Postgraduate Workshop: Linguistics and Language Processing
                                            The University of Queensland, Friday 4 October, 2002.

Call for paper       Presentation Guidelines   Registration    Program    Venue    Linguistics at UQ

Venue: Hawken Engineering Building (building#50)

The Official Welcome: room N201, the workshop: Lecture room 2

Time

Speakers

 Details

Chaired by:

8:30-9:00

Registration

9:00-9:30

Presidents, EMAPS

Prof. Gillian Whitlock, Director Postgraduate Studies, Faculty of Arts

Prof. Alan Lawson, Director, Graduate School

 

Official Welcome

 

 

 

 

 

 

Sacha DeVelle

 

9:30-9:50 Dr. Mary Laughren  Introduction to the workshop
9:50-10:30

Guest speaker:

Dr. Andrew Smith, ARC Key Centre for Human Factors and Applied Cognitive Psychology

Leximancer: A Statistical Path from Words to Meaning
10:30-11:00 Julie Steele The Emergence of Tense
11:00-11:30 Morning Tea
11:30-12:00 Sacha DeVelle Aspectual coercion and on-line processing: Iteration or repair?

Julie Steele

12:00-12:30 Elizabeth Cardell

Well-formedness judgement in Broca’s aphasia: Comparing offline and online performance.

12:30-1:00 Hung PhanVan An English-Vietnamese Machine Translation Web Service.
1:00-2:00 Lunch
2:00-2:30 Baden Hughes Explorations in the Classification and Presentation of Linguistic Data (with special reference to the Flint Archives) Christina Pentland
2:30-3:00 Wenying Jiang

Chinese Students' Perceptions of EFL Teaching in a British University

3:00-3:30 Syd Gould

Phonological Occidentalism: second language interference and the generation of reading difficulties in the mother tongue: a case study

3:30-4:00

Afternoon tea

4:00-4:30 Thu Nguyen The tonal constraints on Vietnamese Perception of English stress Sacha DeVelle
4:30-5:00

Christina Pentland

Prosodic constraints on connected speech processes in Warlpiri

 

 

Dr. Andrew Smith

 ARC Key Centre for Human Factors and Applied Cognitive Psychology

The University of Queensland.  

 

Leximancer: A Statistical Path from Words to Meaning

Leximancer uses statistical word co-ocurrence to discover a surprising amount of information from textual data. The work presented here will describe some preliminary explorations into Chaucer's The Knight's Tale. In particular, an interlinear translation of this work is examined to quickly map the corresponding lexical usage between the original and the translation. This analysis not only builds thesauruses of corresponding terms, but the semantic map shows related terms based on similar usage. Phrasal tendencies of words are also presented.

Next, the two translations of The Knight's Tale are separated, and an analysis technique is devised which can show the alignment of terms in the two languages based on usage with respect to a kernel of common words.

 



Julie Steele

Linguistics Program

The University of Queensland.

The Emergence of Tense

The concept of a functional phrase was introduced into syntax to distinguish inflectional and auxiliary elements (functional items) from lexical items. My thesis proposal seeks to distinguish these diametric items through analysing functional items as emergent structures, which arise from the dynamics of pattern discovery. In this presentation I will focus on the formation of tense in language acquisition.

Crutchfield (1994a and b) has introduced an approach to avoid subjectivity in pattern discovery in which regularities are extracted from the data stream, in this case - an utterance, and the corresponding pattern is represented by a particular computational model. The pattern would be classed as intrinsically emergent as the regularity only has meaning within the system (the data and the model). Functional items would then be represented as semantic and conceptual regularities, broken from the configuration of lexical items.

Functional items would be reliant on the lexical items for their meaning and thus functionality is intrinsically emergent. Emergence is a main characteristic of a complex system and as such language would be viewed as a complex system.

References

Crutchfield, James (1994a) Is Anything Ever New? Considering Emergence Eds. G. Cowan, D. Pines and D Melzner, Integrative Themes Santa Fe Institute Studies in the Sciences of Complexity XIX Addison-Wesley

Crutchfield, James (1994b) The calculi of emergence: computation, dynamics and induction, Physica D 75:11-54


Sacha DeVelle

Linguistics Program

The University of Queensland

Aspectual coercion and on-line processing: Iteration or repair?

Aspectual coercion refers to how word meaning is combined at the sentential level. It specifically describes the semantic interplay between the aspectual properties of the verb and an adverbial modifier:

    1. The light shone for hours / until morning. (activity)
    2. The light flashed for hours / until morning. (point action)

(a) implies that the light shone in a single continuous process, due to the durative process of the activity verb. In comparison (b) implies a repetitive function due to the punctual feature of the verb flash and the bounding feature of the adverbial modifiers for and until. Recent psycholinguistic research has placed a new emphasis on this notion of repetition or iteration at the sentential level (Piñango, Zurif & Jackendoff, 1999; Todorova, Straub, Badecker & Frank, 2000). Results show that the encoding of iteration involves an increased processing load. Two models have been put forward to explain why this is so, however the processing mechanisms of iteration still remain a puzzle.

The present study is a replication and extension of Piñango et al's (1999) original sentence stimuli set. An on-line self-paced reading task specifically tested for processing differences between verb type and adverbial modifier combinations. The results are discussed and compared against the enriched composition model that advocates an extra-linguistic conceptual process (Jackendoff, 2002) and a type-shifting approach (Moens & Steedman, 1988) that results in a non-preferred reading that is then reinterpreted.

References

Jackendoff, R. (2002). Foundations of Language. Oxford: Oxford University Press.

 

Moens, M., & Steedman, M. (1988). Temporal ontology and temporal reference.

Computational Linguistics, 14(2): 15-28.

 

Piñango, M., Zurif, E., & Jackendoff, R. (1999). Real-time processing implications of enriched composition at the syntax-semantics interface. Journal of Psycholinguistic Research, 28(4): 395-414.

 

Todorova, M., Straub, K., Badecker, W., & Frank, R. (2000). Aspectual coercion and the online computation of sentential aspect. Paper presented at the Cognitive Science Society 2000, The University of Pennsylvania.


Elizabeth A. Cardell

Department of Speech Pathology & Audiology

The University of Queensland

 

Well-formedness judgement in Broca’s aphasia:

Comparing offline and online performance.

Judgments of sentence well-formedness have played an important role in attempts to understand the processing impairments underlying the agrammatic comprehension deficit. A strong claim has been made in the literature on the basis of well-formedness data, i.e., that parsing operations are essentially intact in Broca’s aphasia. An offline well-formedness judgment task and its online counterpart were conducted with six Broca’s aphasics and six neurologically-intact adults. The error-types of interest were semantic errors, inflectional (agreement) errors, and thematic-role reversal errors.

Whilst retaining above-chance sensitivity to well-formedness violations, the Broca’s aphasics’ results were significantly poorer than the control group across both the offline and online paradigms. The online results reflected the general pattern from the offline testing, but with a greater proportion of errors. Across both tasks, inflectional errors proved to be the most difficult error-type to detect for the Broca’s aphasics. Thematic–role reversal errors were also compromised, especially in more complex sentence-types. Semantic violations were generally easily detected.

The results will be discussed in relation to three prominent theoretical orientations in the agrammatic comprehension literature, namely, the Trace-Deletion Hypothesis (Grodzinsky, 1995, 2000), the Mapping Hypothesis (Linebarger, Schwartz, & Saffran, 1983), and processing limitation accounts that subscribe to activation and/or timing deficit principles. The claim that parsing processes are generally well-preserved in Broca’s aphasia will also be discussed.

References

Grodzinsky, Y. (1995). A restrictive theory of agrammatic comprehension. Brain and

Language, 50, 27-51.

 

Grodzinsky, Y. (2000). The neurology of syntax: Language use without Broca’s area.

Behavioral and Brain Sciences, 23, 1-71.

 

Linebarger, M.C., Schwartz, M.F., & Saffran, E.M. (1983). Sensitivity to grammatical

structure in so-called agrammatic aphasics. Cognition, 13, 361-392.


Hung Phanvan

School of Information Technology and Electrical Engineering

The University of Queensland

An English-Vietnamese Machine Translation Web Service

This paper presents a prototype of an English-Vietnamese Machine Translation Web Service, which is to translate web pages from English to Vietnamese at the user’s request. The translation is done sentence-by-sentence using the Example-Based Machine Translation approach with examples stored in an XML database. Machine translation has long been one of the most demanding and challenging issues in Natural Language Processing. Although Vietnamese is the 13th most spoken language in the world, very little attention has been paid to language research and processing. A large percentage of the Vietnamese population cannot understand English. There is also no online translation web service currently available so that Vietnamese people can turn to for help in order to access English web pages to gain information and knowledge.

We chose to use the Example-Based Machine Translation (EBMT) approach for some salient features. These advantages include rapidly deployable (don’t need much hard-to-access linguistic experts), easily tuneable to a particular text-type or domain, more extensible than any Rule-Based Machine Translation (RBMT) systems and also more portable to any other language-pairs. In common, most EBMT approaches get the translation for a specific input sentence of a source language by decomposing the sentence into smaller constituents, getting the constituents’ equivalence and then forming the translated sentence into the target language.

The range of proposed techniques includes: segmenting sentence using markers as segment boundaries, using morphologically analysed segments, parsing sentences into dependency tree, or building the sentence patterns on the fly. The significance of our prototype is that we are building a learning model for automatically extracting sentence examples. So, in the future it will be that the more on-line users and the longer the system is running, the better quality the system will be.


Baden Hughes

Linguistics Program EMSAH

University of Queensland

 

Explorations in the Classification and Presentation of Linguistic Data

(with special reference to the Flint Archives)


The Flint Archives consist of written documents and tape recordings compiled by Elwyn Flint predominantly in the 1960's as a part of the Queensland Speech Survey funded by the ARC. The Archives are currently held by the Fryer Library at the University of Queensland. The Flint Archive is of significance primarily because it documents Queensland's Aboriginal languages including a number which have since either died, or
are severely endangered. Although there is some complementary work by other linguists in describing and transcribing these languages, the Flint Archive for the most part has not been classified, indexed, or published in any manner.
This paper addresses a number of theoretical issues related to the classification and presentation of linguistic data, with particular reference to the Flint Archives. A survey of the classification and presentation methods used and recommended by major linguistic data archives is conducted, followed by a review of the classification and
presentation methods used by the Australian Institute of Aboriginal and Torres Strait Islander Studies, specifically for Aboriginal language data. Other approaches specific to Australian Aboriginal language data are surveyed. Finally, the existing classification and presentation methods used for the Flint Archive are evaluated, and in synthesis, a
best practice model developed for future work on this particular archive.


Wenying Jiang

School of Languages and Comparative Cultural Studies

The University of Queensland

Chinese Students' Perceptions of EFL Teaching in a British University

Many cross-cultural issues in pedagogy arise when more and more Chinese students move to the UK to study English. In order to know how Chinese students perceive EFL teaching in the UK to help underpin their cultural assumptions, a questionnaire with 31 items concerning different aspects of EFL teaching was designed for the students. Sixty three copies were collected and 47 copies were considered valid.

An interview with two female students was video recorded to get an in-depth view. The data showed that Chinese EFL students appreciated the teaching given by native English speaker teachers and the English language environment in the UK, and at the same time they felt that the EFL teaching was not always directly aiming at their needs and not really compatible with their learning style either.

This study can help EFL teachers at all levels gain insight into the cultural assumptions and values that lie behind the students' learning style. My hope is that such information will enhance the teaching of students from different cultural backgrounds, improve the assessments of their learning, and create strategies to empower both students and teachers.

References

Cortazzi, M. & Jin, L. (1999) Cultural Mirrors: Materials and methods in the EFL Classroom. In E. Hinkel (Eds.), Culture in second language teaching and learning,196-219. Cambridge: Cambridge University Press.

Fu, D. & Townsend, J. S. (1998) Cross-Cultural Dilemmas in Writing. College Teaching, 46 (4): 128-133.

Kumaravadivelu, B. (1991) Language learning tasks: Teacher intention and learner interpretation. ELT Journal, 45: 98-107.

Scollon, S. (1999) Not to waste words or students: Confucian and Socratic discourse n the tertiary classroom. In E. Hinkel (Eds.), Culture in second language teaching and learning, 13-27. Cambridge: Cambridge University Press.

Wan, G. (2001) The Learning Experience of Chinese Students in American Universities: A cross-cultural perspective. College Student Journal, 35(1): 28-45.


Syd Gould

Department of Studies in Religion

The University of Queensland

 

Phonological Occidentalism: second language interference

and the generation of reading difficulties in the mother tongue:

a case study

The significant terms this paper uses are Kenneth Pike’s neologisms, ‘etic’ and ‘emic’ (derived from the linguistic ‘phonetic/phonemic’ distinction, and widely diffused throughout the behavioural and social sciences) and the converse of Edward Said’s term ‘Orientalism’, that is, ‘Occidentalism’. For Said, ‘Orientalism’ was the particular way westerners constructed and presented oriental society and culture. Its converse, ‘Occidentalism’, then, is how easterners view and construct western society. This concept is applied in the linguistic domain, particularly to the phonology of the second language during SLA, for which I coin the term ‘phonological occidentalism’. Thus ‘phonological occidentalism’ is the phonological construction and phonemic categories subconsciously employed by a non-native speaker using that speaker’s own first language ‘emic’ grid and superimposing it upon the second language.

The case study is taken from Huli, a language of about 100,000 speakers in the Highlands of Papua New Guinea. The field research is based on nearly two decades of participant observation, the writer having been involved in both vernacular literacy in Huli and education in English. The fact that mother tongues influence the way ESL speakers perceive and produce English is widely recognized; and much has been written about the different ‘Englishes’ which have resulted. Little attention, however, has been given to the reflexivity of the process.

This paper presents evidence of how the ‘phonological occidentalism’ of native Huli speakers generates, for the Huli whose primary oracy and language acquisition is Huli, yet whose primary literacy acquisition is in English, grapho-phonic mismatches, and thus confusion for them in attempting subsequently to read their own first language (mother tongue). Particular focus will be the ‘stop consonants’ as analysed by Rule and Rule (1954) .

References

Rule, W. Murrey and Joan E. Rule. Huli Language: Statement of phonemes: Tari, SHP, PNG, APCM, 1954.


Thu Nguyen

Linguistics Program

The University of Queensland

The Tonal Constraints on Vietnamese Perception of English Stress

This experiment investigated the tonal transfer in Vietnamese perception of English stress. Ninety five Vietnamese subjects including 15 advanced Vietnamese speakers of English and 80 beginning learners of English from four regional dialects (20 of each dialect: Hanoi, Saigon, Hue and Nghe An) participated in the perception test. The subjects listened to English words and marked the English syllables with the tones they perceived.

The results indicated that an English syllable could be perceived as a certain Vietnamese tone depending on the syllable structure (a closed syllable ending in an obstruent or a syllable ending in a sonorant) and stress levels (stressed and unstressed). This suggests that there is perceptual tonal transfer, which is constrained by pitch levels and the segmental composition of the syllables.

 


Christina Pentland

Linguistics Program,

University of Queensland

Prosodic Constraints on Connected Speech Proccesses in Warlpiri

Connected speech processes (CSPs) are traditionally viewed as the consequence of reduced articulatory effort (or simplifications) that occur due to physical and communicative constraints on production - such as length of utterance, rate of speech and information load. In this sense CSPs are universal in their expression. However, cross-dialect and cross-language studies have found substantial variability in the expression of CSPs attributable to the phonological system of the speaker’s language. One source of variability derives cross-linguistically from constraints imposed by prosodic structure in blocking or promoting CSPs in various contexts. Another may lie in perceptual demands imposed by the phonological system of a given language or language family which conspire to preserve certain contrasts perhaps at the expense of others (Butcher, 1996).

The aims of this present study were (1) to identify the distribution and phonetic characteristics of CSPs in Warlpiri, a language of Central Australia, and (2) to determine the prosodic context and constraints on their expression. The study examined segmental strengthening and weakening, assimilation processes and boundary phenomena in a database of approximately 180 utterances. The data were obtained from studio recordings of two female Warlpiri speakers made by Mary Laughren at the University of Queensland in 1999 and comprised a range of complex words and short sentences representing different structures and clause types.

The study was carried out in conjunction with a related study of segment duration which found that segment lengthening systematically occurred at certain positions in a word or phrase (Pentland, 2002). A subsidiary aim of the study was therefore to supplement the duration analysis by investigating the relationship to prosodic structure of other strengthening and weakening processes in Warlpiri speech.

References:

Butcher, A. (1996). Some connected speech phenomena in Australian languages: universals and idiosyncracies. MS, Flinders University.

Pentland, C. (2002). Segment duration and word-level prosody in Warlpiri. Paper presented at the Prosody Workshop, University of Melbourne, 15 March.