Yukiko Sasaki Alam

Decision trees for sense disambiguation of prepositions: case ofover

This paper proposes two decision trees for determining the meanings of the prepositional uses of over by using the contextual information. It first examines the meanings of the prepositional uses of over and then aims at identifying the contexts for interpreting the meanings. Some contexts are complementary features, and that makes the decision trees simple. The trees have been tested on a corpus, and the results are encouraging.

Picture-Based e-Learning of English Phrasal Verbs: A Comparative Study of the Use of Pictures and L1 Glosses

English phrasal verbs such as ‘call off’ (a game due to rain) and (the buttons) ‘come off’ are difficult for non-native speakers to learn because the meanings of many phrasal verbs are not compositions of the meanings of each component. It was reported that the use of phrasal verbs was avoided by non-native speakers and that the mastery of collocations was hampered by a strong influence from the learners native language. Therefore, to prevent such interference from learners’ native languages, it would be appropriate to learn phrasal verbs directly through nonverbal media such as pictures and sound. A web application was developed for the picture-based e-learning of phrasal verbs. An experiment was conducted, which consisted of a sequence of two sessions with a week-long break in between. In order to investigate the learning effectiveness, two methods were compared: one with the display of meaning with still pictures, and the other with Japanese word glosses. Two groups participated in the experiment, and...

Algorithm for Identifying the Syntactic and Semantic Categories of Prepositions: Case of Over

This paper proposes an algorithm for assigning the syntactic categories of over, many uses of which are not used as prepositions. The algorithm, enriched for the semantic capacity from earlier studies, identifies four syntactic categories of over and eleven meanings of the prepositional uses. The ability of the algorithm was tested manually by using five hundred instances of over from British National Corpus. The results are encouraging, with over 95 percent of the instances being correctly classified. This study, while pointing to an ideal direction, will reveal many important points to consider in natural language processing

A Software System for Second-Language Vocabulary Learning

The aim of this paper is to introduce a software system for learning English vocabulary. It is a multiuser and multi-platform system, designed on effective learning principles and available for use on the Internet. It provides graphical displays of goals for the student to attain, and of the results that have been achieved. It also offers the instructor a simple means of preparing and editing exercises and modifying them according to pedagogical needs. The system, still at an evolutionary stage, was implemented for an assignment in English courses at a Japanese university for two semesters, and has proved to be an effective tool in teaching second-language vocabulary while a rigorous assessment remains to be made.

Seeing Trees in a Forest for Improving Syntactic and Semantic Parsing

This paper aims to find out what knowledge is needed to improve the quality of syntactic and semantic parsing by manually examining and analyzing individual machine translation output errors that involve syntax.

E-learning of English Phrasal Verbs via Pictorial Elucidation and L1 Glosses

This paper reports on a comparative study of e-learning of English phrasal verbs on the picture and L1 gloss mediums. The results showed that more participants on the picture medium experienced nonverbal conceptual links with concepts, and processed information faster than those on the L1 gloss medium, which seems to suggest important implications for the Dual Coding theory. While the picture medium brought relatively better results than the L1 gloss medium, and was favored by the participants, it alone was not sufficient, and should be supported by other mediums for significant achievement. This study also reminded us of a need for a careful selection of phrasal verbs, and the use of good quality pictures and a variety of question formats.

Algorithms for Identifying the Multiple Syntactic Categories and Meanings of the Word Over

The word over, among others, is associated with a great variety of syntactic categories and meanings. Although over has received attention from scholars in different frameworks for various aspects, there is little research, to my knowledge, to take a comprehensive approach to both syntax and semantics of the word over. This article proposes algorithms for identifying from the context a case of over with one of four syntactic categories and with one of seventeen meanings. The test was carried out manually on five hundred instances of over from British National Corpus. The results are promising, with 95 percent of the classification of the instances being correct. This study, while taking an idealistic approach, brings to light methods that may stretch the limits of natural language processing.

The Organization of the Lexicon: The Polysemy of Grow and Disambiguation

This paper demonstrates that the polysemy of the verb grow is a result of natural extension of individual meanings connoted by its basic literal meaning and that the polysemy of grow, as such, can be disambiguated by applying simple rules of elimination to the argument structures, which are the contexts that make particular senses viable.