Traceless relatives: Agrammatic comprehension of relative clauses with resumptive pronouns

ARTICLE IN PRESS Journal of Neurolinguistics ] (]]]]) ]]] ]]] www.elsevier.com/locate/jneuroling Traceless relatives: Agrammatic comprehension of relative clauses with resumptive pronouns Naama Friedmann,1 School of Education, Language and Brain Lab, Tel Aviv University, Tel Aviv 69978, Israel Received 2 March 2006; received in revised form 29 September 2006; accepted 26 October 2006 Abstract Individuals with agrammatic aphasia fail to interpret reversible movement-derived sentences. According to the Trace Deletion Hypothesis, this impairment in comprehension results from the deletion of traces of phrasal movement. In order to test this hypothesis, we compared object relatives with a trace to identical object relatives that are not derived by syntactic movement and do not include a trace, but rather include a resumptive pronoun at the position of the embedded object. Five Hebrew-speaking individuals with agrammatism and 5 matched controls participated in this study. Comprehension was assessed using a binary sentence picture matching task of 120 reversible relative clauses per participant, 40 subject relatives, 40 object relatives, and 40 object relatives with a resumptive pronoun. The comprehension of subject relatives was significantly above chance, but the comprehension of both types of object relative was at chance. Importantly, the insertion of a resumptive pronoun at the position of the trace did not improve comprehension. The comprehension of object relatives with resumptive pronouns was at chance, and not different from the comprehension of object relatives with traces. Two modifications for the Trace Deletion Hypothesis are considered: a more general deficit in thematic role assignment over an intervening argument, or a deficit in the construction of CP that results in failure to make the syntactic relation between the relative head above CP, the operator in CP and the embedded pronoun or trace within the relative clause. r 2006 Elsevier Ltd. All rights reserved. Keywords: Aphasia; Agrammatism; Comprehension; Relative clauses; Trace Deletion Hypothesis; Tree pruning; Hebrew Tel.: +972 3 6405257. E-mail address: naamafr@post.tau.ac.il. 1 /http://www.tau.ac.il/naamafrs. 0911-6044/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jneuroling.2006.10.005

2 ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] 1. Introduction Individuals with agrammatic aphasia fail to interpret certain movement-derived sentences such as object relatives, object clefts, topicalized structures, and object questions. Data from sentence picture matching and grammaticality judgment tasks indicate that they do not know who did what to whom in reversible sentences when the theme noun phrase (NP) moves to a position before the agent. For example, when they encounter the sentence Here is the girl that the grandmother drew., they fail to identify the agent, and do not know whether the girl drew the grandmother, or the grandmother drew the girl (Friedmann & Shapiro, 2003; Grodzinsky, 1984, 1989; Zurif & Caramazza, 1976; see Grodzinsky, Pin ango, Zurif, & Drai, 1999, for a review). The question of what these impaired structures have in common and what distinguishes them from the intact ones has intrigued many researchers over the years, and several accounts have been suggested for the source of the deficit in agrammatism. One account, the Trace Deletion Hypothesis (TDH) (Grodzinsky, 1990, 1995, 2000), suggested that the interpretation of sentences that are derived by movement of NPs is impaired due to the deletion of traces of movement of referential NPs. This theory accounts for the comprehension deficit in a wide range of noncanonical structures such as object relatives, referential Wh object questions, and topicalization structures, which all include traces of movement of referential NPs. 2 However, this is not the only logical possibility to account for the deficit in these structures. These structures also share the need to transfer thematic roles (or establish a dependency) over another argument of the verb. In fact, they have something else in common: they all involve antecedents in high nodes of the syntactic tree (specifically, in Spec-CP). It is therefore also possible that the deficit is related to an inability to construct the syntactic tree up to its treetop, the CP node, that results in an inability to connect the antecedent and the trace. The current study sought to test whether the deficit in the comprehension of relative clauses relates to trace deletion and a deficit in movement of phrases, or whether the other two possibilities are valid, namely a deficit in crossing another argument or in structures that include CP. This was done using a special type of relative clause that exists in some languages, including Hebrew: relative clauses with a resumptive pronoun at the position of the gap. Object relatives with a resumptive pronoun resemble regular object relatives (with a trace) in everything, except for the fact that they include a pronoun instead of the trace within the relative clause. As seen in Fig. 1, which represents the syntactic tree of the two types of relative clauses (translated from Hebrew), the syntactic structure of object 2 Movement-derived sentences that preserve the canonical order such as subject relatives and subject questions usually yield better performance. The Trace Deletion Hypothesis employs a non-structural strategy to account for the better performance on these sentences. The idea is that although subject relatives in English and Hebrew, for example, are derived by phrasal movement, they yield better performance because they keep the canonical order of an agent as the first NP and a theme as the second. When an NP remains role-less, the strategy assigns the role of an agent to the first NP in the clause. When the first NP is indeed an agent, the sentence is interpreted correctly (though not by the normal syntactic process). However, when a movement takes place and the first NP in the sentence is the theme, it incorrectly receives an agent role. If the sentence includes a real agent in addition to the NP that mistakenly received an agent role from the first-np strategy (as is the case in object relatives), the hearer has to choose who the agent is, and is forced to guess (Grodzinsky, 1995, 2000).

ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] 3 NP NP CP CP The clerk 1 Op 1 that IP The clerk 1 Op 1 that IP Dudu drew VP Dudu drew VP t 1 him 1 Fig. 1. Syntactic trees for relative clauses with and without a resumptive pronoun. relatives with and without a resumptive pronoun is the same in that both of them include the CP node, but whereas the object relative with the trace involves movement (depicted by an arrow in the right hand side figure in Fig. 1), the object relative with the resumptive pronoun does not. If the deficit in the comprehension of relative clauses indeed results from trace deletion, we would predict that similar structures that do not rely on a trace for comprehension would be unimpaired. However, if the deficit is somewhat broader and involves impairment in thematic role assignment that crosses another argument or a deficit in CP, then object relatives with a resumptive pronoun in the embedded object position are expected to be impaired as well. 1.1. On resumption Restrictive relative clauses such as (1) and (2) modify a nominal head (in these examples the cactus and the duck). As such, they must contain a syntactic element in a thematic position within the relative clause which receives the interpretation of a bound variable. The most cross-linguistically pervasive strategy for achieving this involves movement: In sentences (1) and (2), an operator moves from the clause-internal thematic position (in these sentences from the object position) into the left periphery of the relative clause (Spec-CP), leaving behind a trace at the original position (Chomsky, 1986). The operator that moves can be either a Wh-element like which as in example (1), or a phonetically empty operator (Op) as in (2). The trace (t 1 ) at the original position of the operator receives the interpretation of a variable, which is bound by the moved operator namely the trace receives a thematic role and transfers it to the operator. The operator, in turn, is co-indexed with (receives the same index and therefore points to the same entity as) the nominal head.

4 ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] Some languages employ a different strategy that does not involve a trace of movement, alongside or instead of the movement-based strategy. This strategy is known as resumption: a pronominal element, a resumptive pronoun, is generated in the clause-internal thematic position, does not move, and is bound by a co-indexed operator that is base generated in Spec-CP (see the Hebrew example (3)). Thus, two structures are possible for relative clauses: one that involves movement and a trace and one that does not involve movement, and includes a resumptive pronoun. Hebrew allows both strategies in object relatives object relatives with a trace at the object position, and object relatives with a resumptive pronoun at the object position. Crucially, the resumptive pronoun in object relative clauses in Hebrew appears in a sentence that is not derived by movement, and that does not include a trace of movement the operator is not moved from within the relative clause but is rather base generated in CP (Borer, 1984; Shlonsky, 1992 for Hebrew, and Grolla, 2005; McKee & McDaniel, 2001 for similar analyses of other languages. See Aoun, Choueiri, & Hornstein, 2001 for two types of resumptives in Lebanese Arabic, and Friedmann, Novogrodsky, Szterman, & Preminger, in press for arguments for Hebrew resumptive pronouns being true resumptives). Thus, object relatives with a trace and object relatives with a resumptive pronoun form a minimal pair. They include the same syntactic structure, but only one of them involves movement. This makes the comparison between them an interesting and clean test case for the description of agrammatic comprehension: If traces are the only source of difficulty, we expect only object relatives with traces to be impaired. However, if the deficit is broader and related to every type of thematic role assignment that crosses another argument, or related to the construction of the tree up to CP, both object relatives with a trace and object relatives with a resumptive pronoun are expected to be impaired. 2. Experiment 2.1. Participants Five individuals with agrammatism, three women and two men, participated in this study. They were all native speakers of Hebrew. The participants were 20 34 years old (mean age 29;9), they had 12 15 years of education (M ¼ 12.8). The participants were in a stable condition, 18 months to 10 years post-onset. All were right handed and four of them had right hemiplegia. They had a lesion in the left hemisphere, in or involving the frontal lobe, 3 following stroke and 2 following head trauma. They were diagnosed as having Broca s aphasia with agrammatism using the Hebrew versions of the WAB (Kertesz, 1982; Hebrew version by Soroker, 1997) and the BAFLA a test battery for agrammatic comprehension and production (Friedmann, 1998b). All participants had

ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] 5 characteristic agrammatic speech: short, non-fluent, with ungrammatical utterances, use of mainly simple sentences, and ungrammatical production of complex sentences and Wh questions. In addition, we tested five native speakers of Hebrew without language impairment, each matched in age, gender, and education to one of the individuals in the agrammatic group, in order to test the validity of the sentences, pictures, and procedure. 2.2. Procedure Comprehension was assessed using a binary sentence picture matching task. The participant heard a semantically reversible Hebrew sentence, and saw two pictures on the same page, one above the other (see Fig. 2); In one picture the roles matched the sentence; in the other picture the roles were reversed. The participant was asked to point to the picture that correctly represented the sentence. There was no time limit and no limit on the number of repetitions of each sentence. Fig. 2. An example of a picture pair used in the sentence picture matching task.

6 ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] 2.3. Material The test included subject relative clauses (4), object relative clauses without a resumptive pronoun (5), and object relative clauses with a resumptive pronoun (6). Three participants HY, SB, and RA were presented with a total of 120 sentences each, 40 of each type; RN and GR were presented with a total of 90 sentences each, 30 of each type. All verbs were agentive transitives. In order to preclude an agreement cue on the verb (Hebrew verbs agree with the subject in gender, number, and person), the figures in every picture were always of the same gender and number (a little boy and a grandfather, a female nurse and a female soldier, etc.). (4) Subject relative: Tar e li et ha-kof she-mexabek et ha-yeled Show me ACC the-monkey that-hugs ACC the-boy Show me the monkey that hugs the boy. (5) Object relative: Tar e li et ha-kof she-ha-yeled mexabek Show me ACC the-monkey that-the-boy hugs Show me the monkey that the boy hugs. (6) Object relative with a resumptive pronoun: Tar e li et ha-kof she-ha-yeled mexabek oto Show me ACC the-monkey that-the-boy hugs him Show me the monkey that the boy hugs. The sentences were randomly ordered, and presented in three sessions, with the same number of sentences of each type per session and no more than two sentences of the same type appearing consecutively. The participant saw the picture pairs three times; each picture pair appeared with each of the three sentence types, once in each session. The correct picture in each pair was randomized within each session in each session half of the sentences matched the upper picture, and half matched the lower picture, with no more than three consecutive times pointing to the same picture position. 3. Results While the performance on subject relatives was significantly above chance, the performance in both types of object relative was at chance, as seen in Fig. 3. Consistent with previous findings in the literature, the performance on object relatives without resumptive pronouns was significantly poorer than the performance on the subject relatives, the mean performance on object relatives was 59.1%, and on subject relatives 85.2%, t(4) ¼ 4.85, p ¼ 0.004. Importantly, the performance on the object relatives with resumptive pronouns (M ¼ 58.3%) was also significantly poorer than that of subject relatives, t(4) ¼ 7.58, p ¼ 0.0008. The comprehension of object relatives with resumptive pronouns did not differ from the comprehension of object relatives without resumptive pronouns, t(4) ¼ 0.23, p ¼ 0.83. Namely, the addition of a resumptive pronoun to the position of the trace in object relatives did not improve comprehension. Each of the individual participants presented the same pattern of better performance in subject relatives than both types of object relatives (using w 2, po0.02, for every participant

ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] 7 100% 90% 80% 70% 60% 50% GR RN HY RA SB subject rel object rel object rel with resumptive Fig. 3. Comprehension of subject relatives and of object relatives with and without resumptive pronouns. except HY, for whom the difference did not reach significance, p ¼ 0.08), and no difference between object relatives with and without resumptive pronoun (using w 2, p40.37 for every participant). The comparison of each participant s performance to chance level using the binomial distribution yielded above-chance performance for each of the participants on subject relatives, and at-chance performance on each of the types of object relatives (except for RA who was above chance on object relatives without resumptive pronouns). The matched control participants performed at ceiling on all conditions. 4. Discussion Although object relatives with resumptive pronouns do not include movement or traces of movement, they were still comprehended poorly by the individuals with agrammatism, at a level similar to that of object relatives with a trace. Given that object relatives with and without a resumptive pronoun differ only in the involvement of movement of a NP, and in the existence of a movement trace, these results indicate that it is not (or at least not only) the traces that are deleted from the representation of the agrammatic structure that cause the deficit in comprehension of object relatives. The present results rather suggest that a broader application of the TDH might be suitable: the current study tells us that the deficit in agrammatism extends beyond traces to other types of dependencies, at least those that are bound from the CP node. Individuals with agrammatism cannot make the syntactic relation between the relative head, which is above CP, the operator in CP, and the element, be it a resumptive pronoun or a movement trace, within the embedded sentence. Had the deficit been only in traces of Wh-movement, we would expect to see deficits in the comprehension of structures that are derived by movement of a NP, but the addition of a resumptive pronoun at the trace position should have solved the problem. This is actually the case in a different population with a syntactic deficit: orally trained children with hearing impairment who have not been exposed to language at the critical period for the acquisition of syntax. These children show a significant difficulty in the comprehension and

8 ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] production of movement-derived sentences. They fail to understand object relatives, topicalized structures, passives, and object Wh questions (de Villiers, de Villiers, & Hoban, 1994; Friedmann & Szterman, 2006; Power & Quigley, 1973). However, they produce most of their object relatives with a resumptive pronoun, and when given an object relative with a resumptive pronoun their comprehension improves significantly and reaches a normal performance level (Friedmann & Szterman, 2006). It seems that children with hearing impairment are better characterized by the original TDH suggestion of a narrow deficit in movement and in the construction of traces of movement without a general deficit in variable binding or in syntactic structure building. However, the current study suggests that the nature of deficit in agrammatism is different. What type of extension of the TDH can account for the agrammatic deficit in comprehension in light of the current findings? One possible approach would be to say that not only trace-antecedent relations are impaired in agrammatism, but rather all the relations that include binding of a pronoun for thematic role assignment. There are various types of variable binding, but not all of them are impaired. Binding for establishing reference, for example (as in The boy pinched himself ), is not implicated under this account (see Grodzinsky, Wexler, Chien, Marakovitz, & Solomon, 1993 and Ruigendijk, Vasic, & Avrutin, 2006 for unimpaired interpretation of reflexives). However, variable binding that is used for transferring thematic roles, either via a trace and a chain of movement, or via pronoun binding, is impaired. A step in this direction of extending the deficit to other constructions that include variable binding was put forward by Grodzinsky (2005), who suggested that the deficit he reported in VP ellipsis construction actually results from the fact that these structures include variable binding. One theoretical drawback in such a generalization is that it is unclear why variable binding for theta transmission should be selectively impaired. Furthermore, some linguistic analyses consider relative word extraction not to be a genuine operator-variable chain. According to Lasnik and Stowell (1991), the element in Spec-CP that is co-indexed with the trace or the resumptive pronoun is a constant rather than a variable. If this is so, the question is: why should pronominal constants which are bound from CP be impaired while there is no impairment for ordinary pronouns? Two other modifications of the TDH might not suffer from the arbitrary taste of the variable binding approach. One extends it to all types of thematic role assignment that cross another argument of the verb, the other relates to the CP node. A property that object relatives with a resumptive pronoun share with object relatives and with other structures that agrammatic aphasics fail to understand is the fact that these structures require the establishment of a dependency over another argument of the same type, or the assignment of thematic roles over another argument of the same type. Namely, in subject relatives, subject questions and SVO sentences the agent immediately precedes the verb, without any intervening NP, and the theme follows it without intervening NPs, and therefore the establishment of a dependency and the transfer of thematic roles are successful. Unlike them, in object relatives, referential object questions, topicalization structures and object clefts, the assignment of thematic roles to the theme, or the establishment of a dependency that involves the theme, needs to cross another argument of the verb of the same type, in these structures, the agent. 3 For example, in the sentence I 3 The specification of the intervening argument being of the same type needs a more exact specification, of course, but it comes to capture the findings that, for example, non-referential antecedents are not blocked by

ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] 9 know the girl that the grandmother drew _, in order to access the theme the girl, the agent the grandmother has to be crossed. This generalization captures the same data that the TDH captures without assuming a deficit in movement, and thus can also account for the deficit in object relatives that do not involve movement, namely, object relatives with a resumptive pronoun. Such an account has been suggested by Friedmann and Shapiro (2003, p. 295), and has been given a more specific formulation in terms of recent developments of the Relativized Minimality (Rizzi, 1990, 2004) by Grillo (2005). (See Warren and Gibson, 2002, 2005 for a similar account for normal processing.) Another possibility that emerges from the current data is that the deficit is related to the CP node. This possibility accounts equally well for the results reported in the current study, but has the additional advantage of giving a unified account for the deficits in comprehension and production. Note, that the TDH suggests that traces are impaired but does not specify why they are impaired. One reason for why traces are not created in agrammatism might be related to a failure to construct the syntactic tree correctly. If individuals with agrammatism fail to assign the antecedent its position on the syntactic tree, they will not know that they should assume a trace, or where to locate it. For agrammatic production, it has been claimed that the construction of the syntactic tree, and specifically CP (and for some individuals also IP) is impaired (Tree Pruning Hypothesis, Friedmann (1998a, 2001, 2006a) and Friedmann and Grodzinsky (1997)). This can be extended to comprehension as well, to account for the deficit in traces of Wh-movement. If CP is inaccessible for comprehension, then when individuals with agrammatism hear a sentence that should include an operator in CP as an antecedent, they cannot construct the operator and do not place it in Spec-CP, because CP is not projected. They also cannot put the complementizer in its sentential position within CP. Later, when they get to the sentential position in which a trace should be assumed, they do not assume a trace there, because there is no antecedent that hints that a trace should be constructed. Furthermore, the head of the relative clause (like the monkey in sentences (4) (6)) should be situated above CP and be connected to the operator in CP which, in turn, is connected to the trace. If CP is not accessible in comprehension, individuals with agrammatism would also not know where to put the relative head above CP, neither will they be able to connect it to elements in the CP below it or to the trace. Swinney and his colleagues found that during normal on-line processing of traces, there is semantic reactivation of the antecedent at the trace (Love & Swinney, 1996; Nicol & Swinney, 1989; Swinney, Ford, Frauenfelder, & Bresnan, 1988; Swinney, Zurif, & Nicol, 1989). However, in agrammatism, traces are not constructed and therefore no reactivation of the antecedent is found at the trace (Zurif, Swinney, Prather, Solomon, & Bushell, 1993). A very similar explanation can now apply for resumptive pronouns in object relatives: these structures also include an operator at CP as an antecedent. If the CP is not constructed, the antecedent is not constructed, and when the individuals with agrammatism hit the resumptive pronoun, they do not have an antecedent for it, and they therefore do not know which NP they should reactivate at the pronoun. (footnote continued) referential NPs. This is the case for example in non-referential (who) object questions, which agrammatic aphasics understand better than referential (which) object questions. In this case, the intervening NP is usually a referential NP like the duck and the moved NP is non-referential, who. Similarly, at least for some agrammatic aphasics, arbitrary pro does not seem to block overt referential NPs in object relatives.

10 ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] Such an approach can also account for the impaired comprehension of Wh questions and topicalization structures. These structures include an antecedent in CP as well, and are impaired for the same reason the antecedent cannot be constructed in CP, which hinders the further construction of a trace and the reactivation of the antecedent at the trace. Similarly, recent results that indicate a deficit in the comprehension of Hebrew sentences that are derived by movement of the main verb to the second sentential position (Friedmann, Gvion, Biran, & Novogrodsky, 2006) can be accounted for by a deficit in CP, which prevents the movement of the verb to C. Some individuals with agrammatism who have a deficit in the comprehension of relative clauses also show a deficit in the comprehension of passives, whereas others are impaired only in relative clauses but not in passives (see a review in Friedmann, 2006b). Under an account that ascribes the deficit in comprehension to a deficit in syntactic tree building, individuals who have a deficit in both object relatives and passives should have a deficit in the IP node, in addition to the CP node (Friedmann, 2006b). Thus, it might be that the source of the impaired comprehension of various structures that involve the CP node relative clauses, cleft sentences, Wh questions, topicalization, and sentences that include verb movement to C, and possibly also the source of the impaired comprehension of passive sentences, is a deficit in the construction of the syntactic tree for an input sentence. When such a sentence is encountered, individuals with agrammatism are forced to rely on non-syntactic strategies in order to interpret the sentences, such as the agent-first strategy that assigns the first NP an agent role. To summarize, the current study compared the agrammatic comprehension of object relative sentences that are derived by movement to object relatives that include a resumptive pronoun and do not include movement. The results indicate that individuals with agrammatism are also impaired in the comprehension of object relatives without movement. These results suggest that it is not (only) movement that underlies the syntactic deficit in agrammatic comprehension, and the deficit might be wider, and related to the assignment of thematic roles over another argument of the verb, or to an impairment in building the syntactic structure of an input sentence. Acknowledgements The research was supported by the Joint German Israeli Research Program Grant GR01791 (Friedmann). References Aoun, J., Choueiri, L., & Hornstein, N. (2001). Resumption, movement, and derivational economy. Linguistic Inquiry, 32, 371 403. Borer, H. (1984). Restrictive relatives in modern Hebrew. Natural Language and Linguistic Theory, 2, 219 260. Chomsky, N. (1986). Barriers. Cambridge, MA: MIT Press. de Villiers, J., de Villiers, P., & Hoban, E. (1994). The central problem of functional categories in English syntax of oral deaf children. In H. Tager-Flusberg (Ed.), Constraints on language acquisition: Studies of atypical children (pp. 9 47). Hillsdale, NJ: Erlbaum. Friedmann, N. (1998a). Functional categories in agrammatic production: A cross-linguistic study. Unpublished doctoral dissertation, Tel Aviv University. Friedmann, N. (1998b). BAFLA Friedmann s battery for agrammatism. Tel Aviv: Tel Aviv University. Friedmann, N. (2001). Agrammatism and the psychological reality of the syntactic tree. Journal of Psycholinguistic Research, 30, 71 90.

ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] 11 Friedmann, N. (2006a). Speech production in Broca s agrammatic aphasia: Syntactic tree pruning. In Y. Grodzinsky, & K. Amunts (Eds.), Broca s region (pp. 63 82). New York: Oxford University Press. Friedmann, N. (2006b). Generalizations on variations in comprehension and production: A further source of variation and a possible account. Brain and Language, 96, 151 153. Friedmann, N., & Grodzinsky, Y. (1997). Tense and agreement in agrammatic production: Pruning the syntactic tree. Brain and Language, 56, 397 425. Friedmann, N., Gvion, A., Biran, M., & Novogrodsky, R. (2006). Do people with agrammatic aphasia understand verb movement? Aphasiology, 20, 136 153. Friedmann, N., Novogrodsky, R., Szterman, R., & Preminger, O. (in press). Resumptive pronouns as last resort when movement is impaired: Relative clauses in hearing impairment. In S. Armon-Lotem, S. Rothstein, & G. Danon (Eds.), Generative approaches to Hebrew Linguistics, series Linguistics Today. Amsterdam, The Netherlands: John Benjamins. Friedmann, N., & Shapiro, L. P. (2003). Agrammatic comprehension of simple active sentences with moved constituents: Hebrew OSV and OVS structures. Journal of Speech Language and Hearing Research, 46, 288 297. Friedmann, N., & Szterman, R. (2006). Syntactic movement in orally trained children with hearing impairment. Journal of Deaf Studies and Deaf Education, 11, 56 75. Grillo, N. (2005). Minimality effects in agrammatic comprehension. In S. Blaho, L. Vicente, & E. Schoorlemmer (Eds.), Proceedings of Console XIII. Grodzinsky, Y. (1984). Language deficits and linguistic theory. PhD Dissertation, Brandeis University. Grodzinsky, Y. (1989). Agrammatic comprehension of relative clauses. Brain and Language, 37, 480 499. Grodzinsky, Y. (1990). Theoretical perspectives on language deficits. Cambridge: MIT Press. Grodzinsky, Y. (1995). A restrictive theory of trace deletion in agrammatism. Brain and Language, 50, 27 51. Grodzinsky, Y. (2000). The neurology of syntax: Language use without Broca s area. Behavioral and Brain Sciences, 23, 1 71. Grodzinsky, Y. (2005). Ellipsis in deficient language users: Parallels (and divergences) in parallelism. A paper presented in GALA 2005, Workshop on language acquisition and breakdown, Siena, Italy. Grodzinsky, Y., Pin ango, M., Zurif, E., & Drai, D. (1999). The critical role of group studies in neuropsychology: Comprehension regularities in Broca s aphasia. Brain and Language, 67, 134 147. Grodzinsky, Y., Wexler, K., Chien, Y.-C., Marakovitz, S., & Solomon, J. (1993). The breakdown of binding relations. Brain and Language, 45, 396 422. Grolla, E. (2005). Resumptive pronouns as last resort: Implications for language acquisition. In S. Arunachalam, T. Scheffler, S. Sundaresan, & J. Tauberer (Eds.), Proceedings of the 28th annual Penn linguistics colloquium. Penn working papers in linguistics 11. Kertesz, A. (1982). Western aphasia battery. Orlando, FL: Grune and Stratton. Lasnik, H., & Stowell, T. (1991). Weakest crossover. Linguistic Inquiry, 22, 687 720. Love, T., & Swinney, D. (1996). Coreference processing and levels of analysis in object relative constructions: Demonstration of antecedent reactivation with a cross-modal priming paradigm. Journal of Psycholinguistic Research, 25, 5 24. McKee, C., & McDaniel, D. (2001). Resumptive pronouns in English relative clauses. Language Acquisition, 9, 113 156. Nicol, J., & Swinney, D. (1989). The role of structure in coreference assignment during sentence comprehension. Journal of Psycholinguistic Research, 18, 5 19. Power, D. J., & Quigley, S. P. (1973). Deaf children acquisition of the passive voice. Journal of Speech and Hearing Research, 16, 5 11. Rizzi, L. (1990). Relativized minimality. Cambridge, MA: MIT Press. Rizzi, L. (2004). Locality and left periphery. In A. Belletti (Ed.), Structures and beyond: The cartographic syntactic structure, Vol. 2 (pp. 223 251). Oxford: Oxford University Press. Ruigendijk, E., Vasić, N., & Avrutin, S. (2006). Reference assignment: Using language breakdown to choose between theoretical approaches. Brain and Language, 96, 302 317. Shlonsky, U. (1992). Resumptive pronouns as a last resort. Linguistic Inquiry, 23, 443 468. Soroker, N. (1997). Hebrew Western Aphasia Battery. Ra anana, Israel: Loewenstein Hospital Rehabilitation Center (in Hebrew). Swinney, D., Ford, M., Frauenfelder, U. H., & Bresnan, J. (1988). On the temporal course of gap-filling and antecedent assignment during sentence comprehension. Unpublished MS.

12 ARTICLE IN PRESS N. Friedmann / Journal of Neurolinguistics ] (]]]]) ]]] ]]] Swinney, D., Zurif, E., & Nicol, J. (1989). The effects of focal brain damage on sentence processing: An examination of the neurological organization of a mental module. Journal of Cognitive Neuroscience, 1, 25 37. Warren, T., & Gibson, E. (2002). The influence of referential processing on sentence complexity. Cognition, 85, 79 112. Warren, T., & Gibson, E. (2005). Effects of NP type in reading cleft sentences in English. Language and Cognitive Processes, 20, 751 767. Zurif, E., & Caramazza, A. (1976). Psycholinguistic structures in aphasia: Studies in syntax and semantics. In H. Whitaker, & H. A. Whitaker (Eds.), Studies in neurolinguistics, Vol. I. New York: Academic Press. Zurif, E. B., Swinney, D., Prather, P., Solomon, J., & Bushell, C. (1993). An on-line analysis of syntactic processing in Broca s and Wernicke s aphasia. Brain and Language, 45, 448 464.