1. Encoding and Navigating Structure
We can easily diagram sentence structures on a blackboard. But what do those diagrams correspond to in (neuro-)cognitive terms? And what does it mean to “form dependencies” and “manipulate the structures”? This is the research question that our team has worried about the most in recent years.
Two independent developments converged to make this a very productive topic for us.
Selective fallibility and linguistic illusions: around 2000 we started to explore whether detailed grammatical constraints are respected in comprehension. In other words, does the parser construct all and only the representations that are sanctioned by the grammar. The original hope was that the answer would be “yes”, and for a few years the results pointed to that conclusion.
But things got a lot more interesting once we dug into cases of linguistic illusions, i.e., cases where people appear to accept incoherent sentences, e.g., More people have been to Russia than I have, or agreement with an incorrect controller, e.g., The key to the cabinets were on the table, or licensing of a negative polarity item by an inappropriate negation, e.g., The bills that no democratic senators voted for will ever become law. The contrast between the cases of success and failure begs the question of what is responsible for the “selective fallibility”.
Content-addressable memory: Research on memory processes provides strong support for content-addressable memory (CAM), i.e., items in memory are accessed based on retrieval cues rather than on a memory address. This is why it’s hard to remember what you ate for dinner 4 days ago. Psychologists Brian McElree and Peter Gordon highlighted evidence that this memory architecture is also used for human language. CAM-based access has distinctive timing profiles (long ≠ slow) and interference profiles (mis-retrieval of items similar to the target memory). Computational psycholinguists Shravan Vasishth and Rick Lewis captured these properties in an implemented and accessible model (ACT-R). The model makes clear predictions, which only partly fit our selective fallibility profile.
This set the stage for research in which a cycle of modeling and experimentation seeks to explain the selective fallibility profile. This generated clear hypotheses about how structures are encoded and navigated mentally. In our team, Matt Wagers, Ellen Lau, Brian Dillon, and Ming Xiang set us down this path.
Phase 1 – Dependency-wise differences: our first bet was that selective fallibility should be understood at the grain size of dependency types, e.g., subject-verb agreement is prone to illusions but subject-reflexive licensing is not (Dillon et al. 2013). We ran many studies in pursuit of this idea. And we found ways to capture both sides of the selective fallibility profile within a CAM architecture.
Phase 2 – Turning illusions on/off: recent findings by Dan Parker and Dave Kush challenge our earlier framing of the problem. Dan found that robust illusions could be turned off (NPIs: Parker & Phillips 2014a), and that illusions could be turned on in cases that we had thought immune (reflexives: Parker & Phillips 2014b). Meanwhile, Dave’s studies on c-command sensitive phenomena such as bound variable anaphora suggested that comprehenders can show greater structure sensitivity than we had thought possible in a CAM-based architecture.
Getting to the bottom of this will keep us busy for a while yet.
2. Linguistic Architecture
Are locality constraints in grammar (islands etc.) and language processing (longer = harder) ultimately the same? This possibility attracts many in both fields, but it is challenged by our findings from head-final languages such as Japanese and Bengali, whose properties distinguish structural and linear/temporal locality. (Aoshima et al. 2004; Omaki et al. 2014; Chacón et al. 2014)
A routine assumption in linguistics and psycholinguistics alike is that combinatorial semantic interpretation closely follows syntactic structure. Recent ERP findings have been taken as challenges to this orthodoxy, motivating independent semantic composition. Our own ERP studies in English, Chinese, and Spanish support the standard view. Perhaps. (Stroud & Phillips 2012; Chow & Phillips 2013)
Comprehension-production differences present a prima facie argument for a traditional linguistic architecture with a task- and time-independent grammar. The grammar is what links the comprehension and production systems. Shota Momma, Bob Slevc and I are trying to better understand this relation, focusing initially on look-ahead in production. (Momma et al. 2014ab)
If there’s no task-specific comprehension system, distinct from the grammar, then we should be able to explain language processing phenomena by embedding a grammar in a general cognitive architecture, with no mechanisms specific to language processing. We have recently begun to explore the feasibility of this. (Lewis & Phillips 2014; Parker 2014)
We suspect that predictive mechanisms may be important contributors to the speed and robustness of language understanding. When comprehenders can anticipate what is likely to be said, they can process the input more quickly and more successfully in noise.
In earlier work we argued that predictive mechanisms are responsible for the ELAN, an ERP response whose speed is almost too good to be true (Lau et al. 2006). We suggested that predictive mechanisms also underlie some instances of extreme grammatical fidelity in comprehension (Phillips et al. 2011). And we argued that the N400 ERP response is a relatively direct reflection of lexical predictions in comprehension (Lau et al. 2008).
But Wing Yee Chow’s recent findings on the N400’s temporary “blindness” to highly predictive information makes things rather more interesting. Her studies on role-reversed sentences show that some types of predictive information is accessed more quickly than others. This clears a path for explicit models of how predictions are generated. More to come!
4. Variation and Learning
Can differences in language processing abilities aid our understanding of learning outcomes? Perhaps — it depends what you mean.
Explaining failures: language processing abilities may account for some of children’s failings. Errors in the interpretation of anaphora such as the notorious Delay of Principle B Effect may reflect immature reanalysis and cognitive control systems, rather than failure of grammatical learning (Conroy et al. 2009).
Distorted input? In light of the well-documented cases of children’s parsing errors (e.g., Trueswell et al. 1999), there is a danger that children’s misparsing could lead to distortion of crucial distributional evidence for learning. Example: if a child only ever accesses a local interpretation of an ambiguous wh-dependency, might she incorrectly conclude that the target grammar only allows local dependencies? Akira Omaki has paid close attention to this concern (Omaki et al. 2014), and he’ll continue to do so. The jury remains out on how serious this concern is.
Explaining success? Is there any way that population differences in language processing abilities could explain why children are more successful language learners than adults? Good question. Check back soon.
Publications in Language Processing
(including PhD dissertations)
| Xiang, Ming; Dillon, Brian; Phillips, Colin: Illusory licensing effects across dependency types: ERP evidence. In: Brain and Language, vol. 108, no. 1, pp. 40–55, 2009. (Type: Journal Article | | | )|
A number of recent studies have argued that grammatical illusions can arise in the process of completing linguistic dependencies, such that unlicensed material is temporarily treated as licensed, due to the presence of a potential licensor that is semantically appropriate but in a syntactically inappropriate position. A frequently studied case involves illusory licensing of negative polarity items (NPIs) like ever and any, which must appear in the scope (i.e., c command domain) of a negative or similar element. Speakers often show intrusive licensing effects in sentences where an NPI is preceded but not c commanded by a negative element, as in *The restaurants that no newspapers have recommended in their reviews have ever gone out of business. Existing accounts of intrusive licensing have focused on the role of general retrieval processes. In contrast, we propose that intrusive licensing of NPIs reflects semantic/pragmatic processes that are more specific to NPI licensing. As a test of this claim, we present results from an ERP study that presents a structurally matched comparison of intrusive licensing in two types of linguistic dependencies, namely NPI licensing and the binding of reflexive anaphors like himself, herself. In the absence of a potential licensor, both NPIs and reflexives elicit a P600 response, but whereas there is an immediate ERP analog of the intrusion effect for NPI licensing, no such effect is found for reflexive binding. This suggests that the NPI intrusion effect does not reflect general purpose retrieval mechanisms.
| Aoshima, Sachiko; Yoshida, Masaya; Phillips, Colin: Incremental processing of coreference and binding in Japanese. In: Syntax, vol. 12, no. 2, pp. 93–134, 2009. (Type: Journal Article | | | )|
This article presents two on-line self-paced reading studies and three off-line acceptability judgment studies on the processing of backward anaphoric dependencies in Japanese in which a pronoun precedes potential antecedent noun phrases. The studies investigate the real-time formation of coreference relations and operator- variable binding relations to determine whether speakers of head-final languages are able to construct grammatically accurate syntactic structures before they encounter a verb. The results of the acceptability rating studies confirm previous claims that backwards anaphoric dependencies in Japanese are more acceptable in configurations where a pronoun has been fronted via scrambling from a position where it would be c-commanded by its antecedent. The results of the on-line studies demonstrate that these acceptability contrasts have an immediate impact on parsing. Reading-time results showed immediate sensitivity to the semantic congruency between an NP and a preceding pronoun in noncanonical (‘‘scrambled’’) word orders, and no immediate effect of semantic congruency otherwise. This contrast was found both for coreference relations involving the personal pronouns kare/kanojo (experiment 1) and for operator- variable relations involving the demonstrative pronoun soko (experiment 3). These findings go beyond previous evidence for incremental parsing in head-final languages by showing that Japanese speakers build compositional structures (such as anaphoric relations) in a grammatically constrained fashion in advance of encountering a verb in the input.
| Wagers, Matthew W; Phillips, Colin: Multiple dependencies and the role of the grammar in real-time comprehension. In: Journal of Linguistics, vol. 45, no. 02, pp. 395–433, 2009. (Type: Journal Article | | | )|
Wh-dependencies are known to be formed rapidly in real-time comprehension. The parser posits the location of gap sites in advance of the bottom-up evidence for missing constituents, and must therefore have a means of deciding when and where to project dependencies. Previous studies have observed that the parser avoids building ungrammatical wh-dependencies, for example, by restricting the search for gap sites from island domains. This paper tests the stronger claim that constraints are not merely respected, but that grammatical knowledge actively prompts the con- struction of some representations in advance of the input. Three self-paced reading experiments examined patterns of wh-dependency formation in multiple-dependency constructions: obligatory across-the-board (ATB) extraction from coordinated verb phrases, and from optional parasitic gaps in post-verbal adjunct clauses. The key finding is that comprehenders immediately enforce the requirement for extraction from coordinates, and hence actively search for multiple gap sites within a coordinate VP; but they do not search for post-verbal parasitic gaps. This difference cannot be attributed to relative differences in acceptability, as comprehenders rated both of these multiple-gap constructions equally highly, nor can it be explained by general parsing incentives to develop maximal incremental interpretations of partial strings. More plausibly, the difference reflects the deployment of detailed grammatical knowledge in a parser that is motivated to satisfy structural licensing requirements in real time.
| Hsu, Chun-chieh Natalie; Hurewitz, Felicia; Phillips, Colin: Context influences structure generation: evidence from Chinese. 2009, (submitted for publication). (Type: Unpublished | | | )|
This study reports three sets of experiments that investigate the impact of context on structure generation in on-line sentence processing. Unlike previous studies that have addressed the role of context in resolving ambiguities, the current study examined the role of context in helping the parser to identify unambiguous structures that it could not otherwise recognize. The first set of experiments shows that late disambiguation of Chinese head-final relative clauses (RCs) elicits mild garden-path effects, and that a direct cue from the particle suo can facilitate the processing of head-final RCs. The second set of experiments shows that, in contrast to the direct cue, the parser fails to use an indirect-but-unambiguous cue from a mismatching classifier-noun sequence to recognize an upcoming RC structure and to avoid a garden path at the end of the RC. The third set of experiments shows that, when a supporting context for an RC is provided, the parser becomes able to use the indirect cue of the mismatching classifier-noun sequence to recognize an upcoming relative clause. Taken together, the findings suggest that there is a limitation on the parser’s ability to act upon unambiguous-yet-indirect cues in incremental structure building. When such a situation occurs, contextual information becomes crucial and it actively contributes to the structure generation process to help the parser to recognize incoming structures. These findings suggest that structural and non-structural information can interact with each other in the structure generation process, providing a type of evidence for interactivity in structure generation that has proven elusive in previous studies.
| Stroud, Clare: Structural and semantic selectivity in the electrophysiology of sentence comprehension. University of Maryland, 2008. (Type: PhD Thesis | | | )|
This dissertation is concerned with whether the sentence processor can compute plausible relations among a cluster of neighboring open class words without taking into account the relationships between these words as dictated by the structure of the sentence. It has been widely assumed that compositional semantics is built on top of syntactic structures (Heim & Kratzer, 1998; Pollard & Sag, 1994). This view has been challenged by recent electrophysiological findings (Kim and Osterhout, 2005; Kuperberg, 2007; van Herten et al., 2005, 2006) that appear to show that semantic composition can proceed independently of syntactic structure. This dissertation investigates whether the evidence for independent semantic composition is as strong and widespread as has been previously claimed. Recent studies have shown that sentences containing a semantically anomalous interpretation but an unambiguous, grammatical structure (e.g., The meal was devouring...) elicit a P600 response, the component classically elicited by syntactic anomalies, rather than an N400, the component typically elicited by semantic anomalies (Kim and Osterhout, 2005). This has been interpreted as evidence that the processor analyzed meal as a good theme for devour, even though this interpretation is not supported by the sentential structure. This led to the claim that semantic composition can proceed independently of syntactic structure. Two event-related potentials (ERP) studies investigated whether the processor exploits prior structural biases and commitments to restrict semantic interpretations to those that are compatible with that expected structure. A further ERP study and a review of relevant studies reveal that in the majority of studies the P600 is not modulated by manipulations of thematic fit or semantic association between the open class words. We argue that a large number of studies that have been taken as evidence for an independent semantic processing stream can be explained as violations of the verb’s requirement that its subject be agentive. A small number of studies in verb- final languages cannot be explained in this way, and may be evidence of independent semantic composition, although further experimental work is needed. We conclude that the evidence for independent semantic composition is not as extensive as was previously thought.
| Wagers, Matthew: The structure of memory meets memory for structure in linguistic cognition. University of Maryland, 2008. (Type: PhD Thesis | | | )|
This dissertation is concerned with the problem of how structured linguistic representations interact with the architecture of human memory. Much recent work has attempted to unify real-time linguistic memory with a general content-addressable architecture (Lewis & Vasishth, 2005; McElree, 2006). Because grammatical principles and constraints are strongly relational in nature, and linguistic representation hierarchical, this kind of architecture is not well suited to restricting the search of memory to grammatically-licensed constituents alone. This dissertation investigates under what conditions real-time language comprehension is grammatically accurate. Two kinds of grammatical dependencies were examined in reading time and speeded grammaticality experiments: subject-verb agreement licensing in agreement attraction configurations (“The runners who the driver wave to ...”; Kimball & Aissen, 1971, Bock & Miller, 1991), and active completion of wh-dependencies. We develop a simple formal model of agreement attraction in an associative memory that makes accurate predictions across different structures. We conclude that dependencies that can only be licensed exclusively retrospectively, by searching the memory to generate candidate analyses, are the most prone to grammatical infidelity. The exception may be retrospective searches with especially strong contextual restrictions, as in reflexive anaphora. However dependencies that can be licensed principally by a prospective search, like wh-dependencies or backwards anaphora, are highly grammatically accurate.
| Lau, Ellen F; Phillips, Colin; Poeppel, David: A cortical network for semantics:(de) constructing the N400. In: Nature Reviews Neuroscience, vol. 9, no. 12, pp. 920–933, 2008. (Type: Journal Article | | | )|
Measuring event-related potentials (ERPs) has been fundamental to our understanding of how language is encoded in the brain. One particular ERP response, the N400 response, has been especially influential as an index of lexical and semantic processing. However, there remains a lack of consensus on the interpretation of this component. Resolving this issue has important consequences for neural models of language comprehension. Here we show that evidence bearing on where the N400 response is generated provides key insights into what it reflects. A neuroanatomical model of semantic processing is used as a guide to interpret the pattern of activated regions in functional MRI, magnetoencephalography and intracranial recordings that are associated with contextual semantic manipulations that lead to N400 effects.
| Lau, Ellen F; Rozanova, Katya; Phillips, Colin: Syntactic prediction and lexical surface frequency effects in sentence processing. In: University of Maryland Working Papers in Linguistics, vol. 16, pp. 163–200, 2008. (Type: Journal Article | | | )|
This paper presents three experiments which examine the effect of lexical surface frequency on sentence processing and the interaction between surface frequency and syntactic prediction. The first two experiments make use of the self-paced reading paradigm to show that processing time differences due to surface frequency (e.g., the frequency of cats not including occurrences of cat), which have previously been demonstrated in isolated word tasks like lexical decision, also give rise to reaction time differences in sentence processing tasks, in this case for singular and plural English nouns. The second experiment investigates whether a prediction for the number morpheme triggered by the number-marked determiners this and these might counter the surface frequency effect; however, the small size of the surface frequency effect and baseline differences in reaction times to this and these made the results unclear. Results from a third experiment using lexical decision suggest that the difference in the size of the surface frequency effects between the lexical decision experiments and the self- paced-reading experiments are likely due to differences in task demands. Our results have methodological implications for psycholinguistic experiments that manipulate morphology as a means of examining other questions of interest.
| Kazanina, Nina; Lau, Ellen F; Lieberman, Moti; Yoshida, Masaya; Phillips, Colin: The effect of syntactic constraints on the processing of backwards anaphora. In: Journal of Memory and Language, vol. 56, no. 3, pp. 384–409, 2007. (Type: Journal Article | | | )|
This article presents three studies that investigate when syntactic constraints become available during the processing of long-distance backwards pronominal dependencies (backwards anaphora or cataphora). Earlier work demonstrated that in such structures the parser initiates an active search for an antecedent for a pronoun, leading to gender mismatch effects in cases where a noun phrase in a potential antecedent position mismatches the gender of the pronoun [Van Gompel, R. P. G. & Liversedge, S. P. (2003). The influence of morphological information on cataphoric pronoun assignment. Journal of Experimental Psychology: Learning, Memory, and Cognition, 29, 128–139]. Results from three self-paced reading studies suggest that structural constraints on coreference, in particular Principle C of the Binding Theory [Chomsky, N. (1981). Lectures on government and binding. Dordrecht, Foris], exert an influence at an early stage of this search process, such that gender mismatch effects are elicited at grammatically licit antecedent positions, but not at grammatically illicit antecedent positions. The results also show that the distribution of gender mismatch effects is unlikely to be due to differences in the predictability of different potential antecedents. These findings suggest that back- wards anaphora dependencies are processed with a grammatically constrained active search mechanism, similar to the mechanism used to process another type of long-distance dependency, the wh dependency (e.g., [Stowe, L. (1986). Evi- dence for online gap creation. Language and Cognitive Processes, 1, 227–245; Traxler, M. J., & Pickering, M. J. (1996). Plausibility and the processing of unbounded dependencies: an eye-tracking study. Journal of Memory and Language, 35, 454–475.]). We suggest that the temporal priority for syntactic information observed here reflects the predictability of structural information, rather than the need for an architectural constraint that delays the use of non-syntactic information.
| Phillips, Colin; Wagers, Matthew: Relating structure and time in linguistics and psycholinguistics. In: Oxford handbook of psycholinguistics, pp. 739–756, Oxford University Press, 2007. (Type: Incollection | | )|
| Nevins, Andrew; Dillon, Brian; Malhotra, Shiti; Phillips, Colin: The role of feature-number and feature-type in processing Hindi verb agreement violations. In: Brain Research, vol. 1164, pp. 81–94, 2007. (Type: Journal Article | | | )|
This article presents studies of Hindi that investigate whether responses to syntactic agreement violations vary as a function of the type and number of incorrect agreement features, using both electrophysiological (ERP) and behavioral measures. Hindi is well suited to investigation of this issue, since verbs in Hindi mark agreement with the person, number, and gender features of the nominative subject noun phrase. In an ERP study evoked responses were recorded for visually presented verbs appearing at the end of a sentence- initial adverbial clause, comparing responses in a grammatically correct condition with four grammatically incorrect conditions that mismatched the correct agreement on different dimensions (Gender, Number, Gender/Number, Person/Gender). A P600 response was elicited in all grammatically incorrect conditions. No amplitude differences were found among the Gender, Number, and combined Gender/Number violations. This suggests that the feature distance between observed and expected word forms at the morphosyntactic level does not impact ERP responses, contrasting with findings on semantic and auditory processing, and suggests that the P600 response to agreement violations is not additive based on the number of mismatching features and does not reflect top-down, predictive mechanisms. A significantly larger P600 response was elicited by the combined Person/ Gender violation, and two different violations involving the Person feature were judged as more severe and recognized more quickly in the behavioral studies. This effect is attributed to the greater salience of the Person feature at multiple levels of representation.
| Pablos, Leticia: Preverbal structure building in Romance languages and Basque. University of Maryland, 2006. (Type: PhD Thesis | | | )|
The main goal of the work in this dissertation is to investigate pre-verbal structure building effects in languages with different configurations such as Spanish, Galician and Basque, by means of using different pre-verbal cues in order to show that syntactic structure can be both interpreted and anticipated before the verbal head. I examine the syntax of Clitic-Left Dislocations (CLLDs) and other kinds of left-dislocations in Spanish and then analyze their processing. I concentrate on the use of clitic pronouns in Spanish and Galician in CLLD constructions that require the presence of the clitic pronoun to interpret the left-dislocated phrase and where I examine if this left-dislocation is interpreted at the clitic pronoun. Experimental results from three self-paced reading experiments provide evidence that the clitic in these constructions is required and used to interpret the thematic features of the topicalized NP before the verb. Thus, I demonstrate that clitic pronouns are used as pre-verbal cues in parsing and that the active search mechanism is also triggered in long-distance dependencies involving clitic pronouns. I conclude that the active search mechanism is a more general architectural mechanism of the parser that is triggered in all kinds of long-distance dependencies, regardless of whether the search is triggered by gaps or pronouns. In Basque, verbal auxiliaries overtly encode agreement information that reflects the number of arguments of the verbal head. In negatives, auxiliaries are obligatorily fronted and split from the verbal head with which they otherwise form a cluster. Thus, verbal auxiliaries in Basque are a pre-verbal morphological cue that can assist the parser in predicting structure. Specifically, I examine how predictions for the upcoming structure of the sentence are determined by agreement information on the number of arguments specified in the auxiliary and by the mismatch of this auxiliary with the case features of the NP that follows it. I provide results from a self- paced reading experiment to argue that the parser uses the information encoded in the auxiliaries and demonstrate that the mismatch of the auxiliary with the following NP can prevent the reader from following a garden-path analysis of the sentence.
| Yoshida, Masaya: Constraints and mechanisms in long-distance dependency formation. University of Maryland, 2006. (Type: PhD Thesis | | | )|
This thesis aims to reveal the mechanisms and constraints involving in long-distance dependency formation in the static knowledge of language and in real-time sentence processing. Special attention is paid to the grammar and processing of island constraints. Several experiments show that in a head-final language like Japanese global constraints like island constraints are applied long before decisive information such as verb heads and relative heads, are encountered. Based on this observation, the thesis argues that there is a powerful predictive mechanism at work behind real time sentence processing. A model of this predictive mechanism is proposed. This thesis examines the nature of several island constraints, specifically Complex NP Islands induced by relative clauses, and clausal adjunct islands. It is argued that in the majority of languages, both relative clauses and adjunct clauses are islands, but there is a small subset of languages (including Japanese, Korean and Malayalam) where extraction out of adjunct clauses seems to be allowed. Applying well-established syntactic tests to the necessary constructions in Japanese, it is established that dependencies crossing adjunct clauses are indeed created by movement operations, and still the extraction is allowed from adjuncts. Building on previous findings, the thesis turns to the investigation of the interaction between real time sentence processing and island constraints. Looking specifically at Japanese, a head-final language this thesis ask how the structure of sentences are built and what constraints are applied to the structure building process. A series of experiments shows that in Japanese, even before high-information bearing units such as verbs, relative heads or adjunct markers are encountered, the structural skeleton is built, and such pre-computed structures are highly articulated. It is shown that structural constraints on long-distance dependencies are imposed on the pre-built structure. It is further shown that this finding support the incrementality of sentence processing.
| Lieberman, Moti; Aoshima, Sachiko; Phillips, Colin: Nativelike biases in generation of wh-questions by nonnative speakers of Japanese. In: Studies in Second Language Acquisition, vol. 28, pp. 423-448, 2006. (Type: Journal Article | | | )|
A number of studies of second language (L2) sentence processing have investigated whether ambiguity resolution biases in the native language (L1) transfer to superficially similar cognate structures in the L2. When transfer effects are found in such cases, it is difficult to determine whether they reflect surface parallels between the languages or the operation of more abstract processing mechanisms. Wh-questions in English and Japanese present a valuable test case for investigating the relation between L1 and L2 sentence processing. Native speakers (NSs) of English and Japanese both show strong locality biases in processing wh-questions, but these locality biases are realized in rather different ways in the two languages, due to differences in word order and scope marking. Results from a sen- tence generation study with NSs of Japanese and advanced English-speaking L2 learners of Japanese show that the L2 learners show a strongly nativelike locality bias in the resolution of scope ambiguities for in situ wh-phrases, despite the fact that the closest analogue of such an interpretation is impossible in English. This indicates that L2 learners are guided by abstract processing mechanisms and not just by superficial transfer from the L1.
| Phillips, Colin: The real-time status of island phenomena. In: Language, pp. 795–823, 2006. (Type: Journal Article | | | )|
In parasitic-gap constructions an illicit gap inside a syntactic island becomes acceptable in combination with an additional licit gap, a result that has interesting implications for theories of grammar. Such constructions hold even greater interest for the question of the relation between grammatical knowledge and real-time language processing. This article presents results from two experiments on parasitic-gap constructions in English in which the parasitic gap appears inside a subject island,before the licensing gap. An offline study confirms that parasitic gaps are acceptable when they occur inside the infinitival complement of a subject NP, but not when they occur inside a finite relative clause. An on-line self-paced reading study using a plausibility manipulation technique shows that incremental positing of gaps inside islands occurs in just those environments where parasitic gaps are acceptable. The fact that parasitic gaps are constructed incrementally in language processing presents a challenge for attempts to explain subject islands as epiphenomena of constraints on language processing and also helps to resolve apparent conflicts in previous studies of the role of island constraints in parsing.
| Ono, Hajime; Yoshida, Masaya; Aoshima, Sachiko; Phillips, Colin: Real-time computation of Japanese exclamatives and the strength of locality biases in sentence comprehension. In: Cognitive Studies, vol. 13, no. 3, 2006. (Type: Journal Article | | | )|
This paper investigates mechanisms of long-distance dependency formation in language comprehension, using experimental data on the processing of Japanese interrogatives and exclamatives to explore the nature of locality biases in parsing. Findings on the processing of exclamative wh-phrases are compared to previous results involving the processing of interrogative wh-phrases, revealing both similarities and differences. Experiment 1 uses a sentence fragment completion task with in-situ and fronted exclamative and interrogative wh-phrases. Both types of in-situ wh-phrase show a strong bias for local generation of licensing particles. Conditions with fronted wh-phrases show a contrast between interrogative and exclamative wh-phrases: interrogatives show a bias for interpretation in an embedded clause, replicating previous evidence for a long-distance scrambling bias in Japanese (Aoshima, Phillips, & Weinberg, 2004); in contrast, the long-distance scrambling bias is weaker for fronted exclamative wh-phrases. Experiment 2 uses an on-line self-paced reading task to investigate the processing consequences of expectations for a local licensor for in-situ exclamative wh-phrases. Results indicate processing disruption when readers fail to encounter a licensor for an exclamative wh-phrase at the first possible verb position, although the disruption is weaker than the Typing Mismatch Effect shown for interrogatives in previous studies by Miyamoto and Takahashi (2002). Different possible accounts of the parallels and contrasts between processing of interrogatives and exclamatives are discussed.
| Kazanina, Nina: The acquisition and processing of backwards anaphora. University of Maryland, 2005. (Type: PhD Thesis | | | )|
This dissertation investigates long-distance backwards pronominal dependencies (backwards anaphora or cataphora) and constraints on such dependencies from the viewpoint of language development and real-time language processing. Based on the findings from a comprehension experiment with Russian-speaking children and on real-time sentence processing data from English and Russian adults I argue for a position that distinguishes structural and non-structural constraints on backwards anaphora. I show that unlike their non-syntactic counterparts, structural constraints on coreference, in particular Principle C of the Binding Theory (Chomsky 1981), are active at the earliest stage of language development and of real-time processing. In language acquisition, the results of a truth-value judgment task with 3-6 year old Russian-speaking children reveal a striking developmental asymmetry between Principle C, a cross-linguistically consistent syntactic constraint on coreference, and a Russian-specific discourse constraint on coreference. Whereas Principle C is respected by children already at the age of three, the Russian- specific (discourse) constraint is not operative in child language until the age of five. These findings present a challenge for input-driven accounts of language acquisition and are most naturally explained in theories that admit the existence of innately specified principles that underlie linguistic representations. In real-time processing, the findings from a series of self-paced reading experiments on English and Russian show that in backwards anaphora contexts the parser initiates an active search for an antecedent for the pronoun which is limited to positions that are not subject to structural constraints on coreference, e.g. Principle C. This grammatically constrained active search mechanism associated observed in the processing of backwards anaphora is similar to the mechanism found in the processing of another type of a long-distance dependency, the wh-dependency. I suggest that the early application of structural constraints on long-distance dependencies is due to reasons of parsing efficiency rather than due to their architectural priority, as such constraints aid to restrict the search space of possible representations to be built by the parser. A computational parsing algorithm is developed that combines the constrained active search mechanism with a strict incremental left-to-right structure building procedure.
| Phillips, Colin; Kazanina, Nina; Abada, Shani H: ERP effects of the processing of syntactic long-distance dependencies. In: Cognitive Brain Research, vol. 22, no. 3, pp. 407–428, 2005. (Type: Journal Article | | | )|
In behavioral studies on sentence comprehension, much evidence indicates that shorter dependencies are preferred over longer dependencies, and that longer dependencies incur a greater processing cost. However, it remains uncertain which of the various steps involved in the processing of long-distance dependencies is responsible for the increased cost of longer dependencies. Previous sentence comprehension studies using event-related potentials (ERPs) have revealed response components that reflect the construction [J. King, M. Kutas, Who did what and when? Using word- and clause-level ERPs to monitor working memory usage in reading. Journal of Cognitive Neuroscience, 7, (1995) 376–395.] and completion [E. Kaan, A. Harris, E. Gibson, P. Holcomb, The P600 as an index of syntactic integration difficulty. Language and Cognitive Processes, 5, (2000) 159–201.] of long-distance wh-dependencies. This article reports one off-line rating study and one ERP study that manipulated both the presence of wh-dependencies and the length of the dependencies (one clause vs. two clauses), with the aim of clarifying the locus of length-sensitivity and the functional role of associated ERP components. Results of the off- line study confirm that longer wh-dependencies incur greater processing cost. Results of the ERP study indicate that both a sustained anterior negativity that follows the initiation of the wh-dependency and also a late posterior positivity (P600) that marks the completion of the dependency are sensitive to the presence of a wh-dependency, but do not show amplitude variations reflecting the length of the dependency. However, the P600 is delayed when it marks the completion of a longer wh-dependency. This suggests that both the sustained negativity and the P600 reflect length-insensitive aspects of the construction of syntactic dependencies. In addition, an N400 component is elicited in the middle of the two clause wh-dependency, upon encountering a verb with an argument structure that prevents completion of the dependency.
| Phillips, Colin: Electrophysiology in the study of developmental language impairments: Prospects and challenges for a top-down approach. In: Applied Psycholinguistics, vol. 26, no. 01, pp. 79–96, 2005. (Type: Journal Article | | | )|
There is a good deal of interest in the application of neurocognitive techniques to investigate the underpinnings of developmental language impairments (DLIs). Electrophysiological techniques such as electroencephalography and magnetoencephalography offer the promise of the ability to track brain activity with precision in time and space. This article describes a number of findings from studies of normal adults and children that are relevant to neurocognitive studies of developmental language impairments and outlines a series of challenges that should be met in order for electrophysiological measures to realize their promise.
| Phillips, Colin: Linguistics and linking problems. In: Rice, Mabel; Warren, Steven (Ed.): Developmental Language Disorders: From Phenotypes to Etiologies, pp. 241-287, Lawrence Erlbaum Associates, Mahwah, NJ, 2004. (Type: Incollection | | )|
| Aoshima, Sachiko; Phillips, Colin; Weinberg, Amy: Processing filler-gap dependencies in a head-final language. In: Journal of Memory and Language, vol. 51, no. 1, pp. 23–54, 2004. (Type: Journal Article | | | )|
This paper investigates the processing of long-distance filler-gap dependencies in Japanese, a strongly head-final language. Two self-paced reading experiments and one sentence completion study show that Japanese readers associate a fronted wh-phrase with the most deeply embedded clause of a multi-clause sentence. Experiment 1 demonstrates this using evidence that readers expect to encounter a scope-marking affix on the verb of an embedded clause in wh-fronting constructions. Experiment 2 shows that the wh-phrase is already associated with the embedded clause before the em- bedded verb is processed, based on a Japanese counterpart of the Filled Gap Effect (Stowe, 1986). Experiment 3 corroborates these findings in a sentence completion study. These findings clarify the factors responsible for Ôactive fillerÕ effects in processing long-distance dependencies (Crain & Fodor, 1985; Fodor, 1978; Frazier & Clifton, 1989; Stowe, 1986) in ways not possible in head-initial languages. The results provide evidence that the processing of filler-gap dependencies is driven by the need to satisfy thematic role requirements of the fronted phrase, rather than by the need to create a gap as soon as possible. The paper also discusses implications of these findings for theories of reanalysis.
| Phillips, Colin; Lau, Ellen: Foundational issues. In: Journal of Linguistics, vol. 40, no. 03, pp. 571–591, 2004. (Type: Journal Article | | )|
| Aoshima, Sachiko: The grammar and parsing of wh-dependencies. University of Maryland, 2003. (Type: PhD Thesis | | | )|
The aim of this thesis is to explain how grammar relates to real-time comprehension of wh-dependencies. It proposes an explanatory model which makes minimal assumptions to derive the facts seen in both performance and comprehension. The assumptions are twofold. One is that grammatical constraints drive learning and parsing theory. The other is that incrementality and efficiency force left corner analysis. The proposed model can explain the following, heretofore unrelated, facts; (i) the time course of gap-creation for filler-gap dependencies suggests incremental gap-positing, (ii) structure-building is incremental in head-final structures, (iii) the insertion principle forced by the left corner constraint derives unforced reanalysis only for predicted categories, (iv) the constraint on rapid gap interpretation is derived from restrictions on merger enforced by left corner constraint, (v) off-line / on-line contrasts of wh-scope interpretation are also explained by early prediction of question marker given wh-element, and (vi) Subjacency governs (wh-)scrambling, which is an overt movement operation. The psycholinguistics experiments, which take advantage of the strict verb-final property of Japanese, provide evidence that the real-time formation of filler-gap dependencies is driven by the satisfaction of grammatical constraints, rather than simply by the need to create a gap. They also find a filler-gap relation prior to the linearly first verb, suggesting that structure-building operates incrementally even when critical lexical heads are delayed. The experimental findings imply that there is a case where reanalysis is allowed even though it is not forced. This unforced reanalysis contrasts with a view of reanalysis as a last resort operation. A feature-based left-corner parsing model developed in this thesis can not only predict the time course of gap-creation in head-final structures and incremental structure-building, but also can distinguish between contrasting views of reanalysis. In addition, it accounts for why the gap is postulated in a grammatically legitimate position ‘as soon as possible’. Under the current system, furthermore, the same grammatical principles underlying the local relation of a wh-element and a wh-scope marker also predict both on-line and off-line wh-scope interpretations observed here. The model thus shows that the competence grammar operates dynamically as part of the parsing process.
| Aoshima, Sachiko; Phillips, Colin; Weinberg, Amy: Processing of Japanese wh-scrambling constructions. In: McClure, William (Ed.): Japanese/Korean Linguistics, pp. 179–191, CSLI Publications, Stanford, CA, 2003. (Type: Inproceedings | | )|
| Aoshima, Sachiko; Phillips, Colin; Weinberg, Amy: Theoretical implications of the parsing of Japanese wh-scrambling constructions. In: Garding, Gina; Tsujimura, Mimu (Ed.): WCCFL22: Proceedings of the 22nd West Coast Conference on Formal Linguistics, pp. 29-42, Cascadilla Press, Somerville, MA, 2003. (Type: Inproceedings | | )|
| Phillips, Colin: Parsing: Psycholinguistic aspects. In: International Encyclopedia of Linguistics, Oxford University Press, 2003. (Type: Incollection | | )|
| Aoshima, Sachiko; Phillips, Colin; Weinberg, Amy: Active filler effects and reanalysis: A study of Japanese Wh-scrambling constructions. In: vol. 12, pp. 1–24, 2002. (Type: Book Chapter | | | )|
This paper presents evidence for the Active Filler Effect (Frazier & Clifton 1989) in Japanese. The experimental results show that Japanese speakers preferentially associate a fronted wh-phrase with the first potential scope marker, even when that is contained in an embedded clause. This finding allows us to propose two refinements to the Active Filler Strategy's operation for all languages. We suggest that (i) the Active Filler Strategy results from the parser’s rapid attempt to associate an operator with a q-marked variable position, and that (ii) the parser is able to rescind existing commitments in order to satisfy other constraints. Results from a self-paced reading experiment indicate that a fronted wh-phrase is associated with the embedded verb, an association which we argue suggests reanalysis from an initial matrix clause association.
| Schneider, David; Phillips, Colin: Grammatical search and reanalysis. In: Journal of Memory and Language, vol. 45, no. 2, pp. 308-336, 2001. (Type: Journal Article | | | )|
This paper investigates the extent to which existing structural commitments constrain the human parser’s search for grammatical analyses of incoming material, specifically whether a Reanalysis As a Last Resort (RALR) strategy applies to sentence parsing. Two self-paced reading experiments investigate this issue using a structural ambiguity in which a local, easy reanalysis is pitted against a nonlocal attachment requiring no re- analysis. This ambiguity is created by embedding classic noun phrase/sentential complement ambiguities inside a relative clause modifying a subject NP. The results of both experiments indicate that readers’ existing structural commitments do constrain their subsequent parsing decisions: nonlocal analyses which avoid reanalysis are con- sistently favored over local analyses which require an easy reanalysis. This conclusion is confirmed by the results of a subcategorization-bias manipulation in Experiment 2, which shows that readers show a consistent bias to avoid reanalysis, rather than a general bias for either local or matrix clause attachments.
| Schneider, David: Parsing and incrementality. University of Delaware, 1999. (Type: PhD Thesis | | | )|
There is a great deal of evidence that language comprehension occurs very rapidly. To account for this, it is widely, but not universally, assumed in the psycholinguistic literature that every word of a sentence is integrated into a syntactic representation of the sentence as soon as the word is encountered. This means that it is not possible to wait for subsequent words to provide information to guide a word s initial attachment into syntactic structure. In this dissertation I show how syntactic structures can be built on a word-by-word incremental basis. This work is motivated by the desire to explain how structure can be built incrementally. A psycholinguistically plausible theory of parsing should generalize to all languages. In this work I show not only how head-initial languages like English can be parsed incrementally, but also how head-final languages like Japanese and German can be parsed incrementally. One aspect of incremental parsing that is particularly troublesome in head-final languages is that it is not always clear how a constituent should be structured in advance of the phrase-final heads. There is a significant amount of temporary ambiguity in head-final languages related to the fact that heads of constituents are not available until the end of the phrase. In this work I show that underspecification of the features of a head allows for incremental structuring of the input in head-final structures, while still retaining the temporary ambiguity that is so common in these languages. The featural underspecification allowed by this system is extended to categorial features; I do not assume that every head must always be specified for its category. I assume that the incremental parser builds structures in accord with the principles of the grammar. In other words, there should be no need to submit a structure built by the parser to a separate grammar module to determine whether or not the sentence obeys the grammar. As one aspect of this, I show how wh-movement phenomena can be accommodated within the theory. As part of the treatment of wh-movement, constraints on wh-movement are incorporated into the system, thereby allowing the difference between grammatical and ungrammatical wh-movement to be captured in the parse tree. In addition to being incremental and cross-linguistically generalizable, a parsing theory should account for the rest of human parsing behavior. I show that a number of the structurally-motivated parsing heuristics can be accommodated within the general parsing theory presented here. As part of the investigation of the incremental parser, experimental evidence is presented that establishes a preference for structure-preserving operations in the face of temporary ambiguity. In particular, the experiments show that once a commitment has been made to a particular analysis of a verbal argument, there is a preference to avoid reanalyzing the argument. This preference holds even though the reanalysis is not particularly difficult, and the analysis that is adopted in preference to the reanalysis disobeys a general parsing preference for attachments to recent material. Thus, it appears that existing structural assumptions are rejected only as a last resort. Finally, to demonstrate the theory is explicit enough to make specific predictions, I implement portions of the theory as a computer program.
| Phillips, Colin; Gibson, Edward: On the strength of the local attachment preference. In: Journal of Psycholinguistic Research, vol. 26, no. 3, pp. 323–346, 1997. (Type: Journal Article | | | )|
This paper investigates the strength of the local attachment preference in syntactic am- biguity resolution, based on a study of a novel ambiguity for which the predictions of local attachment contrast with the predictions of a wide range of other ambiguityres- olution principles. In sentences of theform ''Because Rose praised the recipe I made ..." we show that the ambiguous clause "I made" is preferentially attached as a relative clause under some circumstances, as predicted by local attachment, and pref- erentially attached as a matrix clause under other circumstances. The implications for accounts of locality in parsing are discussed.
| Phillips, Colin: Order and structure. Massachusetts Institute of Technology, 1996. (Type: PhD Thesis | | | )|
The aim of this thesis is to argue for the following two main points. First, that grammars of natural language construct sentences in a strictly left-to-right fashion, i.e. starting at the beginning of the sentence and ending at the end. Second, that there is no distinction between the grammar and the parser. In the area of phrase structure, I show that the left-to-right derivations forced by the principle Merge Right can account for the apparent contradictions that different tests of constituency show, and that they also provide an explanation for why the different tests yield the results that they do. Phenomena discussed include coordination, movement, ellipsis, binding, right node raising and scope. I present a preliminary account of the interface of phonology and morphology with syntax based on left-to-right derivations. I show that this approach to morphosyntax allows for a uniform account of locality in head movement and clitic placement, explains certain directional asymmetries in phonology-syntax mismatches and head movement, and allows for a tighter connection between syntactic and phonological phrases than commonly assumed. In parsing I argue that a wide range of structural biases in ambiguity resolution can be accounted for by the single principle Branch Right, which favors building right-branching structures wherever possible. Evidence from novel and existing experimental work is presented which shows that Branch Right has broader empirical coverage than other proposed structural parsing principles. Moreover, Branch Right is not a parsing-specific principle: it is independently motivated as an economy principle of syntax in the chapters on syntax. The combination of these results from syntax and parsing makes it possible to claim that the parser and the grammar are identical. The possibility that the parser and the grammar are identical or extremely similar was explored in the early 1960s, but is widely considered to have been discredited by the end of that decade. I show that arguments against this model which were once valid no longer apply given left-to-right syntax and the view of the parser proposed here.
| Phillips, Colin: Right association in parsing and grammar. In: Schütze, Carson; Ganger, Jennifer; Broihier, Keven (Ed.): vol. Papers on Language Acquisition and Proce, pp. 37–93, Department of Linguistics and Philosophy, Massachusetts Institute of Technology, vol. 26, 1995. (Type: Book Chapter | | )|