Douglas Merchant - Academia.edu (original) (raw)

Uploads

Papers by Douglas Merchant

Research paper thumbnail of Processing and empty categories in Brazilian Portuguese

We test the predictions of the Movement Theory of Control (MTC, e.g. Hornstein, 1999; Boeckx & Ho... more We test the predictions of the Movement Theory of Control (MTC, e.g. Hornstein, 1999; Boeckx & Hornstein, 2006) experimentally in order to better understand the online processing of control, raising, and other similar structures as well as the empty categories that form part of their associated accounts. According to the MTC, the coindexed empty categories (EC) in apparent control phenomena (1, 2 from Modesto, 2007) should be analyzed as the product of movement and not as PRO (as in e.g. Chomsky, 1981; Landau, 2000). Notably, Brazilian Portuguese (BP) allows for agreement in object control (2). The MTC predicts that control should behave in a manner consistent with movement traces/copies in generative grammar. Modesto (2010, 2018) observes an asymmetry between exhaustive control (3) and partial control (4) in BP according to matrix predicate type (Landau 2000). The former requires strict identity with the controlling DP (O presidente), while the latter must include the controller. This is unexpected according to the MTC, given that a trace/copy should not be able to alter its identity (i.e. from singular to plural) during the derivation. Experimental evidence for ECs is based on the assumption that they are covert elements analogous to overt pronouns. Cross-modal lexical priming (CMLP) studies, in which sentence primes and semantically-related probe words are presented in different modalities (audio primes and visual probes), have demonstrated that overt anaphors reactivate their antecedents (e.g., Nicol, 1988; Nicol & Swinney, 1989); and that at least some ECs do as well (e.g., Swinney, Ford, & Bresnan, 1989). Converging evidence for this comes from repetition priming studies, in which part of the antecedent phrase serves as a probe (e.g., Bever & McElree, 1988); such studies have found reactivation in raising contexts as well. Under the MTC, control is analyzed as movement analogous to raising; we would therefore expect a similar pattern of activation in raising and control sentences. In the current study, we employ a cross-modal repetition priming (CMRP) paradigm, similar to the CMLP paradigm, but with the probe being taken directly from the antecedent phrase, in order to examine the timecourse of processing of phonologically overt pronouns/anaphora and theoretically-predicted ECs in BP, with a particular interest in whether ECs reactivate their antecedents at the predicted gap positions. Native speakers (n=52) of BP participated in a three-part procedure (CMRP, bimodal acceptability judgment, linguistic history/handedness questionnaire). In the CMRP, we measured reactivation at three testpoints: 1) following the matrix predicate, 2) at the predicted gap position, and 3) at the end of the sentence (short [S], medium [M], or long [L] latency, respectively) as in (5). We report reaction time results for raising, overt anaphor, partial control, and exhaustive control sentences at the three different latency intervals. Table 1 shows evidence of similar reactivation in sentence-medial position (i.e. at the [M] gap) in raising sentences and in those with overt pronouns, thus supporting the standard account of movement in raising constructions, where a trace/copy of the raised element reactivates it in the gap position. No such reactivation was found for exhaustive control sentences in the [M] gap; response times were significantly slower in these sentences than in raising sentences and in those with overt pronouns (p=0.045 in both instances). We found an unexpected asymmetry in processing of exhaustive and partial control, and limited evidence for sentence-final reactivation in exhaustive control sentences. We suspect this is due to the parser positing the EC PRO not in the predicted gap position prior to the embedded infinitive, but in a theta-position following the infinitive (as in Larsen & Johansson, 2020). Notwithstanding, accuracy greatly decreased at the [L] position, suggesting that the EC (i.e. PRO) does not fully reactivate its antecedent phrase. This differentiation suggests that 1) the parser makes active use of empty categories in comprehension, contra traceless accounts (e.g. Pickering and Barry, 1991), and 2) different linguistic mechanisms are involved in the processing of raising and control structures, a finding presumably inconsistent with the MTC. Examples (1) Os meninosj querem (ECj) nadar(*em). the boys want.PRS.3pl swim.INF(3pl) 'The boys want to swim' (2) Euj convenci os meninosk a (EC*j/k) tomar(em) banho. I convince.PST.1sg the boys to take.INF(3pl) bath. 'I convinced the boys to take a bath.' (3) *O presidentej conseguiu [ECj+k] se reunirem às 6 the president achieve.PST.3sg SE meet.INF.3pl at.the six intended: 'The president managed for them to meet at 6.' (4) O presidentej preferia [ECj+k] se reunirem às 6. the president prefer.IMPFV.3sg SE meet.INF.3pl at.the six 'The president preferred for them to meet at 6.' (5) Matrix-clause raising predicate (Target probe: recolhidos; non-target probe: falsos) Os dados recolhidos resultaram [S] surpreendentemente [M] ser muito problemáticos para as hipóteses existentes [L] 'The data gathered turned out [S] surprisingly [M] to be very problematic for current hypotheses [L]'

Research paper thumbnail of What’s in the bucket? Aspectual (non)compositionality in phrasal idioms

Natural Language & Linguistic Theory, 2021

I consider here the question of whether the aspectual properties of idioms are compositionally de... more I consider here the question of whether the aspectual properties of idioms are compositionally determined with respect to the normal literal meanings of their components. This issue is unresolved in the literature, with some (Marantz 1997; McGinnis 2002, 2005) arguing for systematic literal compositionality, some noting mismatches in the aspectual properties of some idioms’ literal and figurative readings and therefore arguing for compositionality with respect to idioms’ meanings only (Glasbey 2003, 2007), and others attempting to reconcile the two views, maintaining systematic literal compositionality while accounting for mismatches by adding aspectual information post-syntactically (Leivada 2017) or by nonlinguistic conceptual processes (Mateu and Espinal 2007). Working within a monotonic model of aspectual composition in which telicity, durativity, and dynamicity are privative features inherent to verbs (Olsen 1997), I present further evidence for Glasbey’s position, arguing that that the aspectual properties of idioms are predictable on the basis of their figurative meanings, and that the literal readings of their components are irrelevant to aspectual composition, at least in the domain of language production. I further argue that some phrasal idioms are more highly specified than their commonly paraphrased meanings, and that idiomaticity represents a motivated exception to Olsen’s model in that idioms may be negatively specified for telicity, durativity, and dynamicity. I conclude that this exception falls out directly from the assumption that idioms may lexicalize a larger structure than verbs, speculating that this may include an inner aspect head in the syntax.

Research paper thumbnail of Idioms at the Interface(S): Towards a Psycholinguistically Grounded Model of Sentence Generation

The primary goal of this dissertation is to demonstrate that a close analysis of the syntactic an... more The primary goal of this dissertation is to demonstrate that a close analysis of the syntactic and semantic properties of phrasal idioms leads directly to the conclusion that the phonological forms of idioms are not present in narrow syntax. The secondary goal of this dissertation is to demonstrate that from the psycholinguistic perspective, syntactic structure-building in the context of language production cannot plausibly be modelled as a bottom-up process powered by iterative Merge, as is standardly assumed. Instead, I propose a provisional model grounded in both theory-external and theory-internal evidence, wherein the computational component operates over semantically individuated but phonologically null Roots-as-nodes (N-Roots), and structures are built top-down by iterative Branch. In this model, which I refer to as LIT-D (Late-Insertion, Top-Down), N-Roots are inserted in functional and discourse positions, and head chains terminating in theta positions. Cyclically transferred to the interfaces, these N-Roots are then matched with Roots-as-exponents (E-Roots), via syntactic and semantic feature mapping. Idioms are exceptional in this model only in that they are complex E-Roots, with multiple components that can be mapped into multiple N-Roots.

Research paper thumbnail of What’s in the bucket? Aspectual (non)compositionality in phrasal idioms.

Natural Language and Linguistic Theory, 2021

I consider here the question of whether the aspectual properties of idioms are compositionally de... more I consider here the question of whether the aspectual properties of idioms are compositionally determined with respect to the normal literal meanings of their components. This issue is unresolved in the literature, with some (Marantz 1997; McGinnis 2002, 2005) arguing for systematic literal compositionality, some noting mismatches in the aspectual properties of some idioms’ literal and figurative readings and therefore arguing for compositionality with respect to idioms’ meanings only (Glasbey 2003, 2007), and others attempting to reconcile the two views, maintaining systematic literal compositionality while accounting for mismatches by adding aspectual information post-syntactically (Leivada 2017) or by nonlinguistic conceptual processes (Mateu and Espinal 2007). Working within a monotonic model of aspectual composition in which telicity, durativity, and dynamicity are privative features inherent to verbs (Olsen 1997), I present further evidence for Glasbey’s position, arguing that that the aspectual properties of idioms are predictable on the basis of their figurative meanings, and that the literal readings of their components are irrelevant to aspectual composition, at least in the domain of language production. I further argue that some phrasal idioms are more highly specified than their commonly paraphrased meanings, and that idiomaticity represents a motivated exception to Olsen’s model in that idioms may be negatively specified for telicity, durativity, and dynamicity. I conclude that this exception falls out directly from the assumption that idioms may lexicalize a larger structure than verbs, speculating that this may include an inner aspect head in the syntax.

Research paper thumbnail of Processing and empty categories in Brazilian Portuguese

We test the predictions of the Movement Theory of Control (MTC, e.g. Hornstein, 1999; Boeckx & Ho... more We test the predictions of the Movement Theory of Control (MTC, e.g. Hornstein, 1999; Boeckx & Hornstein, 2006) experimentally in order to better understand the online processing of control, raising, and other similar structures as well as the empty categories that form part of their associated accounts. According to the MTC, the coindexed empty categories (EC) in apparent control phenomena (1, 2 from Modesto, 2007) should be analyzed as the product of movement and not as PRO (as in e.g. Chomsky, 1981; Landau, 2000). Notably, Brazilian Portuguese (BP) allows for agreement in object control (2). The MTC predicts that control should behave in a manner consistent with movement traces/copies in generative grammar. Modesto (2010, 2018) observes an asymmetry between exhaustive control (3) and partial control (4) in BP according to matrix predicate type (Landau 2000). The former requires strict identity with the controlling DP (O presidente), while the latter must include the controller. This is unexpected according to the MTC, given that a trace/copy should not be able to alter its identity (i.e. from singular to plural) during the derivation. Experimental evidence for ECs is based on the assumption that they are covert elements analogous to overt pronouns. Cross-modal lexical priming (CMLP) studies, in which sentence primes and semantically-related probe words are presented in different modalities (audio primes and visual probes), have demonstrated that overt anaphors reactivate their antecedents (e.g., Nicol, 1988; Nicol & Swinney, 1989); and that at least some ECs do as well (e.g., Swinney, Ford, & Bresnan, 1989). Converging evidence for this comes from repetition priming studies, in which part of the antecedent phrase serves as a probe (e.g., Bever & McElree, 1988); such studies have found reactivation in raising contexts as well. Under the MTC, control is analyzed as movement analogous to raising; we would therefore expect a similar pattern of activation in raising and control sentences. In the current study, we employ a cross-modal repetition priming (CMRP) paradigm, similar to the CMLP paradigm, but with the probe being taken directly from the antecedent phrase, in order to examine the timecourse of processing of phonologically overt pronouns/anaphora and theoretically-predicted ECs in BP, with a particular interest in whether ECs reactivate their antecedents at the predicted gap positions. Native speakers (n=52) of BP participated in a three-part procedure (CMRP, bimodal acceptability judgment, linguistic history/handedness questionnaire). In the CMRP, we measured reactivation at three testpoints: 1) following the matrix predicate, 2) at the predicted gap position, and 3) at the end of the sentence (short [S], medium [M], or long [L] latency, respectively) as in (5). We report reaction time results for raising, overt anaphor, partial control, and exhaustive control sentences at the three different latency intervals. Table 1 shows evidence of similar reactivation in sentence-medial position (i.e. at the [M] gap) in raising sentences and in those with overt pronouns, thus supporting the standard account of movement in raising constructions, where a trace/copy of the raised element reactivates it in the gap position. No such reactivation was found for exhaustive control sentences in the [M] gap; response times were significantly slower in these sentences than in raising sentences and in those with overt pronouns (p=0.045 in both instances). We found an unexpected asymmetry in processing of exhaustive and partial control, and limited evidence for sentence-final reactivation in exhaustive control sentences. We suspect this is due to the parser positing the EC PRO not in the predicted gap position prior to the embedded infinitive, but in a theta-position following the infinitive (as in Larsen & Johansson, 2020). Notwithstanding, accuracy greatly decreased at the [L] position, suggesting that the EC (i.e. PRO) does not fully reactivate its antecedent phrase. This differentiation suggests that 1) the parser makes active use of empty categories in comprehension, contra traceless accounts (e.g. Pickering and Barry, 1991), and 2) different linguistic mechanisms are involved in the processing of raising and control structures, a finding presumably inconsistent with the MTC. Examples (1) Os meninosj querem (ECj) nadar(*em). the boys want.PRS.3pl swim.INF(3pl) 'The boys want to swim' (2) Euj convenci os meninosk a (EC*j/k) tomar(em) banho. I convince.PST.1sg the boys to take.INF(3pl) bath. 'I convinced the boys to take a bath.' (3) *O presidentej conseguiu [ECj+k] se reunirem às 6 the president achieve.PST.3sg SE meet.INF.3pl at.the six intended: 'The president managed for them to meet at 6.' (4) O presidentej preferia [ECj+k] se reunirem às 6. the president prefer.IMPFV.3sg SE meet.INF.3pl at.the six 'The president preferred for them to meet at 6.' (5) Matrix-clause raising predicate (Target probe: recolhidos; non-target probe: falsos) Os dados recolhidos resultaram [S] surpreendentemente [M] ser muito problemáticos para as hipóteses existentes [L] 'The data gathered turned out [S] surprisingly [M] to be very problematic for current hypotheses [L]'

Research paper thumbnail of What’s in the bucket? Aspectual (non)compositionality in phrasal idioms

Natural Language & Linguistic Theory, 2021

I consider here the question of whether the aspectual properties of idioms are compositionally de... more I consider here the question of whether the aspectual properties of idioms are compositionally determined with respect to the normal literal meanings of their components. This issue is unresolved in the literature, with some (Marantz 1997; McGinnis 2002, 2005) arguing for systematic literal compositionality, some noting mismatches in the aspectual properties of some idioms’ literal and figurative readings and therefore arguing for compositionality with respect to idioms’ meanings only (Glasbey 2003, 2007), and others attempting to reconcile the two views, maintaining systematic literal compositionality while accounting for mismatches by adding aspectual information post-syntactically (Leivada 2017) or by nonlinguistic conceptual processes (Mateu and Espinal 2007). Working within a monotonic model of aspectual composition in which telicity, durativity, and dynamicity are privative features inherent to verbs (Olsen 1997), I present further evidence for Glasbey’s position, arguing that that the aspectual properties of idioms are predictable on the basis of their figurative meanings, and that the literal readings of their components are irrelevant to aspectual composition, at least in the domain of language production. I further argue that some phrasal idioms are more highly specified than their commonly paraphrased meanings, and that idiomaticity represents a motivated exception to Olsen’s model in that idioms may be negatively specified for telicity, durativity, and dynamicity. I conclude that this exception falls out directly from the assumption that idioms may lexicalize a larger structure than verbs, speculating that this may include an inner aspect head in the syntax.

Research paper thumbnail of Idioms at the Interface(S): Towards a Psycholinguistically Grounded Model of Sentence Generation

The primary goal of this dissertation is to demonstrate that a close analysis of the syntactic an... more The primary goal of this dissertation is to demonstrate that a close analysis of the syntactic and semantic properties of phrasal idioms leads directly to the conclusion that the phonological forms of idioms are not present in narrow syntax. The secondary goal of this dissertation is to demonstrate that from the psycholinguistic perspective, syntactic structure-building in the context of language production cannot plausibly be modelled as a bottom-up process powered by iterative Merge, as is standardly assumed. Instead, I propose a provisional model grounded in both theory-external and theory-internal evidence, wherein the computational component operates over semantically individuated but phonologically null Roots-as-nodes (N-Roots), and structures are built top-down by iterative Branch. In this model, which I refer to as LIT-D (Late-Insertion, Top-Down), N-Roots are inserted in functional and discourse positions, and head chains terminating in theta positions. Cyclically transferred to the interfaces, these N-Roots are then matched with Roots-as-exponents (E-Roots), via syntactic and semantic feature mapping. Idioms are exceptional in this model only in that they are complex E-Roots, with multiple components that can be mapped into multiple N-Roots.

Research paper thumbnail of What’s in the bucket? Aspectual (non)compositionality in phrasal idioms.

Natural Language and Linguistic Theory, 2021

I consider here the question of whether the aspectual properties of idioms are compositionally de... more I consider here the question of whether the aspectual properties of idioms are compositionally determined with respect to the normal literal meanings of their components. This issue is unresolved in the literature, with some (Marantz 1997; McGinnis 2002, 2005) arguing for systematic literal compositionality, some noting mismatches in the aspectual properties of some idioms’ literal and figurative readings and therefore arguing for compositionality with respect to idioms’ meanings only (Glasbey 2003, 2007), and others attempting to reconcile the two views, maintaining systematic literal compositionality while accounting for mismatches by adding aspectual information post-syntactically (Leivada 2017) or by nonlinguistic conceptual processes (Mateu and Espinal 2007). Working within a monotonic model of aspectual composition in which telicity, durativity, and dynamicity are privative features inherent to verbs (Olsen 1997), I present further evidence for Glasbey’s position, arguing that that the aspectual properties of idioms are predictable on the basis of their figurative meanings, and that the literal readings of their components are irrelevant to aspectual composition, at least in the domain of language production. I further argue that some phrasal idioms are more highly specified than their commonly paraphrased meanings, and that idiomaticity represents a motivated exception to Olsen’s model in that idioms may be negatively specified for telicity, durativity, and dynamicity. I conclude that this exception falls out directly from the assumption that idioms may lexicalize a larger structure than verbs, speculating that this may include an inner aspect head in the syntax.