Research

Dates: Wed, 08/09/2006 - 12:00

This dissertation addresses a foundational debate regarding the role of structure and abstraction in linguistic representation, focusing on representations at the lexical level. Under one set of views, positing abstract morphologically-structured representations, words are decomposable into morpheme-level basic units; however, alternative views now challenge the need for abstract structured representation in lexical representation, claiming non-morphological whole-word storage and processing either across-the-board or depending on factors like transparency/productivity/surface form. Our cross-method/cross-linguistic results regarding morphological-level decomposition argue for initial, automatic decomposition, regardless of factors like semantic transparency, surface formal overlap, word frequency, and productivity, contrary to alternative views of the lexicon positing non-decomposition for some or all complex words. Using simultaneous lexical decision and time-sensitive brain activity measurements from magnetoencephalography (MEG), we demonstrate effects of initial, automatic access to morphemic constituents of compounds, regardless of whole-word frequency, lexicalization and length, both in the psychophysical measure (response time) and in the MEG component indexing initial lexical activation (M350), which we also utilize to test distinctions in lexical representation among ambiguous words in a further experiment. Two masked priming studies further demonstrate automatic decomposition of compounds into morphemic constituents, showing equivalent facilitation regardless of semantic transparency. A fragment-priming study with spoken Japanese compounds argues that compounds indeed activate morphemic candidates, even when the surface form of a spoken compound fragment segmentally-mismatches its potential underlying morpheme completion due to a morpho-phonological alternation (rendaku), whereas simplex words do not facilitate segment-mismatching continuations, supporting morphological structure-based prediction regardless of surface-form overlap. A masked priming study on productive and non-productive Japanese de-adjectival nominal derivations shows priming of constituents regardless of productivity, and provides evidence that affixes have independent morphological-level representations. The results together argue that the morpheme, not the word, is the basic unit of lexical processing, supporting a view of lexical representations in which there are abstract morphemes, and revealing immediate, automatic decomposition regardless of semantic transparency, morphological productivity, and surface formal overlap, counter to views in which some/all complex words are treated as unanalyzed wholes. Instead, we conclude that morphologically-complex words are decomposed into abstract morphemic units immediately and automatically by rule, not by exception.

Dates: Wed, 08/09/2006 - 12:00

The dissertation consists of theoretical and experimental studies of multiple interrogatives (i.e., sentences containing more than one wh-phrase, like Who bought what?). First, I examine the status of Superiority effects in contexts with and without subject-aux(iliary) inversion cross-linguistically. The relevant contrast from English is between Who bought what?, ??What did who buy?, and *I wonder what who bought., where (*) indicates a greater degree of unacceptability by native speakers than (??). I argue that the presence of subject-aux inversion in main clauses in English is responsible for the given asymmetry, and I attribute the degraded status of ??What did who buy? to the independent semantic properties of questions. Next, I explore the semantic properties of multiple interrogatives in detail. I develop an analysis that does not rely on covert wh-movement, relying instead on the syntactic position of the Question morpheme. I also explore the nature of complex wh-phrases (e.g., what boy, which book). I propose that choice functions are part of complex wh-phrases but not bare wh-phrases. I then explore the behavior of multiple interrogatives under Sluicing (i.e., clausal ellipsis). I observe that, in Slavic, it is possible to have multiple wh-phrases as well as focused referential expressions as remnants of sluicing. Based on this data, I argue that clausal ellipsis is licensed by focus in general. I also explore the apparent Superiority effects under sluicing in Russian and Polish and conclude that those are, in fact, parallelism effects, and not minimality effects. Finally, I present the results of several language acquisition studies on at what age and how English-, Russian-, and Malayalam- speaking children acquire the language-specific syntactic and semantic properties of multiple interrogatives, given the limited evidence in the input. I report the results of the corpus studies of parental speech with respect to the frequency of occurrence of multiple interrogatives, as well as the results of the studies, where multiple interrogatives were elicited from children and adults in specific contexts. I conclude that young children acquire syntax and semantics of multiple interrogatives quite successfully. I then discuss what evidence in the input they might be using.

Dates: Fri, 08/04/2006 - 12:00

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.

Dates: Fri, 07/14/2006 - 12:00

This dissertation discusses children's understanding of semantic contribution of the universal quantifier every and its interactions with negation from a logico-semantic perspective. The universal quantifier every evokes an asymmetric entailment pattern in its first and second arguments (e.g., Ladusaw, 1979), which influences both sentential meanings and inferential relations among them. Whereas several studies have revealed children's sophisticated ability to compose the meanings of sentences containing every (e.g., Gualmini 2005), far less is known about whether children's knowledge about every can be extended to the level of meaning comparison, i.e., to the computation of the inferential relations among every-sentences. We thus investigate whether children are able to apply their knowledge about every to the calculation of the inference relations between every-sentences. In particular, this dissertation aims to experimentally examine children's ability to evaluate the inferences between every-sentences. We first report an experiment featuring the Truth Value Judgment Task (e.g., Crain and Thornton 1998), reconfirming children's adult-like ability to compose individual sentence meanings involving every. We then introduce two novel experimental methodologies, the Prediction-Rephrase Task and the Demand-Fulfillment Task, designed to assess children's ability to evaluate inferences between the entailing and entailed sentences. Three experiments utilizing these new tasks demonstrate that children's highly sophisticated knowledge about every is appropriately applied in comparing meaning relations involving every across-propositionally. Additionally, we present experiments that reveal children's adult-like knowledge regarding the semantic interaction between every and negation (e.g., Ludlow 2002) in both composing and comparing sentential meanings; these findings provide the evidence showing children's adult-like linguistic representations of the sentences, in which the structural relation between every and negation determines the patterns of inferences. Taken together, these studies demonstrate children's adult-like knowledge regarding the semantics of every that is applied both in the composition and comparison of sentential meanings, as well as their adult-like knowledge about the interaction between every and negation. In addition, our development of the two new experimental methodologies has made possible further steps toward the full understanding of semantic competence in child language, not only at the level of meaning composition but also at the level of meaning comparison.

Dates: Thu, 04/27/2006 - 12:00

This dissertation investigates aspects of auditory scene analysis such as the detection of a new object in the environment. Specifically I try to learn about these processes by studying the temporal dynamics of magnetic signals recorded from outside the scalp of human listeners, and comparing these dynamics with psychophysical measures. In total nine behavioral and Magneto-encephalography (MEG) brain-imaging experiments are reported. These studies relate to the extraction of tonal targets from background noise and the detection of change within ongoing sounds. The MEG deflections we observe between 50-200 ms post transition reflect the first stages of perceptual organization. I interpret the temporal dynamics of these responses in terms of activation of cortical systems that participate in the detection of acoustic events and the discrimination of targets from backgrounds. The data shed light on the statistical heuristics with which our brains sample, represent, and detect changes in the world, including changes that are not the immediate focus of attention. In particular, the asymmetry of responses to transitions between 'order' and 'disorder' within a stimulus can be interpreted in terms of different requirements for temporal integration. The similarity of these transition-responses with commonly observed onset M50 and M100 auditory-evoked fields allows us to suggest a hypothesis as to their underlying functional significance, which so far has remained unclear. The comparison of MEG and psychophysics demonstrates a striking dissociation between higher level mechanisms related to conscious detection and the lower-level, pre-attentive cortical mechanisms that sub-serve the early organization of auditory information. The implications of these data for the processes that underlie the creation of perceptual representations are discussed. A comparison of the behavior of normal and dyslexic subjects in a tone-in-noise detection task revealed a general difficulty in extracting tonal objects from background noise, manifested by a globally delayed detection speed, associated with dyslexia. This finding may enable us to tease apart the physiological and behavioral corollaries of these early, pre-attentive processes. In conclusion, the sum of these results suggests that the combination of behavioral and MEG investigative tools can provide new insights into the processes by which perceptual representations emerge from sensory input.

Dates: Wed, 04/26/2006 - 12:00

The present thesis is concerned with the syntax of constructions variously referred to as 'applicative', 'ditransitive', or 'multiple object' constructions: constructions that contain arguments that transcend the traditional subject-object characterization. The present thesis is also concerned with how the syntax of such constructions yields the interpretive effects that previous research has identified. In this thesis I try to remedy the inadequacies and limitations of previous accounts. As far as the syntax of applicatives is concerned, my analysis necessitates the rejection of phase-based derivation, and requires an emphasis on anti-locality, a rethinking of the phenomenon of successive cyclicity, and a renewed appreciation for the relevance of case and category in the context of multiple object constructions. The system I end up with is more relativized than previous accounts, as it makes use of more factors to capture the syntax of applicatives. In addition to providing a more adequate chracterization of the syntax of applicative constructions, I develop a semantic analysis of double-object/low applicative constructions. Specifically, I argue that such constructions involve object-sharing, captured via theta-driven movement, a derivational process that they share with serial verbs and resultative constructions. If correct, the present thesis offers empirical arguments for various theoretical options currently entertained in the minimalist program, among which movement into theta-position, multiple agree, anti-locality, and early successive cyclic movement (i.e., movement taking place before the final landing site is introduced into the structure).

Dates: Tue, 04/18/2006 - 12:00

A number of recent electrophysiological studies of sentence processing have shown that a subclass of syntactic violations elicits very rapid ERP responses, occurring within around 200 ms of the onset of the violation. Such ﬁndings raise the question of how it is possible to diagnose violations so quickly. This paper suggests that very rapid diagnosis of errors is possible speciﬁcally in situations where the diagnosis problem is tightly constrained by speciﬁc expectations generated before the critical word is presented. In an event-related potentials (ERP) study of visual sentence reading participants encountered violations of a word order constraint (...Max’s of...) that has elicited early ERP responses in previous studies. Across conditions the illicit sequence was held constant, while sentence context was used to manipulate the expectation for a noun following the possessor Max’s, by manipulating the possibility of ellipsis of the head noun. Results showed that the anterior negativity elicited by the word category violation was attenuated when the availability of ellipsis reduced the expectation for a noun in the position of the oﬀending preposition of, with divergence between conditions starting around 200 ms after the onset of the violation. This suggests a role for structural expectations in accounting for very fast syntactic diagnosis processes.

Dates: Mon, 12/12/2005 - 12:00

In this thesis, I propose a syntactic structure for verbs which directly encodes their event complexities. I present a model that is 'internalist' in the Chomskyan sense: Aktionsart properties of predicates are not a real-world affair, but the interpretation of a mind structure. For this purpose, I base my proposal on the Dimensional Theory of Uriagereka (2005, forthcoming). Syntactic constructs are in this view the results of operations that create increasingly complex objects, based on an algorithm that is homo-morphic with the structure of numerical categories. First, I propose that Aktionsart can be read off from structural complexities of syntactic objects and their associated 'theta-roles'. Specifically, I present the SAAC Hypothesis: Syntactic complexity in a verb is reflected in the number of syntactic arguments it takes. This approach, within the confines of the Dimensional Theory, results in an emergent 'thematic hierarchy': Causer > Agent > Locative > Goal > Theme. I test the accuracy of this hierarchy and concomitant assumptions through paradigms like the control of implicit arguments, selectional properties of verbs, extractions, aspect-sensitive adverbials, etc. Second, I argue that the verbal structure I propose is syntactically and semantically real, by extending the proposal in Lasnik (1999) on VP ellipsis from inflectional to derivational morphology. I discuss two contrasting methods of morphological amalgamation in English and Japanese, executed in Syntax and PF, respectively. This demonstrates a tight network of entailment patterns that holds of verbs, derived crucially from the architecture I argue for. Third, an analogous point is made through the structural positionings of causative and inchoative derivational morphemes in Japanese. There, each order of structural complexity has a profound impact on the class of eventualities a derivational morpheme can describe. 'Dimensional talks' are observed between certain derivational morphemes, which presumably find their roots in operations of the computational system within the Dimensional Theory. I show that the verbal structure in Japanese reflects directly an underlying bi-clausality that I argue for, in terms of derivational morphemes, further supporting a natural mapping between syntax and semantics. This is, in the end, an attempt for a 'Minimalist' theory of Aktionsart.

Dates: Thu, 12/08/2005 - 12:00

Turkish relative clauses display a subject/non-subject asymmetry. The subject relative (SR) is licensed for relativization from [Spec, TP]. Whereas the non-subject relative (NSR) is never acceptable for subject relativization, the SR is licensed in clauses where there is no external argument, and when relativizing a non-subject in clauses where the subject is non-specific. Within the framework of the Minimalist Program, Turkish RCs are explained in terms of satisfaction of the EPP of T by a D feature and Minimality effects. As long as no nominal expression intervenes between the relative head and [Spec, TP], the SR is licensed. The SR, then, can be used as a diagnostic for movement through TP. Minimality effects are incurred when there is an intervening nominal between T° and the RC head, and the SR becomes unacceptable. The proposal is that in Turkish, specific nominals, +human nominals, and Experiencers of psych verbs all contain a DP projection. Non-specifics are NPs which cannot satisfy the EPP. NP subjects cannot move to [Spec, TP], and thus permit the SR form for relativization of non-subjects. NPs create intervention effects, as does PRO, with the exception of subject control PRO which is perhaps a trace of movement. Scrambling ameliorates intervention effects. Once scrambled, expressions are frozen but remain porous for movement of a subconstituent. Differences between inherent and structural Case are suggested with structural case assignment limited to DPs and in a Spec-Head configuration. Structurally case-marked DPs are barred from moving to case-assigning positions unless there is a morphological match. Further proposals include structures for verb classes, including Psych verbs, and structures for infinitivals and +human DPs. Contrastive focus is briefly addressed. Though superficially complex, relativization in Turkish can be accounted for with a minimum of technology. The suggestions here have implications for the theory of the EPP, Case, its assignment and interface conditions, feature satisfaction, and movement.

Dates: Fri, 12/02/2005 - 12:00

Until recently the focus of the Natural Language Processing (NLP) community has been on a handful of mostly European languages. However, the rapid changes taking place in the economic and political climate of the world precipitate a similar change to the relative importance given to various languages. The importance of rapidly acquiring NLP resources and computational capabilities in new languages is widely accepted. Statistical NLP models have a distinct advantage over rule-based methods in achieving this goal since they require far less manual labor. However, statistical methods require two fundamental resources for training: (1) online corpora (2) manual annotations. Creating these two resources can be as difficult as porting rule-based methods. This thesis demonstrates the feasibility of acquiring both corpora and annotations by exploiting existing resources for well-studied languages. Basic resources for new languages can be acquired in a rapid and cost-effective manner by utilizing existing resources cross-lingually. Currently, the most viable method of obtaining online corpora is converting existing printed text into electronic form using Optical Character Recognition (OCR). Unfortunately, a language that lacks online corpora most likely lacks OCR as well. We tackle this problem by taking an existing OCR system that was desgined for a specific language and using that OCR system for a language with a similar script. We present a generative OCR model that allows us to post-process output from a non-native OCR system to achieve accuracy close to, or better than, a native one. Furthermore, we show that the performance of a native or trained OCR system can be improved by the same method. Next, we demonstrate cross-utilization of annotations on treebanks. We present an algorithm that projects dependency trees across parallel corpora. We also show that a reasonable quality treebank can be generated by combining projection with a small amount of language-specific post-processing. The projected treebank allows us to train a parser that performs comparably to a parser trained on manually generated data.

Featured Research News

Mayfest 2025 - Constraints on Meaning

Lexical Structure and the Nature of Linguistic Representations

An MEG study arguing that the morpheme and not the word, is the basic unit of lexical processing.

Multiple Interrogatives: Syntax, Semantics, and Learnability

On the syntax, semantics and acquisition of multiple interrogatives, cross-linguistically.

Pre-verbal Structure Building in Romance Languages and Basque

An experimental study of preverbal cues to predicting verbal structure in Galician, Spanish and Basque.

Everyone Knows, Therefore Every Child Knows: An Investigation of Logico-semantic Competence in Child Language

Children's understanding of scope interactions between negation and universal quantification.

Auditory edge detection: the dynamics of the construction of auditory perceptual representations

An MEG study of auditory scene analysis.

The Landscape of Applicatives

A minimalist analysis of ditransitives and applicatives.

The role of prediction in rapid syntactic analysis

ERP study shows rapid predictive use of structural expectations in online comprehension.

A Syntactic Structure of Lexical Verbs

A minimalist theory of verbal aspect and argument structure.

Minimality and Turkish Relative Clauses

An analysis of subject/object asymmetries in the syntax of Turkish relative clauses.

Rapid resource transfer for multilingual natural language processing

How to build annotated corpora for less-studied languages by exploiting existing resources for well-studied languages.