v’
tV VP
tV FocP
tj Foc’
Foc TP
T’
ha FocP
tj Foc’
Foc vP
v’
revisado VP
tV tj
ti Foc’
Foc TP
T’
ha FocP
dicho vP
Marıa v’
tV VP
tV FocP
ti Foc’
Foc TP
T’
ha FocP
ti Foc’
Foc vP
v’
leıdo VP
tV ti
4.4. CYCLICITY IN WH-DETERMINER SHARING 209
In the last structure (419), no verbal element is left above coordination. This means
that both main and embedded verbal complexes need to be inside coordination. Because
wh-determiner sharing is present in the sentence, the DET Projection needs to be outside
coordination. This is why the only element left outside coordination is the wh-phrase in
the first conjunct which is situated in Spec, DETP. In this case, copies of wh-movement
appear at the Focus Projection, the highest in the whole sentence. Copies of wh-movement
in this case are at the final landing sites. The overt copy of the second conjunct wh-phrase
is at the regular final landing site for wh-movement. The overt copy of the wh-phrase
is at Spec, DETP because after moving to Spec, FocP, this wh-phrase needs to move
to the DET Projection so that its wh-D is spelled out. Wh-determiner sharing provides
evidence for these landing sites, both final and intermediate. Evidence is also in favor of
the proposal that intermediate landing sites are between vP and TP as in Chomsky (1986).
The four possible wh-movement landing sites for which wh-determiner sharing provides
evidence are represented in the following schema (420).
(420) wh-copy– Main CP – wh-copy– Main vP – wh-copy– Embed. CP – wh-copy–
Embed. vP
One main point that can be remarked about the representation in (420) is its clear con-
nection to phases. All wh-copies appear in a straight relation to phases, at the edges of
phases. This adjusts to the proposal for my analysis of determiner sharing, which relies on
the phases notion in both declarative sentences and questions. Let us recall some of the
details of the proposal so that the adjustment between (420) and the proposal is explicitly
stated. I propose that DET projections are located in positions that are not completely
fixed. Their locations vary according to phases. DETPs need to always appear at the
edges of phases, either at the edge of the DP-phase, the vP-phase or the CP-phase, when
needed. In this respect, these DET projections are similar to other information structure
projections –which appear at the edges of phases when needed (Butler (2004))–. Also,
the DP that needs to get its Determiner spelled out needs to always move to Spec, DETP,
the position in which that spelling out is guaranteed. In wh-determiner sharing, the wh-
phrase needs to move to an intermediate landing site available for wh-movement before
getting to Spec, DETP. In (420), intermediate wh-copies are at edges of phases. From
these positions, in wh-determiner sharing sentences, wh-phrases move to Spec, DETP for
the spelling out of the wh-Determiner, which is also at the phase edge, right above the
intermediate landing site. This is why (420) perfectly adjusts to the phases notion in my
proposal.
210 CHAPTER 4. WH-DETERMINER SHARING
The idea that information structure may be at every phase level in the structure, the
CP-phase level included, makes the analysis of wh-determiner sharing sentences with no
sharing of Tense available. Although this has already been vaguely introduced in paradigm
(412)-(415), as well as its structures (416)-(419), I analyze this type of determiner sharing
here so that the details are carefully considered in the structures. Because non T sharing
structures are similar among them, I provide the structures of two of the non T sharing
sentences that I have presented in this chapter. These are sentences (370) and (384), which
I repeat here as (421) and (422) respectively. In the first sentence (421), the wh-phrases
refer to the subjects of the conjuncts. In the second sentence (422), the wh-phrases refer
to the objects of the conjuncts.
(421) ¿Cuantoshow.many
estudiantesstudents
acudengo
ato
lathe
claseclass
deof
sintaxissyntax
yand
[ ]D[ ]D
profesoresprofessors
faltanmiss
ato
lathe
reunionmeeting
deof
cadaeach
mes?month
‘How many students go to the syntax class and how many professors miss the
monthly meeting?’
(422) ¿Cuantoshow.many
librosbooks
hanhave
leıdoread
losthe
estudiantesstudents
yand
[ ]D[ ]D
revistasmagazines
hanhave
revisadoreviewed
losthe
profesores?professors
‘How many books have students read and how many magazines have professors
reviewed?’
Example (421) is a wh-determiner sharing sentence where Tense is not shared. The
main Verb in each conjunct is a lexical Verb that is overt in each conjunct. The analysis
of sentence (421) is (423) below.
4.4. CYCLICITY IN WH-DETERMINER SHARING 211
(423) DETP
cuantos
estudi.j
DET’
DET CoP
FocP Co’
Co
y
FocP
(-)profes.i Foc’
Foc TP
ti T’
faltan FocP
ti Foc’
Foc vP
ti v’
tV PP
a la reunion
de cada mes
tj Foc’
Foc TP
tj T’
acuden FocP
tj Foc’
Foc vP
tj v’
tV PP
a la clase
de sintaxis
This structure (423) reflects all features related to determiner sharing which have been
introduced in this work. It also reflects the features which wh-movement displays in Span-
ish. What is noticeable about this structure is that conjuncts leave space for the final
landing sites of the wh-phrases. The cyclic condition of these wh-movements is repre-
sented by the traces in the conjuncts in (423). The overt copy in the second conjunct
cuantos profesores ‘how many professors’ is the final landing site of this phrase. In the
first conjunct, the wh-phrase cuantos estudiantes ‘how many students’ needs to move
further out of the conjunct but because of a reason unrelated to wh-movements. This
wh-phrase cuantos estudiantes ‘how many students’ needs to move further to the specifier
of DETP so that its wh-Determiner is spelled out. In this case, the specifier of DETP is
212 CHAPTER 4. WH-DETERMINER SHARING
the position in which the wh-phrase in the first conjuncts is overtly spelled out.7
Example (422) is another wh-determiner sharing with Tense present in each conjunct.
Both the auxiliary and the lexical Verb are present in both conjuncts. I analyze the
sentence in (424).
(424) DETP
cuantos
librosj
DET’
DET CoP
FocP Co’
Co
y
FocP
(-)revistasi Foc’
Foc TP
T’
han FocP
ti Foc’
revisado vP
los
profesores
v’
tV ti
tj Foc’
Foc TP
T’
han FocP
tj Foc’
leıdo vP
los
estudiantes
v’
tV tj
This structure (424) is similar to the previous one (423). Conjuncts are large in the
sense that final wh-movement steps take place inside them. The wh-phrase in the second
7In terms of location this position has already been used in Arregi and Centeno (2005) and Centeno(2007b), Centeno (2007a). However, there are differences in the features of the DET projections betweenthose works and this one. In Arregi and Centeno (2005), DET positions are located in fixed positionsin the structure above different projections. Among those, the one DET position that would correspondto the one in question here, DETWh, is above CP in that work. Centeno (2007) already suggested toconsider information structure in more detail and proposed to have the DET position, DETFoc, aboveFocP, which is similar to what I propose in this work. An important difference, however, that existsamong works is that DETs are fixed in other works. In the previous chapter, I have shown, that DETPhrases are not fixed in the structure. These are not completely rigid in their location. They are onlylimited to phase edges, which is something that has been proven to be useful in both declarative sentencesand questions.
4.4. CYCLICITY IN WH-DETERMINER SHARING 213
conjunct is overtly pronounced in the specifier of the highest Focus Phrase. However,
these conjuncts are not sentential because there is a shared part to both conjoined parts.
To this shared part, the wh-phrase in the first conjunct cuantos libros ‘how many books’
moves so that the wh-D is spelled out. The trace of the first wh-phrase that is the highest
in the conjunct represents the final wh-movement step of this wh-phrase. The rest of the
traces in this conjunct represent the cyclic steps which I have provided evidence for in
this work. This also applies to the traces in the second conjunct.
The main idea behind this section, is that the close analysis of wh-determiner sharing
provides evidence for cyclicity. The analysis that I provide in this work, which relies not
only on the notion of cyclicity, but also on the notion of phases, is also useful to determine
which phrases are phases. This is what I focus on in the next subsection, summarizing
the work in Centeno and Vicente (2009).
4.4.4 Centeno and Vicente (2009)
I focus on the work by Centeno and Vicente (2009). Their goal is to determine which
categories are phases by means of the analysis of wh-determiner sharing. In doing so, they
identify categories that are phases, and categories that are not. Since they also rely on the
notions of phases and cyclicity, and their analysis is able to distinguish among categories
that are and are not phases, their data can be evidence in favor of the proposal in this
work which also relies on those notions. The work in Centeno and Vicente is relevant to
this work in two ways: (i) it shows that wh-determiner sharing can be used as evidence to
distinguish among phases and (ii) it can be considered evidence in favor of my proposal
on determiner sharing, which relies on the notions of phases and cyclicity.
Since the introduction of the phases idea (Chomsky (2000), Chomsky (2001), et seq.)),
one part of the debate has been focused on identifying categories that create phase bound-
aries. At the beginning, the claim in Chomsky’s work is that phase boundaries are the
CP projection and the vP projection, the latter only when transitive. However, Legate
(2003) and Sauerland (2003) provide evidence that vPs are also phase boundaries when
predicates are passive, unaccusative, and raising. Identification of phase boundaries is
done by means of checking whether movement can create an intermediate copy at the
edge of the category that needs to be defined. Legate (2003) and Sauerland (2003) find
evidence for the existence of this copy at this position via reconstruction phenomena.
Centeno and Vicente (2009) introduce further evidence in favor of their conclusion by
means of empirical data which is all composed of wh-determiner sharing sentences.
214 CHAPTER 4. WH-DETERMINER SHARING
Based on the cyclicity properties that are displayed in wh-determiner sharing sen-
tences, their prediction is that if passive, unaccusative, and raising vPs create phase
boundaries, it will be possible to place the overt intermediate copies of wh-phrases in the
Spec, vP position. Recall that these overt intermediate copies of wh-phrases are located
in final conjuncts, at the edges, and without a spelled out wh-D –because of the sharing
of the Determiner–. Centeno and Vicente check their prediction against paradigms of
sentences with the different types of predicates. The first paradigm of examples that they
provide is composed of wh-determiner sharing sentences with raising predicates. I include
this paradigm in (425)-(427).
(425) Enin
elthe
mundoplace
laboral,of.work
¿cuantashow.many
ventajasadvantages
hanhave
resultadoturned.out
tenerhave
siemprealways
losthe
hombresmen
yand
[ ]D[ ]D
desventajasdisadvantages
[ ]T[ ]T
parecidoseemed
sufrirsuffer
lasthe
mujeres?women
‘In the work place, how many advantages have men always turned out to have and
how many disadvantages have women seemed to suffer from?’
(426) *Enin
elthe
mundoplace
laboral,of.work
¿cuantashow.many
ventajasadvantages
hanhave
resultadoturned.out
tenerhave
siemprealways
losthe
hombresmen
yand
[ ]T[ ]T
parecidoseemed
sufrirsuffer
lasthe
mujereswomen
[ ]D[ ]D
desventajas?disadvantages
‘In the work place, how many advantages have men always turned out to have and
how many disadvantages have women seemed to suffer from?’
(427) *Enin
elthe
mundoplace
laboral,of.work
¿cuantashow.many
ventajasadvantages
hanhave
resultadoturned.out
tenerhave
siemprealways
losthe
hombresmen
yand
[ ]T[ ]T
parecidoseemed
[ ]D[ ]D
desventajasdisadvantages
sufrirsuffer
lasthe
mujeres?women
‘In the work place, how many advantages have men always turned out to have and
how many disadvantages have women seemed to suffer from?’
As can be seen in the paradigm (425)-(427), the three sentences are constructed so that
the word order in the final conjunct varies from sentence to sentence. All these sentences
have main predicates that are headed by the Spanish raising Verbs resultar ‘turn out’ and
parecer ‘seem’. The only grammatical word order is that in sentence (425). The object
in the final conjunct of this sentence, cuantas desventajas ‘how many disadvantages’, is
at the edge of the vP headed by the raising Verb parecer ‘seem’ (428).
4.4. CYCLICITY IN WH-DETERMINER SHARING 215
(428) FocP
c. ventajasi Foc’
Foc TP
T’
han DETP
ti DET’
DET CoP
FocP Co’
Co
yFocP
(-)desventajasj Foc’
parecido sufrir
siempre las mujeres tj
ti Foc’
resultado tener
siempre los hombres ti
If the possibility exists that this intermediate step is done at the edge of this raising vP,
this vP projects a phase boundary. Notice that not only this word order is a possibility
but it is the only one which is grammatical because the other two sentences (426, 427),
that have different word orders, are ungrammatical.
The same applies to passive predicates in the same kinds of wh-determiner sharing
sentences. The following two examples (429, 430) illustrate it.
(429) ¿Cuantoshow.many
librosbooks
fueronwere
vendidossold
yand
[ ]D[ ]D
revistasmagazines
[ ]V[ ]V
robadas?stolen
‘How many books were sold and how many magazines were stolen?’
(430) *¿Cuantoshow.many
librosbooks
fueronwere
vendidossold
yand
[ ]V[ ]V
robadasstolen
[ ]D[ ]D
revistas?magazines
‘How many books were sold and how many magazines were stolen?’
These examples (429, 430) show that with the passive Verbs fueron vendidos ‘were sold’
and fueron robadas ‘were stolen’ the cyclic step needs to also be at the edge of vP. This can
216 CHAPTER 4. WH-DETERMINER SHARING
be seen because of word order and because intermediate copies in wh-determiner sharing
are overtly spelled out in final conjuncts. When the wh-phrase in the final conjunct
cuantas revistas ‘how many magazines’ is not occupying a position at the edge of vP, like
in (430), the sentence is ungrammatical. Example (429) is grammatical because the overt
intermediate copy is occupying the position in question. This is illustrated in (431).
(431) cuantos librosi fueron [ ti [vP vendidos] ] y [ (-) revistas [vP robadas] ]
The fact that these intermediate steps are done at the edge of these predicates is evidence
that passive vPs do also create phase boundaries. So far, according to this evidence, the
conclusion is that raising and passive vPs create phase boundaries. I turn to unaccusative
predicates in the following. Centeno and Vicente (2009) construct the following sentences
(432, 433) to check the behavior of movement in wh-determiner sharing with unaccusative
predicates.
(432) ¿Cuantoshow.many
hombresmen
hanhave
entradowalked
eninto
elthe
barbar
yand
[ ]D[ ]D
mujereswomen
[ ]T[ ]T
salidowalked.out
defrom
el?it
‘How many men have walked into the bar and how many women have walked out
of it?’
(433) *¿Cuantoshow.many
hombresmen
hanhave
entradowalked
eninto
elthe
barbar
yand
[ ]T[ ]T
salidowalked.out
[ ]D[ ]D
mujereswomen
defrom
el?it
‘How many men have walked into the bar and how many women have walked out
of it?’
The grammaticality contrast between (432) and (433) leads to the same conclusion in
reference to these types of predicates. The predicates in these examples are headed by
the unaccusative Verbs entrar ‘walk in’ and salir ‘walk out’. And when the wh-phrases
move upwards, the intermediate step needs to be at the edge of vP, like in sentence (432).
The word order in example (433) is different. In this example, the wh-phrase in the second
conjunct cuantas mujeres ‘how many women’ is not at the edge of vP and because of this
the sentence is ungrammatical. The schema in (434) illustrates the intermediate step at
the edge of vP in sentence (432).
(434) cuantos hombresi han [ ti [vP entrado en el bar] ] y [ (-) mujeres [vP salido de el] ]
The need for these intermediate steps at the edge of vP in examples with unaccusative
predicates leads to the conclusion that unaccusative vPs create phase boundaries too.
The final conclusion with respect to the notion of phases and cyclicity in wh-determiner
sharing is that this construction brings evidence that raising, passive, and unaccusative
4.5. CONCLUSION 217
vPs are also phase boundaries.
The characteristics of the location of these intermediate steps need further specifi-
cation. Following the theory of determiner sharing that I have proposed in this work,
these intermediate steps are located at the edge of phases. In the previous chapter, I
have shown that a proper analysis of determiner sharing structures requires DETPs to be
located at the edge of phases. It can be seen that, because DETP are proposed to be at
the edges, and because the intermediate steps of wh-movement are right below the DETP
projection, intermediate copies are also always at the edge of phases.
4.4.5 Interim conclusion
The analysis in this work of wh-determiner sharing can be used to provide evidence in
favor of cyclicity in wh-movement as in Chomsky (1986). I have introduced this evidence
in this section. The reliance by this analysis on the notion of phases together with the
cyclicity feature of wh-movement can also be used in order to investigate which categories
project phase boundaries. In this section I have also shown the way this can be done.
The conclusion is in line with the works by Legate (2003) and Sauerland (2003): vPs,
which is the category on which the debate is centered, projects a phase boundary when
predicates are transitive, raising, passive, and unaccusative.
4.5 Conclusion
In this chapter, I have investigated the properties of determiner sharing in Spanish with
respect to a specific type of sentence, namely, wh-questions. I first showed that the be-
havior of wh-determiner sharing displays similar features to that of determiner sharing in
declarative sentences. Because of this, one of the interim conclusions is that the labeling
for sentences of this type is correct. Afterwards, I show that my proposal on determiner
sharing is also suitable to account for wh-determiner sharing.
All properties of my analysis are independently motivated and the addition of wh-
movement characteristics in the structure does not disturb any of the major assumptions
introduced in previous chapters. The features of wh-movement are in tune with my
proposal. This is favorable as it shows that my analysis can adapt to different types
of sentences. One important feature of my analysis of wh-determiner sharing is that it
brings evidence in favor of cyclicity and that vPs are phases also in raising, passive and
unaccusative predicates.
Chapter 5
Conclusion
5.1 Introduction
My main goals in writing this dissertation are: (i) to investigate and to find a proper syn-
tactic analysis of silent elements in Spanish gapping and determiner sharing; (ii) to create
an analysis that generates further instructive discussion within the debate that surrounds
the investigation of silent elements in syntax. In this dissertation, the analysis that is
provided focuses on the positive points of previously existing theories on gapping and
determiner sharing and adds elements that are needed to account for the covered data.
According to the conclusions of my investigation, the eclectic analysis that I provide has
more potential in its scope than other analyses that have been previously provided. I
believe that the reason why resides in the possibility to have both deletion and move-
ment and both large and small conjuncts in order to account for the structures. Notice,
furthermore, that I have shown that this devices have to be used in a controlled fashion,
not freely. I have shown that both tests and specific features that affects the Spanish
language are needed to be considered in the decision towards each device.
I comment on several directions for future research in the sections that follow con-
sidering the aspects that I believe are most important for my theory at the moment. In
section 5.2, I express the need for comparison across languages attending to the classi-
fication of gapping and determiner sharing in my analysis. In section 5.3, I comment
on the difficulty about grammaticality judgments in this work and on a potential way
to solve the difficulty from a methodological point of view. In section 5.4, I enumerate
some of the remaining targets that this analysis can aim for. Some other gapping and
determiner sharing types remain to be tackled and I list some of them here. In section
5.5, the relation of my determiner sharing proposal with already existing proposals on the
219
220 CHAPTER 5. CONCLUSION
syntax of DPs is established with the goal of focusing future research on the placement
of my theory among these other syntactic analyses of Determiner Phrases.
5.2 Cross-linguistic comparison
Further investigation upon these structures and this analysis should come from cross-
linguistic comparison. Some of the challenges that I have encountered in this dissertation
come from features that are characteristic of the syntax of Spanish. The same challenges,
possibly with different consequences, may be found in other languages. One of the next
steps then for future investigation is to deepen into other languages and analyze whether
the theory that works for Spanish works in other languages too. Though gapping and
determiner sharing have already been studied in some of these languages, the kind of
investigation I am proposing here needs to be done in a different way with respect to
the data and to the analysis. The several types of gapping that I have mentioned in
this dissertation are crucial in replicating this work cross-linguistically. Comparison with
other languages should take the gapping classification into account, check similarities and
differences with Spanish, analyze each type of gapping according to the tests and the
syntax of each language. The reason for that is that I have shown that a closer look at
elements such as negation than the investigation that has been done so far is important.
And the classification also responds to such aspects.
In this dissertation, I have taken as starting point studies that are focused on English
or German, among others. Cross-linguistic comparison within gapping and determiner
sharing could start with these languages. The research that has already been done in
these other languages, like English for instance, shows that there is a good basis to tar-
get gapping and determiner sharing data like in this dissertation. In previous literature,
the need for small conjuncts has already been proven in English. With this in hand, a
theory with the additional possibility of large conjuncts is not a big compromise. The
same applies to the processes that account for missing elements. The literature on En-
glish gapping has already proven the need of movement as one of those processes. And
PF-deletion has being well founded in theories that account for other ellipsis phenomena.
Putting these two processes together in the analysis of gapping, the same way which I
have done in this dissertation could be considered a natural extension to this language.
The semantics of negation is other aspect that leads the direction of future research to
where I am explaining. The three different reading that negation displays in the con-
joined contexts that I am considering occur in English and German, which are actually
the languages upon which Repp (2009) does her investigation first. Hence, the extension
5.3. GRAMMATICALITY JUDGMENTS 221
of the classification of gapping, the use of small and large conjuncts, and the proposal
that movement and deletion may be needed in gapping to other languages is well founded.
This applies to determiner sharing in the same way. Previous literature has mainly
focused on English, and what applies to gapping with respect of the size of conjuncts and
movement and deletion applies to determiner sharing too. The study on the phenomena
that is strictly related to the Determiner in these sharing contexts is also founded on
studies which are mainly focused on English, like in Sportiche’s DP-Split. This is a good
basis upon which projecting future research on cross-linguistic investigation. 1
5.3 Grammaticality judgments
Grammaticality judgments deserve special attention when analyzing gapping and deter-
miner sharing in the contexts that I have considered. I have based my analysis on Spanish
gapping and determiner sharing on the judgements which I have represented all through-
out. With respect to judgments, the methodology has been the same as in other syntax
works, and as such, it is as correct as it is in other works. I elicit judgments and I rep-
resent those that correspond to the majority of speakers. In addition, in my case, I have
taken into account judgments of speakers that where consistent all throughout because
I consider consistency across different sentences of the same type is a strong factor in
deciding upon grammaticality.
In an attempt to pursue a picture as near to reality as I can, I make a note here
which may also be useful for future research. While writing the dissertation, I have
observed that some of the judgments do present a certain variability. A possibility for
future work is to create a study which is more detailed and more controlled in order to
obtain more security data wise. In this respect, I would like to draw the attention of
the reader to the discussion that exists in the literature in relation to the elicitation of
grammaticality judgments. I here mention three of the many works that are of interest
to this methodological matter: Phillips (2009) –sections 4 and 6 are specifically relevant
1The comparison could also be with languages with which Spanish is in contact, like Basque orCatalan, for instance. The relevance of this type of cross-linguistic comparison is twofold. First, thestudy of gapping and determiner sharing within the context of ellipsis is interesting because investigationupon these areas is needed in these languages. Second, the study of these structures in Basque or Catalancould also help bring conclusions upon possible language influences in specific judgments and structures ofSpanish data. Previous literature is clear about the influence of one language towards other languages ordialects in language contact situations. One example of such literature is Landa and Franco (1999), whichtreats the Basque Spanish dialect in comparison to other Spanish dialects in contact to other languageslike Quechua, and which enumerates ways in which Spanish gets affected by Basque along the years.This way, a certain influence can be expected in similar language contact situations, Basque-Spanish,Catalan-Spanish, in relation to the constructions I specifically study, gapping and determiner sharing.
222 CHAPTER 5. CONCLUSION
to this work–; Sprouse and Almeida (n.d.) –sections 3.3, 4.2-4.4, and 5 are also more
specifically relevant to this work–; and Phillips and Lasnik (2003).
5.4 Further types of gapping and determiner sharing
One of the points that are easy to conclude after writing this dissertation is that the
label ‘gapping’ has the potential to unify many different types of sentences and that
each sentence has to be carefully inspected so that all details are taken into account
in the analysis. This makes the analysis ‘complex’ and ‘basic’ at the same time. It is
‘complex’ if one sees the behavior of elements in one sentence of this dissertation under
the microscope. There are many aspects going on in each sentence, even in the simplest
ones. And still, the analysis is also ‘basic’ because it only attends to the most basic
gapping types: sentences where either one Verb, or one verbal complex, or one Verb and
its object are omitted. In the case of determiner sharing, only the Determiner is further
omitted. Future research should take further steps to depart from this basic level. Notice,
however, that this basic step is crucial in that, without it, all questions for future research
would be premature. The complexity which is hidden behind each analysis provided here
makes the latter point even stronger. Further investigation for the future should be also
centered on other types of gapping and determiner sharing. Take for instance some of the
examples that Culicover and Jackendoff (2005) (P. 273) mention in their gapping section.
Sentences (435), (436), and (437) are their examples (62d), (62e), and (63b), respectively.
(435) Robin wants to wake up in the morning and be able to speak French, and Leslie,
German.
(436) Robin believes that everyone pays attention to you when you speak French, and
Leslie, German.
(437) Robin thinks that Ferraris are cool, and Leslie Maseratis.
These examples (435)-(437) are dubbed under the gapping label in the literature. In
these examples, the omitted material is more than the omitted material in the example
that I have analyzed so far in this dissertation. In sentence (435), the string ‘wants to
wake up in the morning and be able to speak’ is omitted. There are two Verbs and more
dependent material. In sentence (436), the omitted material is ‘believes that everyone
pays attention to you when you speak’. There are three different verbs in this case. And
in sentence (437), the omitted strings of words are ‘thinks that’, ‘are cool’. Here, there
are two Verbs omitted which are quite separate structure wise because here is a clause
boundary in between. These same examples can be paralleled in Spanish (438)-(440).
5.4. FURTHER TYPES OF GAPPING AND DETERMINER SHARING 223
(438) RobinRobin
quierewants
levantarseto.wake.up
ain
lathe
mananamorning
yand
poderbe.able
hablarto.speak
francesFrench
yand
LeslieLeslie
aleman.German.‘Robin wants to wake up in the morning and be able to speak French and Leslie
German.’
(439) RobinRobin
piensabelieves
quethat
todoseveryone
teCL
escuchanpays.attention
cuandowhen
tuyou
hablasspeak.2SG
francesFrench
yand
LeslieLeslie
aleman.German
‘Robin believes that everyone pays attention to you when you speak French and
Leslie German.’
(440) RobinRobin
piensathinks
quethat
losthe
FerrarisFerraris
sonare
chuloscool
yand
LeslieLeslie
losthe
Maseratis.Maseratis
‘Robin thinks that Ferraris are cool and Leslie Maseratis.’
Examples of this kind with more complexity inside the omitted parts, constitutes the
study of potential future steps in the understanding of gapping. The Spanish examples
(438)-(440) present the exact same complexity as the English counterparts (435)-(437).
One of the strengths of the analysis that I provide in this dissertation is the ample
number of devices that provides in order to analyze the syntax of gapping sentences. That
is why, it may not be difficult to extend the analysis to sentences such as (438), (439),
and (440) within the standard frame of minimalist syntax. Even within the study of the
paradigm that I have analyzed in this dissertation, one could think of ways of completing
the paradigm. One aspect that could also constitute a future step in the investigation of
these types of sentences is to check the behavior of sentences where, for instance, ‘complex
gapping for objects’ is put together with what takes place in ‘complex gaping for verbal
complexes’. This kind of sentence would be a gapping example where omitted elements
are the auxiliary or the modal Verb, the lexical Verb, and one of the arguments of this
Verb. A priori, the behavior of such sentence should be the same as in other ‘complex
gapping contexts’ but it is important to check the details of the analysis very closely.
Other permutations are easy to imagine and checking the different types of gapping will
bring a more complete picture of this structure.
In the case of determiner sharing, future steps may be founded in the same type of
potential complexity and also in aspects specifically related to Determiners. There is a
possible structure within determiner sharing that I label ‘multiple determiner sharing’.
224 CHAPTER 5. CONCLUSION
(441) Muchosmany
chicosboys
bebierondrank
demasiadostoo.many
vasosglasses
deof
aguawater
yand
[ ]D[ ]D
chicasgirls
[ ]V[ ]V
[ ]D[ ]D
tazascups
deof
te.tea
‘Many boys drank too many glasses of water and many girls drank too many cups
of tea’.
In this example (441), sharing of the Determiner happens at the level of both arguments.
Both arguments, though missing an overt Determiner, they are interpreted as if both Ds
were present. It is interesting for future research to investigate sentences of this type
in all the different types of the classification that I have provided. The two DET-D
relationships, at the level of the subject and at the level of the object, need to also be
investigated closely in order to get to appropriate conclusions.
5.5 The syntax of Determiner Phrases
On the one hand, my proposal on determiner sharing is, of course, intimately related to
Sportiche’s proposal of the DP-Split. Notice that the proposal in Sportiche’s work is to
have two fixed Determiner related positions in the structure, one for the subject and one
for the object. The analysis of determiner sharing in this dissertation is a novel adapta-
tion of the DP-Split with respect to previous literature. I have claimed in my analysis of
determiner sharing that Determiner related positions need to project their own projection
and that these DETP need to be at the edges of phases: DP phases in standard sentences;
CP or vP phases in determiner sharing sentences. The interaction of these two theories,
the DP-Split proposal and my determiner sharing proposal, and its consequences should
also be part of future research.
On the other hand, part of my claims about determiner sharing are relevant for the
wider context of the syntax of DPs. Since Abney (1987), many proposals regarding the
internal structure of the Determiner Phrase and its internal projections have been put
forth. See Butler (2004) and references therein for exemplifications of syntactic configu-
rations that have been proposed for the Determiner Phrase from the views of the DP as
a phase and as a regular phrase. Better understanding of the condition and the location
of DETPs in relation to the DP and the elements inside the latter is interesting for future
investigation.
Bibliography
Abney, S.: 1987, The English Noun Phrase in its Sentential Aspect, PhD thesis, M.I.T.
Ackema, P. and Szendroi, K.: 2002, Determiner sharing as an instance of dependent
ellipsis, Journal of Comparative Germanic Linguistics 5, 3–34.
Arregi, K. and Centeno, N.: 2005, Determiner sharing and cyclicity in wh-movement,
in R. Gess and E. Rubin (eds), Theoretical and Experimental Approaches to Romance
Linguistics, John Benjamins, Amsterdam and Philadelphia.
Boeckx, C. and Stjepanovic, S.: 2001, Head-ing toward PF, Linguistic Inquiry 32(2), 345–
355.
Bosque, I. and Demonte, V.: 1999, Gramatica Descriptiva de la Lengua Espanola, Espasa
Calpe, Madrid.
Brody, M.: 2000, Mirror theory, Linguistic Inquiry 31, 29–56.
Brucart, J. M.: 1987, La Elision Sintactica en Espanol, Bellaterra, Barcelona.
Butler, J.: 2004, Phase Structure, Phrase Structure and Quantification, PhD thesis, Uni-
versity of York.
Centeno, N.: 2007a, Coordination of Focus Phrases and DETFoc positions, BIDE 2007,
Bilbao.
Centeno, N.: 2007b, A Determiner position above CP, 37th Linguistic Symposium on
Romance Languages, Pittsburgh.
Centeno, N. and Vicente, L.: 2009, An argument in favor of a vP phase boundary in
raising, passive, and unaccusative predicates, Linguistic Society of America Meeting,
San Francisco.
Chomsky, N.: 1957, Syntactic Structures, Mouton, The Hague.
Chomsky, N.: 1986, Barriers. (= Linguistic Inquiry Monographs, 13.), MIT Press, Cam-
bridge, Mass.
225
226 BIBLIOGRAPHY
Chomsky, N.: 1995, The Minimalist Program, MIT Press, Cambridge, Mass.
Chomsky, N.: 2000, Minimalist inquiries: The framework, in J. U. Roger Martin,
David Michaels (ed.), Step by Step: Essays on Minimalist Syntax in Honor of Howard
Lasnik, MIT Press, Cambridge, Mass., pp. 89–156.
Chomsky, N.: 2001, Derivation by phase, in M. Kenstowicz (ed.), Ken Hale: A Life in
Linguistics, MIT Press, Cambridge, Mass., pp. 1–52.
Chomsky, N.: 2004, Beyond explanatory adequacy, in A. Belletti (ed.), Structures and Be-
yond. The Cartography of Syntactic Structure, Vol. 3, Oxford OUP, New York, pp. 104–
131.
Chung, S., Ladusaw, W. A. and McCloskey, J.: 1995, Sluicing and Logical Form, Natural
Language Semantics 3, 239–282.
Citko, B.: 2007, Determiner sharing from a crosslinguistic perspective, in P. Pica (ed.),
Linguistic Variation Yearbook 2006, John Benjamins, Amsterdam and Philadelphia,
pp. 73–96.
Coppock, E.: 2001, Gapping: In defense of deletion, CLS 37: The Main Session Papers
from the 37th Meeting of the Chicago Linguistics Society, pp. 133–148.
Culicover, P. W. and Jackendoff, R.: 2005, Simpler Syntax, Oxford University Press, New
York.
Depiante, M.: 2000, The Syntax of Deep and Surface Anaphora: A Study of Null Com-
plement Anaphora and Stripping/Bare Argument Ellipsis, PhD thesis, University of
Connecticut.
Depiante, M. and Vicente, L.: 2010, El movimiento y la morfologıa del Verbo, in J. M. B.
y Angel Gallego (ed.), El Movimiento de Constituyentes, Visor Libros, Madrid.
Elbourne, P.: 2008, Ellipsis sites as definite descriptions, Linguistic Inquiry (39), 191–220.
Fox, D.: 2000, Economy and Semantic Interpretation. (= Linguistic Inquiry Monographs,
35.), MIT Press and MIT Working Papers in Linguistics, Cambridge, Mass.
Franchini, E.: 1986, Las Condiciones Gramaticales de la Coordinacion Copulativa en
Espanol, Vol. 102, Romanica Helvetica, Bern, Francke.
Fredin, R.: 1999, Cyclicity and minimalism, in S. D. Epstein and N. Hornstein (eds),
Working Minimalism, MIT Press, Cambridge, Mass., pp. 95–126.
BIBLIOGRAPHY 227
Gallego, A.: 2009, Ellipsis by phase, XIX Colloquium on Generative Grammar, University
of the Basque Country, Vitoria-Gasteiz.
Gleitman, L.: 1965, Coordinating conjunctions in english, Language 41, 260–293.
Goodall, G.: 1987, Parallel Structures in Syntax: Coordination, Causatives and Restruc-
turing, Cambridge University Press, Cambridge.
Grodzinsky, Y. and Finkel, L.: 1998, The neurology of empty categories: Aphasics’ failure
to detect ungrammaticality, Journal of Cognitive Neuroscience 10, 281–292.
Hankamer, J.: 1971, Constraints on Deletion in Syntax, PhD thesis, Yale University.
Hankamer, J.: 1972, On the non-existence of mirror image rules in syntax, in J. Kimball
(ed.), Syntax and Semantics, Vol. 1, Seminar Press, New York and London, pp. 199–212.
Hankamer, J. and Sag, I.: 1976, Deep and surface anaphora, Linguistic Inquiry (7), 391–
428.
Hartmann, K.: 2000, Right Node Raising and Gapping: Interface Conditions on Prosodic
Deletion, John Benjamins, Amsterdam.
Jayaseelan, K. A.: 1990, Incomplete VP deletion and gapping, Linguistic Analysis
(20), 64–81.
Jayaseelan, K. A.: 2001, IP internal Topic and Focus Phrases, Studia Linguistica
55(1), 39–75.
Jimenez-Julia, T.: 1984, La llamada coordinacion negativa en espanol, Verba (11), 213–
244.
Johannessen, J. B.: 1998, Coordination, Oxford University Press, Oxford.
Johnson, K.: 1991, Object positions, Natural Language and Linguistic Theory (9), 577–
636.
Johnson, K.: 2000, Few dogs eat Whiskas or cats Alpo, in K. Kusumoto and E. Villalta
(eds), UMOP 23: Issue in Semantics and its Interface, GLSA Publications, Amherst,
MA, pp. 59–82.
Johnson, K.: 2001, What VP ellipsis can do, what it can’t, but not why, in M. Baltin and
C. Collins (eds), The Handbook of Contemporary Syntactic Theory, Blackwell Publish-
ers, pp. 439–479.
Johnson, K.: 2004, In search of the English middle field.
228 BIBLIOGRAPHY
Johnson, K.: 2009, Gapping is not (VP-) ellipsis, Linguistic Inquiry 40(2), 289–328.
Kayne, R.: 1994, The Antisymmetry of Syntax, MIT Press, Cambridge, Mass.
Kim, J.-S.: 1997, Syntactic Focus Movement and Ellipsis. A Minimalist Approach, PhD
thesis, University of Connecticut.
Lakoff, G. and Peters, S.: 1969, Phrasal conjunction and symmetric predicates, in D. A.
Reibel and S. A. Schane (eds), Modern Studies in English: Readings in Transforma-
tional Grammar, Prentice-Hall, Englewood Cliffs, New Jersey, pp. 113–142.
Landa, A. and Franco, J.: 1999, Converging and diverging grammars, Anuario del Semi-
nario de Filologıa Vasca “Julio de Urquijo”. International Journal of Basque Linguistics
and Philology 33(2), 569–581.
Lasnik, H.: 1995, A note on pseudogapping, in R. Pensalfini and H. Ura (eds), MIT
Working Papers in Linguistics 27: Papers on Minimalist Syntax, pp. 143–163.
Lasnik, H.: 1999, On feature strength: Three minimalist approaches to overt movement,
Linguistic Inquiry (30), 197–217.
Lasnik, H.: 2006, Multiple sluicing in English?, UC Los Angeles.
Lechner, W.: 2000, Bivalent coordination in German, Snippets .
Legate, J.: 2003, Some interface properties of the phase, Linguistic Inquiry 34, 506–516.
Lin, V.: 2000, Determiner sharing, in R. Billerey and B. D. Lillehaugen (eds), Proceedings
of the 19th West Coast Conference on Formal Linguistics, Cascadilla Press, Somerville,
MA, pp. 274–287.
Lin, V.: 2002, Coordination and Sharing at the interfaces, PhD thesis, M.I.T.
Lobeck, A.: 1999, VP ellipsis and the Minimalist Program: Some speculations and pro-
posals, in S. Lappin and E. Benmamoun (eds), Fragments: Studies in Ellipsis and
Gapping, Oxford University Press, Oxford, pp. 98–123.
McCawley, J. D.: 1993, Gapping with shared operators, in D. A. Peterson (ed.), Pro-
ceedings of Berkeley Linguistics Society (BLS), Vol. 19, Berkeley Linguistics Society,
Berkeley: University of California, pp. 245–253.
Merchant, J.: 2001, The Syntax of Silence: Sluicing, Islands, and the Theory of Ellipsis,
Oxford University Press, Oxford.
Moltmann, F.: 1992a, Coordination and Comparatives, PhD thesis, MIT.
BIBLIOGRAPHY 229
Moltmann, F.: 1992b, On the interpretation of three-dimensional syntactic trees, Pro-
ceedings of SALT II.
Moro, A.: 1997, Dynamic antsymmetry: Movement as a symmetry-breaking phenomenon,
Studia Linguistica 51, 50–76.
Moro, A.: 2000, Dynamic Antisymmetry, MIT Press, Cambridge, Mass.
Muadz, H.: 1991, A Planar Theory of Coordination, PhD thesis, University of Arizona,
Tucson.
Neijt, A.: 1979, Gapping: A Contribution to Sentence Grammar, Foris, Dordrecht.
Nissenbaum, J.: 2000, Investigations of Covert Phrase Movement, PhD thesis, MIT.
Oehrle, R.: 1987, Boolean properties in the analysis of gapping, in G. J. Huck and A. E.
Ojeda (eds), Syntax and Semantics 20: Discontinuous Constituency, Academic Press,
San Diego, California, pp. 203–240.
Phillips, C.: 2009, Should we impeach armchair linguists?, in P. C. S. Hiwasaki, H. Hoji
and S.-O. Sohn (eds), Japanese/Korean Linguistics 19, CSLI Publications, Stanford,
CA.
Phillips, C. and Lasnik, H.: 2003, Linguistics and empirical evidence: Reply to Edelman
and Christiansen, Trends in Cognitive Sciences 7, 61–62.
Repp, S.: 2009, Negation in Gapping, Oxford University Press, Oxford.
Richards, N.: 1997, What moves Where When in Which Language?, PhD thesis, MIT,
Cambridge, Mass.
Romero, M.: 1998, Focus and Reconstruction Effects in Wh-Phrases, PhD thesis, Univer-
sity of Massachusetts, Amherst.
Ross, J. R.: 1967, Constraints on Variables in Syntax, PhD thesis, MIT, Cambridge,
Mass.
Ross, J. R.: 1970, Gapping and the order of constituents, in M. M. Bierwisch and K. E.
Heidolph (eds), Progress in Linguistics, Mouton, The Hague, pp. 249–259.
Sag, I.: 1976, Deletion and Logical Form, PhD thesis, MIT.
Sauerland, U.: 2003, Intermediate adjunction with A-movement, Linguistic Inquiry
34, 308–314.
230 BIBLIOGRAPHY
Siegel, M. E. A.: 1984, Gapping and interpretation, Linguistic Inquiry (15), 523–530.
Siegel, M. E. A.: 1987, Compositionality, case, and the scope of auxiliaries, Linguistics
and Philosophy (10), 53–76.
Sportiche, D.: 1996, Clitic constructions, in J. Rooryck and L. Zaring (eds), Phrase
Structure and the Lexicon, Kluwer, Dordrecht, pp. 213–276.
Sportiche, D.: 1997, Reconstruction and constituent structure, talk delivered as an MIT
Linguistics Department Colloquium.
Sportiche, D.: 2005, Division of labor between Merge and Move: Strict locality selec-
tion and apparent reconstruction paradoxes, in N. Klinedinst and G. Kobele (eds),
Proceedings of the Workshop Divisions of Linguistic Labor, The Labretesche Workshop.
Sprouse, J. and Almeida, D.: n.d., A quantitative defense of linguistic methodology.
submitted.
Wilder, C.: 1994, Coordination, ATB and ellipsis, in J.-W. Zwart (ed.), Minimalism and
Kayne’s Asymmetry Hypothesis, Vol. 37 of Groninger Arbeiten zur Germanistischen
Linguistik, pp. 291–331.
Wilder, C.: 1997, Some properties of ellipsis in coordination, in A. Alexiadou and T. A.
Hall (eds), Studies in Universal Grammar and Typological Variation, number 13 in
Linguistik Aktuell/Linguistics Today, John Benjamins, Amsterdam, pp. 59–107.
Williams, E.: 1997, Blocking and anaphora, Linguistic Inquiry 28, 577–628.
Wurmbrand, S.: 2008, Nor: Neither disjunction nor paradox, Linguistics Inquiry
39(3), 511–522.