Subcategorization Information Research Articles

Cross-linguistic differences in predicting L2 sentence structure: The use of categorical and gradient verb constraints

AbstractThis study investigates whether cross-linguistic differences affect how adult second language (L2) learners use different types of verb subcategorization information for prediction in real-time sentence comprehension. Using visual world eye-tracking, we tested if first language (L1) German and L1 Turkish intermediate-to-advanced learners of L2 English make use of categorical and gradient probabilistic selectional information of ditransitive verbs to predict whether the verbs would be followed by prepositional-object or double-object dative constructions. L1 German learners used both categorical (“pay/*donate the woman the money”) and gradient (“pay/#send the woman the money”) constraints for prediction in a target-like manner. In contrast, L1 Turkish learners were delayed in recruiting categorical verb information and were only selectively sensitive to gradient verb information. We argue that target-like predictive processing across categorical and gradient verb information is attainable for L2 learners, but differences in L1-L2 word order may curtail the utility of prediction by verb subcategorization information in L2 processing.

L2 Learners Do Not Ignore Verb's Subcategorization Information in Real-Time Syntactic Processing.

This study addressed the question of whether L2 learners are able to utilize verb’s argument structure information in online structural analysis. Previous L2 research has shown that L2 learners have difficulty in using verb’s intransitive information to guide online syntactic processing. This is true even though L2 learners have grammatical knowledge that is correct and similar to that of native speakers. In the present study, we contrasted three hypotheses, the initial inaccessibility account, the intransitivity overriding account, and the fuzzy subcategorization frame account, to investigate whether L2 learner’s knowledge of intransitive verbs is in fact ignored in L2 online structural analysis. The initial inaccessibility account and the fuzzy subcategorization frame account predicted that L2 learners cannot access intransitivity information in building syntactic structures in any situation. The intransitivity overriding account predicted that intransitivity information is accessed in L2 parsing, but this process is overridden by the strong transitivity preference when a verb is followed by a noun phrase. Importantly, the intransitivity overriding account specifically predicted that L2 learners would be able to use intransitive information in online syntactic processing when a noun phrase does not appear immediately following a verb. We tested the three accounts in an eye-tracking reading experiment using filler-gap dependency structures. We manipulated verb’s transitivity information and lexically based plausibility information and tested English native speakers as a control L1 group (N = 29) and Japanese-English L2 participants (N = 32). The results showed that L2 learners as well as native speakers processed sentences differently depending on the subcategorization information of the verb, and adopted transitive analysis only when the verb was optionally transitive, providing support for the intransitivity overriding. The results further demonstrated that L2 learners had strong expectations for the transitive structure, which is consistent with the view proposed by the hyper-active gap-filling hypothesis. In addition, the results showed that the semantic mismatch in the incorrect transitive analysis facilitated native speaker’s processing but caused difficulty for L2 learners. Together, the current study provides evidence that L2 learners use intransitive information of the verbs to guide their structural analysis when there are no overriding constraints.

Bilinguals on the garden-path: Individual differences in syntactic ambiguity resolution.

Syntactic parsing plays a central role in the interpretation of sentences, but it is unclear to what extent non-native speakers can deploy native-like grammatical knowledge during online comprehension. The current eye-tracking study investigated how Chinese-English bilinguals and native English speakers respond to syntactic category and subcategorization information while reading sentences with OBJECT-SUBJECT ambiguities. We also obtained measures of English language experience, working memory capacity, and executive function to determine how these cognitive variables influence online parsing. During reading, monolinguals and bilinguals showed similar GARDEN-PATH EFFECTS related to syntactic reanalysis, but native English speakers responded more robustly to VERB SUBCATEGORIZATION cues. Readers with greater language experience and executive function showed increased sensitivity to verb subcategorization cues, but parsing was not influenced by working memory capacity. These results are consistent with exposure-based accounts of bilingual sentence processing, and they support a link between syntactic processing and domain-general cognitive control.

Integration of Verb Subcategorization Information in L2 English Sentence Processing by Korean Learners : An Eye-Tracking Study

The present study tested whether adult Korean learners of English are able to immediately integrate verb-specific structural information (i.e. subcategorization restrictions imposed by individual verbs) as they read sentences in English. Seventeen native speakers of English and 22 Korean learners of English participated in an eye-tracking experiment, in which they read English sentences that contained a wh-dependency between the head NP and the verb of a relative clause. The relative clause verbs consisted of three types that varied in terms of subcategorization properties: transitive verbs that take an NP complement more often than a PP complement, transitive verbs that take a PP complement more often than an NP complement, and unaccusative verbs that are strictly intransitive. The results suggest that whereas native speakers use verb subcategorization information in making initial parsing decisions regarding wh-dependency formation, Korean learners are only able to integrate the information in later stages of sentence processing. Possible accounts of this discrepancy in the time course of verb information integration are discussed.

Difference between L1 and L2 Language Processing in the Use of Subcategorization Information Evidence form Syntactic Priming

Difference between L1 and L2 Language Processing in the Use of Subcategorization Information Evidence form Syntactic Priming

A Comparative Study of Native and Non-native Speakers’ Use of Verb Subcategorization Information in Written English Production

The present study investigated how Korean L2 learners of English used their knowledge of verb subcategorization information in written production....

Verb Subcategorization Information During Sentence Comprehension by Japanese EFL Learners : Evidence From Self-Paced Sentence Anomaly Task

Verb Subcategorization Information During Sentence Comprehension by Japanese EFL Learners : Evidence From Self-Paced Sentence Anomaly Task

세종 전자사전과 준지도식 학습 방법을 이용한 용언의 어의 중의성 해소

The Sejong Electronic(machine-readable) Dictionary, developed by the 21st century Sejong Plan, contains systematically organized information on Korean words. It helps to solve problems encountered in the electronic formatting of the still-commonly-used hard-copy dictionary. The Sejong Electronic Dictionary, however has a limitation relate to sentence structure and selection-restricted nouns. This paper discuses the limitations of word-sense disambiguation(WSD) that uses subcategorization information suggested by the Sejong Electronic Dictionary and generalized selection-restricted nouns from the Korean Lexico-semantic network. An alternative method that utilized semi-supervised learning, the chi-square test and some other means to make WSD decisions is presented herein.

한국어 어휘의미망(UWordMap)을 이용한 동형이의어 분별 개선

Disambiguation of homographs is an important job in Korean semantic processing and has been researched for long time. Recently, machine learning approaches have demonstrated good results in accuracy and speed. Other knowledge-based approaches are being researched for untrained words. This paper proposes a hybrid method based on the machine learning approach that uses a lexical semantic network. The use of a hybrid approach creates an additional corpus from subcategorization information and trains this additional corpus. A homograph tagging phase uses the hypernym of the homograph and an additional corpus. Experimentation with the Sejong Corpus and UWordMap demonstrates the hybrid method is to be effective with an increase in accuracy from 96.51% to 96.52%.

A Corpus-based Study of To-Infinitive Errors in Korean College Freshmen’s Writing

This study aims to investigate how accurately to-infinitives are used by Korean learners of English who have completed primary and secondary education in Korea. Data were retrieved from a learner corpus consisting of essays written by 815 incoming freshmen at a university in Seoul, Korea. Of the 2,309 tokens of to-infinitives from the corpus, only 171 tokens were found to be errors, a finding which suggests that Korean students with high-intermediate proficiency in English possess a firm understanding of the use of to-infinitives. In addition to those 171 error tokens, 65 tokens of omission errors were manually identified; of these 236 error tokens, substitution was the most common type of error (103 tokens, 43.6%), which was closely followed by omission errors (99 tokens, 42.0%). A close examination of all the errors revealed that several types of errors derive from the learner’s lack of knowledge regarding subcategorization, i.e., information about selecting proper complementation types. Teachers should therefore provide L2 learners with detailed subcategorization information, especially when new verbs are introduced.

세종 전자사전과 한국어 어휘의미망을 이용한 용언의 어의 중의성 해소

The Sejong Electronic(machine readable) Dictionary, which was developed by the 21 century Sejong Plan, contains a systematic of immanence information of Korean words. It helps in solving the problem of electronical presentation of a general text dictionary commonly used. Word sense disambiguation problems can also be solved using the specific information available in the Sejong Electronic Dictionary. However, the Sejong Electronic Dictionary has a limitation of suggesting structure of sentences and selection-restricted nouns. In this paper, we discuss limitations of word sense disambiguation by using subcategorization information as suggested by the Sejong Electronic Dictionary and generalize selection-restricted noun of argument using Korean Lexico-semantic network.

Patrixa: a unification-based parser for Basque and its aplication to the automatic analysis of verbs

In this paper we describe a computational grammar for Basque, and the first results obtained using it in the process of automatically acquiring subcategorization information about verbs and their associated sentence elements (arguments and adjuncts). The first part of this article (section 1) will be devoted to the description of Basque syntax, and to present the grammar we have developed. The grammar is partial in the sense that it cannot recognize every sentence in real texts, but it is capable of describing the main syntactic elements, such as noun-phrases (NPs), prepositional phrases (PPs), and subordinate and simple sentences. This can be useful for several applications. Next, the syntactic grammar will be used by a syntactic analyzer (or parser) to automatically acquire information on verbal subcategorization from texts (section 2). The results will later be used by a linguist or processed by statistical filters.

The use of verb-specific information for prediction in sentence processing

Recent research has shown that language comprehenders make predictions about upcoming linguistic information. These studies demonstrate that the processor not only analyses the input that it received but also predicts upcoming unseen elements. Two visual world experiments were conducted to examine the type of syntactic information this prediction process has access to. Experiment 1 examined whether the verb's subcategorisation information is used for predicting a direct object, by comparing transitive verbs (e.g., punish) to intransitive verbs (e.g., disagree). Experiment 2 examined whether verb frequency information is used for predicting a reduced relative clause by contrasting verbs that are infrequent in the past participle form (e.g., watch) with ones that are frequent in that form (e.g., record). Both experiments showed that comprehenders used lexically specific syntactic information to predict upcoming syntactic structure; this information can be used to avoid garden paths in certain cases, as Experiment 2 demonstrated.

ERPs reveal comparable syntactic sentence processing in native and non-native readers of English

L2 syntactic processing has been primarily investigated in the context of syntactic anomaly detection, but only sparsely with syntactic ambiguity. In the field of event-related potentials (ERPs) syntactic anomaly detection and syntactic ambiguity resolution is linked to the P600. The current ERP experiment examined L2 syntactic processing in highly proficient L1 Spanish-L2 English readers who had acquired English informally around the age of 5 years. Temporary syntactic ambiguity (induced by verb subcategorization information) was tested as a language-specific phenomenon of L2, while syntactic anomaly resulted from phrase structure constraints that are similar in L1 and L2. Participants judged whether a sentence was syntactically acceptable or not. Native readers of English showed a P600 in the temporary syntactically ambiguous and syntactically anomalous sentences. A comparable picture emerged in the non-native readers of English. Both critical syntactic conditions elicited a P600, however, the distribution and latency of the P600 varied in the syntactic anomaly condition. The results clearly show that early acquisition of L2 syntactic knowledge leads to comparable online sensitivity towards temporal syntactic ambiguity and syntactic anomaly in early and highly proficient non-native readers of English and native readers of English.

Inducing Fuzzy Classes for Chinese Polysemic Verbs via Subcategorization Information

This paper describes the application of Fuzzy k-Means, a derivant of k-Means that may assign an item to more than one cluster, in the task of inducing fuzzy classes for Chinese polysemic verbs. The probability distributions over subcategorization frames of 60 Chinese verbs, among which there are 40 polysemic ones and 20 monosemic ones are first acquired, and then these verbs are clustered into fuzzy classes. Evaluation and post-hoc analysis show that a combined measure of purity and pairwise precision can better estimate the clustering performance, and although to a certain extent syntactic behaviors of verbs have their counterparts of meaning components underlying, syntactic behaviors of verbs cannot be easily predicted from a single semantic level, at least for Chinese polysemic verbs.

Automatic acquisition of verb subcategorization information by exploiting mininal linguistic resources

A set of well known statistical filtering methods (binomial hypothesis testing, log-likelihood ratio, t-test, thresholds on relative frequencies) is used on Modern Greek and English corpora in order to automatically acquire verb subcategorization frames that are not limited in number and are not known beforehand. As sophisticated linguistic resources and tools are not available for most languages (including Modern Greek), pre-processing of our corpora reaches merely the stage of elementary, intrasentential, non-embedded phrase chunking. By forming, permutating and counting subsets of the verb's neighboring set of phrases, and by applying the statistical filters mentioned previously, valid syntactic frames of verbs are detected. The results achieved were comparable to and, in several cases, better than the ones of previous approaches, even approaches utilizing richer resources. Incorporating the extracted list of frames into a shallow parser, the performance of the latter increases by almost 6%, showing thereby the importance of the acquired knowledge.

Application of finite-state transducers to the acquisition of verb subcategorization information

This paper presents the design and implementation of a finite-state syntactic grammar of Basque that has been used with the objective of extracting information about verb subcategorization instances from newspaper texts. After a partial parser has built basic syntactic units such as noun phrases, prepositional phrases, and sentential complements, a finite-state parser performs syntactic disambiguation, determination of clause boundaries and filtering of the results, in order to obtain a verb occurrence together with its associated syntactic components, either complements or adjuncts. The set of occurrences for each verb is then filtered by statistical measures that distinguish arguments from adjuncts.

Lexical guidance in sentence processing: a note on Adams, Clifton, and Mitchell (1998).

Mitchell (1987) conducted a self-paced reading experiment that showed that people experienced difficulty reading a noun phrase when it immediately followed an intransitive verb. From this, he argued for a two-stage theory of parsing, in which verb subcategorization information is initially ignored. In response, Adams, Clifton, and Mitchell (1998) found no evidence to support this claim in an eye-tracking experiment and argued that Mitchell's segmentation procedure distorted the parsing process. We report an eye-tracking experiment, in which materials similar to those in Adams et al., but with longer noun phrases, were used, that showed a pattern of difficulty similar to Mitchell's findings. Hence, Mitchell's results did not depend on the use of an artificial method of presentation. The results cast further doubt on the adequacy of constraint-based accounts of parsing.

Dictionary-Driven Semantic Look-up

1. IntroductionThe French Semantic Dictionary Look-up (SDL) uses dictionary information aboutsubcategorization and collocates to perform Word Sense Disambiguation (WSD).The SDL is fully integrated in a multilingual comprehension system which uses theOxford Hachette French-English bilingual dictionary (OUP-H). Although the SDLworks on all words both for French and English, Romanseval results are relevantfor French verbs only because subcategorisation and collocate information is richerfor this part of speech in the OUP-H. The SDL uses dictionaries as semanticallytagged corpora of different languages, making the methodology reusable for anylanguage with existing on-line dictionaries.This paper first describes the system architecture as well as its components andresources. Second, it presents the work we did within Romanseval, namely sensemapping and results analysis.2. Semantic Dictionary Look-Up: Goal, Architecture and ComponentsThe SDL selects the most appropriate translation of a word appearing in a givencontext. It reorders dictionary entries making use of dictionary information. It isbuilt on top of Locolex,

Lexical guidance in sentence processing?

An eyetracking experiment was conducted to explore a self-paced reading effect reported by Mitchell (1987). Mitchell found that a noun phrase (NP) was read slowly when it immediately followed an intransitive verb, as long as the verb and NP appeared in the same presentation region. This effect has been used to support the claim that verb subcategorization information is not used initially in sentence parsing. However, the effect did not appear in the eyetracking experiment reported in the present paper, supporting criticisms that Mitchell’s segmentation procedure distorted the parsing process.