An on-line study of verb knowledge in individuals diagnosed with AD and FTD. Brain and

Brain and Language xxx (2005) xxx–xxx

www.elsevier.com/locate/b&l

YBRLN 3237 No. of Pages 15; DTD=5.0.1ARTICLE IN PRESS

25 January 2005 Disk Used Jaya (CE) / Hemavathy (TE)

ECTED PROOF

Verb agreements during on-line sentence processing in Alzheimer’s disease and frontotemporal dementia�

Catherine C. Pricea,¤, Murray Grossmanb

a Clinical and Health Psychology, University of Florida, Gainesville, FL, USAb Department of Neurology, University of Pennsylvania, Philadelphia, PA, USA

Accepted 13 December 2004

Abstract

An on-line “word detection” paradigm was used to assess the comprehension of thematic and transitive verb agreements duringsentence processing in individuals diagnosed with probable Alzheimer’s Disease (AD, nD 15) and Frontotemporal Dementia (FTD,nD 14). AD, FTD, and control participants (nD 17) were asked to listen for a word in a sentence. Unbeknownst to the participants,the target word followed an agreement involving a verb’s transitivity or thematic role component. Control participants took signiW-cantly longer to respond to a target word only when it immediately followed a violation of a thematic role agreement or a transitivityagreement, relative to target word detection immediately following the corresponding correct agreement. AD patients were selec-tively insensitive to thematic role agreement violations, although they demonstrated a normal processing pattern for transitivityagreements. This is consistent with previous observations showing selective diYculty with the thematic role component of a verb inAD. FTD patients were insensitive to violations of thematic role and transitivity agreements. FTD patients’ impairment for bothtransitivity and thematic role agreements may reXect a broader degradation of verb knowledge that involves both grammatical andsemantic representations, or diYculty processing sentence structure that also causes a thematic role deWcit. 2005 Published by Elsevier Inc.

Keywords: Sentence comprehension; On-line processing; Verb; Thematic; Transitivity; Alzheimer’s disease; Frontotemporal dementia

1. Introduction investigating on-line sentence processing proWles in 37

UNCORRVerbs impart information about transitivity and the-

matic roles. Although there has been much theorizingabout these two verb elements (Chomsky, 1965; Grim-shaw, 1990; JackendoV, 1990; Shapiro & Levine, 1990),the neurocognitive basis for comprehending these twocomponents of verb knowledge remains unclear. In thecurrent investigation, we attempted to determinewhether these two facets of a verb are dissociable by

� This work was completed in partial fulWllment of the Doctor ofPhilosophy degree in the Department of Psychology, Drexel Universi-ty, Philadelphia, PA. This work was supported in part by Grants fromthe US Public Health Service (AG17586 and AG15116).

¤ Corresponding author. Fax: +1 352 392 6893.E-mail address: [email protected] (C.C. Price).

0093-934X/$ - see front matter 2005 Published by Elsevier Inc.doi:10.1016/j.bandl.2004.12.009

patients with probable Alzheimer’s disease (AD) andFrontotemporal Dementia (FTD).

Verb transitivity and thematic roles represent two corecomponents of verb knowledge. Transitivity is widelythought to be a grammatical feature associated with verbknowledge (Hopper & Thompson, 1980), conveying infor-mation about direct and indirect object use (Levin, 1993).For our purposes, transitive verbs are those verbs thatrequire a direct object for a sentence to be correct. Con-sider the verb ‘put.’ In the sentence “The boy puts the cupon the table,” the verb “put” requires a direct object (i.e.,‘the cup’). A transitive verb like “put” without a directobject is incorrect (e.g., “The boy puts on the table”).Intransitive verbs, by comparison, may have an indirectobject (e.g., “The boy sleeps on the bed”) but should nothave a direct object (e.g., “The boy sleeps the bed”).

38394041424344454647484950515253

1

2

3

45

6

7

8910111213141516171819

20

21222324252627282930313233343536

mailto: [email protected]




2 C.C. Price, M. Grossman / Brain and Language xxx (2005) xxx–xxx



UNCORREC

Verbs also convey meaning about ‘who can do whatto whom.’ This is called “thematic role knowledge.” The-matic role knowledge is widely thought to be con-strained by the conceptual or semantic structure of averb (JackendoV, 1972, 1990). Thematic roles refer tofunctions that participants and objects play in a sen-tence, such as ‘agent,’ ‘patient,’ ‘theme,’ ‘goal,’ ‘instru-ment,’ and ‘location’ (JackendoV, 1994; Levin, 1993).For example, “The girl puts the cup on the table.” In thissentence, ‘The girl’ is the agent because she is performingthe event, ‘the cup’ is the patient because the event isbeing performed on it, and ‘the table’ is the locationbecause it is the place where the cup is being put.

Sentence processing diYculty is a prominent featureof FTD patients with progressive aphasia (Grossman,D’Esposito, et al., 1996; Mesulam, 2001; Snowden,GriYths, & Neary, 1994; Snowden, Neary, et al., 1992;Thompson, Ballard, et al., 1997; Weintraub, Rubin, et al.,1990). With rare exceptions, assessments of sentence pro-cessing have involved traditional, oV-line measures.Task-related resource demands involved with oV-linemeasures (e.g., judging the grammaticality of a sentenceor pointing to one of four pictures based on a sentence)confound our ability to assess the relative contributionof syntactic, semantic, and executive components to sen-tence processing. To circumvent this problem, Tyler,Moss, et al. (1997) used an on-line word-monitoringtechnique to assess sensitivity to a grammatical agree-ment violation in a patient with Progressive Non-XuentAphasia (PNFA). In this technique, subjects are asked todetect a target word in a sentence. Unbeknownst to sub-jects, the target word is strategically placed immediatelyafter the grammatical agreement. Healthy subjects ordi-narily take several milliseconds longer to respond to thetarget word when it immediately follows a grammaticalagreement violation compared to its detection followinga correct grammatical agreement. Evidence to supportthe claim that this delay is related to the location of thetarget word in the temporal window for grammaticalprocessing comes from the absence of the delayedresponse when the target word follows an agreement byseveral words. The patient studied by Tyler, Moss, et al.(1997) was insensitive to the grammatical agreement vio-lation with this on-line technique, as demonstrated bythe absence of a delay in target word detection immedi-ately following a grammatical agreement violation. In asubsequent study using the same technique (Grossman,Rhee, et al., 2004), PNFA patients were insensitive togrammatical agreements such as pluralization, deter-miner–noun agreement, and Q-Xoat in the immediatetemporal window following an agreement violation.However, they were sensitive to a grammatical agree-ment violation following a delay, suggesting a pattern ofslowed activation of grammatical processes. ThesePNFA patients also were signiWcantly impaired in theiroV-line sentence comprehension, and their performance

TED PROOF

correlated with a measure of working memory. Gross-man and colleagues hypothesized that knowledge criticalto long-distance syntactic dependencies and retained inworking memory may have degraded during the slowedgrammatical activation.

Several studies have demonstrated that sentence com-prehension is compromised in patients with Alzheimer’sdisease (AD) as well. The basis for this deWcit, however,has been unclear. Using traditional measures such assentence-picture matching and sentence probe tasks,some investigators attributed impaired sentence compre-hension to a grammatical deWcit (Emery, 1985; Grober& Bang, 1995; Kontiola, Laaksonen, et al., 1990). Morerecent work has focused on a limitation in the cognitiveresources that contribute to sentence comprehensionsuch as working memory (Almor, Kempler, et al., 1999;Croot, Hodges, et al., 1999; Grossman & White-Devine,1998; Rochon, Waters, et al., 1994; Waters, Caplan, &Rochon, 1995; Waters & Caplan, 1997).

Another potential source of comprehension diYcultyin AD may be an impairment at a semantic level of pro-cessing that parallels their diYculty with single wordcomprehension. Although it has been suggested that ADpatients experience some grammatical diYculty asshown by their deWcit matching a brief sentence contain-ing a quantiWer violation such as “The glass containsmuch milk” or “The bowl has many cereal” to one offour pictures (Grossman, Mickanin, et al., 1995), manyother studies demonstrate semantic diYculties. Forexample, AD patients have shown an impaired ability tocorrectly identify incorrect semantic agreements such as“The door is singing,” in which the agent ‘door’ violatesthe selection restriction associated with the predicate ‘issinging’ (Grossman, Mickanin, et al., 1996). In anotherstudy examining the ability to learn the grammatical andthematic aspects of the new verb “wamble,” AD patientswere found to acquire information about its grammati-cal properties (e.g., the new word is a verb that plays aparticular grammatical role in a sentence), but had diY-culty learning about the new word’s thematic role prop-erties (Grossman, Mickanin, et al., 1997). Another recentstudy used an on-line word detection task to demon-strate that AD patients have impaired sensitivity tosemantic agreement violations such as “The handsomeyoung man pointed the yellow in the right direction”(Grossman & Rhee, 2001; Rhee, Antiquena, et al., 2001).AD patients nevertheless showed relatively normal sen-sitivity to grammatical agreement violations involvingpluralization and negation agreements. This sensitivityoccurred, however, in a manner that was delayed beyondthe normally rapid temporal processing window for thetarget grammatical agreement. These Wndings empha-sized the contribution of a semantic limitation toimpaired sentence processing in AD, and related a gram-matical processing deWcit to resource limitations likeslowed information processing speed.

1111111111111111111112121212121212121212131313131313131313131414141414141414141415151515151515151515161616161616

54555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109

01234567890123456789012345678901234567890123456789012345

C.C. Price, M. Grossman / Brain and Language xxx (2005) xxx–xxx 3



UNCORREC

While previous work has examined sentence compre-hension in FTD and AD from the perspective of pro-cessing deWcits associated with long-distance syntacticdependencies, another possibility is that these patientshave impaired sentence comprehension because of adeWcit related to impaired verb knowledge. Verbs are atthe core of a sentence, guiding sentence features such asthematic role assignment and transitivity. It appears thatFTD patients and AD patients have diYculty with verbs.Patients with FTD are reported to have impaired nam-ing of verbs relative to nouns (Cappa, Binetti, et al.,1998; Rhee, Antiquena, et al., 2001). A study by Rhee,Antiquena, et al. (2001) compared noun and verb com-prehension alone and during a concurrent dual task inpatients diagnosed with FTD. These patients were sig-niWcantly less accurate and required signiWcantly longerto make word-picture matching judgments about verbsrelative to nouns at baseline. During concurrent perfor-mance on a secondary task, accuracy decreased andresponse latencies became prolonged for nouns to thepoint that these measures equaled performance withverbs at baseline. Several studies have shown that ADpatients have a small but consistent disadvantage intheir comprehension and naming of verbs compared tonouns (Cappa, Binetti, et al., 1998; Robinson, Grossman,et al., 1996; White-Devine et al., 1996), although excep-tions to verb diYculty have been seen in individualpatient analyses in these studies and in other reportedcases (Fung, Chertkow, et al., 2001; Robinson, Rossor,et al., 1999). In one study that taught AD patients a newverb, diYculty was seen in acquiring thematic informa-tion, although the patients were able to acquire knowl-edge such as major grammatical form class (Grossman,Mickanin, et al., 1997). Work noting AD patients’impairment for function features (Johnson & Hermann,1995; Lambon Ralph, Graham, et al., 1998) implies animpairment in verb thematic relations representing “whodoes what to whom.”

In the present study, transitivity and thematic rolecomponents of verb knowledge were investigated in indi-viduals diagnosed with AD and FTD while using an“on-line” sentence processing procedure. We used theon-line word-detection technique pioneered by Marslen-Wilson and Tyler (1980) to assess processing of theseverb elements while minimizing confounds associatedwith the executive resource limitations observed in AD(Almor, Kempler, et al., 1999; Croot, Hodges, et al.,1999; Grossman & White-Devine, 1998; Kempler,Almor, et al., 1998; Rochon, Waters, et al., 1994; Waterset al., 1995; Waters & Caplan, 1997) and FTD (Elfgren,Passant, et al., 1993; Hodges, Patterson, et al., 1999;Miller, Cummings, et al., 1991; Pachana, Boone, et al.,1996). In the present study investigating verb-relatedsentence agreements, we hypothesized diVerent patternsof sensitivity to thematic and transitivity agreements. Weexpected AD patients to show limited sensitivity to the-

TED PROOF

matic role agreements because of the rich semantic com-ponent of this agreement, but normal sensitivity totransitivity agreements. However, FTD patients wereexpected to demonstrate limited sensitivity to thematic(i.e., semantically related) and transitivity (i.e., grammati-cal) agreements associated with verbs because of theirbroader deWcit with verbs.

2. Methods

2.1. Participants

A total of 46 right handed native English-speakerswere included in this investigation. Fifteen individualswith AD and 14 individuals with FTD were recruitedfrom the Cognitive Neurology Clinic at the Hospital ofthe University of Pennsylvania. These patients werecompared with 17 non-depressed, right-handed, highschool-educated, native English-speakers who were neu-rologically intact.1 The demographic features of the par-ticipants are summarized in Table 1. The three groupswere matched for education [F (2, 43)D0.24; ns] and age[F (2, 43)D3.21; ns]. The AD and FTD groups wereequivalent in their overall dementia severity, as mea-sured by the Mini Mental State Exam (Folstein, Fol-stein, et al., 1975) ([t (27)D 0.45; ns]). There was nodiVerence in male/female ratios across groups[�2 (2)D .70; ns]. The control subjects were recruited fromamong the spouses of the patients. The diagnosis of ADwas made according to the National Institute of Neuro-logic and Communicative Disorders and Stroke-Alzhei-mer’s Disease and Related Disorders Associationcriteria (McKhann, Drachman, et al., 1984). Diagnosisfor FTD was made according to published criteria(Lund/Manchester, 1994; McKhann, Albert, et al., 2001).Patients with FTD were placed into three subgroups(Progressive Non-Fluent Aphasia, PNFA; SemanticDementia, SD; Executive dysfunction, EXEC) based onpublished criteria (Neary, Snowden, et al., 1998), asmodiWed by Davis, Price, et al. (2001) and Price, Davis, etal. (2001) to improve reliability. A consensus mechanismwas used to assign patients to diagnostic groups. At leasttwo trained reviewers considered a semi-structured his-tory, a detailed mental status exam, and a full neurologic

1 Fifty-nine participants were originally tested (19 controls, 17 AD,and 23 FTD patients). This participant dataset that we are reportingwas reduced in size for the following reasons: CONTROLS: 1 scoredin the severely depressed range on the Geriatric Depression Scale(Brink et al., 1982), 1 scored below the normal cut-oV of 24 on theMMSE; AD: 1 reported ambidexterity, 1 was unable to comprehendinstructions for the on-line sentence task; FTD: 3 developed clinicalfeatures of Corticobasal Degeneration with longitudinal follow-up, 4were unable to eVectively comprehend sentence task, 1 did not haveEnglish as a Wrst language, and 1 did not complete the protocol due toa time constraint.

22222222222222222222222222222222222222222222222222222222

166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221

2223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677




UNCORREC

exam to classify patients. When there was a disagree-ment (11% of patients), the full committee resolved thediVerence through discussion. Exclusion criteriaincluded the presence of other causes of dementia suchas vascular disease [Hachinski ischemia scores (Rosen,Terry, et al., 1980) were 2 or less in all patients], primarypsychiatric disorders such as depression or psychosis,medical illnesses or metabolic derangements that mayresult in an encephalopathy, infection diseases that mayresult in a progressive intellectual decline, and other neu-rological conditions aVecting the central nervous systemthat may impact cognitive performance. None of thepatients were taking sedating medications at the time oftesting. Additionally, because our investigation assessedauditory sentence comprehension, all patients wereassessed with an auditory discrimination test for words,Wepman’s Auditory Discrimination Test (Wepman &Reynolds, 1987).

2.2. Materials and procedure

2.2.1. Word detection procedureThe sentence stimuli for the sentence comprehension

tasks were designed around 30 verbs (15 transitive and15 intransitive). These verbs were selected from a groupof 187 intransitive and transitive verbs that were initiallyidentiWed from the American Heritage Dictionary (1999)based on their ‘pure’ transitive and intransitive proper-ties, and then judged for familiarity by 64 individuals (42undergraduates, 22 healthy seniors). Only those verbsendorsed as familiar by 100% of these individuals werecandidates for inclusion in this study. The transitive andintransitive verbs were matched for English corpus fre-quency using Francis and Kucera (1982) norms[t (28)D 0.67; ns], syllable length [t (28)D 0.75; ns], andletter length [t (28)D 0.13; ns].

Each verb was represented in six types of sentencesthat used a simple present tense Noun Phrase–VerbPhrase format limited to an average of eight words persentence [mean # of words per sentenceD 8.06;SDD 0.62]. The transitive and thematic violations, andthe corresponding control sentences (i.e., containingcoherent agreements), were created from judgmentresponses of 55 undergraduate students who were

TED PROOFrecruited to rate each sentence for plausibility. For this

judgment task, each undergraduate was asked to readeach sentence and identify whether the sentence was‘plausible,’ ‘somewhat plausible,’ or ‘not plausible.’ Thistask was conducted to ensure that only the control sen-tences contained coherent semantic and grammaticalagreements relative to the sentences that containedagreement violations. For example, it was imperativethat the thematic control sentences had a coherent sub-ject–verb relationship (such as ‘The cat scratchesƒ’), incontrast to sentences containing a thematic violation(such as ‘The air scratchesƒ’). Appendix A lists the sen-tence stimuli.

Based on a paradigm introduced by Marslen-Wilsonand Tyler (1980) and Tyler (1985), each trial began withthe aural presentation of a target word. Then a briefauditory warning signal was heard 500 ms prior to theaural presentation of a sentence. Participants wereinstructed to press the space bar on the computer assoon as the target word was heard in the sentence. Thisbutton press stopped the computer’s clock that had beeninitiated at the beginning of the target word in the sen-tence. Without informing the participants, half of thesentences contained an agreement violation prior to thetarget word. Within these sentence violations, 33% con-tained a transitive/intransitive violation, and 33% con-tained a thematic agreement violation, and 33% of theremaining sentences served as Wllers containing otherkinds of violations. Each violation condition also had anequal number of control sentences that consisted ofidentical sentences—the only diVerence was that they didnot contain a violation.

Two time window conditions for each violation andcontrol sentence were used to establish the temporalenvelope for agreement processing. These two time win-dow conditions included an immediate condition (theagreement under investigation was immediately fol-lowed by the target word) and a four-syllable delay con-dition (the target word followed the agreement ofinterest with four intervening syllables). Pilot investiga-tions showed that unstressed grammatical morphemessuch as ‘a’ or ‘the’ are more diYcult to identify than con-tent words such as adjectives, adverbs, or nouns duringthis on-line auditory word detection task. For this rea-

343434343435353535353535353535363636363636363636363737373737373737373738383838383838383838

Table 1Mean (§SD) demographic features in the Control Group (Con), Alzheimer Disease (AD), Frontotemporal Dementia (FTD overall), and FTD sub-groups (Progressive NonXuent Aphasia, PNFA, Semantic Dementia, SD; Executive dysfunction, EXEC)

a For the FTD subgroup analyses, the EXEC group was not examined due to the small number of patients representing this group diagnosis.

Age (years) Education (years) MMSE (max D 30)

Con (nD 17) 73.24 (§10.45) 14.06 (§3.00) 28.94 (§1.25)AD (n D 15) 75.67 (§5.72) 14.73 (§2.52) 21.67 (§4.81)FTD (n D 14) 67.79 (§8.51) 14.29 (§2.84) 22.57 (§5.96)PNFA (n D 6) 74.17 (§4.83) 13.50 (§1.97) 21.00 (§7.48)SD (n D 6) 64.33 (§7.15) 15.33 (§3.72) 23.00 (§5.29)EXEC (n D 2)a 59.00 (§9.90) 13.50 (§2.12) 26.00 (§1.41)

3333333333333434343434

278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333

45678901234567890123456789012345678901234567890123456789




UNCORREC

son, both the immediate and delay target words for allsentences were content words (i.e., nouns, adverbs, andadjectives) rather than articles (i.e., ‘a,’ ‘the’). Each targetword was also tested for predictability within its sen-tence context using a cloze procedure. Immediate anddelay target word predictability was assessed by a groupof undergraduate students (total nD 44) with a cut-oV

rule requiring that the target could not be judged as pre-dictable by more than 10% of the undergraduates. Inaddition to these criteria, all target words in each sen-tence type were matched for frequency of use in theEnglish corpus (Francis & Kucera, 1982) and the typesof sentence violations contained an equal number ofnoun, prepositions, adverb, and adjective targets[�2 (6)D 4.38; ns].

Overall, across all of these conditions, a total of 360sentences were presented to each participant. The 360sentences were divided into four balanced blocks ofstimuli. Each block contained 90 sentences with equalnumbers of randomly ordered violations and controlitems. Each sentence block took approximately 15–20minto complete. A stimulus sentence containing an agree-ment violation and its paired control sentence featuringthe corresponding coherent agreement were placed indiVerent blocks, and these were distributed so that pairsof blocks containing these pairs of sentences wereadministered in diVerent sessions separated by 2–4weeks. To ensure that participants were listening to sen-tences and not merely performing a vigilance test for asingle word, each participant’s knowledge of the contentof 10% of the correct sentences in each block was probedrandomly after a target word response was elicited: Theywere asked a question about a fact in the sentence andrequired to provide an answer. Patients who were unableto complete this requirement were excluded from thestudy.

A training procedure was designed with six sentencesto introduce the participants to the word detection para-digm, to familiarize them with the response modality,and to prepare them for random question probes duringthe course of stimulus sentence presentation. Each par-ticipant was given up to Wve repetitions of this trainingprocedure before the experimental testing procedurebegan. Patients who could not complete the training pro-cedure were excluded from the study.

The stimuli were digitized by SoundEdit v2.0 softwareusing 16 bit sound Wles, stimulus presentation was con-trolled by PsyScope v1.2.5 software (Cohen, MacWhin-ney, Flatt, & Provost, 1993), and a Macintosh G3 laptopcomputer was used to administer the stimuli and recordthe responses. Latencies to respond to the target wordswere analyzed once very long (>10 s) and very short(<100 ms) responses had been eliminated, and theremaining responses had been screened within each par-ticipant using a two-standard deviation Wlter based oneach participant’s mean response latency. This resulted

TED PROOF

in the elimination of equal percentages of items from theprotocol for control participants [mean§SDD 7.14§4.20%], AD patients [mean§SDD5.91§ 4.26%], andFTD patients [mean§SDD 8.57§4.58%; F (2, 43)D1.37, ns].

2.2.2. Additional cognitive background tasksMeasures of language and executive function were

administered to all participants. These included:Test for the Reception of Grammar (TROG; Bishop,

1989)—The TROG is an 80-item four-choice sentence-picture-matching test. It is divided into 20 blocks of fouritems each, with each block testing a diVerent lexical,morphosyntactic or syntactic construction. A block isscored as passed only if all four items within it areanswered correctly. Later blocks are generally more diY-cult than earlier blocks. The dependent variable includedtotal number of items correct.

Pyramid and Palm Tree Test (PPT; Howard, 1992)—This test is designed to measure semantic associates forconcrete noun words and pictured objects. For the pur-pose of this experiment, individuals were tested on theirability to judge the associativity of 52 word sets in aforced-choice, two-alternative format (e.g., given the tar-get ‘Glass,’ which of these two words goes best with thetarget: ‘Bottle,’ ‘Mug,’). The dependent variable includedtotal number of items correct.

Verb Similarity Task (VST)—The VST uses aforced-choice, 2-alternative format (paralleling the Pyra-mid and Palm Tree Test) to assess low and high synonymassociativity. This task presents a target verb visuallycentered on a computer screen and above two otherverbs. The participant is required to choose which ofthese two words is most similar in meaning to the targetword. For this task, we attempted to control for diVer-ences in verb semantic representation (verb of motion,verb of cognition, and verb of perception) and argumentstructure (transitive vs. intransitive). To identify thesemantic representation of each target word, werecruited 58 individuals (English Wrst language; 35undergraduate students; 23 healthy seniors) to catego-rize 187 verbs as either a verb of motion or a verb of per-ception/cognition. From this set, 50 verbs were identiWed(25 verbs of motion, 25 perception/cognition verbs).Together, these 50 verbs included 25 transitive verbs, 19intransitive, and 6 verbs that could be either transitive orintransitive depending on sentence role (as identiWedfrom the American Heritage Dictionary, 1999). Theforced-choice responses for the task were developedbased on a normative database in which undergraduatestudents (nD 86; all native English-speakers) were askedto listen to each target verb and immediately list at leastWve associated verbs. The participants were providedwith practice examples prior to administration of thetask (i.e., “The verb SNEEZE might generate the verbsBLOW, SNIFFƒ”). The correct answer and competing

44444444444444444444444444444444444444444444444444444455

390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445

4647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798990001




NCORREC

foils were selected according to the frequency ofresponse. The original 50 triads were reduced to 48 whenwe found that healthy seniors were performing at chanceon two of the items. Items were presented in a randomorder, and the correct force-choice responses were bal-anced in their right vs. left location. For the verb task, alltargets and forced-choice responses were matched forEnglish corpus frequency and for their use as a verb withat least a 5:1 verb/noun ratio using Francis and Kucera(1982) norms [frequency F (2, 141)D1.80, ns; verb/nounratio F (2, 141)D0.71, ns]. In addition, all targets andresponses were matched in English corpus frequency tothat of the Pyramid and Palm Tree Test triads [frequencyt (298)D 0.18, ns].

Trail Making Test Part B (Trails B test; Reitan,1958)—A measure of planning and inhibitory controlthat requires participants to trace a line between anascending series of alternating numbers and letters ran-domly arrayed on a printed page. Our dependent vari-able included the amount of time required to completethe task (one AD patient refused to complete the Trail Btest).

2.2.3. Statistical considerationsRaw reaction times were recorded for the number of

milliseconds it took a person to respond to a target wordfollowing a verb agreement that is correct or incorrect.From these raw reaction times, diVerence scores werecomputed. The equation for the “diVerence score” was:[ms reaction time to a target word following an incorrectagreement¡ms reaction time to a target word followinga correct agreement]. Sensitivity to an agreement viola-tion involves a delay in reaction time to target worddetection immediately after the abnormal agreement rel-ative to target word detection immediately following acorrect agreement. However, a return to baseline follow-ing completion of processing for the grammatical agree-ment is demonstrated by equivalent latencies to targetword detection following an incorrect agreement and acorrect agreement by about four syllables downstreamfrom the violation. To assess this normal processing pat-tern, an immediate “diVerence score” and a delay “diVer-ence score” were computed. The equation for the

TED PROOF

immediate “diVerence score” was: [ms reaction time to atarget word immediately following an incorrectagreement¡ms reaction time to a target word immedi-ately following a correct agreement]. The equation forthe delay “diVerence score” was: [ms reaction time to atarget word located four syllables downstream from theincorrect agreement¡ms reaction time to a target wordlocated four syllables downstream from the correctagreement]. If normal processing is demonstrated, thenthe “diVerence score” for the immediate position shouldbe greater than the “diVerence score” occurring down-stream from the verb’s agreement. We conducted arepeated measures ANOVA to examine performancediVerences between participant groups. Within-groupanalyses were conducted for each participant group aswell, to conWrm that normal processing with the sen-tences took place for the health senior control group andto examine our speciWc hypotheses about AD and FTDperformance patterns for the transitive and thematicconditions. For the within-group analyses, two sets ofpair-wise comparisons were conducted for each partici-pant group. First, for each sentence condition, we con-ducted pair-wise comparisons of immediate and delayed“diVerence scores.” Second, to conWrm sensitivity pat-terns, follow-up planned comparisons were conductedbetween raw reaction times to target words followingincorrect agreements and raw reaction times to correctagreements. Tables 2 and 3 provide the raw reactiontimes for both the thematic and transitive conditions.

3. Results

3.1. Repeated measures ANOVA

A 3 (Group: Control, AD, FTD)£ 2 Sentence Condi-tion (Thematic, Transitive)£Time (Immediate, Delay)Repeated Measures ANOVA was used to examinediVerences in group sensitivity patterns to each of thesentence conditions. The between-group main eVect wasnot signiWcant [F (2, 43)D 2.53, pD .09] but suggested atrend for the FTD patients to be less sensitive regardlessof sentence type of violation position (millisecond over-

5555565656565656565656565757575757575757575758585858585858585858595959595959595959596060
UTable 2
Mean (§SD) millisecond reaction time for the thematic conditions (control/violation) by target position (immediate/delay) for the Control (Con,n D 17), Alzheimer Disease (AD, n D 15), Frontotemporal Dementia (FTD overall, n D 14), and FTD subgroups (Progressive NonXuent Aphasia,PNFA, nD 6; Semantic Dementia, SD, n D 6; Executive/social dysfunction, EXEC, nD 2)

Thematic control Thematic violation

Immediate Delay Immediate Delay

Con 493.14 (§84.24) 451.07 (§69.93) 536.55 (§89.65) 453.76 (§63.60)AD 653.57 (§220.45) 585.53 (§251.27) 682.12 (§178.67) 600.33 (§251.44)FTD 814.09 (§210.91) 737.78 (§298.31) 844.77 (§194.54) 695.08 (§200.99)PNFA 854.82 (§191.49) 811.97 (§353.78) 862.06 (§209.20) 753.19 (§260.55)SD 777.60 (§257.80) 677.86 (§286.07) 843.47 (§221.06) 640.78 (§171.41)EXEC 810.38 (§210.65) 694.93 (§252.31) 796.77 (§149.68) 600.33 (§251.44)

606060606060606061616161

502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557

89012345678901234567890123456789012345678901234567890123




UNCORREC

all diVerence score mean§SD by group: FTDD¡17.80§12.83; ADD 15.79§12.30; ControlD17.65§11.64). In regard to main eVects, a signiWcant eVect ofTime [F (1, 43)D 8.64, p < .005] conWrms that all partici-pants, regardless of group diagnosis, demonstratedgreater reaction times to the targets immediately follow-ing any sentence violation (mean§SDD28.83§ 10.59)relative to the four-syllable delay position (mean§SDD¡18.40§ 10.85). There were no signiWcant two-way orthree-way interactions [Group£Time: F (2, 43)D .27, ns;Group£Sentence type: F (2, 43)D .24; Group£SentenceCondition£Time: F (2, 43)D .28, ns]. We interpret theseWndings as supportive of the validity of our sentenceprocessing paradigm in general.

3.2. Within group comparisons for transitive and thematic conditions

3.2.1. Healthy seniorsHealthy seniors demonstrated normal sensitivity to

both transitivity and thematic agreements. SpeciWcally,for the transitivity agreement condition, the healthyseniors had a greater “diVerence score” in the immediatetemporal window [mean§SDD32.24§54.91ms] relativeto the “diVerence score” during the delayed temporalwindow [mean§SDD¡7.71§30.02 ms; t (16)D 3.23,p < .006]. These diVerence scores are illustrated in Fig. 1.In the immediate “diVerence score” position, follow-upplanned analyses showed that healthy seniorsdemonstrate a prolonged reaction time latency to atarget word following the incorrect transitivity agree-ment [mean§SDD548.32§ 74.33] relative to a correcttransitivity agreement [mean§SDD 516.08§ 85.92 ms;t (16)D 2.42, p < 0.03]. This reaction time latency patternwas not observed in the delayed “diVerence score”position [delayed incorrect agreement mean§SDD465.83§74.80 ms; delayed correct agreement mean§SDD473.55§72.90 ms; t (16)D1.06, ns]. These Wndingssuggest that the healthy seniors exhibited normal sensi-tivity for the transitivity component of a verb’s represen-tation.

Healthy seniors also had a normal processing patternfor the thematic agreements. As shown in Fig. 1, these

TED PROOF

participants had a greater “diVerence score” for the the-matic agreements in the immediate temporal window[mean§SDD43.41§ 36.97 ms] relative to the “diVer-ence score” for the thematic agreements in the delayedtemporal window [mean§SDD2.68§31.14 ms; t (16)D3.87, p < .002]. SpeciWcally, the greater immediate “diVer-ence score” involved longer latencies to the target wordsfollowing an incorrect thematic agreement [mean§SDD536.55§89.65 ms] relative to a correct thematicagreement [mean§SDD 493.14§ 84.24; t (16)D 4.84,p < .001]. A diVerence in latency following incorrect andcorrect thematic agreements was not observed for targetwords in the four-syllable delay position [delay incorrectagreement mean§SDD 453.76§63.59 ms; delay correctmean§SDD 451.07§69.93; t (16)D0.36; ns]. TheseWndings suggest that the healthy seniors exhibited nor-mal processing for the component of verb representationinvolving thematic agreements.

3.2.2. AD patientsAD patients were sensitive to transitivity agreements

but not thematic agreements, as shown in Fig. 1. Thus,resembling healthy seniors, the AD patients demon-

Fig. 1. Group diVerence scores for the thematic and transitive agree-ment conditions by target word position (Immediate/delayed). Theequation for the “diVerence score” was: [ms reaction time to a targetword following an incorrect agreement ¡ ms reaction time to a targetword following a correct agreement].

66666666666666666677777777777777777777777777

Table 3Mean (§SD) millisecond reaction time for the thematic conditions (control/violation) by target position (immediate/delay) for the Control (Con,n D 17), Alzheimer Disease (AD, nD 15), Frontotemporal Dementia (FTD overall, n D 14), and FTD subgroups (Progressive NonXuent Aphasia,PNFA, n D 6; Semantic Dementia, SD, n D 6; Executive/social dysfunction, EXEC, n D 2)

Transitive control Transitive violation

Immediate Delay Immediate Delay


666666666666

614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669

7071727374757677787980818283848586878889909192939495969798990001020304050607080910111213141516171819202122232425




UNCORREC

strated a greater “diVerence score” for the transitivityagreement in the immediate temporal window [mean§SDD 34.65§50.86 ms] compared to the “diVerencescore” in the delayed temporal window [mean§SDD¡14.86§ 67.02 ms; t (14)D 3.93, p < .002]. Planned com-parisons conWrmed that the AD patients’ “diVerencescore” for transitivity agreements in the immediate tem-poral window involve longer latencies to the targetwords following incorrect agreements [mean§SDD676.73§136.35 ms] compared to correct agreements[mean§SDD 642.08§ 146.71 ms; t (14)D 2.64, p < .02].This diVerence between latencies for incorrect and cor-rect agreements was not observed for target words in thefour-syllable delay position [delay incorrect agreementmean§SDD594.16§154.99 ms; delay correct mean§SDD 609.02§ 177.24 ms; t (14)D0.86, ns]. These Wndingsindicate that the AD participants normally processed theverb grammatical agreement involving transitivity.

For the thematic agreement, however, the ADpatients did not demonstrate a greater “diVerence score”in the immediate temporal window relative to the“diVerence score” in the delayed temporal window[immediate mean§SDD 28.56§64.14 ms; delay mean§SDD 14.80§41.80 ms; t (14)D0.79; ns]. Moreover, ADpatients did not demonstrate a longer latency to respondto a target word following the incorrect thematic agree-ments, relative to the correct thematic agreements, ineither the immediate temporal window [incorrect agree-ment mean§SDD682.12§178.67 ms; correct agree-ment mean§SDD653.57§ 220.45 ms; t (14)D 1.72; ns]or the delayed temporal window [incorrect agreementmean§SDD600.33§251.44 ms; correct agreementmean§SDD585.53§251.27 ms; t (14)D1.37; ns]. ThisWnding indicates that the participants diagnosed withAD were insensitive to thematic agreement violations.

3.2.3. FTD patientsAs a group, Fig. 1 shows that FTD patients were

insensitive to both transitivity and thematic agreements.For the transitivity agreement, the FTD patients did notdemonstrate a greater “diVerence score” in the immedi-ate temporal window relative to the “diVerence score” inthe delayed temporal window [immediate mean§SDD 3.46§167.65 ms; delay mean§SDD¡62.63§129.96 ms; t (13)D 1.04; ns]. Moreover, the FTD patientsdid not demonstrate a longer latency to detect a targetword following incorrect agreements, relative to correctagreements, in either the immediate temporal window[incorrect agreement mean§SDD831.78§ 247.81 ms;correct agreement mean§SDD828.31§ 212.94 ms;t (13)D0.08; ns] or the delayed temporal window [incor-rect agreement mean§SDD 702.92§ 218.01 ms; correctagreement mean§SDD 765.54§ 296.34 ms; t (13)D1.80; ns]. This Wnding indicates that, unlike healthyseniors and AD patients, the FTD patients were insensi-tive to transitivity agreements.

TED PROOF

For the thematic condition, FTD patients did notdemonstrate a greater “diVerence score” in the immedi-ate temporal window relative to the “diVerence score” inthe delayed temporal window [immediate mean§SDD 30.67§124.30 ms; delay mean§SDD¡42.69§148.01 ms; t (13)D1.15; ns]. Moreover, FTD patients didnot demonstrate a longer latency to detect a target wordfollowing the incorrect thematic agreements, relative tothe correct thematic agreements, in either the immediatetemporal window [incorrect agreement mean§SDD 844.77§ 194.54 ms; correct agreement mean§SDD 814.09§ 210.91 ms; t (13)D0.92; ns] or the delayedtemporal window [incorrect agreement mean§SDD695.08§200.99 ms; correct agreement mean§SDD737.78§ 298.31 ms; t (13)D1.08; ns]. This Wnding indi-cates that the participants diagnosed with FTD, likepatients with AD, were insensitive to thematic agree-ments.

As with the Control, AD, and FTD groups, within-group analyzes were conducted for the PNFA (nD6)and SD (nD6) subgroups of FTD patients to examinetheir performance on the transitive and thematic viola-tions. Due to the small number of participants in thePNFA and SD subgroups of FTD, the within-group per-formance on the transitive and thematic violations wereanalyzed with the nonparametric Wilcoxon SignedRanks Test. We did not analyze the EXEC subgroupseparately because of the small number of these patients.

For the transitive condition, the PNFA subgroup ofFTD patients did not demonstrate a discrepancybetween the “diVerence score” immediately following atransitive agreement [mean§SDD¡4.14§ 131.55 ms]compared to a four-syllable delay [mean§SDD¡103.53§ 186.26 ms; ZD .31; ns]. None of the latenciesto target words following correct vs. incorrect agree-ments diVered in either temporal window [all p values atleast >0.10, according to t tests]. For the thematic condi-tion, the PNFA subgroup did not demonstrate a signiW-

cant discrepancy between the “diVerence score”immediately following a thematic agreement [mean§SDD7.24§125.50 ms] compared to a four-syllabledelay [mean§SDD¡58.78§ 131.07 ms; ZD .31; ns].None of the latencies to target words following correctvs. incorrect agreements diVered in either temporal win-dow [all p values at least >0.10, according to t tests].These Wndings indicate that PNFA patients are insensi-tive to transitive and thematic agreement violations.

SD patients in the transitive condition did not diVerfor the “diVerence score” immediately following atransitive agreement [mean§SDD75.75§ 159.27 ms]compared to a four-syllable delay [mean§SDD¡36.39§ 27.19 ms; ZD1.36; ns]. None of the latencies totarget words following correct vs. incorrect agreementsdiVered in either temporal window [all p values at least>0.10, according to t tests]. The SD subgroup also didnot diVer for the “diVerence score” immediately follow-

7878787878787878797979797979797979798080808080808080808081818181818181818181828282828282828282828383838383838383

726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781

23456789012345678901234567890123456789012345678901234567




UNCORREC

ing a thematic agreement (mean§SDD65.86§145.04 ms) compared to a four-syllable delay (mean§SDD¡37.08§169.19 ms; ZD1.15; ns). None of thelatencies to target words following correct vs. incorrectagreements diVered in either temporal window [all p val-ues at least >0.10, according to t tests]. These Wndingsindicate insensitivity to transitivity and thematic agree-ments in SD, as well.

An analysis of individual participant performanceproWles showed that three of the six (50%) of the PNFAand three of the six (50%) of the SD patients had agreater (violation-correct) discrepancy in their latency torespond to a target word immediately following a transi-tive agreement compared to the “diVerence score” fol-lowing a four-syllable delay. Moreover, three of the six(50%) of the PNFA and Wve of the six (83%) of the SDpatients had a greater (violation-correct) discrepancy intheir latency to respond to a target word immediatelyfollowing a thematic agreement. Although we did notWnd a signiWcant eVect in SD because of the small num-ber of patients and the wide variability in performance,this observation begins to suggest an associationbetween thematic role knowledge in verbs and lexicalsemantic knowledge that is often impaired in SD.

3.2.4. Correlations between agreement sensitivity and measures of language and executive functioning

Table 4 provides group scores for the additional lan-guage and executive background tasks. To improve ourunderstanding of AD and FTD patients’ sensitivity toverb roles, we correlated each patient groups’ measure ofoverall agreement sensitivity with measures of languageand executive functioning. A summary diVerence scorewas created for transitivity sensitivity and thematic sen-sitivity with the following equation: [Immediate “DiVer-ence Score”¡Delay “DiVerence Score”]. Patients whodid not complete one or more of the additional cognitive

TED PRO

tasks were removed from the analyses. For the ADpatients, correlation analyses revealed that sensitivity tothe thematic agreement correlated with our experimental‘verb similarity task’ (rD .59, pD .02). No other correla-tions approached signiWcance. This Wnding suggests thatthematic role knowledge may be important for theunderstanding of verb meaning. As verb meaningbecomes diYcult in AD, thematic role knowledge alsobecomes impaired. For the FTD patients, no signiWcantcorrelations were seen between verb agreement sensitiv-ity and measures of language and executive functioning.

4. Discussion

Verbs play important roles in sentences. Verbs relaygrammatical information such as transitivity—a gram-matical role organized around a verb’s direct and indi-rect object use. Verbs also relay semantic informationinvolving thematic roles—such as who can do an actionand what type of patient can receive the action. Wehypothesized that sentence processing would be diVeren-tially compromised in AD and FTD due in part to dis-tinct patterns of impaired processing of thematic andtransitive agreements involving verbs. The on-line tech-nique pioneered by Marslen-Wilson and Tyler (1980)allowed us to examine the processing of these two verbagreements while minimizing task-related resourcedemands. Healthy seniors processed the transitivity andthematic agreements normally. Thus, they were rela-tively delayed in their detection of a target word encoun-tered in the temporal envelope of processing anagreement violation relative to a coherent agreement.This diVerence was not seen when the occurrence of thetarget word was delayed, consistent with the conclusionthat the delay in the immediate vicinity of the agreementviolation is due to the processing of the agreement. By

99999999999999999999999999999999999999

OF

Table 4Mean (§SD) for the additional language and executive function background tasks by Control Group (Con, n D 17), Alzheimer Disease (AD, n D 15),Frontotemporal Dementia (FTD overall, nD 14), and FTD subgroup (Progressive NonXuent Aphasia, PNFA, n D 6; Semantic Dementia, SD, nD 6;Executive/social dysfunction, EXEC, nD 2)a

a Group Comparisons: Separate 3 (Group: AD, FTD, Con) £Task One-Way ANOVAs were conducted and are reported by test below. Indepen-dent-sample t tests show the PNFA and SD subgroups performed equally on all tasks. The EXEC subgroup was not compared due to the smallgroup number.

b Pyramid and Palm Tree Test (Howard, 1992); F (2, 43)D 6.27, p < .01 with the AD and FTD equally less accurate relative to the Control group.c Verb Similarity Task—an experimental word task that models the format of the Pyramid and Palm Tree Test; F (2, 43) D 7.44, p < .01 with only

the FTD group less accurate relative to Control group.d Test for the Reception of Grammar (Bishop, 1989); F (2, 41)D 6.93, p < .01 with only the FTD group less accurate relative to the Control group.e Trail Making Test—Part B (Reitan, 1958); F (2, 42) D 7.29, p < .01 with the AD and FTD groups equally slower relative to the Control group.

PPTb (% correct) VSTc (% correct) TROGd (% correct) TMT Part Be (seconds)


888888999999999999

838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893

9495969798990001020304050607080910111213141516171819202122232425262728293031323334353637383940414243444546474849




UNCORREC

comparison, AD patients were impaired at processingthe thematic agreements, and FTD patients appeared tobe insensitive to transitivity and thematic role agree-ments. We argue below that impaired thematic role pro-cessing in AD may be related to impaired verb semanticknowledge. FTD patients’ impairment for both transi-tivity and thematic role agreements may reXect abroader degradation of verb knowledge that involvesboth grammatical and semantic representations, or diY-culty processing sentence structure that also causes athematic role deWcit.

4.1. Alzheimer’s disease and verb comprehension

AD patients, like the healthy seniors, demonstratedintact sensitivity to agreements involving transitivity.However, they did not demonstrate the normal patternfor thematic role agreements. It is unlikely that impairedthematic role agreement sensitivity in AD is due to task-related resource demands, because the on-line compre-hension task used to assess verb role sensitivity minimizesexecutive resource demands needed to perform the task.Moreover, this would not explain AD patients’ normalperformance for transitivity agreements using the sametechnique. These Wndings are consistent with reports thatverb thematic information is processed diVerently fromverb transitivity information (Shapiro, Zurif, et al., 1987;Shapiro, Zurif, & Grimshaw, 1989). This dissociationinstead may indicate that AD patients are selectivelyimpaired comprehending the semantic roles associatedwith verbs. Previous work has shown that individualsdiagnosed with AD are impaired with the thematic, butnot grammatical roles, of verbs (Grossman, Mickanin,et al., 1997). This diYculty with semantic informationrepresented in verbs is consistent with the view that indi-viduals with AD have semantic impairments (Chan,Salmon, et al., 1995; Chertkow & Bub, 1990; Cummings,Benson, & Hill, 1985; Glosser, Kohn, et al., 1997; Hodges,Salmon, et al., 1990; Nebes, Martin, et al., 1984).

Some investigators argue that AD results in a loss ofsemantic knowledge (i.e., semantic degradation hypothe-sis). One possibility thus suggests that impaired thematicagreement sensitivity results from semantic network deg-radation. Although not a universal Wnding (Fung, Chert-kow, et al., 2001; Robinson, Rossor, et al., 1999;Williamson, Adair, et al., 1998) many studies have shownthat individuals with AD have greater diYculty withverbs than nouns (Bushell & Martin, 2002; Cappa,Binetti, et al., 1998; Robinson, Grossman, et al., 1996;White-Devine et al., 1996). Semantic knowledge associ-ated with verbs may be more susceptible to degradationbecause it is embedded in a sparser, less redundant andless hierarchically organized semantic matrix than nouns(Levin, 1993). Indeed, some researchers have hypothe-sized that random damage associated with the diseasereduces the redundancy of the semantic network under-

TED PROOF

lying a word, and this may make a word more suscepti-ble to comprehension diYculty in AD (Devlin,Gonnerman, et al., 1998; Gonnerman, Andersen, et al.,1997). From another perspective, semantic knowledge issaid to deteriorate in a bottom-up manner with detailedsubordinate concepts lost before superordinate concepts(Chan, Butters, et al., 1993; Martin, 1987; Schwartz,Marin, et al., 1979; Warrington, 1975; Weingartner,Kawas, et al., 1993). Although these studies were basedon semantic memory for nouns, the impoverishedsemantic networks associated with verbs may makethese words even more susceptible to degradation.Regardless of the speciWc nature of the semantic knowl-edge that is degraded, the observations of the presentstudy may be consistent with degradation of verb the-matic role knowledge in AD.

Others argue that the semantic deWcit in AD is due toimpaired aspects of semantic processing. SpeciWcally,this processing deWcit could be due to a deWcit ofretrieval, such as inhibitory breakdown resulting in com-petition among semantic nodes thus delaying retrieval,or fast decay of activation that aVects the Xow of infor-mation within the semantic network (Dell, Schwartz,et al., 1997; Kolk, 1995). Individuals with AD may per-form poorly on tests of semantic knowledge from thisperspective because they have deWcits in initiating andmaintaining organized retrieval strategies (Bonilla &Johnson, 1995; Nebes, Boller, et al., 1986; Nebes &Brady, 1990; Nebes, Martin, et al., 1984). Support forthis theory has largely come from priming paradigms(Nebes and colleagues). That is, when the eVort of asemantic task is reduced to automatic processing (viapriming), AD patients are said to perform normally.Problems with this approach include the fact thatsemantic priming for associative knowledge does notnecessarily mean that semantic knowledge is contribut-ing to the association (e.g., Glosser, Friedman, et al.,1998). In this priming study, AD patients primed for lex-ical associates like “cottage–cheese” but not for semanticcategory coordinates like “peach–plum.” In the currentstudy, we too used a task believed to assess the auto-matic aspects of language processing (Tyler, Moss, et al.,1997), yet still did not observe sensitivity to thematic roleknowledge. These Wndings suggest that the thematic roleinsensitivity we observed may be due to something otherthan this kind of processing impairment.

Another possible account for impaired thematic rolesensitivity in AD is concerned with the categorizationprocesses needed for verb comprehension. This model ofsemantic memory proposes two components: Semanticfeature representations; and processes such as categori-zation that make use of this knowledge for semantic deci-sions. There are several types of categorization processes.One type is ‘rule-based.’ This process evaluates wordmeaning relative to a set of diagnostic criteria. From thisperspective, the thematic role of a verb is evaluated in a

1010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010

950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005

0607080910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061




UNCORREC

rule-like manner with reference to a set of features repre-senting the verb’s meaning (Grossman, Smith, et al.,2002, 2003). Selective attention identiWes the relevant fea-tures; inhibitory control manages irrelevant features; andthe criteria must be kept active in working memory whileworking memory is up-dated to keep track of the resultsof the evaluation. Evidence to support this approachcomes from categorization judgments of brief objectdescriptions. However, these did not involve verbs ortheir associated thematic roles. There was no correlationbetween executive background tasks and sensitivity tothematic role violations in AD patients in the currentstudy, moreover, suggesting that this model may be moreappropriate for representations of concepts rather thanverb-mediated agreements in sentences.

An alternative processing theory is that individualsdiagnosed with AD have semantic diYculty due to a fail-ure to inhibit competing associates (Hasher, Stoltzfus,et al., 1991; Hasher & Zacks, 1979). According to theseresearchers, eYcient semantic processing requires theability to inhibit competing semantic associates thatbecome active during semantic processing. Theseresearchers postulate that there is a breakdown in inhibi-tion at the semantic network level in AD that results inan extended search time associated with long reactiontime latencies for both correct and incorrect associates.Such would Wt a proposed model of thematic role pro-cessing described by Shapiro and Levine (1990). Thismodel states that all of a verb’s thematic possibilities areactivated when a verb is encountered. The thematic pro-cessor then assigns thematic roles to noun phrases asthey are encountered in real-time. According to this the-ory, if AD patients activate all of the thematic matrixpossibilities and at the same time they suVer from inhibi-tory breakdown, then the search process should be moreeVortful and result in extended reaction time latenciesfor both correct and incorrect associates. Although anintriguing possibility, data from the current investigationare not consistent with this theory. We found that thereaction time latency at the four-syllable time windowfor the AD patients is not signiWcantly diVerent fromthat of the control participants. This suggests that theAD patients are not engaged in a prolonged semanticsearch process. Additionally, there was no correlationbetween executive processing and thematic violationsensitivity—a Wnding that could have been supportive ofthe theory that semantic processing is related to inhibi-tory semantic demands.

Fast decay of activation in the semantic network mayalso be inXuencing AD patient sensitivity to the thematicrole agreements. Kolk and Van Grusven (1985) origi-nally posed the concept of fast decay to explain varia-tions in processing among agrammatic patients.According to Kolk (1995), two changes to the normalprocessing situation are possible. First, slow activationcan occur in that it takes longer for an element to reach

TED PROOF

its critical level of activation. Second, fast decay mayoccur which makes elements unavailable because theyfall below their critical level of activation too soon to becombined with other elements in the sentence represen-tation. The current study did not observe slow activationin processing thematic information. Thus, possibly thefailure to process thematic roles is associated with rapiddecay of activation. More work investigating the possi-bility of fast decay in semantic networks in AD patientsis needed to evaluate this account of impaired thematicrole knowledge.

Regardless of the cognitive processes explaining the-matic role diYculty in AD, the semantic component ofverb meaning may also be associated with a particularanatomic distribution of disease in these patients. Whilethere has been no work assessing the neural basis forverb thematic roles in AD, a recent fMRI study investi-gated the neural basis of verb meaning in AD relative tohealthy seniors (Grossman, Koenig, et al., 2003). Patientswere asked to judge the “pleasantness” of a verb, includ-ing verbs of motion and verbs of cognition. Both healthyseniors and AD patients activated the left posterolateraltemporal region across both semantic categories. How-ever, a direct comparison revealed signiWcantly reducedactivation in this area in AD relative to healthy seniors.Activation was instead displaced to an adjacent region.A similar activation pattern was also seen for nouns(Grossman, Koenig, et al., 2003). The authors stated thatthis reduced and displaced activation may be due to thehistopathological distribution of the disease in left pos-terolateral temporal cortex in AD. Other work hasimplicated left posterolateral temporal cortex in lexical–semantic processing in structural imaging studies ofpatients with damage to this region (Chertkow, Bub,et al., 1997; Hart & Gordon, 1990; Hillis, Wityk, et al.,2001). Correlations of cortical atrophy with lexicalsemantic judgments of nouns/objects in AD have impli-cated left temporal–parietal cortex as well (Baron, Chet-elat, et al., 2001).

The data from the current investigation thus are con-sistent with a larger picture suggesting that verbs playmultiple roles in sentence processing, and observations ofAD patients suggest that some of these roles are dissocia-ble. Because the AD patients can comprehend transitiv-ity, it is clear that they have an understanding of at leastone component of a verb. In contrast, they have diYcultyprocessing a verb’s thematic roles. There is much specula-tion about the precise basis for this impairment, but addi-tional work is needed to determine whether this is relatedto degradation of verb thematic role knowledge or a deW-

cit in processing this component of a verb.

4.2. Frontotemporal dementia and verb comprehension

The pattern of performance in FTD diVers from thatseen in AD. Unlike AD, we did not observe sensitivity to

11111111111111111111111111111111111111111111111111111111

10621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117

118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173




UNCORREC

thematic or transitive agreements in FTD. This wasdespite the relatively automatic nature of the on-linetask. FTD patients’ impairment thus cannot be attrib-uted easily to task-related resource demands. There wasalso no evidence of slowed processing, and we found nocorrelation between agreement sensitivity and executiveand language comprehension abilities.

Verb naming and comprehension appears to beimpaired in FTD. Cappa, Binetti, et al. (1998) reportedthat both AD and FTD patients were more impairedwith action naming relative to object naming. However,the discrepancy between object and action naming wassigniWcantly greater in FTD than in AD. Rhee, Antiqu-ena, et al. (2001) reported that FTD patients are moreimpaired for verbs than nouns on a picture-word match-ing task. This study showed diVerent patterns of compre-hension diYculty in PNFA and non-aphasic patientswith an executive disorder. The PNFA patients’ compre-hension of verbs may have been related to degraded verbknowledge since concurrent performance of a secondarytask did not worsen their performance further. Bak et al.also suggested that the neural representation of verbknowledge is degraded in some FTD patients (Bak,O’Donovan, et al., 2001). In the present study, we triedto determine whether there were consequences of this forverb agreements during sentence processing. In factPNFA patients appeared to be broadly impaired in pro-cessing verb transitivity and thematic role agreements. Itis important to note that there is some variability inPNFA patients’ performance, with some patients show-ing grammatical diYculty and others presenting primar-ily with dysarthria (Thompson, Ballard, et al., 1997). Itmay be that individual diVerences such as these maskedsigniWcant eVects in FTD, and we were unable to bringout eVects within the PNFA patient subgroup because ofthe small number of patients. Although we know lessabout verbs in SD, there is much evidence suggestingthat SD patients have degraded knowledge of individualword meanings (Hodges, Graham, et al., 1995; Hodges,Patterson, et al., 1992, 1994; Lambon Ralph, Graham,et al., 1998; Lambon Ralph, McClelland, et al., 2001;Snowden, Goulding, et al., 1989). The Wnding thatPNFA patients and SD patients had similar patterns ofimpairment may indicate that there is a single formof Primary Progressive Aphasia, where diVerent aspectsof language diYculty emerge at diVerent points in thecourse of the condition (Karbe, Kertesz, et al., 1993;Mesulam, 2001; Mesulam, Grossman, et al., 2003) How-ever, this is diYcult to determine based on the small num-ber of patients participating in this study. Additional workis necessary to determine the role of verb diYculty in thesentence processing deWcits of FTD patients.

An alternate possibility is that processing verb knowl-edge in a sentence could be due to independent deWcitsinvolving each subcomponent of a verb, rather than deg-radation of all verb-associated knowledge, and that the

TED PROOF

deWcit in both verb components is due to the dependenceof one verb component on the other. An impairment intransitivity in FTD could be due to degradation withinthe grammatical system (Grodzinsky, 1986, 1990, 1995),or there could be a failure to process grammatical knowl-edge for the purpose of understanding a sentence (Frazier& Friederici, 1991; Kolk, 1995; Swinney & Zurif, 1995;Zurif, Swinney, et al., 1993). Transitivity is an example ofa core structure-building device necessary to support sen-tence comprehension (Grodzinsky, 1986, 1990, 1995;Shapiro & Levine, 1990). For example, transitivity signalswhether a verb requires a direct or indirect object. Fromthis perspective, transitivity provides a structure for thesemantic mapping of thematic roles, and a deWcit appre-ciating transitivity agreements thus may hinder the pro-cessing of thematic roles. It may be that appreciation ofthematic role agreements is itself not impaired, but ratherits dependence on transitivity structure may result inimpaired thematic role processing. Grodzinsky (1995)proposes, for example, that the structural traces that linkthematic roles to the grammatical referent of a verb canbe lost or “deleted.” Insensitivity to thematic roles thuscould involve a decay of the verb’s trace that would limitreactivation of the agent role’s node following processingof the verb. Alternately, there could be an impairment inverb processing such that grammatical structure informa-tion does not remain active long enough to allow process-ing of transitivity or mapping of thematic roles (Frazier& Friederici, 1991; Kolk, 1995; Swinney & Zurif, 1995;Zurif, Swinney, et al., 1993).

4.3. Study limitations

There are shortcomings associated with this studythat must be kept in mind when interpreting our Wnd-ings. First, the study assessed patients with mild to mod-erate disease severity, and the results should begeneralized with caution to other AD and FTD cohorts.Second, the diagnosis in these patients has not been vali-dated at autopsy. Although our clinic has histopatho-logic conWrmation of AD and FTD in many patientsresembling those who participated in this study, thediagnosis of FTD and AD has been based on clinicalevaluation conWrmed by structural and functional imag-ing investigations. Third, the small sample size of thepatient samples and, in particular, the FTD subgroupsmay have contributed to insensitivity of the on-line task.Finally the word detection task is not entirely free ofresource demands since divided attention is needed todetect a target word while listening to a sentence forcomprehension.

With these caveats in mind, the current study demon-strates that verb agreement processing is compromisedin AD and FTD. Comprehension of thematic role agree-ments, but not transitivity agreements, is impaired inAD. This may be related to their broader semantic deW-

1212121212121212121212121212121212121212121212121212121212121212121212121212121212121212121212121212121212121212

11741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229

3031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485




UNCORREC

cit. FTD patients, in contrast, were insensitive to boththematic and transitivity agreements. This may be asso-ciated with a broad-based degradation in verb knowl-edge, or diYculty with structure-building that alsoimpaired comprehension of thematic roles in a sentence.

Appendix A

Thematic condition*

The [chickens/jokes] appear quickly from behindƒThe [trains/houses] bring charcoal to the tinyƒThe [buckets/colors] capture nearly every dropƒThe [lizards/feathers] cling next to the cageƒThe [trucks/brushes] compress rotten smelly bagsƒThe [turtles/birds] crawl through dry grass towardƒThe [vines/acorns] creep beside the oldƒThe [radar/desk] detects changes in motionƒThe [stars/rocks] emerge between clouds on clearƒThe [plaster/thought] falls onto the carpetƒThe [teacher/tree] gazes away from disabledƒThe [plumber/pillow] injects bright yellow liquidƒThe [worker/cow] installs gray bricks with blackƒThe [duck/snake] limps toward every personƒThe [police/Wres] listen closely from behindƒThe [chef/sauce] mashes buckets of overlyƒThe [academies/lights] mold ladies into nationalƒThe [eagle/Xame] notices movement on the canyonƒThe [wind/robber] peeks into the room throughƒThe [owl/rain] peers beneath old maple treeƒThe [butler/table] perceives problems with manyƒThe [waiter/chair] put cheese samples nearƒThe [submarines/sponges] rescue dolphins and otherƒThe [roaches/snails] scurry loudly across this kitchenƒThe [thermometer/curtain] senses minor changes betweenƒThe [student/soup] slumps onto the tableƒThe [arrow/worm] soars through the trees intoƒThe [butcher/lion] sorts red meat from behindƒThe [Wreman/fountain] stumbles across the thresholdƒThe [passengers/blankets] view movies during crossƒ

Transitive Condition*

The bugs [appear/appear water]nightly under porchƒThe hamster [brings/brings beside] chewed carrot chipsƒThe bubbles [cling/cling food] with cold water aroundƒThe girls [compress/compress through] dough using largeƒThe child [crawls/crawls grass] through mud and leavesƒThe convicts [creep/creep dusk] nightly into localƒThe device [detects/detect into] tide levels and changesƒThe tadpoles [emerge/emerge caves] brown and very muddyƒThe clown [gazes/gazes face] toward the circus buildingƒThe artist [injects/injects within] rust colored inkƒThe salesman [installs/installs behind] new and antique kitchenƒThe donkey [limps/limps straw] toward water bucketsƒThe jury [listens/listens lawyer] quietly for moreƒThe children [mash/mash within] rocks into the dryƒThe reporters [mold/mold toward] boring town eventsƒThe travelers [notice/notice along] angry country nativesƒThe mouse [peeks/peeks cheese] bravely through bedroomƒThe machines [peer/peer light] inside small darkƒThe researchers [perceive/perceive within] changes in the oceanƒThe secretary [puts/puts around] coVee stains alongƒThe navy [rescues/rescues within] injured baby blackƒThe crab [scurries/scurries sand] through the iron pipeƒ

TED PROOF

[ ], brackets contain the correct and incorrect agreements. Italicizedwords, target word used for the immediate position. Underlined words,target words used for the delay position. Each sentence ended after theparticipant identiWed the target word.

References

Almor, A., Kempler, D., et al. (1999). Why do Alzheimer patients havediYculty with pronouns. Working memory, semantics, and refer-ence in comprehension and production in Alzheimer’s disease.Brain and Language, 67(3), 202–227.

Bak, T. H., O’Donovan, D. G., et al. (2001). Selective impairment ofverb processing associated with pathological changes in Brodmannareas 44 and 45 in the motor neurone disease-dementia-aphasiasyndrome. Brain, 124(Pt 1), 103–120.

Baron, J.-C., Chetelat, G., et al. (2001). In vivo mapping of gray matterloss with voxel-based morphometry in mild Alzheimer’s disease.Neuroimage, 14, 298–309.

Bishop, D. V. M. (1989). Test for the Reception of Grammar (TROG).Chapel Press: Medical Research Council.

Bonilla, J. L., & Johnson, M. K. (1995). Semantic space in Alzheimer’sdisease patients. Neuropsychology, 9(3), 345–353.

Bushell, C., & Martin, A. (2002). Automatic semantic priming of nounsand verbs in patients with Alzheimer’s disease. Neuropsychologia,35, 298–309.

Cappa, S. F., Binetti, G., et al. (1998). Object and action naming in Alz-heimer’s disease and frontotemporal dementia [see comment]. Neu-rology, 50(2), 351–355.

Chan, A. S., Butters, N., et al. (1993). Dimensionality and clustering inthe semantic network of patients with Alzheimer’s disease. Psychol-ogy and Aging, 8(3), 411–419.

Chan, A. S., Salmon, D. P., et al. (1995). Semantic network abnormalitypredicts rate of cognitive decline in patients with probable Alzhei-mer’s disease. Journal of the International Neuropsychological Soci-ety, 1(3), 297–303.

Chertkow, H., & Bub, D. (1990). Semantic memory loss in dementia ofAlzheimer’s type. What do various measures measure? Brain, 113(Pt2), 397–417.

Chertkow, H., Bub, D., et al. (1997). On the status of object concepts inaphasia. Brain and Language, 58(2), 203–232.

Chomsky, N. (1965). Aspects of the theory of syntax. MIT Press:Cambridge.

Cohen, J. D., MacWhinney, B., Flatt, M. R., & Provost, J. (1993). Psy-Scope: A new graphic interactive environment for designing psy-chology experiments. Behavioral Research Methods forInstrumentation and Computation, 25, 101–113.

Croot, K., Hodges, J. R., et al. (1999). Evidence for impaired sentencecomprehension in early Alzheimer’s disease. Journal of the Interna-tional Neuropsychological Society, 5(5), 393–404.

Cummings, J. L., Benson, F., & Hill (1985). Aphasia in dementia of theAlzheimer type. Neurology, 35(3), 394–397.

Davis, K. L., Price, C. C., et al. (2001). Evaluating the clinical diagnosisof Frontotemporal degeneration: A re-examination of Neary e al.,1998. Neurology, 56, A144–A145.

Dell, G. S., Schwartz, M. F., et al. (1997). Lexical access in aphasic andnonaphasic speakers. Psychological Review, 104(4), 801–838.

Devlin, J. T., Gonnerman, L. M., et al. (1998). Category-speciWc seman-tic deWcits in focal and widespread brain damage: a computationalaccount. Journal of Cognitive Neuroscience, 10(1), 77–94.

The twins [sense/sense toward] exciting and traumaticƒThe cowboy [slumps/slumps walls] behind the oldƒThe ball [soars/soars dirt] beyond the crowdedƒThe king [sorts/sorts in] rare Asian jadeƒThe horses [stumble/stumble rock] along the newlyƒThe strangers [view/view onto] ocean seafood marketsƒ

111111111

1

1111111111111111111111111111111111111111111111111111

12861287128812891290

1291

12921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338

339340341342343344345346347

348

349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400




UNCORREC

Elfgren, C., Passant, U., et al. (1993). Neuropsychological Wndings infrontal lobe dementia. Dementia, 4(3–4), 214–219.

Emery, O. B. (1985). Language and Aging. Experimental AgingResearch, 11, 3–60.

Folstein, M. F., Folstein, S. E., et al. (1975). Mini-mental state. A prac-tical method for grading the cognitive state of patients for the clini-cian. Journal of Psychiatry Research, 12(3), 189–198.

Francis, W. N., & Kucera, H. (1982). The frequency analysis of Englishusage. Boston: Houghton-MiZin.

Frazier, L., & Friederici, A. D. (1991). On deriving properties of apha-sic comprehension. Brain and Language, 40, 51–66.

Fung, T. D., Chertkow, H., et al. (2001). The spectrum of categoryeVects in object and action knowledge in dementia of the Alzhei-mer’s type. Neuropsychology, 15(3), 371–379.

Glosser, G., Friedman, R. B., et al. (1998). Lexical semantic and associa-tive priming in Alzheimer’s disease. Neuropsychology, 12(2), 218–224.

Glosser, G., Kohn, S. E., et al. (1997). Repetition of single words andnonwords in Alzheimer’s disease. Cortex, 33(4), 653–666.

Gonnerman, L. M., Andersen, E. S., et al. (1997). Double dissociationof semantic categories in Alzheimer’s disease. Brain and Language,57(2), 254–279.

Grimshaw, J. B. (1990). Argument structure. Cambridge, MA, London,England: MIT Press.

Grober, E., & Bang, S. (1995). Sentence comprehension in Alzheimer’sdisease. Developmental Neuropsychology, 11, 95–107.

Grodzinsky, Y. (1986). Language deWcits and the theory of syntax.Brain and Language, 27(1), 135–159.

Grodzinsky, Y. (1990). Theoretical perspectives on language deWcits.Cambridge, MA: MIT Press.

Grodzinsky, Y. (1995). Trace deletion, theta-roles, and cognitive strate-gies. Brain and Language, 51(3), 469–497.

Grossman, M., D’Esposito, M., et al. (1996). Language comprehensionproWles in Alzheimer’s disease, multi-infarct dementia, and fronto-temporal degeneration. Neurology, 47(1), 183–189.

Grossman, M., Koenig, P., et al. (2003). Neural basis for semanticmemory diYculty in Alzheimer’s disease: An fMRI study. Brain,126(Pt 2), 292–311.

Grossman, M., Mickanin, J., et al. (1995). An aspect of sentence pro-cessing in Alzheimer’s disease: QuantiWer-noun agreement. Neurol-ogy, 45(1), 85–91.

Grossman, M., Mickanin, J., et al. (1997). Lexical acquisition in proba-ble Alzheimer’s disease. Brain and Language, 60(3), 443–463.

Grossman, M., Mickanin, J., et al. (1996). Anomaly judgments of sub-ject–predicate relations in Alzheimer’s disease. Brain and Language,54(2), 216–232.

Grossman, M., & Rhee, J. (2001). Cognitive resources during sentenceprocessing in Alzheimer’s disease. Neuropsychologia, 39(13), 1419–1431.

Grossman, M., Rhee, J., et al. (2004). Sentence processing in frontotem-poral dementia. Cortex (in press).

Grossman, M., Smith, E. E., et al. (2002). The neural basis for categori-zation in semantic memory. Neuroimage, 17(3), 1549–1561.

Grossman, M., Smith, E. E., et al. (2003). Categorization of objectdescriptions in Alzheimer’s disease and frontotemporal dementia:Limitation in rule-based processing. Cognitive AVect and Behav-ioural Neurosciences, 3(2), 120–132.

Grossman, M., & White-Devine, T. (1998). Sentence comprehension inAlzheimer’s disease. Brain and Language, 62(2), 186–201.

Hart, J., Jr., & Gordon, B. (1990). Delineation of single-word semanticcomprehension deWcits in aphasia, with anatomical correlation.Annals of Neurology, 27(3), 226–231.

Hasher, L., Stoltzfus, E. R., et al. (1991). Age and inhibition. Journal ofExperimental Psychology: Learning, Memory and Cognition, 17(1),163–169.

Hasher, L., & Zacks, R. T. (1979). Automatic and eVortful processes inmemory. Journal of Experimental Psychology: General, 108, 356–388.

TED PROOF

Hillis, A. E., Wityk, R. J., et al. (2001). Hypoperfusion of Wernicke’sarea predicts severity of semantic deWcit in acute stroke. Annals ofNeurology, 50, 561–566.

Hodges, J. R., Graham, K. S., et al. (1995). Charting the progression ofsemantic dementia: Implications for the organization of semanticmemory. Memory, 3, 363–395.

Hodges, J. R., Patterson, K., et al. (1992). Semantic dementia: Progres-sive Xuent aphasia with temporal lobe atrophy. Brain, 115, 1783–1806.

Hodges, J. R., Patterson, K., et al. (1994). Semantic dementia: A studyof modularity. Cognitive Neuropsychology, 11, 206–245.

Hodges, J. R., Patterson, K., et al. (1999). The diVerentiation of seman-tic dementia and frontal lobe dementia (temporal and frontal vari-ants of frontotemporal dementia) from early Alzheimer’s disease:A comparative neuropsychological study. Neuropsychology, 13(1),31–40.

Hodges, J. R., Salmon, D. P., et al. (1990). DiVerential impairment ofsemantic and episodic memory in Alzheimer’s and Huntington’sdiseases: a controlled prospective study. Journal of Neurology, Neu-rosurgery, and Psychiatry, 53(12), 1089–1095.

Hopper, P. J., & Thompson, S. (1980). Transitivity in grammar and dis-course. Language, 56(2), 251–299.

Howard, D. P. K. (1992). The Pyramid and Palm Trees Test: A Test ofSemantic Access from Words and Pictures. Bury, St. Edmunds:Thames Valley Test Company.

JackendoV, R. (1972). Semantic interpretation in generative grammar.Cambridge: MIT Press.

JackendoV, R. (1990). Semantic structures. Cambridge, MA: MITPress.

JackendoV, R. (1994). Patterns in the Mind: Language and humannature. Harvester: Wheatsheaf.

Johnson, M. K., & Hermann, A. M. (1995). Semantic relations and Alz-heimer’s disease: An early and disproportionate deWcit in func-tional knowledge. Journal of the International NeuropsychologicalSociety, 1(6), 568–574.

Karbe, H., Kertesz, A., et al. (1993). ProWles of language in primaryprogressive aphasia. Archives of Neurology, 50, 193–201.

Kempler, D., Almor, A., et al. (1998). Sentence comprehension deWcitsin Alzheimer’s disease: A comparison of oV-line vs. on-line sentenceprocessing. Brain and Language, 64(3), 297–316.

Kolk, H. (1995). A time-based approach to agrammatic production.Brain and Language, 50(3), 282–303.

Kontiola, P., Laaksonen, R., et al. (1990). Pattern of language impair-ment is diVerent in Alzheimer’s disease and multi-infarct dementia.Brain and Language, 38, 364–383.

Lambon Ralph, M. A., Graham, K. S., et al. (1998). Naming in seman-tic dementia—what matters. Neuropsychologia, 36, 775–784.

Lambon Ralph, M. A., McClelland, J. L., et al. (2001). No right tospeak. The relationship between object naming and semanticimpairment: Neuropsychological evidence and a computationalmodel. Journal of Cognitive Neuroscience, 13, 341–356.

Levin, B. A. (1993). English verb classes and alternations: A preliminaryinvestigation. Chicago: The University of Chicago Press.

Lund/Manchester (1994). Clinical and neuropathological criteria forfrontotemporal dementia. Journal of Neurology, Neurosurgery, andPsychiatry, 57(4), 416–418.

Marslen-Wilson, W., & Tyler, L. K. (1980). The temporal structure ofspoken language understanding. Cognition, 8(1), 1–71.

Martin, A. (1987). Representations of semantic and spatial knowledgein Alzheimer’s disease: Implications for models of preserved learn-ing in amnesia. Journal of Clinical and Experimental Neuropsychol-ogy, 9, 191–224.

McKhann, G., Drachman, D., et al. (1984). Clinical diagnosis ofAlzheimer’s disease: Report of the NINCDS-ADRDA WorkGroup under the auspices of Department of Health and HumanServices Task Force on Alzheimer’s Disease. Neurology, 34(7),939–944.

14141414141414141414141414141414141414141414141414141414141414141515151515151515151515151515151515151515151515151515151515151515151515

1401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467

68697071727374757677787980818283848586878889909192939495969798990001020304050607080910111213141516171819202122232425262728293031323334




CORREC

McKhann, G. M., Albert, M. S., et al. (2001). Clinical and pathologicaldiagnosis of frontotemporal dementia: Report of the Work Groupon Frontotemporal Dementia and Pick’s Disease. Archives of Neu-rology, 58(11), 1803–1809.

Mesulam, M.-M. (2001). Primary progressive aphasia. Annals of Neu-rology, 49, 425–432.

Mesulam, M.-M., Grossman, M., et al. (2003). The core and halo of pri-mary progressive aphasia and semantic dementia. Annals of Neurol-ogy, 54, S11–S14.

Miller, B. L., Cummings, J. L., et al. (1991). Frontal lobe degeneration:Clinical, neuropsychological, and SPECT characteristics. Neurol-ogy, 41(9), 1374–1382.

Neary, D., Snowden, J. S., et al. (1998). Frontotemporal lobar degener-ation: a consensus on clinical diagnostic criteria. Neurology, 51(6),1546–1554.

Nebes, R. D., Boller, F., et al. (1986). Use of semantic context by patientswith Alzheimer’s disease. Psychology and Aging, 1(3), 261–269.

Nebes, R. D., & Brady, C. B. (1990). Preserved organization of seman-tic attributes in Alzheimer’s disease. Psychology and Aging, 5(4),574–579.

Nebes, R. D., Martin, D. C., et al. (1984). Sparing of semantic memory inAlzheimer’s disease. Journal of Abnormal Psychology, 93(3), 321–330.

Pachana, N. A., Boone, K. B., et al. (1996). Comparison of neuropsy-chological functioning in Alzheimer’s disease and frontotemporaldementia. Journal of the International Neuropsychological Society,2(6), 505–510.

Price, C. C., Davis, K. L., et al. (2001). Clinical diagnosis of frontotem-poral dementia (FTD). Neurology, 56, A176.

Reitan, R. M. (1958). Validity of the trail making test as an indicationof organic brain damage. Perceptual and Motor Skills, 8, 271–276.

Rhee, J., Antiquena, P., et al. (2001). Verb comprehension in fronto-temporal degeneration: The role of grammatical, semantic andexecutive components. Neurocase, 7(2), 173–184.

Robinson, G., Rossor, M., et al. (1999). Selective sparing of verb nam-ing in a case of severe Alzheimer’s disease. Cortex, 35(3), 443–450.

Robinson, K. M., Grossman, M., et al. (1996). Category-speciWc diY-culty naming with verbs in Alzheimer’s disease. Neurology, 47(1),178–182.

Rochon, E., Waters, G. S., et al. (1994). Sentence comprehension inpatients with Alzheimer’s disease. Brain and Language, 46(2), 329–349.

Rosen, W. G., Terry, R. D., et al. (1980). Pathological veriWcation ofischemic score in diVerentiation of dementias. Annals of Neurology,7(5), 486–488.

Schwartz, M. F., Marin, O. S., et al. (1979). Dissociations of languagefunction in dementia: A case study. Brain and Language, 7(3), 277–306.

Shapiro, L. P., & Levine, B. A. (1990). Verb processing during sentencecomprehension in aphasia. Brain and Language, 38(1), 21–47.

UN

TED PROOF

Shapiro, L. P., Zurif, E., et al. (1987). Sentence processing and the men-tal representation of verbs. Cognition, 27(3), 219–246.

Shapiro, L. P., Zurif, E. B., & Grimshaw, J. (1989). Verb processing dur-ing sentence comprehension: Contextual Impenetrability. Journalof Psycholinguistic Research, 18(2), 223–243.

Snowden, J., GriYths, H., & Neary, D. (1994). Semantic dementia:Autobiographical contribution to preservation of meaning. Cogni-tive Neuropsychology, 11(3), 265–288.

Snowden, J. S., Goulding, P. J., et al. (1989). Semantic dementia: Aform of circumscribed cerebral atrophy. Behavioral Neurology, 2,167–182.

Snowden, J. S., Neary, D., et al. (1992). Progressive language disorderdue to lobar atrophy. Annals of Neurology, 31(2), 174–183.

Swinney, D., & Zurif, E. (1995). Syntactic processing in aphasia. Brainand Language, 50(2), 225–239.

Thompson, C. K., Ballard, K. J., et al. (1997). Patterns of languagedecline in non-Xuent primary progressive aphasia. Aphasiology, 11,297–331.

Tyler, L. K. (1985). Real-time comprehension processes in agramma-tism: A case study. Brain and Language, 26(2), 259–275.

Tyler, L. K., Moss, H. E., et al. (1997). The gradual deterioration of syn-tax and semantics in a patient with progressive aphasia. Brain andLanguage, 56(3), 426–476.

Warrington, E. K. (1975). The selective impairment of semantic mem-ory. Quarterly Journal of Experimental Psychology, 27, 635–657.

Waters, G. S., & Caplan, D. (1997). Working memory and on-line sen-tence comprehension in patients with Alzheimer’s disease. Journalof Psycholinguistic Research, 26(4), 377–400.

Waters, G. S., Caplan, C., & Rochon, E. (1995). Processing capacityand sentence comprehension in patients with Alzheimer’s disease.Cognitive Neuropsychology, 12(1), 1–30.

Weingartner, H. J., Kawas, C., et al. (1993). Changes in semantic mem-ory in early stage Alzheimer’s disease patients. Gerontologist, 33(5),637–643.

Weintraub, S., Rubin, N. P., et al. (1990). Primary progressive aphasia.Longitudinal course, neuropsychological proWle, and language fea-tures. Archives of Neurology, 47(12), 1329–1335.

Wepman, J. M., & Reynolds, W. M. (1987). Wepman’s Auditory Dis-crimination Test (second ed.). Los Angeles: Western PsychologicalServices.

White-Devine, T., Grossman, M., Robinson, K. M., Onishi, K., Bias-sou, N., & D’Esposito, M. (1996). Verb confrontation naming andword-picture matching in Alzheimer’s disease. Neuropsychology,10(4), 495–503.

Williamson, D. J., Adair, J. C., et al. (1998). Object and action namingin Alzheimer’s disease. Cortex, 34(4), 601–610.

Zurif, E. B., Swinney, D., et al. (1993). An on-line analysis of syntacticprocessing in Broca’s and Wernicke’s aphasia. Brain and Language,45, 448–464.

1111111111111111111111111111111111111111111111111

1535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583

584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632

An on-line study of verb knowledge in individuals diagnosed with AD and FTD. Brain and

Documents

Transcript of An on-line study of verb knowledge in individuals diagnosed with AD and FTD. Brain and