- Email: [email protected]
Verb Movement in Acquisition and Aphasia: Same Problem, Different Solutions—Evidence from Dutch
Brain and Language 77, 449–458 (2001) doi:10.1006/brln.2000.2414, available online at http://www.idealibrary.com on
Verb Movement in Acquisition and Aphasia: Same Problem, Different Solutions—Evidence from Dutch Shalom Zuckerman, Roelien Bastiaanse, and Ron van Zonneveld Graduate School for Behavioral and Cognitive Neurosciences, University of Groningen, The Netherlands Published online March 9, 2001
This paper focuses on verb movement in agrammatism and child language. We present data from a sentence completion experiment with 6 Dutch agrammatic aphasics and 21 Dutchspeaking children. The experiment compares completion of matrix clauses (which require verb movement) and embedded clauses (where such movement is not required) in these two populations. The results reveal a clear asymmetry: Both agrammatics and children do very well with embedded clauses but fail in 50% with the matrix clauses. It is concluded that the problem which both populations are facing is one of verb movement rather than verb inflection. An error analysis of the responses reveals that, although both agrammatics and children try to avoid movement, they apply different strategies to achieve this goal. 2001 Academic Press Key Words: agrammatism; language acquisition; syntax; verb movement; Dutch.
INTRODUCTION
The idea that aphasia and language acquisition might share properties goes back to the writings of Jakobson (1971), who suggested that language loss, such as seen in aphasic patients, is a mirror image of the course of acquisition seen in normally developing children. The idea is that operations which are the last ones to be acquired by children are the first to be lost by aphasic patients. Looking at the literature on child language and agrammatic aphasic speech reveals that similar theories and ideas are independently proposed for each of these two fields, for example, lack of functional categories, truncation, underspecification, problems with chain formation, processing difficulties, and problems with movement [for instance, for acquisition, Radford (1990), Wexler (1994), Rizzi (1995), Platzack (1996), Avrutin (1999); for Agrammatism: Grodzinsky (1990), Platzack (1996), Friedmann and Grodzinsky (1997), Avrutin (1999)]. In the current paper we raise the similarity/difference question regarding an operation which is rather well documented in the literature: verb movement in Dutch. We present results from an experimental sentence completion task conducted with agrammatic aphasics and three groups of normally developing children. The results reveal that agrammatic aphasics and young children share the problem of raising the finite verb to I 0 in matrix clauses, whereas performance in embedded clauses (where the finite verb remains in a lower position) is virtually
Address reprint requests to S. Zuckerman, OTS, Trans 10 3512 JK, Utrecht, The Netherlands. 449 0093-934X/01 $35.00 Copyright 2001 by Academic Press All rights of reproduction in any form reserved.
450
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
normal. A qualitative error analysis shows, however, that children and aphasics use different strategies to overcome this problem. VERB MOVEMENT IN DUTCH
In Dutch, there is a matrix/embedded clause asymmetry regarding verb placement, as shown in example (1) below: (1)
a. De man leest i een boek t i The man reads a book. b. De man die een boek leest The man who a book reads.
It is generally accepted that in the matrix clause (1a) the finite verb has undergone movement, while in the embedded clause (1b) it remains in its base-generated position. It is important to point out here that the asymmetry in example (1) is one of verb movement and not of verb inflection, since the verb ‘‘leest’’ (reads) is inflected in both matrix and embedded contexts while it is also moved only in the matrix clause. Movement and inflection have often been treated as two sides of the same coin, since it is assumed that the motivation for verb movement to I is to check (or collect) inflection; however, the asymmetry in example (1) allows us to focus on one of these two properties, reaching more fine-grained conclusions regarding the limitations that agrammatics and children face. The purpose of this paper is to answer, through an experimental investigation, the following three questions regarding the movement operation demonstrated in example (1): —How well do agrammatic aphasics perform this movement operation? —How well do children perform this movement operation? —To what extent can the ‘‘errors’’ that these two population make be compared or differentiated? We decided to conduct an experimental investigation rather than an analysis of spontaneous speech since, apart from the known advantages of experimental investigation, spontaneous speech of young children and of agrammatics contains virtually no embedded clauses and thus a real investigation of the asymmetry in example (1) can be done only via an experiment that will direct the subjects to use embedded clauses. VERB MOVEMENT IN APHASIA AND CHILD LANGUAGE
Spontaneous speech data of Dutch agrammatic aphasic patients and Dutch-speaking children reveal a lack of verb inflection: that is, in many cases a nonfinite verb appears as the main verb of a clause as in example (2) below: (2)
boek lezen book read.
Nevertheless, both populations are reported to make no errors with regard to the position of the verb; i.e., finite verbs do not appear in final position and infinitives do not appear in V2 position (Poepel and Wexler, 1993; Wexler, 1994; Bastiaanse & Van Zonneveld, 1998). The question that arises from these paradoxical findings (on the one hand, failure to inflect, and on the other hand, knowledge of the relation between inflection and structural position) is whether the problem here is primarily
VERB MOVEMENT IN ACQUISITION AND APHASIA
451
related to inflection or to movement. In order to answer this question a comparison between matrix and embedded clauses must be made. Unfortunately, spontaneous data of both agrammatics and children (of the relevant age) do not include embedded clauses and thus a direct comparison between verb inflection in matrix and embedded clauses cannot be achieved by spontaneous speech analysis. Bastiaanse and Van Zonneveld (1998) (henceforth B&vZ) used an experimental insertion/completion task to compare the production of finite and nonfinite verbs in matrix and embedded clauses in a Dutch agrammatic population. The method and results of B&vZ are described below. B&vZ insertion task. Ten agrammatic patients were presented with pictures accompanied by sentences in which the verb was missing. The task was to complete/ insert the missing verb in the correctly inflected form (thus, a one-word answer was sufficient). The sentences, divided into the following four conditions, were presented to the subjects both orally and in writing: (3)
a. De boer de koe (⫹fin V2, matrix) the cow the farmer (⫹fin final, embedded) b. Ik zie dat de boer de koe I see that the farmer the cow (⫺fin final, matrix) c. De boer wil de koe the farmer wants the cow (⫺fin final, embedded) d. Ik zie dat de boer de koe wil I see that the farmer the cow wants
answer: melkt milks answer: melkt milks answer: melken milk(inf) answer: melken milk(inf)
The results of their study are shown in Table 1. The first column is the only one which involves movement and it is also the only one that reflects a significant error rate. A comparison of the first two columns, which are parallel to the asymmetry in example (1), above, is particularly striking: the patients inflected the verb almost perfectly in the embedded clause (where movement was not required) but failed in 42% of the items in the matrix clause that required movement. These results led B&vZ to the following conclusion: agrammatic aphasics have a problem with verb movement rather than with verb inflection. Although their results were clear and showed statistical significance, two objections can be made with regard to the design of the experiment in B&vZ: (i) Asymmetry between insertion and completion might have been caused by an artifact. Because the task was to produce an isolated verb, the V2 condition (example (3a), above) was actually an insertion task while the other three V-final conditions (examples (3b)–(3d), above) were completion tasks. Since the results showed a difference between the V2 condition to the other three V-final conditions, it might be claimed that the difference here is one between insertion and completion and not one between movement and nonmovement, as the authors concluded. (ii) The patient gives a one-word answer (‘‘melkt,’’ in sentence (2a)) from which a direct conclusion in regard to the position of the verb (V2 vs. final) cannot be made;
TABLE 1 Bastiaanse and Van Zonneveld (1998) Results of 10 Patients— Number of Correctly Inflected Verbs/Number of Correctly Retrieved Verbs ⫹Finite V2 (matrix) 24/41
⫺Finite Final (embedded) 50/58
Final (matrix) 60/60
Final (embedded) 50/51
452
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
i.e., we know what the target position for the verb was but we do not know what the intention of the patient was when he gave the one-word answer. In fact, the position was dictated and the patient had only to match the correct inflection, rather than choose both the inflection and the position. For the purposes of the present paper, a new test suitable for both aphasics and children was developed, in which the subjects were asked to complete matrix and embedded clauses with a finite verb and an object. In this way the possible objections to B&vZ (1998) mentioned above are taken into account since the task is one of sentence completion for all conditions and the answers which the subjects are asked to give include a finite verb and an object, thus indicating the position of the verb, as well as the inflection. The experimental research has three objectives: (1) comparing the production of finite verbs in matrix and embedded clauses in agrammatic aphasics; (2) comparing the production of finite verbs in matrix and embedded clauses in children; (3) finding similarities/differences (both qualitative and quantitative) between the two populations.
METHODS Subjects. Four groups of subjects were tested: six agrammatic Broca’s aphasics and three groups of normally developing Dutch-speaking children (group 1, five children (ages 3;2–3;11) attending a part time day-care center; group 2, 14 children (ages 4;8–5) attending full-time kindergarten; group 3, two children (age 8 years) attending the third grade). Materials. A sentence completion test was developed in which the subject had to produce a finite verb and an object. Two pictures were presented to the subject (see Fig. 1). The experimenter presented the pictures through a coordination structure in which the first conjunct was fully produced by the experimenter and the second conjunct was truncated. The subjects were asked to complete the sentence (that is, to produce the verb and the object). The sentences were divided to two conditions: matrix sentences (VO condition), where the correct answer was verb–object, and embedded sentences (OV condition), where the correct answer was object–verb.
FIG. 1.
Example of two pictures: a man cuts a loaf of bread and a man cuts a tomato.
453
VERB MOVEMENT IN ACQUISITION AND APHASIA
TABLE 2 Performance of the Agrammatics and the Young Children Group (Ages, 3–4) Children (n ⫽ 5)
Matrix (VO) Relative (OV)
Agrammatics (n ⫽ 6)
Correct
Incorrect
Correct
Incorrect
37 (58%) 55 (89%)
27 (42%) 7 (11%)
51 (57%) 74 (82%)
39 (43%) 16 (18%)
(4) a. Experimenter: Dit is de man die het brood snijdt en dit is de man die de tomaat snijdt. Dus deze man snijdt het brood en deze man . . . [subject: snijdt de tomaat: VO] This is the man who the bread cuts and this is the man who the tomato cuts. So, this man cuts the bread and this man. . . . [subject: cuts the tomato: VO] b. Experimenter: Deze man snijdt het brood en deze man snijdt de tomaat. Dus dit is de man die het brood snijdt en dit is de man die. . . . [subject: de tomaat snijdt: OV] This man cuts the bread and this man cuts the tomato. So, this is the man who the bread cuts and this is the man who. . . [subject: the tomato cuts: OV] In the verb–object (VO) condition (4a), the verb has been moved (Verb Second); in the object–verb (OV) condition (4b), it is in its base-generated position. The aphasics and the children were presented with the same pictures and same test sentences; 32 picture pairs were presented, half in the VO-condition and half in the OV-condition.
RESULTS
The results for the aphasics and the young children are given in Table 2. In this table each answer that was deviant from the expected answer (as specified in examples (4a)–(4b), above) was considered as incorrect. Both the aphasics and the children performed very well in the embedded clauses (OV condition): the aphasics responded correctly to 82% of these sentences and the children to 89%. However, both groups failed with the matrix clauses (VO), which received only 57% of correct responses by the aphasics and 58% by the children. These differences are significant for both groups (aphasics, t ⫽ ⫺2.79, df ⫽ 5, p ⬍ .05; children, z ⫽ ⫺2.023, p ⬍ .05). The results of the older children are presented in Table 3. Only 4 of the 14 children of group 2 made an error. The results from group 2 show no difference between matrix and embedded clauses. The two 8-year-olds made no errors. Despite the similarity between the agrammatics and the children of the first group, an analysis of the incorrect responses shows an interesting difference between the two populations (see Table 4). In the matrix clause, the aphasics’ incorrect responses are mainly finite verbs in a final position (henceforth V fin-final error: deze man snijdt het brood en deze man TABLE 3 Results of the Older Children Group (Group 2) (n ⫽ 14)
Matrix (VO) Relative (OV)
⫹
⫺
198 (94%) 208 (99%)
12 (6%) 2 (1%)
454
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
TABLE 4 ‘‘Error’’ Types in Matrix Clauses for Aphasics and the Young Children
Aphasics Children
V fin-final
Aux-insertion
V-omission
Other
21 1
0 10
14 10
4 6
. . . aphasic: de tomaat snijdt: this man cuts the bread and this man . . . the tomato cuts), while most of the children’s incorrect responses concern insertion of an inflected auxiliary, thus leaving the (nonfinite) verb in final position (aux-insertion ‘‘error’’: child: . . . doet de tomaat snijden: . . . does the tomato cut). A third type of error is verb omission ‘‘error’’: (deze man snijdt het brood and deze man. . . . answer: de tomaat: this man cuts the bread and this man . . . the tomato). This error is produced by both aphasics and children. Omission of the verb in the second conjunct of a coordination is grammatical in Dutch (and many other languages) matrix clauses, but nevertheless it was counted as an incorrect response for reasons to be specified under Discussion. A fourth type of incorrect response, labeled as ‘‘other,’’ indicates that the subject did not respond or made an irrelevant response. The ‘‘errors’’ of the children of group 2 in the Matrix (VO) condition were six aux-insertions, six V-omissions; in the embedded (OV) condition: two V-omissions. DISCUSSION
Before turning to a discussion of the results, we must justify the decision to count the V-omission response as an incorrect one. By omitting the verb of the second conjunct in a coordination (deze man snijdt het brood and deze man. . . . answer: de tomaat: this man cuts the bread and this man . . . the tomato), the subjects created a structure which seems to be similar to what is usually referred to as ‘‘gapping’’ which is grammatical in Dutch (as well as in English) as in example (6) below: (6)
John ate an apple and Bill a pear.
Nevertheless, four reasons convinced us that this response should not be counted as an ‘‘innocent’’ adult-like response but rather as a failure to produce the required VO order: • First, such gapping-structures are considered to be more complex than the full coordination structures and children of the tested ages as well as agrammatic patients are said to not use them spontaneously. Tager-Flusberg, de Villiers, and Hakuta (1982) tested the ability of children to produce various coordination structures experimentally and found that gapping structures as in example (6) above were ‘‘most rarely produced, correctly imitated and understood’’ (see also Lust, Pinhas, and Flynn (1980) for the same conclusion). • Second, the results of the older children groups (groups 2 and 3) show a significant decrease in the use of this V-omission response (16% for group 1, 3% for group 2, and 0% for group 3). This fact supports the claim that the V-omission response should not be seen as an innocent adult-like response but as a strategy used to overcome a problematic operation (namely, overt V to I movement). When the operation ceases to be problematic, there is no use for the strategy and the V-omission ‘‘errors’’ disappear. • Third, the V-omission response and the aux-insertion response seem to disappear in the same time (the 4 children of 14 in group 2 who made any ‘‘error’’ pro-
VERB MOVEMENT IN ACQUISITION AND APHASIA
455
duced both responses and the rest of the children, including the two 8-year-olds, made none of them), again indicating that these responses represent an early stage of trying to avoid overt movement and not an adult-like behavior. • Fourth, one child of group 1 and two children of group 2 omitted the verb in the embedded clause (dit is de man die het brood snijdt en dit is de man die . . . de tomaat: this is the man who cuts the bread and this is the man who . . . the tomato), a response which is ungrammatical. Further research will determine whether this decision was correct (e.g., a similar experiment with different verbs in the two conjuncts or a similar experiment in a language that does not show an asymmetry between matrix and embedded clauses). The data support the conclusion of Bastiaanse and Van Zonneveld (1998) that the problem the Dutch agrammatics are facing is one of V-movement: they have the means to inflect the verb but they often fail to raise it in matrix clauses, resulting in a matrix/embedded asymmetry. The results of the current experiment show that B&vZ’s (1998) results were not due to an artifact (a difference between insertion task and completion task) but rather reflect a real syntactic asymmetry between matrix and embedded clauses, as originally concluded by the authors. What is interesting and new is that for child language a similar conclusion can be drawn: although children are equipped with a fully projected phrase marker, they choose to not raise the verb in matrix clauses, but instead insert an inflected auxiliary in I 0 to allow the main verb to remain in its base-generated position as an infinitive. It is important to point out here that such ‘‘aux-insertion’’ is allowed in the dialect spoken in the region where the children live and that it is known to be characteristic of child language (both in Dutch and in English; see Roeper (1991), Hollebrandse and Roeper (1996), and Van Kampen (1997). Nevertheless, in this dialect, do-insertion is allowed also in embedded clauses, whereas in the results presented above, children choose to insert the aux only in matrix clauses, where movement is required. This observation is crucial—it shows that indeed the aux-insertion strategy is not used simply in an optional, random manner but only in cases where overt movement of the verb is obligatory. Van Kampen (1997), who describes the acquisition process in terms of economy of derivation, observed the same phenomena (aux-insertion in matrix clauses, but not in embedded clauses) based on the analysis of spontaneous speech corpora. Therefore, the data above confirm Van Kampen’s prediction and indicate that children choose to insert the auxiliary only in cases where the other alternative is less economical. It is often claimed that Dutch children master verb movement from the early stages onward. The data presented here indicate, however, that although the children master one aspect of the inflection–movement equation, namely the knowledge that in matrix clauses finite verbs appear in V2 position and nonfinite verbs in final position, they do not yet master the other aspect of this equation, i.e., the requirement that in the absence of an auxiliary the lexical verb is moved. If such structures of aux-insertion (especially ones with auxiliaries/modals such as ‘‘gaan,’’ see Van Kampen 1997) are analyzed as grammatical progressive structures and not as indications for failing to perform V-movement, they may lead to the conclusion that Dutch children master V-movement from early ages onward. Two unexpected results emerge from the present study with respect to the agrammatic patients: the first is the large number of finite verbs in final position in matrix clauses, which are not attested in spontaneous speech, and the second is the lack of root infinitives (there were no responses of object followed by an infinitive: deze man snijdt het brood en deze man. . . . Answer -#tomaat snijden; this man cuts the bread and this man . . . -# tomato cut (inf.)). Root infinitive errors might have been expected based both on findings in spontaneous speech of agrammatics and on the results of B& vZ. These paradoxical results are actually two sides of the same phenomenon; it seems
456
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
that in the current experiment the aphasic patients used the Vfin-final structure instead of the root infinitive structure which was reported for spontaneous speech and in B& vZ. This is indicated by the observation that the rate of the agrammatics’ V fin-final error in the current experiment (43%) is similar to the rate of root infinitive errors in B&vZ (42%). One possible explanation for the lack of root infinitive errors and for the existence of V fin-final errors in the current experiment is that the coordination structures used in the current experiment triggered the subjects to prefer the V fin-final structure over the root infinitive structure. A demonstration for the effect that a coordination structure has on the finiteness of the verb is shown below: In Dutch, several contexts allow root infinitives in normal-language use, e.g.: (7) —
Wat doen jullie elke avond? What do you(pl.) every evening? [what are you doing every evening] Ik T.V. kijken en mijn vrouw een boek lezen I T.V. watch(inf ) and my wife a book read(inf )
Nevertheless, there is a restriction on these structures: the verbs of the two conjuncts must be identical with respect to their finiteness; thus, example (8) below is ungrammatical: (8) —
Wat doen jullie elke avond? What do you every evening? *Ik kijk T.V. en mijn vrouw een boek lezen I watch(fin) T.V. and my wife a book read(inf )
Answer (8) is ungrammatical since it violates the restriction on coordination structures; the verb of the first conjunct (kijk ‘‘watch’’) is finite while the verb of the second conjunct (lezen ‘‘read’’) is nonfinite. If we assume that agrammatics are aware of this restriction, then the use of V fin-final structure instead of root infinitives is explained: since the verb of the first conjunct (given by the experimenter) is finite, the agrammatics’ natural choice of an infinitive is ruled out by the coordination restriction; and the use a finite verb is forced. If this analysis is correct, it shows that a violation of a movement rule (namely, the rule that requires that in matrix clauses finite verbs must be moved) is considered, in the eyes of the agrammatic, to be a weaker violation than a violation of the restriction on coordination mentioned above. Further, this fact shows again that aphasics are able to inflect a verb when they feel it is necessary for the convergence of the derivation. Friedmann (1999), who tested verb movement in agrammatism in several languages, concluded that the preference of agrammatics is not necessarily for the infinitive but might be for other forms that do not require raising to high functional positions, dependant on language-specific requirements. The current results imply, as an addition to Friedmann’s conclusion that not only the nature of the specific language determines the choice of form but also the specific requirements of the structure (in this case coordination) that is tested. Regarding the comparison between the two populations, an interesting conclusion emerges: While agrammatics and children share the same problem—failing to raise the finite verb overtly—they choose different strategies to overcome this problem. The aphasics apply a ‘‘nonmovement’’ strategy and the children a ‘‘merge’’ strategy. Both strategies are available in UG and result in a structure that is more economical than the target structure. Why should agrammatics and children use different strategies to overcome the same problem? A possible answer to this question is that while children have a problem with V-movement alone, agrammatics’ problem is a combination of a problem with V-movement and the underspecification of Tense. Thus the
457
VERB MOVEMENT IN ACQUISITION AND APHASIA
children are able to overcome the V-movement problem by inserting an auxiliary in I but the agrammatics are forced to leave the verb in the final position. In conclusion, the three research questions presented above received an answer: both agrammatics and children have a problem with V-movement which creates an asymmetry between matrix clauses (which require overt V-movement) and embedded clauses (which lack V-movement); nevertheless this mutual problem is solved through different strategies by the two populations, as children can project a full tree which allows them to insert the aux in I, while agrammatics have an underspecified tense projection and are thus forced to leave the verb in the final position. In addition it is concluded that, although the natural preference of Dutch agrammatics is for the infinitive, in the case of coordination with a finite verb in the first conjunct, they are able to produce a finite verb in final position in the second conjunct to satisfy the local demands of the coordination structure.
APPENDIX: INDIVIDUAL RESULTS
Individual Results: Children Relative Clause (OV) Total Correct Rowin Bo Nils Teun Thom Total
17 8 17 11 9 62
V2 (VO)
15 6 16 10 6 53
Aux-insertion
V-omission
1
Doet ⫹V-omission
Other
1 2 1 1
1 (!!) 1
1
1 1
2
1 4
Doet ⫹V-omission
Other
Matrix Clause (VO)
Rowin Bo Nils Teun Thom Total
Total
Correct
17 9 17 12 9 64
11 2 11 4 4 32
V fin final
Aux-insertion
V-omission
6 6 3 1 1 10
1
2 2 10
1 3 2 5
3 2 6
Individual Results—Agrammatics Relative Clause (OV)
S1 S2 S3 S4 S5 S6 Total
Total
Correct
V2 (VO)
Aux-insertion
V-omission
Other
15 15 15 15 15 15 90
11 8 14 13 14 14 74
1 6 1 — 1 1 10
— — — — — — 0
— — — — — — 0
3 1 — 2 — — 6
458
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
Matrix Clause (VO)
S1 S2 S3 S4 S5 S6 Total
Total
Correct
V fin-final (OV)
Aux-insertion
V-omission
Other
15 15 15 15 15 15 90
5 10 7 7 10 12 51
3 2 3 7 5 1 21
— — — — — — 0
7 — 5 — — 2 14
— 3 — 1 — — 4
REFERENCES Avrutin, S. (1999). Development of the syntax–discourse interface. Dordrecht: Kluwer. Bastiaanse, R., & van Zonneveld, R. (1998). On the relation between verb inflection and verb position in Dutch agrammatic aphasics. Brain and Language, 64, 165–181. Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Friedmann, N., & Grodzinky, Y. (1997). Tense and agreement in agrammatic production: Pruning the syntactic tree. Brain and Language, 56, 397–425. Friedmann, N. (1999). Moving verbs in agrammatic production. In R. Bastiaanse and Y. Grodzinsky (Eds.), Grammatical disorders in aphasia: A neurolinguistic perspective. London: Whurr. Graetz, P., de Bleser, R., & Willmes, K. (1992). Akense Afasie Test. Lisse: Swets Zeitlinger. Grodzinsky, Y. (1990). Theoretical perspectives on language deficits. Cambridge, MA: MIT Press. Hollebrandse, B., and Roeper, T. (1996). The concept of DO-insertion and the theory of INFL in acquisition. In Charlotte Koster & Frank Wijnen (Eds.), Proceedings of the Groningen Assembly on Language Acquisition, Netherlands: Centre Language and Cognition Groningen, 1996 (pp. 261–271). Jakobson, R. (1971). Studies on child language and aphasia. The Hague: Mouton. Van Kampen, J. (1997). First steps in WH, Doctoral dissertation, Utrecht, the Netherlands. Lust, B., Pinhas, J., & Flynn, S. (1980). The acquisition of gapped coordination structures. Paper presented at BUCLD, Boston, 1980. Platzack, C. (1996). The initial hypothesis of syntax. In H. Clahsen (Ed.), Generative perspective on language acquisition (pp. 369–414). Amsterdam: Benjamins. Poeppel, D., & Wexler, K. (1993). The full competence hypothesis of clause structure in early German. Language, 69(1), Mar, 1–33. Radford, A. (1990). Syntactic theory and the acquisition of English syntax: The nature of early child grammars of English. Oxford, UK: Basil Blackwell. Rizzi, L. (1995). Some notes on linguistic theory and language development: The case of root infinitives. Language Acquisition, 3(4), 371–393. Roeper, T. (1991). How a marked parameter is chosen: Adverbs and do-insertion in the IP of child grammar. In T. L. Maxfield & B. Plunkett (Eds.), Papers in the acquisition of WH, Proceedings of the Umass Roundtable, May 1990, Amherst, pp. 175–202. Tager-Flusberg, H., de Villiers, J., & Hakuta, K. (1982). ‘‘The Development of sentence coordination.’’ In S. Kuczaj (Ed.), Language development (pp. 201–243). New Jersey: LEA. Wexler, K. (1994). Optional infinitives, head movement and the economy of derivation. In D. Lightfoot and N. Hornstein (Eds.), Verb movement. Cambridge, UK: Cambridge University Press.
Verb Movement in Acquisition and Aphasia: Same Problem, Different Solutions—Evidence from Dutch Shalom Zuckerman, Roelien Bastiaanse, and Ron van Zonneveld Graduate School for Behavioral and Cognitive Neurosciences, University of Groningen, The Netherlands Published online March 9, 2001
This paper focuses on verb movement in agrammatism and child language. We present data from a sentence completion experiment with 6 Dutch agrammatic aphasics and 21 Dutchspeaking children. The experiment compares completion of matrix clauses (which require verb movement) and embedded clauses (where such movement is not required) in these two populations. The results reveal a clear asymmetry: Both agrammatics and children do very well with embedded clauses but fail in 50% with the matrix clauses. It is concluded that the problem which both populations are facing is one of verb movement rather than verb inflection. An error analysis of the responses reveals that, although both agrammatics and children try to avoid movement, they apply different strategies to achieve this goal. 2001 Academic Press Key Words: agrammatism; language acquisition; syntax; verb movement; Dutch.
INTRODUCTION
The idea that aphasia and language acquisition might share properties goes back to the writings of Jakobson (1971), who suggested that language loss, such as seen in aphasic patients, is a mirror image of the course of acquisition seen in normally developing children. The idea is that operations which are the last ones to be acquired by children are the first to be lost by aphasic patients. Looking at the literature on child language and agrammatic aphasic speech reveals that similar theories and ideas are independently proposed for each of these two fields, for example, lack of functional categories, truncation, underspecification, problems with chain formation, processing difficulties, and problems with movement [for instance, for acquisition, Radford (1990), Wexler (1994), Rizzi (1995), Platzack (1996), Avrutin (1999); for Agrammatism: Grodzinsky (1990), Platzack (1996), Friedmann and Grodzinsky (1997), Avrutin (1999)]. In the current paper we raise the similarity/difference question regarding an operation which is rather well documented in the literature: verb movement in Dutch. We present results from an experimental sentence completion task conducted with agrammatic aphasics and three groups of normally developing children. The results reveal that agrammatic aphasics and young children share the problem of raising the finite verb to I 0 in matrix clauses, whereas performance in embedded clauses (where the finite verb remains in a lower position) is virtually
Address reprint requests to S. Zuckerman, OTS, Trans 10 3512 JK, Utrecht, The Netherlands. 449 0093-934X/01 $35.00 Copyright 2001 by Academic Press All rights of reproduction in any form reserved.
450
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
normal. A qualitative error analysis shows, however, that children and aphasics use different strategies to overcome this problem. VERB MOVEMENT IN DUTCH
In Dutch, there is a matrix/embedded clause asymmetry regarding verb placement, as shown in example (1) below: (1)
a. De man leest i een boek t i The man reads a book. b. De man die een boek leest The man who a book reads.
It is generally accepted that in the matrix clause (1a) the finite verb has undergone movement, while in the embedded clause (1b) it remains in its base-generated position. It is important to point out here that the asymmetry in example (1) is one of verb movement and not of verb inflection, since the verb ‘‘leest’’ (reads) is inflected in both matrix and embedded contexts while it is also moved only in the matrix clause. Movement and inflection have often been treated as two sides of the same coin, since it is assumed that the motivation for verb movement to I is to check (or collect) inflection; however, the asymmetry in example (1) allows us to focus on one of these two properties, reaching more fine-grained conclusions regarding the limitations that agrammatics and children face. The purpose of this paper is to answer, through an experimental investigation, the following three questions regarding the movement operation demonstrated in example (1): —How well do agrammatic aphasics perform this movement operation? —How well do children perform this movement operation? —To what extent can the ‘‘errors’’ that these two population make be compared or differentiated? We decided to conduct an experimental investigation rather than an analysis of spontaneous speech since, apart from the known advantages of experimental investigation, spontaneous speech of young children and of agrammatics contains virtually no embedded clauses and thus a real investigation of the asymmetry in example (1) can be done only via an experiment that will direct the subjects to use embedded clauses. VERB MOVEMENT IN APHASIA AND CHILD LANGUAGE
Spontaneous speech data of Dutch agrammatic aphasic patients and Dutch-speaking children reveal a lack of verb inflection: that is, in many cases a nonfinite verb appears as the main verb of a clause as in example (2) below: (2)
boek lezen book read.
Nevertheless, both populations are reported to make no errors with regard to the position of the verb; i.e., finite verbs do not appear in final position and infinitives do not appear in V2 position (Poepel and Wexler, 1993; Wexler, 1994; Bastiaanse & Van Zonneveld, 1998). The question that arises from these paradoxical findings (on the one hand, failure to inflect, and on the other hand, knowledge of the relation between inflection and structural position) is whether the problem here is primarily
VERB MOVEMENT IN ACQUISITION AND APHASIA
451
related to inflection or to movement. In order to answer this question a comparison between matrix and embedded clauses must be made. Unfortunately, spontaneous data of both agrammatics and children (of the relevant age) do not include embedded clauses and thus a direct comparison between verb inflection in matrix and embedded clauses cannot be achieved by spontaneous speech analysis. Bastiaanse and Van Zonneveld (1998) (henceforth B&vZ) used an experimental insertion/completion task to compare the production of finite and nonfinite verbs in matrix and embedded clauses in a Dutch agrammatic population. The method and results of B&vZ are described below. B&vZ insertion task. Ten agrammatic patients were presented with pictures accompanied by sentences in which the verb was missing. The task was to complete/ insert the missing verb in the correctly inflected form (thus, a one-word answer was sufficient). The sentences, divided into the following four conditions, were presented to the subjects both orally and in writing: (3)
a. De boer de koe (⫹fin V2, matrix) the cow the farmer (⫹fin final, embedded) b. Ik zie dat de boer de koe I see that the farmer the cow (⫺fin final, matrix) c. De boer wil de koe the farmer wants the cow (⫺fin final, embedded) d. Ik zie dat de boer de koe wil I see that the farmer the cow wants
answer: melkt milks answer: melkt milks answer: melken milk(inf) answer: melken milk(inf)
The results of their study are shown in Table 1. The first column is the only one which involves movement and it is also the only one that reflects a significant error rate. A comparison of the first two columns, which are parallel to the asymmetry in example (1), above, is particularly striking: the patients inflected the verb almost perfectly in the embedded clause (where movement was not required) but failed in 42% of the items in the matrix clause that required movement. These results led B&vZ to the following conclusion: agrammatic aphasics have a problem with verb movement rather than with verb inflection. Although their results were clear and showed statistical significance, two objections can be made with regard to the design of the experiment in B&vZ: (i) Asymmetry between insertion and completion might have been caused by an artifact. Because the task was to produce an isolated verb, the V2 condition (example (3a), above) was actually an insertion task while the other three V-final conditions (examples (3b)–(3d), above) were completion tasks. Since the results showed a difference between the V2 condition to the other three V-final conditions, it might be claimed that the difference here is one between insertion and completion and not one between movement and nonmovement, as the authors concluded. (ii) The patient gives a one-word answer (‘‘melkt,’’ in sentence (2a)) from which a direct conclusion in regard to the position of the verb (V2 vs. final) cannot be made;
TABLE 1 Bastiaanse and Van Zonneveld (1998) Results of 10 Patients— Number of Correctly Inflected Verbs/Number of Correctly Retrieved Verbs ⫹Finite V2 (matrix) 24/41
⫺Finite Final (embedded) 50/58
Final (matrix) 60/60
Final (embedded) 50/51
452
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
i.e., we know what the target position for the verb was but we do not know what the intention of the patient was when he gave the one-word answer. In fact, the position was dictated and the patient had only to match the correct inflection, rather than choose both the inflection and the position. For the purposes of the present paper, a new test suitable for both aphasics and children was developed, in which the subjects were asked to complete matrix and embedded clauses with a finite verb and an object. In this way the possible objections to B&vZ (1998) mentioned above are taken into account since the task is one of sentence completion for all conditions and the answers which the subjects are asked to give include a finite verb and an object, thus indicating the position of the verb, as well as the inflection. The experimental research has three objectives: (1) comparing the production of finite verbs in matrix and embedded clauses in agrammatic aphasics; (2) comparing the production of finite verbs in matrix and embedded clauses in children; (3) finding similarities/differences (both qualitative and quantitative) between the two populations.
METHODS Subjects. Four groups of subjects were tested: six agrammatic Broca’s aphasics and three groups of normally developing Dutch-speaking children (group 1, five children (ages 3;2–3;11) attending a part time day-care center; group 2, 14 children (ages 4;8–5) attending full-time kindergarten; group 3, two children (age 8 years) attending the third grade). Materials. A sentence completion test was developed in which the subject had to produce a finite verb and an object. Two pictures were presented to the subject (see Fig. 1). The experimenter presented the pictures through a coordination structure in which the first conjunct was fully produced by the experimenter and the second conjunct was truncated. The subjects were asked to complete the sentence (that is, to produce the verb and the object). The sentences were divided to two conditions: matrix sentences (VO condition), where the correct answer was verb–object, and embedded sentences (OV condition), where the correct answer was object–verb.
FIG. 1.
Example of two pictures: a man cuts a loaf of bread and a man cuts a tomato.
453
VERB MOVEMENT IN ACQUISITION AND APHASIA
TABLE 2 Performance of the Agrammatics and the Young Children Group (Ages, 3–4) Children (n ⫽ 5)
Matrix (VO) Relative (OV)
Agrammatics (n ⫽ 6)
Correct
Incorrect
Correct
Incorrect
37 (58%) 55 (89%)
27 (42%) 7 (11%)
51 (57%) 74 (82%)
39 (43%) 16 (18%)
(4) a. Experimenter: Dit is de man die het brood snijdt en dit is de man die de tomaat snijdt. Dus deze man snijdt het brood en deze man . . . [subject: snijdt de tomaat: VO] This is the man who the bread cuts and this is the man who the tomato cuts. So, this man cuts the bread and this man. . . . [subject: cuts the tomato: VO] b. Experimenter: Deze man snijdt het brood en deze man snijdt de tomaat. Dus dit is de man die het brood snijdt en dit is de man die. . . . [subject: de tomaat snijdt: OV] This man cuts the bread and this man cuts the tomato. So, this is the man who the bread cuts and this is the man who. . . [subject: the tomato cuts: OV] In the verb–object (VO) condition (4a), the verb has been moved (Verb Second); in the object–verb (OV) condition (4b), it is in its base-generated position. The aphasics and the children were presented with the same pictures and same test sentences; 32 picture pairs were presented, half in the VO-condition and half in the OV-condition.
RESULTS
The results for the aphasics and the young children are given in Table 2. In this table each answer that was deviant from the expected answer (as specified in examples (4a)–(4b), above) was considered as incorrect. Both the aphasics and the children performed very well in the embedded clauses (OV condition): the aphasics responded correctly to 82% of these sentences and the children to 89%. However, both groups failed with the matrix clauses (VO), which received only 57% of correct responses by the aphasics and 58% by the children. These differences are significant for both groups (aphasics, t ⫽ ⫺2.79, df ⫽ 5, p ⬍ .05; children, z ⫽ ⫺2.023, p ⬍ .05). The results of the older children are presented in Table 3. Only 4 of the 14 children of group 2 made an error. The results from group 2 show no difference between matrix and embedded clauses. The two 8-year-olds made no errors. Despite the similarity between the agrammatics and the children of the first group, an analysis of the incorrect responses shows an interesting difference between the two populations (see Table 4). In the matrix clause, the aphasics’ incorrect responses are mainly finite verbs in a final position (henceforth V fin-final error: deze man snijdt het brood en deze man TABLE 3 Results of the Older Children Group (Group 2) (n ⫽ 14)
Matrix (VO) Relative (OV)
⫹
⫺
198 (94%) 208 (99%)
12 (6%) 2 (1%)
454
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
TABLE 4 ‘‘Error’’ Types in Matrix Clauses for Aphasics and the Young Children
Aphasics Children
V fin-final
Aux-insertion
V-omission
Other
21 1
0 10
14 10
4 6
. . . aphasic: de tomaat snijdt: this man cuts the bread and this man . . . the tomato cuts), while most of the children’s incorrect responses concern insertion of an inflected auxiliary, thus leaving the (nonfinite) verb in final position (aux-insertion ‘‘error’’: child: . . . doet de tomaat snijden: . . . does the tomato cut). A third type of error is verb omission ‘‘error’’: (deze man snijdt het brood and deze man. . . . answer: de tomaat: this man cuts the bread and this man . . . the tomato). This error is produced by both aphasics and children. Omission of the verb in the second conjunct of a coordination is grammatical in Dutch (and many other languages) matrix clauses, but nevertheless it was counted as an incorrect response for reasons to be specified under Discussion. A fourth type of incorrect response, labeled as ‘‘other,’’ indicates that the subject did not respond or made an irrelevant response. The ‘‘errors’’ of the children of group 2 in the Matrix (VO) condition were six aux-insertions, six V-omissions; in the embedded (OV) condition: two V-omissions. DISCUSSION
Before turning to a discussion of the results, we must justify the decision to count the V-omission response as an incorrect one. By omitting the verb of the second conjunct in a coordination (deze man snijdt het brood and deze man. . . . answer: de tomaat: this man cuts the bread and this man . . . the tomato), the subjects created a structure which seems to be similar to what is usually referred to as ‘‘gapping’’ which is grammatical in Dutch (as well as in English) as in example (6) below: (6)
John ate an apple and Bill a pear.
Nevertheless, four reasons convinced us that this response should not be counted as an ‘‘innocent’’ adult-like response but rather as a failure to produce the required VO order: • First, such gapping-structures are considered to be more complex than the full coordination structures and children of the tested ages as well as agrammatic patients are said to not use them spontaneously. Tager-Flusberg, de Villiers, and Hakuta (1982) tested the ability of children to produce various coordination structures experimentally and found that gapping structures as in example (6) above were ‘‘most rarely produced, correctly imitated and understood’’ (see also Lust, Pinhas, and Flynn (1980) for the same conclusion). • Second, the results of the older children groups (groups 2 and 3) show a significant decrease in the use of this V-omission response (16% for group 1, 3% for group 2, and 0% for group 3). This fact supports the claim that the V-omission response should not be seen as an innocent adult-like response but as a strategy used to overcome a problematic operation (namely, overt V to I movement). When the operation ceases to be problematic, there is no use for the strategy and the V-omission ‘‘errors’’ disappear. • Third, the V-omission response and the aux-insertion response seem to disappear in the same time (the 4 children of 14 in group 2 who made any ‘‘error’’ pro-
VERB MOVEMENT IN ACQUISITION AND APHASIA
455
duced both responses and the rest of the children, including the two 8-year-olds, made none of them), again indicating that these responses represent an early stage of trying to avoid overt movement and not an adult-like behavior. • Fourth, one child of group 1 and two children of group 2 omitted the verb in the embedded clause (dit is de man die het brood snijdt en dit is de man die . . . de tomaat: this is the man who cuts the bread and this is the man who . . . the tomato), a response which is ungrammatical. Further research will determine whether this decision was correct (e.g., a similar experiment with different verbs in the two conjuncts or a similar experiment in a language that does not show an asymmetry between matrix and embedded clauses). The data support the conclusion of Bastiaanse and Van Zonneveld (1998) that the problem the Dutch agrammatics are facing is one of V-movement: they have the means to inflect the verb but they often fail to raise it in matrix clauses, resulting in a matrix/embedded asymmetry. The results of the current experiment show that B&vZ’s (1998) results were not due to an artifact (a difference between insertion task and completion task) but rather reflect a real syntactic asymmetry between matrix and embedded clauses, as originally concluded by the authors. What is interesting and new is that for child language a similar conclusion can be drawn: although children are equipped with a fully projected phrase marker, they choose to not raise the verb in matrix clauses, but instead insert an inflected auxiliary in I 0 to allow the main verb to remain in its base-generated position as an infinitive. It is important to point out here that such ‘‘aux-insertion’’ is allowed in the dialect spoken in the region where the children live and that it is known to be characteristic of child language (both in Dutch and in English; see Roeper (1991), Hollebrandse and Roeper (1996), and Van Kampen (1997). Nevertheless, in this dialect, do-insertion is allowed also in embedded clauses, whereas in the results presented above, children choose to insert the aux only in matrix clauses, where movement is required. This observation is crucial—it shows that indeed the aux-insertion strategy is not used simply in an optional, random manner but only in cases where overt movement of the verb is obligatory. Van Kampen (1997), who describes the acquisition process in terms of economy of derivation, observed the same phenomena (aux-insertion in matrix clauses, but not in embedded clauses) based on the analysis of spontaneous speech corpora. Therefore, the data above confirm Van Kampen’s prediction and indicate that children choose to insert the auxiliary only in cases where the other alternative is less economical. It is often claimed that Dutch children master verb movement from the early stages onward. The data presented here indicate, however, that although the children master one aspect of the inflection–movement equation, namely the knowledge that in matrix clauses finite verbs appear in V2 position and nonfinite verbs in final position, they do not yet master the other aspect of this equation, i.e., the requirement that in the absence of an auxiliary the lexical verb is moved. If such structures of aux-insertion (especially ones with auxiliaries/modals such as ‘‘gaan,’’ see Van Kampen 1997) are analyzed as grammatical progressive structures and not as indications for failing to perform V-movement, they may lead to the conclusion that Dutch children master V-movement from early ages onward. Two unexpected results emerge from the present study with respect to the agrammatic patients: the first is the large number of finite verbs in final position in matrix clauses, which are not attested in spontaneous speech, and the second is the lack of root infinitives (there were no responses of object followed by an infinitive: deze man snijdt het brood en deze man. . . . Answer -#tomaat snijden; this man cuts the bread and this man . . . -# tomato cut (inf.)). Root infinitive errors might have been expected based both on findings in spontaneous speech of agrammatics and on the results of B& vZ. These paradoxical results are actually two sides of the same phenomenon; it seems
456
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
that in the current experiment the aphasic patients used the Vfin-final structure instead of the root infinitive structure which was reported for spontaneous speech and in B& vZ. This is indicated by the observation that the rate of the agrammatics’ V fin-final error in the current experiment (43%) is similar to the rate of root infinitive errors in B&vZ (42%). One possible explanation for the lack of root infinitive errors and for the existence of V fin-final errors in the current experiment is that the coordination structures used in the current experiment triggered the subjects to prefer the V fin-final structure over the root infinitive structure. A demonstration for the effect that a coordination structure has on the finiteness of the verb is shown below: In Dutch, several contexts allow root infinitives in normal-language use, e.g.: (7) —
Wat doen jullie elke avond? What do you(pl.) every evening? [what are you doing every evening] Ik T.V. kijken en mijn vrouw een boek lezen I T.V. watch(inf ) and my wife a book read(inf )
Nevertheless, there is a restriction on these structures: the verbs of the two conjuncts must be identical with respect to their finiteness; thus, example (8) below is ungrammatical: (8) —
Wat doen jullie elke avond? What do you every evening? *Ik kijk T.V. en mijn vrouw een boek lezen I watch(fin) T.V. and my wife a book read(inf )
Answer (8) is ungrammatical since it violates the restriction on coordination structures; the verb of the first conjunct (kijk ‘‘watch’’) is finite while the verb of the second conjunct (lezen ‘‘read’’) is nonfinite. If we assume that agrammatics are aware of this restriction, then the use of V fin-final structure instead of root infinitives is explained: since the verb of the first conjunct (given by the experimenter) is finite, the agrammatics’ natural choice of an infinitive is ruled out by the coordination restriction; and the use a finite verb is forced. If this analysis is correct, it shows that a violation of a movement rule (namely, the rule that requires that in matrix clauses finite verbs must be moved) is considered, in the eyes of the agrammatic, to be a weaker violation than a violation of the restriction on coordination mentioned above. Further, this fact shows again that aphasics are able to inflect a verb when they feel it is necessary for the convergence of the derivation. Friedmann (1999), who tested verb movement in agrammatism in several languages, concluded that the preference of agrammatics is not necessarily for the infinitive but might be for other forms that do not require raising to high functional positions, dependant on language-specific requirements. The current results imply, as an addition to Friedmann’s conclusion that not only the nature of the specific language determines the choice of form but also the specific requirements of the structure (in this case coordination) that is tested. Regarding the comparison between the two populations, an interesting conclusion emerges: While agrammatics and children share the same problem—failing to raise the finite verb overtly—they choose different strategies to overcome this problem. The aphasics apply a ‘‘nonmovement’’ strategy and the children a ‘‘merge’’ strategy. Both strategies are available in UG and result in a structure that is more economical than the target structure. Why should agrammatics and children use different strategies to overcome the same problem? A possible answer to this question is that while children have a problem with V-movement alone, agrammatics’ problem is a combination of a problem with V-movement and the underspecification of Tense. Thus the
457
VERB MOVEMENT IN ACQUISITION AND APHASIA
children are able to overcome the V-movement problem by inserting an auxiliary in I but the agrammatics are forced to leave the verb in the final position. In conclusion, the three research questions presented above received an answer: both agrammatics and children have a problem with V-movement which creates an asymmetry between matrix clauses (which require overt V-movement) and embedded clauses (which lack V-movement); nevertheless this mutual problem is solved through different strategies by the two populations, as children can project a full tree which allows them to insert the aux in I, while agrammatics have an underspecified tense projection and are thus forced to leave the verb in the final position. In addition it is concluded that, although the natural preference of Dutch agrammatics is for the infinitive, in the case of coordination with a finite verb in the first conjunct, they are able to produce a finite verb in final position in the second conjunct to satisfy the local demands of the coordination structure.
APPENDIX: INDIVIDUAL RESULTS
Individual Results: Children Relative Clause (OV) Total Correct Rowin Bo Nils Teun Thom Total
17 8 17 11 9 62
V2 (VO)
15 6 16 10 6 53
Aux-insertion
V-omission
1
Doet ⫹V-omission
Other
1 2 1 1
1 (!!) 1
1
1 1
2
1 4
Doet ⫹V-omission
Other
Matrix Clause (VO)
Rowin Bo Nils Teun Thom Total
Total
Correct
17 9 17 12 9 64
11 2 11 4 4 32
V fin final
Aux-insertion
V-omission
6 6 3 1 1 10
1
2 2 10
1 3 2 5
3 2 6
Individual Results—Agrammatics Relative Clause (OV)
S1 S2 S3 S4 S5 S6 Total
Total
Correct
V2 (VO)
Aux-insertion
V-omission
Other
15 15 15 15 15 15 90
11 8 14 13 14 14 74
1 6 1 — 1 1 10
— — — — — — 0
— — — — — — 0
3 1 — 2 — — 6
458
ZUCKERMAN, BASTIAANSE, AND VAN ZONNEVELD
Matrix Clause (VO)
S1 S2 S3 S4 S5 S6 Total
Total
Correct
V fin-final (OV)
Aux-insertion
V-omission
Other
15 15 15 15 15 15 90
5 10 7 7 10 12 51
3 2 3 7 5 1 21
— — — — — — 0
7 — 5 — — 2 14
— 3 — 1 — — 4
REFERENCES Avrutin, S. (1999). Development of the syntax–discourse interface. Dordrecht: Kluwer. Bastiaanse, R., & van Zonneveld, R. (1998). On the relation between verb inflection and verb position in Dutch agrammatic aphasics. Brain and Language, 64, 165–181. Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Friedmann, N., & Grodzinky, Y. (1997). Tense and agreement in agrammatic production: Pruning the syntactic tree. Brain and Language, 56, 397–425. Friedmann, N. (1999). Moving verbs in agrammatic production. In R. Bastiaanse and Y. Grodzinsky (Eds.), Grammatical disorders in aphasia: A neurolinguistic perspective. London: Whurr. Graetz, P., de Bleser, R., & Willmes, K. (1992). Akense Afasie Test. Lisse: Swets Zeitlinger. Grodzinsky, Y. (1990). Theoretical perspectives on language deficits. Cambridge, MA: MIT Press. Hollebrandse, B., and Roeper, T. (1996). The concept of DO-insertion and the theory of INFL in acquisition. In Charlotte Koster & Frank Wijnen (Eds.), Proceedings of the Groningen Assembly on Language Acquisition, Netherlands: Centre Language and Cognition Groningen, 1996 (pp. 261–271). Jakobson, R. (1971). Studies on child language and aphasia. The Hague: Mouton. Van Kampen, J. (1997). First steps in WH, Doctoral dissertation, Utrecht, the Netherlands. Lust, B., Pinhas, J., & Flynn, S. (1980). The acquisition of gapped coordination structures. Paper presented at BUCLD, Boston, 1980. Platzack, C. (1996). The initial hypothesis of syntax. In H. Clahsen (Ed.), Generative perspective on language acquisition (pp. 369–414). Amsterdam: Benjamins. Poeppel, D., & Wexler, K. (1993). The full competence hypothesis of clause structure in early German. Language, 69(1), Mar, 1–33. Radford, A. (1990). Syntactic theory and the acquisition of English syntax: The nature of early child grammars of English. Oxford, UK: Basil Blackwell. Rizzi, L. (1995). Some notes on linguistic theory and language development: The case of root infinitives. Language Acquisition, 3(4), 371–393. Roeper, T. (1991). How a marked parameter is chosen: Adverbs and do-insertion in the IP of child grammar. In T. L. Maxfield & B. Plunkett (Eds.), Papers in the acquisition of WH, Proceedings of the Umass Roundtable, May 1990, Amherst, pp. 175–202. Tager-Flusberg, H., de Villiers, J., & Hakuta, K. (1982). ‘‘The Development of sentence coordination.’’ In S. Kuczaj (Ed.), Language development (pp. 201–243). New Jersey: LEA. Wexler, K. (1994). Optional infinitives, head movement and the economy of derivation. In D. Lightfoot and N. Hornstein (Eds.), Verb movement. Cambridge, UK: Cambridge University Press.