Post on 03-Nov-2019
A CONSTRAINT ON LINEARISATIONAND THE DOUBLE CONFIGURATION OF IE COORDINATION
MORENO MITROVICjesus college, cambridge
1. Introduction & Preliminaries
1.1. Plan & Desiderata.
(1) The plan[analysisIE data
]−→
[refinementof model
]−→
[constraints
on the derivation
](2) Desiderata
i. Provision a derivational analysis within the standard antisymmetricsyntactic model of coordinate structure.
ii. Unification of the coordinate configurations: coordinate allostructuresderive from an underlyingly single configuration
iii. Refining of the model to account for head-initial pleonasmiccoordination (morphologically compound coordinate heads)
iv. To invalidate the Coordinate Structure Constraint (csc) (Ross, 1967)
v. To show that coordinate structures are subject to a universallinearisation constraint, namely the Final-over-Final Constraint (fofc)
1.2. Three Sets of Background Assumptions.
1.2.1. Coordination Structure. Following Kayne (1994) and Zhang (2010), inter alia,1
I am assuming a relatively traditional template for coordinate structure.
Date: 21 December, 2011 · University of Nova Gorica.1A structure like the one in (3) has been argued for by Blumel (1914), Bloomfield (1933), Bach
(1964), Chomsky (1965), Dik (1968), Dougherty (1969), Gazdar et al. (1985), Goodall (1987) andMuadz (1991), among many others.
1
moreno mitrovic
(3) &P
βP↑
(external) coordinand &0
↑coordinator
αP↑
(internal) coordinand
• Zhang’s (2010) conclusions: the derivation of coordinate construction
· does not create any special syntactic configuration, other than the gen-eral binary complement and specifier/adjunct configuration,
· does not resort to any special syntactic category,2
· is not subject to any special constraint on syntactic operations,
· does not require any special type of syntactic operations, other thanMerge and the step-by-step, one-tail-one-head chains of Move.
1.2.2. Antisymmetry.
• Linear Correspondence Axiom (lca), as per Kayne (1994):
(4) d(A) is a linear ordering of T 3
• spec 〉 head 〉 complent is a universal (underlying) configurational base/order
· the only way in which word-oder differences can emerge is from differ-ential movement-triggering properties of functional categories (let, then,[µ] be a formal movement-trigering/linearising feature)
1.2.3. Syntax of Clitics.
• clitics are defective goals (Roberts, 2010a)
• Spec-Head configuration maps onto encliticisation structure at PF
2I am thus referring to a coordinate head concatenating a categorial complex as &0X, where X is
a variable over syntactic categories. The reasons behind this will become apparent.3Where A is a set of non-terminals and T the set of terminals. For a more detailed formalisation,
see, for instance, (Nilsen, 2003, 19)
2
a constraint on linearisation & the double configuration of ie coordination
2. The Double Configuration of IE Coordination
2.1. The Basic Facts: An Overview of IE Coordination. Early Indo-European(IE) dialects operated a double system of coordinate configuration, in that theyboasted two series of coordinators
• Two kinds of coordinators → [±en/clitic]
• Two kinds of configurations → [±medial]
The plan for this section
• Overview of IE data: Latin, Ancient Greek, Hittite
– show head-final &-configuration of simplex structures↪→ e.g., one prosodic unit (ω0) +coordinator is linearised as[ωi&
0ti]
– show second position (2P) &-configuration (&2) of complex structures↪→ e.g., more than one prosodic unit (ωP) + coordinator is linearised as[ωi&
0[ωP ...ti...]]
– show that linear configuration of coordinate complexes is sensitive to /dependent on the categorial makeup of those complexes↪→ distinction between clausal and subclausal categories
• Closer inspection of Vedic
– Based on problematic data, show violations of csc and refine the theo-retical model of coordination
2.1.1. Latin.
• two sets of &0s
– one postpositive and enclitic (viz. que ‘and’, ue ‘or’) and
– the other prepositive (viz. et, acque/atque ‘and’, aut, uel ‘or’), whichconsistently occupies the medial, that is head-initial, position
(5) [&P [ domihome.loc
] [ duelliiwar.loc
] [&′ queand
[ ti ]]] malebadly
fecisitdid.2.sg.pst
‘In peace and in warfare you have performed villainously.’(Pl., Asin. 559; Ledgeway in press, 32)
3
moreno mitrovic
• loss of postpositive connectors at a relatively early period. (Torrego, 2009,455)4
• que coordinates shorter units, while its function as &0C is very rare
↪→(Agbayani and Golston 2010, Ledgeway in press, inter alia)
• By contrast, as Ledgeway and others5 observe, the non-medial (postpositive)que persists longer, still constituting the principal copulative coordinator inearly writers such as Cato (7), though not Plautus where it had already beenreplaced by et (11).
Subclausal:
(6) uestemgarment.acc
etand
nominanames.acc
‘Garment and names.’ (Capt., 37; Torrego 2009, 458)
(7) defrom
dominoowner.abl
bonogood.abl
bonogood.abl
queand
aedificatorebuilder.abl
‘From a good owner and a good builder.’ (Agr., 1; Ledgeway in press, 32)
(8) plusmore
minusiless
ueor
ti
‘More or less.’ (Capt., 995; Torrego 2009, 456f.)
(9) iniusteunjustly
autor
grauiterseverely
‘Unjustly or severely.’ (Capt., 308; Torrego 2009, 456f.)
Clausal:
• clausal coordination only when headed by an initial &0 like et
(10) [consulemconsul
interficerat]had.killed
etand
[eiushis
exercitumarmy
subunder
iugumyoke
miserat]sent
‘He had killed the consul and sent his army under the yoke.’(Caesar, Bello Gallico 1.12.5; Agbayani and Golston 2011)
4For further discussion on and dating of the disappearance of the non-medial disjunctive coordi-nator in Latin, see Hofmann and Szantyr (1965, 502f.), Janson (1979, 103) and, for a concise andcontemporary approach, Ledgeway (in press).
5cf. (Hofmann and Szantyr, 1965, 473-5), (Gildersleeve and Lodge, 1997, 300), (Torrego, 2009,457)
4
a constraint on linearisation & the double configuration of ie coordination
2.1.2. Ancient Greek.
• Ancient Greek also shows very similar, if not identical, diachronic patternswith regards to competition between medial (prepositive) conjoining coordi-nator kaı and the non-medial (postpositive) te.
• In post-Classical Greek te dies out completely and is replaced by kaı. (Clack-son, p.c.)
Subclausal:
(11) skeeptronscepter
tiimaasihonours
te&
ti
‘Scepter and honours.’ (Aeschylus,Prometheus 171; Agbayani and Golston2011)
(12) tethe
de&
praksaimaking
ti
‘and the making.’ (Aristotle,Poetics 1454a; Agbayani and Golston 2011)
Clausal:
(13) [entauthathere
emeinanwaited.3.pl.pst
heemeraasdays
treis]three
kaı&
[eekecame
Menoon]Menon
‘They waited there three days and Menon came.’ (Xenophon,Anabasis1.2.6; Agbayani and Golston 2011)
• Klein (1992, 10; Tab. 3) provides provisional syntagmatic analysis of HomericGreek coordination.6
– a fairly clear system of coordination emerges from this data.
– In terms of conjunctive coordinators, Homeric Greek clearly operated adouble system of coordination: while kaı is a medial (head-initial andprepositive) coordinator and te is non-medial (postpositive) and, thelinear position of de is unclear. Their syntactic distribution is shown inTable 1.7
• The Classical Greek System: the competition—Tab. 1
– sub-clausal coordinator (te)
6His data base consists of 610 lines of the first book of the Iliad, representing only about 2.2%of the entire text.
7My statistical analysis ignores ambiguous configurations and mixed sequences of coordination.See Klein (1992, 10-44) for a detailed discussion.
5
moreno mitrovic
Table 1. Syntactic distribution of kaı, te and de in Homeric Greek
[medial] clausal sub-clausal
kaı + 41.67% (N = 35) 58.33% (N = 49)
te − 11.11% (N = 5) 88.89% (N = 40)
de ± 89.34% (N = 176) 10.66% (N = 21)
– clausal coordinator (de)
– sitting on the fence (kaı)
• Post-Classical period
– kaı takes over clausal coordination
– te dies out
2.1.3. Hittite. Hittite, as our third comparative IE candidate, also shows the canon-ical IE &2, as shown in examples (14) and (15).
(14) [ ginuwas gad.hi.afor.kneesveils
] patanniof.feet
aand
[ tigisgır.gubstool
]
‘Veils for the knees and a stool for the feet.’ (StBoT. 25.25.i.10)
(15) [ ansu.kur..ra.mescharioteers
] lu.mesis.guskinigolden.grooms
yaand
[ ti humandanall
]
‘Charioteers and all the golden grooms.’ (StBoT. 24.ii.60-61)
Interestingly, clausal coordination in Hittite is always head-initial8 as seen from thefollowing polysyndetic example.
(16) [ [αP nuand
[kanprt
MursilinMursilis.acc
kuennir]βPthey.killed
]αP nuand
[esharblood
ieir]βPshed.3.pl
] ]αP
nuand
[HantilisHantilis
nahsariyatati]βPfeared.3.sg.m
‘And they killed Mursilis and they shed blood and Hantilis was afraid.’ (2BoTU.23.1.33-35)
8See Agbayani and Golston (2010, 8-11) for discussion and evidence.
6
a constraint on linearisation & the double configuration of ie coordination
Table 2. The double system of coordination in early IE languages:[±initial] &0s
[+initial] [−initial]
Sanskrit uta ‘and’ =ca ‘and’=va ‘or’
Avestan uta ‘and’ =ca ‘and’=va ‘or’
Ancient Greek kaı ‘and’ te ‘and’atar ‘but’ de ‘and’
Latin et ‘and’ =que ‘and’at ‘but’ =ve ‘but’
Old Irish ocus ‘and’ =ch ‘and’
Gothic thande ‘and’ =uh ‘and’
Hittite nu ‘and’ =ya ‘and’at ‘but’ =ma ‘but’
Old Church Slavonic i ‘and’ =ze ‘but’
2.2. A Closer Look at Vedic.
2.2.1. Violations of CSC.
(17) [im´ankthese.acc.sg
ca&0
lok´anworld.acc.sg
tk ]i upa-hvayatesummon.2.sg.pres
[(ti)
et´anijthese.acc.pl
ca&0
s´amanichants.acc.pl
tj ] ti
‘He summons these worlds and these chants.’(Sathapathabrahman. a, 1.8.1.19)
7
moreno mitrovic
(18) TopP
DP
im´an ca lok´anTop0 TP
···
VP
V[acc] &P[uCase:acc]
&P
&0
ca[µ]
DP
D
et´ani
NP
s´amani
&0 〈&P〉
&0
ca[µ]
DP
D
im´an
NP
lok´an
8
a constraint on linearisation & the double configuration of ie coordination
Vedic, along with other early IE languages, operated a double system of coordination,whereby coordinate structures fall within two types:
→A one in which the coordinator (&0) is placed in the medial, head-initial (non-dislocated) surface position, e.g., the configurational status of uta in (19);
→B in another type, the coordinator (e.g., ca, va, tu) is placed in a non-medialand dislocated surface position, i.e. placement of ca in (19).
(19) &P[
yasminupon.whom.m
vısvasall
cars.anayamen
utaand.&0
[cyautn´aachievements.pl.nom
jrayansiiregions.n.pl.nom
caand.&0
ti]B ]A
‘He upon whom all men depend [andA], all regions, [andB] allachievements, [he takes pleasure in our wealthy chiefs.]’ (R. gveda, 8.2.33ab)
↪→ · As Klein (1985a, 88) observes, ca in R. gveda normally functions as an in-ner coordinator signalling tighter nexus between shorter units, while utaserves as an outer, higher level concatenator conjoining longer stretchesof discourse.
↪→ The same is found across a wide range of early IE dialects.
2.3. Analysis I: Head-movement (Mitrovic, 2011).
• We may posit that one &0 (ca, va, tu) triggers (head/X0) movement of, andcliticises onto, its complement/internal coordinand (20b), while the other(uta) does not (20a).
(20) a. medial configuration (uta)
&P
αP&0 βP
b. non-medial configuration (ca)
&P
αP&0 βP
ti
• The analysis needs refining as we need to derive
· polysyndetic coordination (n = m) 9
9where n := # of coordinators;m := # of coordinands, as per Zwart (2005)
9
moreno mitrovic
· elements larger than X0 may precede (=be moved by) &0 (cf. Slovenianpa)
2.4. Analysis II: Fine-graining The Head.
• Polysyndetic coordination
(21) . . . dharmedharma/law.loc
ca&
arthecommerce.loc
ca&
kamepleasure.loc
ca&
moks.eliberation.loc
ca&
bharataBharata
r.s.abhabull/giant
yadwhich
ihahere
astiis.3.sg
tadthat
anyatraelsewhere
yadwhich
nanot
ihahere
astiis.3.sg
nanot
tatthat
kvacitanywhere
‘. . .Giant among Bharatas whatever is here on Law, on commerce, onpleasure, on liberation is found elsewhere, but what is not here is nowhereelse.’ (Mahabharata, 1.56.34)
• We need to distinguish between phonologically realised and unrealised coor-dinators
↪→ i. # of phonologically (φ) realised coordinators = nφ = m
ii. # of syntactically (σ) present coordinators = nσ = 2m− 1
∴ The coordinate head is therefore recursive: in monosyndeticcoordination of two arguments, a single &0 projects, whereasthe head re-projects in polysyndetic constructions.
10
a constraint on linearisation & the double configuration of ie coordination
(22) &P
&P
&P
&P
&0
ca[µ]
DP
dharme
&0 &P
&0
ca[µ]
DP
arthe
&0 &P
&0
ca[µ]
DP
kame
&0 &P
&0
ca[µ]
DP
moks.e
• There is also sound empirical evidence in support of the structure proposed in(22) and the number of coordinators amounting to 2m−1. In Avar (NorthernDaghestan), all three coordinators may be phonologically realised in the linearorder predicted by (22).
(23) Ketocat
vaand.&0
hvedog
‘A cat and a dog.’
(24) Ketogicat-and.&0
hvegidog-and.&0
‘A cat and a dog.’
(25) Ketogicat-and.&0
vaand.&0
hvegidog-and.&0
‘A cat and a dog.’
• Are &0φ and &0
σ in IE, or -gi and va respectively in Avar, categories of thesame status? (No.)
solution: a fine-grained set of functional heads
• A cross-structural parallel:
11
moreno mitrovic
· in polysyndetic construction,
· only the coordinator immediately c-commanding/c-selecting its co-ordinand receives phonological realisation (=&0
σ)
· there are no ‘immediately lexical’ external coordinands, i.e thoseexternal arguments merged to [Spec, &0
σ∧¬φ]
(26)
&0 x
Spec
&0 y
Spec
&0 z
Spec
&0 w
· monosyndetic construction (‘x and y’) — structurally symmetric (internal
and the external coordinands, in a sense, balance each other out)
· On the other hand, polysyndetic constructions (‘and x and y’), cf. (26), areasymmetric since they involve internal coordinands alone
↪→ structurally the first internal coordinand raises to the external positionby virtue of merger of the second internal coordinand. (object-to-subjectraising parallel?)
· therefore: one (=lower) &0 (of the two) is responsible for the merger of theinternal argument/coordinand, while the higher head handles the merger ofthe external argument
↪→ let’s dub the former head &0 (=lexical base) and the latter ∧0 (=lighthead/functional layer)
· the coordinate domain is thus theoretically/structurally unified with otherdomains, such as the verbal (V ∼ v), adjectival (A ∼ a) or nominal (N ∼ n).
12
a constraint on linearisation & the double configuration of ie coordination
(27) vna
P
ext. arg. v0
n0
a0
V
NA
P
V0
N0
A0
int. arg.
(28) ∧P
ext. arg.∧0 &P
&0 int. arg.
Outcomes of Analysis:
• cross-structural parallel (internal structure of vP ‖ nP ‖ ∧P)
• analysis structurally captures the difference between IE covert (=&0σ+φ) and
overt (=&0φ) coordinators
· or, the difference between -gi and va in Avar
• the model also allows a non-head/full constituent landing site in [Spec, &P](sub)extracted material, e.g. [∧P [CP ... ] XPi &0 [CP ti ] ]
· cf. Slovenian pa; see Marusic et al. (2011)
13
moreno mitrovic
(29) ∧P
ext. arg.∧0 &P
&0int. arg.
¬ Agree: fcat
[±µ]
Derivation:
¬ &0 c-Selects, that is, it Agrees with fcat, the categorial features of its co-ordinand/s, which it inherits (as per premiliminary assumption 1.2.2 / fn.2)
i. iff fcat < C such that the categorial feature of the coordinand/s is notC ⇒ [µ] is present on the head
↪→ movement of simplex complement ∼ head-final linearisation (orsubextraction of head from within the complex complement)
ii. iff fcat = C such that the categorial feature of the coordinand/s is C ⇒[µ] is not present on the head
↪→ no movement (or subextraction) ∼ no head-final linearisation
↪→ the requirement that [Spec, &P] be filled (∼ [epp]) is satisfiedpleonasmically: an orthotone ∧0 is phonologically realised: a pleaonas-mic head-initial coordinator is generated via head-movement of &0
to ∧0, qua [∧0 [∧−1 u ] + [&0 ta ] ]
• Two heads for the price of one: pleonasm
– reconstructing coordinate heads—Dunkel (1982), summarised in Tab. 3
• The head-initial (configurationally medial) coordinators are not simply thepleonastic forms of enclitic heads:
14
a constraint on linearisation & the double configuration of ie coordination
Table 3. Two sets of IE &0s
enclitic orthotone
?-kwe ?kwo / ?kw ı?-Au ?Aew?-yo ?yo?-te ?to
– Ved. uta, Gr. aute, Lat. aut = ?Au + ?-te
– Ved. u ca, Goth. uh = ?Au + ?-kwe
– Goth. jau = ?yo + ?-Au
– Hit. takku, OIr. toch = ?to + ?-kwe
– freestanding/initial [+C] head = &0 + &0
∴ Head-initial coordinators in IE are morphologically (etymologically) pleonas-mic
(30) &0C/intial = 〈 &0
[orthotonic]
, &0
[enclitic]
〉
2.5. A Bigger Picture: Disharmonic Word-order.
• Vedic/Sanskrit word-order is disharmonic
· predominantly verb-final configuration (V/v-domain)
· clause-level elements, such as Cs, are head-initial (discourse/C-domain)
• Derivation of disharmonic word-orders:
· Some complements, and/or elements contained in those complements,undergo movement and others do not. (Biberauer, Holmberg and Roberts,2010, 63)
· Given the predominantly SOV configuration,
15
moreno mitrovic
i. verbs in Sanskrit carry a ‘movement-triggering’10 feature ([µ]),which causes its objects to undergo movement,
ii. complementisers, on the other, do not carry [µ], hence its comple-ments remain in situ, as shown in (31) & (32), and generalised inTab. 1.
(31) yadwhy/that.C
evapah.part-water.acc
pran. ayatibring.3.sg.act
‘[The reason why/] as he brings water.’ (Sathapathabrahman. a, 1.1.1.14.1)
(32)[
cpC0 . . . NPi V0
[µ] [ 〈NPi〉obj]]
Table 4. Linear Configurationality in Vedic sub/clausal domains
clausal domain subclausal domain= {C} = {T, V, N, A}
initial +final/non-initial +
2.6. Coordination & c-Selection.
• Assuming a feature inheritance in coordination
· whereby a coordinator inherits the categorial and formal features of itscoordinand(s), the differences in linearisation of coordinate complexes inVedic can therefore be analysed as resulting from different c-selectionalproperties of two different &0s:
i. uta-type &0s c-select for (head-initial) clausal elements,
ii. ca-type &0s c-select for (head-final) sub-clausal elements, as perTab. 4.11
10 The Movement-Triggering Feature [µ] is a formal and arbitrary feature (with no se-mantic contents) that triggers movement. It originates on heads and passes monotonically ontotheir Extended Projections (eps).
11See Klein 1985a; 1985b for statistical and evidentiary support for this fact.
16
a constraint on linearisation & the double configuration of ie coordination
2.7. Previous Analysis of Final Position: Against Hale (1987; 1996; 2007).
• This finding also invalidates the phonological/prosodic accounts of coordinatelinearisation and places this phenomenon in narrow syntax:
· since the linear position of non-medial coordinators (ca/va/tu) is sensi-tive to categories they coordinate, c-selection is clearly at work and aphonological account (∼ prosodic flip) of coordination (Hale 1987; 1996;2007) cannot be maintained.
· Another argument in favour of a syntactic analysis of (the double systemof) coordination in Vedic comes from the syntactic constraints that applyto coordinate complexes.
3. The Final-over-Final Constraint (fofc)
3.1. The Basic Facts. The Final-over-Final Constraint (fofc) is as an invariantsyntactic principle that rules out the possibility of a head-final (fn) phrase dominat-ing a categorially alike head-initial (in) phrase (33).
↪→ See Biberauer, Holmberg and Roberts 2010, 63 for details.12
(33) a. X βp
β↑in
αp
α↑in
γp
b. X βp
αp
γp α↑
fn
β↑
fn
c. X βp
β↑in
αp
γp α↑
fn
d. ? βp
αp
α↑in
γp
β↑
fn
= v 〉 obj 〉 aux
(34) The Final-over-Final Constraint (fofc):If α is a head-initial phrase and β is a phrase immediately dominating α,then β must be head-initial. If α is a head-final phrase, and β is a phraseimmediately dominating α, then β can be head-initial or head-final, where:
a. α and β are in the same Extended Projection (Grimshaw 2000)
12For further theoretical developments and empirical validity of fofc, see also Biberauer, Newtonand Sheehan (2010), Biberauer, Holmberg, Roberts and Sheehan (2010), Biberauer and Sheehan(to appear), Roberts (2010b), Sheehan (2010b), Sheehan (2010a), Biberauer, Holmberg and Roberts(2009), Biberauer, Newton and Sheehan (2009), Biberauer et al. (2008), Mobbs (2008), Walkden(2009).
17
moreno mitrovic
b. αp has not been A-moved to [Spec, βp].
18
a constraint on linearisation & the double configuration of ie coordination
3.2. The Immunity of Coordination & Syncategorematicity.
syncategorematicity. Biberauer, Holmberg and Roberts (2010, 81–87) adopt thenotion of syncategorematicity and argue that syncategorematic elements, such aschiefly negators and coordinators, are immune to fofc.
· The reasons, why syncategorematic elements may violate fofc and may ap-parently violate consistent word-order patterns of the language, are shown in(35).
(35) Syncategorematic elements:
a. are not c-selected
b. do not c-select
c. have surface scope determined by their position
• in Vedic—and other early IE languages—coordination, as we have seen, in-volved c-selection
– cross-linguistic evidence suggest that there is a clear c-selection operatingwithin coordination.
↪→ As Zhang (2010) reports, overt coordinators that require conjunctsto be of specific categories in many languages, employ differentcoordinators for coordinate conjuncts of different categories.
↪→ see Appendix in §4
3.3. A True Constraint on Coordination.
• The observation that ...
– the head-final phrases (generally subclausal elements belonging to cate-gories T, V, N, A, etc.) are coordinated by a head-final (or in complexphrases, head-non-initial/2P) coordinator,
– and [the observation] that head initial phrases (generally clausal C-elements) are coordinated by a head-initial &0,
... is consistent with the prediction of fofc, which thus predicts thathigher a X0 is (in the EP), the likelier the X0 is to be initial/on the left.
19
moreno mitrovic
3.4. Diachrony & Paths of Change: Rigidification & Loss of the DoubleSystem.
• fofc predictions
– the higher an element is in the EP, the more likely it is to be initial
– head-final −→ head-initial must proceed top down
↪→ starting with CP (in verbal EP) / PP (in nominal EP)
(36) A model of change in IE coordination
+rigid change
CP
CT
vV PP
PD
N
∴ this explains why final Cs are rarer than OV order
Table 5. The IE Coordination League
Language competition winner in post-period
Ancient Greek kai vs te kai(Vedic) Sanskrit uta vs ca uta-like (!)
Latin et vs que et
generally [+orthotonic] vs [−orthotonic] [+orthotonic][+initial] vs [−initial] [+initial]
20
a constraint on linearisation & the double configuration of ie coordination
Table 6. Single/Double System of coordination in IE
system of coordination
doublesingle
non-medial medial
*IIr. +Indic
t
y Vedic +Classical +Synchronic +
Iranian
t
y Avestan (+) +Old Persian +Synchronic +
Hittite +
Italic
t
y Latin +Synchronic +
Celtic
t
y Old Irish +Synchronic +
Greek
t
y Ancient Greek +Synchronic +
Germanic
t
y Gothic +Synchronic +
Slavonic
t
y OCS +Synchronic (+) +
21
moreno mitrovic
4. Appendix: Evidence for c-Selection
In Mandarin Chinese, for instance, the coordinators gen, tong, yu, and ji coordinatenominals only, whereas the coordinators erqie and you cannot coordinate nominals.Tab. 7, shows cross-linguistic evidence for coordinate c-selection. Further cross-linguistic evidence for categorial compatibility between coordinators and coordinandsis shown in Tab. below, taken from Zhang (2010, 47, ex. 3.5).
(37) DaiDai
JiaoshouProfessor
xihuanlike
hedrink
pijiubeer
{gen/?you}and/and
lu-cha.green-tea
‘Prof. Dai likes to drink beer and green-tea.’ (Zhang, 2010, 46, ex. 3.3a)
(38) DaiDai
JiaoshouProfessor
shanliangkind
{you/?gen}and/and
youmo.humorous
‘Prof. Dai is kind and humorous.’ (Zhang, 2010, 46, ex. 3.3b)
Table 7. Coordination and c-Selection
language coordinator coordinand
fujian kei nominalska non-nominals
japanese to nominalssi finite clauses, Adj’ste non-finite Vs
korean (k)wa nominalsko clauses
malagasy sy VPsary clauses
sissala ka clausesa VPs
somali iyo nominalsoo VPsna clauses
turkish la nominalsıp VPs
yapese ngea nominalsma clauses
22
a constraint on linearisation & the double configuration of ie coordination
References
Agbayani, B. and Golston, C. (2010), ‘Second-position is first-position: Wacker-nagel’s law and the role of clausal conjunction’, Indogermanische Forschungen:Zeitschrift fur Indogermanistik und allgemeine Sprachwissenschaft 115, 1–21.
Agbayani, B. and Golston, C. (2011), Second position is first position, in S. Embleton,G. Giannakis and K. Koerner, eds, ‘Festchrift for Raimo Anttila’, Amsterdam:John Benjamins.
Bach, E. (1964), An Introduction to Transformational Grammars, New York: Holt,Rinehart & Winston.
Biberauer, T., Holmberg, A. and Roberts, I. (2008), Structure and linearization indisharmonic word orders, in C. B. Chang and H. J. Haynie, eds, ‘Proceedings ofthe 26th West Coast Conference on Formal Linguistics’.
Biberauer, T., Holmberg, A. and Roberts, I. (2009), Linearization and the architec-ture of grammar: A view from the final-over-final constraint, in V. Moscati andE. Servidio., eds, ‘Proceedings of the Incontro di Grammatica Generativa XXXV’.
Biberauer, T., Holmberg, A. and Roberts, I. (2010), A Syntactic Universal and itsConsequences. Ms. University of Cambridge.
Biberauer, T., Holmberg, A., Roberts, I. and Sheehan, M. (2010), Parametric Vari-ation, Cambridge: Cambridge University Press.
Biberauer, T., Newton, G. and Sheehan, M. (2009), ‘Limiting synchronic and di-achronic variation and change: the final-over-final constraint.’, Language and Lin-gusitics 10(4), 701–743.
Biberauer, T., Newton, G. and Sheehan, M. (2010), Impossible changes and im-possible borrowings: the final-over-final constraint, in A. Breitbarth, C. Lucas,S. Watts and D. Willis, eds, ‘Continuity and Change in Grammars’, Amsterdam:John Benjamins.
Biberauer, T. and Sheehan, M. (to appear), Disharmony, antisymmetry, and thefinal-over-final constraint, in M. Uribe-Etxebarria and V. Valmala, eds, ‘Ways ofStructure Building’, Oxford: Oxford University Press.
Bloomfield, L. (1933), Language, New York: Holt, Rinehart & Winston.
Blumel, R. (1914), Einfuhrung in die Syntax, Heidelberg: C. Winter.
Chomsky, N. (1965), Aspects of the Theory of Syntax, Cambridge, MA: MIT Press.
23
moreno mitrovic
Dik, S. C. (1968), Coordination: its implications for the theory of general linguistics,Amsterdam: North-Holland.
Dougherty, R. C. (1969), ‘Review of Coordination: Its Implications for the Theoryof General Linguistics by simon c. dik’, Language 45, 624–636.
Dunkel, G. E. (1982), ‘IE conjunctions: pleonasm, ablaut, suppletion’, Zeitschriftfur vergleichende Sprachforschung 96(2), 178–199.
Gazdar, G., Klein, E., Pullum, G. and Sag, I. (1985), Generalized Phrase StructureGrammar, Oxford: Basil Blackwell.
Gildersleeve, B. L. and Lodge, G. (1997), Gildersleeve’s Latin Grammar, third edn,Wauconda, Ill.: Bolchazy-Carducci Publishers.
Goodall, G. (1987), Parallel Structures in Syntax: Coordination, Causatives andRestructuring, Cambridge: Cambridge University Press.
Grimshaw, J. (2000), Extended projection and locality, in P. Coopmans, M. Ev-eraert and J.Grimshaw, eds, ‘Lexical Specification and Insertion’, Amsterdam:Benjamins.
Hale, M. (1987), Notes on Wackernagel’s Law in the language of the Rigveda, inC. Watkins, ed., ‘Studies in memory of Warren Cowgill (1929–1985)’, New York:De Gruyter, pp. 38–50.
Hale, M. (1996), Deriving wackernagel’s law: Prosodic and syntactic factors de-termining clitic placement in the language of the Rigveda, in A. Halpern andA. Zwicky, eds, ‘Approaching Second: Second position clitics Approaching Sec-ond: Second position clitics and related phenomena.’, Stanford: SCLI Publica-tions, pp. 165–197.
Hale, M. (2007), Historical Linguistics: Theory and Method, Oxford: Blackwell.
Hofmann, J. B. and Szantyr, A. (1965), Lateinische Syntax und Stilistik, Munich:Beck.
Janson, T. (1979), Mechanisms of language change in Latin, Stockholm: Almqvist& Wiksell.
Kayne, R. (1994), The Antisymmetry of Syntax, Cambridge, MA: MIT Press.
Klein, J. S. (1985a), Toward a Discourse Grammar of the Rigveda. Part 1., Vol. I,Heidelberg: Carl Winter Universitatsverlag.
Klein, J. S. (1985b), Toward a Discourse Grammar of the Rigveda. Part 2., Vol. II,Heidelberg: Carl Winter Universitatsverlag.
24
a constraint on linearisation & the double configuration of ie coordination
Klein, J. S. (1992), ‘Some Indo-European Systems of Conjunction: Rigveda, OldPersian, Homer’, Harvard Studies in Classical Philology 94, 1–51.
Ledgeway, A. (in press), From Latin to Romance: Morphosyntactic Typology andChange [Chapter 5], Oxford: Oxford University Press.
Marusic, F., Mismas, P. and Zaucer, R. (2011), Some notes on the slovenian secondposition conjunction pa. Paper presented at GLiP-7.
Mitrovic, M. (2011), The syntax of coordination in sanskrit, Master’s thesis, Univer-sity of Cambridge.
Mobbs, I. (2008), ‘functionalism’, the design of the language faculty, and (dishar-monic) typology, Master’s thesis, University of Cambridge.
Muadz, H. (1991), Coordinate structure: a planar representation, PhD thesis, Uni-versity of Arizona.
Nilsen, O. (2003), Eliminating Positions, PhD thesis, University of Utrecht.
Roberts, I. (2010a), Agreement and Head Movement, Linguistic Inquiry Monographs,Cambridge, MA: MIT Press.
Roberts, I. (2010b), Fofc and the realisation of argument structure. Ms. Universityof Cambridge.
Ross, J. R. (1967), Constraints on variables in syntax, PhD thesis, MassachusettsInstitute of Technology.
Sheehan, M. (2010a), Extraposition and antisymmetry. Ms. [under review] Universityof Cambridge.
Sheehan, M. (2010b), ‘Formal and functional approaches to disharmonic word orders’,Newcastle Working Papers in Linguistics (16), 146–166.
Torrego, M. E. (2009), Coordination, in P. Baldi and P. Cuzzolin, eds, ‘New per-spectives on historical Latin syntax’, Vol. 180.1 of Trends in linguistics, New York:Mouton de Gruyter.
Walkden, G. (2009), ‘Deriving the final-over-final constraint from third factor con-siderations’, Cambridge Occasional Papers in Linguistics (5), 67–72.
Zhang, N. N. (2010), Coordination in Syntax, Cambridge Studies in Linguistics,Cambridge: Cambridge University Press.
25
moreno mitrovic
Zwart, J.-W. (2005), Some notes on coordination in head-final languages, in J. Doet-jes and J. van de Weijer, eds, ‘Linguistics in The Netherlands’, Amsterdam: JohnBenjamins, pp. 232–241.
University of CambridgeDepartment of Theoretical& Applied Linguisticsand Jesus CollegeCambridgecb5 8bluk
mm821@cam.ac.uk
http://people.pwf.cam.ac.uk/mm821
26