- Email: [email protected]
The isomorphic mapping hypothesis: Evidence from Korean
226
TENNET XI
-25
Motor Potential (Group x condition interaction
-20
\
-15
<
-10 - ~ Autlmated ~ - Con~olled
-5
Controls
Complex
Simple
FIG. 1. Motor potential amplitude for control, simple, and complex tic disorders groups averaged over automated and controlled response conditions.
automatic and nonvoluntary. We suggest that future research focus on aspects of motor control and how this control is regulated under changing task demand.
REFERENCES Georgiou, N., Bradshaw, J. L., Phillips, J. G., Cunnington, R., & Rogers, M. (1997). Functional asymmetries in the movements kinematics of patients with Tourette's syndrome. Journal of Neurology, Neurosurgery and Psychiatry, 63(2), 188-195. O'Connor, K. P., Serataway, M., & Stip, E. (1999). Simple and complex motor processing in chronic tic disorders. Brain and Cognition, 37(1). Silverman, J. S., & Loychick, S. G. (1990). Brain-mapping abnormalities in a family with three obsessive compulsive children. Journal of Neuropsychiatry and Clinical Neurosciences, 2, 319-322. Zieman, U., Paulus, W., & Rothenberger, A. (1997). Decreased motor inhibition in Tourette's disorder: Evidence from transcranial magnetic stimulation. American Journal of Psychiatry, 154(9), 12771284.
The Isomorphic Mapping Hypothesis: Evidence from Korean William O'Grady and Miseon Lee Department of Linguistics, University of Hawai 'i This paper evaluates the relative merits of the trace deletion hypothesis, which attributes agrammatic comprehension difficulties to the loss of traces, and the isomorphic mapping hypothesis, which proposes that agrammatics have difficulty understanding sentences in which there is a nonisomorphic mapping between the syntactic representation and the corresponding event in the real world. (The two are isomorphic if the order of NPs reflects the place of entities in the event's 'action chain.' Since agents act on themes and transmit them to goals, the agent-theme-goal order is isomorphic with the corresponding event but the agent-goaltheme order is not.) The two hypotheses contrast in the predictions they make concerning goal-theme and theme-goal patterns in Korean: the TDH predicts degraded performance on
TENNET XI
227
the theme-goalstructure(whichis derivedfrom the goal-themestructure);the IMH predicts difficultieson the goal-themestructure,sinceit is nonisomorphic.Resultsof a comprehension study involvingfour Korean Broca's aphasics provide strong support for the IMH over the TDH. ©2001 Academic Press
Introduction Theories of agrammatism can be roughly divided into two groups--those that posit a deficit in syntactic representations (such as the loss of traces, as proposed by Grodzinsky, 1995, in press) and those that posit a defect in the interpretive processes that operate on those representations (e.g., the 'mapping hypothesis,' in the sense of Linebarger, 1995). This paper uses data from Korean Broca's aphasics to support a version of the mapping hypothesis that focuses on the relationship between syntactic representations and the events they denote. We will begin by outlining our hypothesis and contrasting it with Grodzinsky's Trace Deletion Hypothesis.
The Isomorphic Mapping Hypothesis Grodzinsky (e.g., in press) attributes a significant portion of the comprehension deficit associated with Broca's aphasia to the loss of traces, proposing the Trace Deletion Hypothesis summarized in (1).
(1) The Trace Deletion Hypothesis (TDH) • Traces are deleted from the syntactic representations of Broca's aphasics. • Phrasal constituents with no thematic role are assigned one by a default linear strategy (NPI = Agent) In the absence of traces, two agents are identified in a passive sentence such as (2)--the sentence-initial NP because of the default linear strategy and the sentencefinal NP by the usual grammatical mechanisms.
(2) The Lion Was Killed by the Cheetah The resulting 'thematic competition,' Grodzinsky suggests, accounts for the chance-level performance often observed on these patterns. The essence of our proposal is that agrammatics encounter difficulty with sentences in which the order of NPs is not isomorphically aligned with their referents in the corresponding event. We formulate our hypothesis as follows.
(3) The Isomorphic Mapping Hypothesis (IMH) Agrammatics have difficulty establishing a nonisomorphic mapping between syntactic representations and the corresponding event in the real world. In a clause built around a prototypical action-denoting verb, isomorphic mapping places the agent before the theme, consistent with the fact that the corresponding event involves an action that originates with the former entity and is directed toward the latter one (e.g., Talmy, 1988; Van Voorst, 1988, p. 27; Dik, 1989; Pinker, 1989, p. 193; Croft, 1991, p. 262; Langacker, 1995, pp. 18-20; Ritter & Rosen, 1998). The agent-theme order thus presents an event's participants in a conceptually advantageous manner, with the initiator of the action being mentioned before the entity to which the action is directed. In many cases, the TDH and the IMH make identical predictions with respect to
228
TENNET X!
comprehension deficits. Space does not permit a full review of these cases, but to take one obvious example, both hypotheses predict that Broca's aphasics will encounter difficulty with syntactic passives such as (2)--the TDH for the reasons outlined above and the IMH because the order of the NPs (theme-agent) is not isomorphic with the corresponding situation, l To choose between the IMH and the TDH, it is necessary to seek out nonisomorphic structures that do not contain traces and isomorphic structures that do. The IMH predicts poor performance on the former structure and good performance on the latter; the TDH makes the opposite predictions. We focus in this paper on a structure of this sort in Korean, a head-final language with relatively free ordering of its argument NPs. Patterns such as the following are particularly relevant: (4) a. agent ~ theme Ai-ka kay-eykey tol-ul tenci-ess-ta. child-Nom dog-Dat stone-Ac throw-Pst-Decl 'The child threw a stone to the dog.' b.
agent
theme
Ai-ka tol-ul kay-eyekey tenci-ess-ta. child-Nora stone-Ac dog-Dat throw-Pst-Decl 'The child threw a stone to the dog.' The agent-goal-theme pattern in (4a) represents the neutral or unmarked order (Sohn, 1996, p. 86), and it is commonly assumed that the theme-goal order in (4b) is derived from it via movement. 2 This is significant, for it is the derived theme-goal order that is isomorphic with the corresponding situation: the agent (the child) acts upon the theme (the stone) causing it to move to the goal (the dog). The IMH therefore predicts that agrammatics will encounter difficulty with the goal-theme construction, but not its theme-goal counterpart. The TDH makes the opposite prediction: only the trace-containing theme-goal pattern should be problematic for agrammatics.
An Experiment To test the validity of these predictions, we conducted a comprehension experiment with four Broca's aphasics (three males and one female). All patients were righthanded monolingual speakers of Korean with 12 to 16 years of education; they ranged in age from 40 to 50 years (mean age, 43.75). The patients had been diagnosed as Broca's aphasics based on the Korean version of the Boston Diagnostic Aphasia Examination (as adapted by S. J. Kim) and clinical evaluation by two speech thera-
The IMH predicts the same degree of difficulty on agentless passives such as The lion was killed, which is also correct (e.g., Martin et al. 1989). z Evidence for this analysis includes the contrasts observed in the following sentences: (5)
Sample test item (theme-goal order): Kom-ul so-eykey milecwul-lay-yo? bear-Ac cow-Dat push -Fut-SentEnder 'Will you push the bear to the cow?'
The fact that coreference in the (b) pattern is less natural than in its (a) counterpart suggests a weak 'reconstruction effect': the accusative-marked NP originates in a position lower than the dative and returns there in Logical Form. Takano (1998) offers extensive arguments along these lines for the typologically identical Japanese.
TENNET Xl
229
TABLE 1 Results on the C o m p r e h e n s i o n Task" Subject
Theme-goal pattern
Goal-theme pattern
A B C D
Above chance Above chance Above chance Chance
Chance Chance Chance Chance
"Chance, at the .01 level, as determined by a one-sample t test with a population mean of .5.
pists; CT scans performed at the onset of symptoms (6 to 69 months prior to our study) demonstrated a unilateral left-sided lesion. The experiment was built around 40 semantically reversible sentence types such as (5), half employing the theme-goal order, and the other half the goal-theme order. (6) a. A dative antecedent preceding an accusative anaphor Nay-ka kewul-lo John-eykeyi caki-luli pichwuepoyecwu-ess-ta. I-Nora mirror in John-Dat self-Acc show -Pst-Decl 'I showed to John himself in the mirror.' b. An accusative antecedent preceding a dative anaphor ?*Nay-ka kewul-lo John-uli caki-eyekeyi pichwuepoyecwu-ess-ta. I-Nora mirror in John-Acc self-Dat show -Pst-Decl 'I showed John to himself in the mirror.' The subjects, who were tested individually in a quiet place, were asked to respond to the test sentences with the help of stuffed toys and other props provided by the experimenter) To establish a baseline for evaluating our results, we administered the experiment to nine right-handed monolingual native speakers of Korean living in Seoul, Korea (four males and five females, all with 16 to 18 years of education and ranging in age from 31 to 59 years; mean age 39.9 years). As expected, they encountered no difficulty with either pattern: the mean proportion of correct responses for both structures exceeded 98%. Table 1 reports on our findings for the agrammatics. As can be seen here, three of our four subjects performed above chance on the isomorphic theme-goal pattern and at chance on the nonisomorphic goal-theme structures. The fourth performed at a chance level on both patterns. Discussion and Conclusion Our results support the IMH--where there was a contrast, the agrammatics were better at interpreting the pattern in which the order of the NPs was isomorphic with the structure of the corresponding event than they were at comprehending the syntactically more basic pattern from which it is apparently derived. We interpret this as evidence that at least part of the deficit associated with Broca's aphasia can be traced to difficulties in dealing with nonisomorphic relationships between a sentence's word order and the structure of the corresponding event rather than to the loss of traces per se. The IMH has conceptual advantages as well, since it provides a principled explana3 None of the subjects were suffering from a physical disability.
230
TENNET XI
tion for the word order preferences associated with Broca's aphasia, tracing them to the structure of an event's 'action chain' (to use Langacker's term), which runs from agent to theme to goal. In this it differs both from the TDH and from many mapping theories (e.g., Caplan et al., 1985), which associate thematic roles with linear positions on the basis of simple frequency of occurrence. Once again, the Korean facts are crucial: the goal-theme order is apparently more frequent (Sohn, 1994, p. 232; Cho et al., 1998), but is problematic for agrammatics due to its lack of isomorphism. Finally, it is worth noting that the effects of the isomorphism preference are independently attested in another phenomenon that involves a type of linguistic incapaci t y - n a m e l y language acquisition. There too better performance is observed on structures that are isomorphic with the corresponding event than on those that are not. Thus, children learning English do better on active sentences than on passives, better on subject relative clauses and subject wh questions (both agent-initial) than on direct object relatives and direct object wh questions (both nonagent initial), and so forth (see O'Grady, 1997, Chapters 7, 9, and 10, for a review). The parallel extends even to word order in Korean: Cho et al. (1998) report a strong preference for the isomorphic theme-goal order among Korean-speaking children even though it is less common than the goal-theme order in maternal speech. We anticipate that further exploration of these and other parallels will shed additional light on the extent to which the Isomorphic Mapping Hypothesis can contribute to our understanding Broca's aphasia. REFERENCES Caplan, D., Baker, C., & Dehaut, F. (1985). Syntactic determinants of sentence comprehension in aphasia. Cognition, 21, 117-75. Cho, S., Lee, M., O'Grady, W., Song, M., & Suzuki, T. (1998). Word order preferences for direct and indirect objects in children learning Korean. In B. Park & J. Yoon (Eds.), Selected Papers from the Eleventh International Conference on Korean Linguistics (pp. 735-743). Seoul: International Circle of Korean Linguistics. Croft, W. (1991). Syntactic categories and grammatical relations. Chicago: University of Chicago Press. Dik, S. (1989). The theory of functional grammar. L The structure of the clause--Functional grammar series 9. Dordrecht: Foris. Grodzinsky, Y. (1995). Trace deletion, I-roles, and cognitive strategies. Brain and Language, 51, 469497. Grodzinsky, Y. (in press). The neurology of syntax: Language use without Broca's area. Behavioral and Brain Sciences. Langacker, R. (1995). Raising and transparency. Language, 71, 1-62. Linebarger, M. (1995). Agrammatism as evidence about grammar. Brain and Language, 50, 52-91. Martin, R. C., Wetzel, F., Blossom-Stack, C., & Feher, E. (1989). Syntactic loss versus processing deficit. Cognition, 32, 157-191. O'Grady, W. (1997). Syntactic development. Chicago: University of Chicago Press. Pinker, S. (1989). Learnability and cognition: The acquisition of argument structure. Cambridge, MA: MIT Press. Ritter, E., & Rosen, S. (1998). Delimiting events in syntax. In W. Geuder & M. Butt (Eds.), The projection of arguments: Lexical and syntactic constraints. Stanford, CA: CSLI. Sohn, H. (1994). Korean. New York: Routledge. Takano, Y. (1998). Object shift and scrambling. Natural Language and Linguistic Theory, 16, 817-889. Talmy, L. (1988). Force dynamics in language and cognition. Cognitive Science 12, 49-100. Van Voorst, J. (1988). Event structure. Philadelphia: John Benjamins.
TENNET XI
-25
Motor Potential (Group x condition interaction
-20
\
-15
<
-10 - ~ Autlmated ~ - Con~olled
-5
Controls
Complex
Simple
FIG. 1. Motor potential amplitude for control, simple, and complex tic disorders groups averaged over automated and controlled response conditions.
automatic and nonvoluntary. We suggest that future research focus on aspects of motor control and how this control is regulated under changing task demand.
REFERENCES Georgiou, N., Bradshaw, J. L., Phillips, J. G., Cunnington, R., & Rogers, M. (1997). Functional asymmetries in the movements kinematics of patients with Tourette's syndrome. Journal of Neurology, Neurosurgery and Psychiatry, 63(2), 188-195. O'Connor, K. P., Serataway, M., & Stip, E. (1999). Simple and complex motor processing in chronic tic disorders. Brain and Cognition, 37(1). Silverman, J. S., & Loychick, S. G. (1990). Brain-mapping abnormalities in a family with three obsessive compulsive children. Journal of Neuropsychiatry and Clinical Neurosciences, 2, 319-322. Zieman, U., Paulus, W., & Rothenberger, A. (1997). Decreased motor inhibition in Tourette's disorder: Evidence from transcranial magnetic stimulation. American Journal of Psychiatry, 154(9), 12771284.
The Isomorphic Mapping Hypothesis: Evidence from Korean William O'Grady and Miseon Lee Department of Linguistics, University of Hawai 'i This paper evaluates the relative merits of the trace deletion hypothesis, which attributes agrammatic comprehension difficulties to the loss of traces, and the isomorphic mapping hypothesis, which proposes that agrammatics have difficulty understanding sentences in which there is a nonisomorphic mapping between the syntactic representation and the corresponding event in the real world. (The two are isomorphic if the order of NPs reflects the place of entities in the event's 'action chain.' Since agents act on themes and transmit them to goals, the agent-theme-goal order is isomorphic with the corresponding event but the agent-goaltheme order is not.) The two hypotheses contrast in the predictions they make concerning goal-theme and theme-goal patterns in Korean: the TDH predicts degraded performance on
TENNET XI
227
the theme-goalstructure(whichis derivedfrom the goal-themestructure);the IMH predicts difficultieson the goal-themestructure,sinceit is nonisomorphic.Resultsof a comprehension study involvingfour Korean Broca's aphasics provide strong support for the IMH over the TDH. ©2001 Academic Press
Introduction Theories of agrammatism can be roughly divided into two groups--those that posit a deficit in syntactic representations (such as the loss of traces, as proposed by Grodzinsky, 1995, in press) and those that posit a defect in the interpretive processes that operate on those representations (e.g., the 'mapping hypothesis,' in the sense of Linebarger, 1995). This paper uses data from Korean Broca's aphasics to support a version of the mapping hypothesis that focuses on the relationship between syntactic representations and the events they denote. We will begin by outlining our hypothesis and contrasting it with Grodzinsky's Trace Deletion Hypothesis.
The Isomorphic Mapping Hypothesis Grodzinsky (e.g., in press) attributes a significant portion of the comprehension deficit associated with Broca's aphasia to the loss of traces, proposing the Trace Deletion Hypothesis summarized in (1).
(1) The Trace Deletion Hypothesis (TDH) • Traces are deleted from the syntactic representations of Broca's aphasics. • Phrasal constituents with no thematic role are assigned one by a default linear strategy (NPI = Agent) In the absence of traces, two agents are identified in a passive sentence such as (2)--the sentence-initial NP because of the default linear strategy and the sentencefinal NP by the usual grammatical mechanisms.
(2) The Lion Was Killed by the Cheetah The resulting 'thematic competition,' Grodzinsky suggests, accounts for the chance-level performance often observed on these patterns. The essence of our proposal is that agrammatics encounter difficulty with sentences in which the order of NPs is not isomorphically aligned with their referents in the corresponding event. We formulate our hypothesis as follows.
(3) The Isomorphic Mapping Hypothesis (IMH) Agrammatics have difficulty establishing a nonisomorphic mapping between syntactic representations and the corresponding event in the real world. In a clause built around a prototypical action-denoting verb, isomorphic mapping places the agent before the theme, consistent with the fact that the corresponding event involves an action that originates with the former entity and is directed toward the latter one (e.g., Talmy, 1988; Van Voorst, 1988, p. 27; Dik, 1989; Pinker, 1989, p. 193; Croft, 1991, p. 262; Langacker, 1995, pp. 18-20; Ritter & Rosen, 1998). The agent-theme order thus presents an event's participants in a conceptually advantageous manner, with the initiator of the action being mentioned before the entity to which the action is directed. In many cases, the TDH and the IMH make identical predictions with respect to
228
TENNET X!
comprehension deficits. Space does not permit a full review of these cases, but to take one obvious example, both hypotheses predict that Broca's aphasics will encounter difficulty with syntactic passives such as (2)--the TDH for the reasons outlined above and the IMH because the order of the NPs (theme-agent) is not isomorphic with the corresponding situation, l To choose between the IMH and the TDH, it is necessary to seek out nonisomorphic structures that do not contain traces and isomorphic structures that do. The IMH predicts poor performance on the former structure and good performance on the latter; the TDH makes the opposite predictions. We focus in this paper on a structure of this sort in Korean, a head-final language with relatively free ordering of its argument NPs. Patterns such as the following are particularly relevant: (4) a. agent ~ theme Ai-ka kay-eykey tol-ul tenci-ess-ta. child-Nom dog-Dat stone-Ac throw-Pst-Decl 'The child threw a stone to the dog.' b.
agent
theme
Ai-ka tol-ul kay-eyekey tenci-ess-ta. child-Nora stone-Ac dog-Dat throw-Pst-Decl 'The child threw a stone to the dog.' The agent-goal-theme pattern in (4a) represents the neutral or unmarked order (Sohn, 1996, p. 86), and it is commonly assumed that the theme-goal order in (4b) is derived from it via movement. 2 This is significant, for it is the derived theme-goal order that is isomorphic with the corresponding situation: the agent (the child) acts upon the theme (the stone) causing it to move to the goal (the dog). The IMH therefore predicts that agrammatics will encounter difficulty with the goal-theme construction, but not its theme-goal counterpart. The TDH makes the opposite prediction: only the trace-containing theme-goal pattern should be problematic for agrammatics.
An Experiment To test the validity of these predictions, we conducted a comprehension experiment with four Broca's aphasics (three males and one female). All patients were righthanded monolingual speakers of Korean with 12 to 16 years of education; they ranged in age from 40 to 50 years (mean age, 43.75). The patients had been diagnosed as Broca's aphasics based on the Korean version of the Boston Diagnostic Aphasia Examination (as adapted by S. J. Kim) and clinical evaluation by two speech thera-
The IMH predicts the same degree of difficulty on agentless passives such as The lion was killed, which is also correct (e.g., Martin et al. 1989). z Evidence for this analysis includes the contrasts observed in the following sentences: (5)
Sample test item (theme-goal order): Kom-ul so-eykey milecwul-lay-yo? bear-Ac cow-Dat push -Fut-SentEnder 'Will you push the bear to the cow?'
The fact that coreference in the (b) pattern is less natural than in its (a) counterpart suggests a weak 'reconstruction effect': the accusative-marked NP originates in a position lower than the dative and returns there in Logical Form. Takano (1998) offers extensive arguments along these lines for the typologically identical Japanese.
TENNET Xl
229
TABLE 1 Results on the C o m p r e h e n s i o n Task" Subject
Theme-goal pattern
Goal-theme pattern
A B C D
Above chance Above chance Above chance Chance
Chance Chance Chance Chance
"Chance, at the .01 level, as determined by a one-sample t test with a population mean of .5.
pists; CT scans performed at the onset of symptoms (6 to 69 months prior to our study) demonstrated a unilateral left-sided lesion. The experiment was built around 40 semantically reversible sentence types such as (5), half employing the theme-goal order, and the other half the goal-theme order. (6) a. A dative antecedent preceding an accusative anaphor Nay-ka kewul-lo John-eykeyi caki-luli pichwuepoyecwu-ess-ta. I-Nora mirror in John-Dat self-Acc show -Pst-Decl 'I showed to John himself in the mirror.' b. An accusative antecedent preceding a dative anaphor ?*Nay-ka kewul-lo John-uli caki-eyekeyi pichwuepoyecwu-ess-ta. I-Nora mirror in John-Acc self-Dat show -Pst-Decl 'I showed John to himself in the mirror.' The subjects, who were tested individually in a quiet place, were asked to respond to the test sentences with the help of stuffed toys and other props provided by the experimenter) To establish a baseline for evaluating our results, we administered the experiment to nine right-handed monolingual native speakers of Korean living in Seoul, Korea (four males and five females, all with 16 to 18 years of education and ranging in age from 31 to 59 years; mean age 39.9 years). As expected, they encountered no difficulty with either pattern: the mean proportion of correct responses for both structures exceeded 98%. Table 1 reports on our findings for the agrammatics. As can be seen here, three of our four subjects performed above chance on the isomorphic theme-goal pattern and at chance on the nonisomorphic goal-theme structures. The fourth performed at a chance level on both patterns. Discussion and Conclusion Our results support the IMH--where there was a contrast, the agrammatics were better at interpreting the pattern in which the order of the NPs was isomorphic with the structure of the corresponding event than they were at comprehending the syntactically more basic pattern from which it is apparently derived. We interpret this as evidence that at least part of the deficit associated with Broca's aphasia can be traced to difficulties in dealing with nonisomorphic relationships between a sentence's word order and the structure of the corresponding event rather than to the loss of traces per se. The IMH has conceptual advantages as well, since it provides a principled explana3 None of the subjects were suffering from a physical disability.
230
TENNET XI
tion for the word order preferences associated with Broca's aphasia, tracing them to the structure of an event's 'action chain' (to use Langacker's term), which runs from agent to theme to goal. In this it differs both from the TDH and from many mapping theories (e.g., Caplan et al., 1985), which associate thematic roles with linear positions on the basis of simple frequency of occurrence. Once again, the Korean facts are crucial: the goal-theme order is apparently more frequent (Sohn, 1994, p. 232; Cho et al., 1998), but is problematic for agrammatics due to its lack of isomorphism. Finally, it is worth noting that the effects of the isomorphism preference are independently attested in another phenomenon that involves a type of linguistic incapaci t y - n a m e l y language acquisition. There too better performance is observed on structures that are isomorphic with the corresponding event than on those that are not. Thus, children learning English do better on active sentences than on passives, better on subject relative clauses and subject wh questions (both agent-initial) than on direct object relatives and direct object wh questions (both nonagent initial), and so forth (see O'Grady, 1997, Chapters 7, 9, and 10, for a review). The parallel extends even to word order in Korean: Cho et al. (1998) report a strong preference for the isomorphic theme-goal order among Korean-speaking children even though it is less common than the goal-theme order in maternal speech. We anticipate that further exploration of these and other parallels will shed additional light on the extent to which the Isomorphic Mapping Hypothesis can contribute to our understanding Broca's aphasia. REFERENCES Caplan, D., Baker, C., & Dehaut, F. (1985). Syntactic determinants of sentence comprehension in aphasia. Cognition, 21, 117-75. Cho, S., Lee, M., O'Grady, W., Song, M., & Suzuki, T. (1998). Word order preferences for direct and indirect objects in children learning Korean. In B. Park & J. Yoon (Eds.), Selected Papers from the Eleventh International Conference on Korean Linguistics (pp. 735-743). Seoul: International Circle of Korean Linguistics. Croft, W. (1991). Syntactic categories and grammatical relations. Chicago: University of Chicago Press. Dik, S. (1989). The theory of functional grammar. L The structure of the clause--Functional grammar series 9. Dordrecht: Foris. Grodzinsky, Y. (1995). Trace deletion, I-roles, and cognitive strategies. Brain and Language, 51, 469497. Grodzinsky, Y. (in press). The neurology of syntax: Language use without Broca's area. Behavioral and Brain Sciences. Langacker, R. (1995). Raising and transparency. Language, 71, 1-62. Linebarger, M. (1995). Agrammatism as evidence about grammar. Brain and Language, 50, 52-91. Martin, R. C., Wetzel, F., Blossom-Stack, C., & Feher, E. (1989). Syntactic loss versus processing deficit. Cognition, 32, 157-191. O'Grady, W. (1997). Syntactic development. Chicago: University of Chicago Press. Pinker, S. (1989). Learnability and cognition: The acquisition of argument structure. Cambridge, MA: MIT Press. Ritter, E., & Rosen, S. (1998). Delimiting events in syntax. In W. Geuder & M. Butt (Eds.), The projection of arguments: Lexical and syntactic constraints. Stanford, CA: CSLI. Sohn, H. (1994). Korean. New York: Routledge. Takano, Y. (1998). Object shift and scrambling. Natural Language and Linguistic Theory, 16, 817-889. Talmy, L. (1988). Force dynamics in language and cognition. Cognitive Science 12, 49-100. Van Voorst, J. (1988). Event structure. Philadelphia: John Benjamins.