Invited Speakers

Dafydd Gibbon (Universität Bielefeld)

Gesture Theory is Linguistics: On Modelling Multimodality as Prosody

In modelling sign languages of the hearing-impaired, the metaphor of gesture patterns as 'phonology' has been very fruitful. On the other hand, although many disciplines from anthropological linguistics to robotics are currently occupied very productively with conversational gesture studies, the field is still highly fragmented. Starting from (computational) linguistic models as metaphors for the forms and functions of gestural signs, it can be shown that similarities between gesture and speech (specifically: prosody), go much further than metaphors, opening up avenues for integrating speech and gesture theory and developing a shared ontology for speech and gesture resource creation and retrieval.

Biography

Dafydd Gibbon is affiliated as full professor with Universität Bielefeld, was educated at the Universities of London, Erlangen and Göttingen and has also held positions at the Cologne University of Applied Sciences and Göttingen University, guest professorships at the Université de Cocody, Abidjan, Côte d'Ivoire and the University of Uyo, Akwa Ibom State, Nigeria, and Fellowships at the Jawaharlal Nehru Institute of Advanced Studies, New Delhi, India. He has worked and published in several areas of descriptive and computational linguistics, including forms and functions of prosody, prosody synthesis, lexicon theory and computational lexicography. He is especially interested in applications of computational linguistics in speech technology and language documentation, particularly for African languages, and most recently in the Tibeto-Burman area. His research has been supported by EU (SAM, EAGLES, ISLE, FlaReNet) and German (ASL, VerbMobil, DoBeS, DAAD) projects, and he has worked with the NSF projects EMELD and LEGO.

Anna Korhonen (University of Cambridge)

Automatic Lexical Classification - Balancing between Machine Learning and Linguistics

Lexical classes have attracted considerable interest in both linguistics and computational linguistics. Capturing useful generalizations about a range of
(cross-)linguistic properties, such classes have been used to support a number of practical tasks and applications, including parsing, information extraction, question-answering, and machine translation. However, large-scale exploitation of lexical classes in real-world or domain-sensitive tasks has not been possible because existing manually built classifications are incomprehensive. In this talk, I will describe recent and on-going research on extending and acquiring lexical classifications automatically. The automatic approach is attractive since it is cost-effective and opens up the opportunity of learning and tuning lexical classifications for the application and domain in question. However, the development of an optimal approach is challenging, and requires not only expertise machine learning but also a good understanding of the linguistic principles of lexical classification.

Biography

Anna Korhonen is a Royal Society University Research Fellow in the Computer Laboratory and RCEAL (Research Centre for English and Applied Linguistics) at the University of Cambridge, UK. She has an MA in Linguistics from the University in Reading and an MPhil and a PhD in Computer Science from the University of Cambridge. During her doctoral and postdoctoral work in research institutions in the UK, USA and Japan, she has mainly worked on natural language processing (automatic lexical acquisition, computational semantics, text mining) and linguistics (syntax, lexical semantics), and on the application of natural language processing to aid practical tasks in biomedicine and cognitive sciences. She has recent and on-going collaborations with researchers in Cambridge, Colorado, Stockholm, Tel Aviv, Tokyo and Paris. An area chair / programme committee member for more than 20 international conferences and workshops, she has published over 40 articles in the area of computational linguistics.

Helen Meng (Chinese University of Hong Kong)

Developing Speech Recognition Synthesis Technologies to Support Computer-Aided Pronunciation Training for Chinese Learners of English

We describe ongoing research in the development of speech technologies that strives to raise the efficacy of computer-aided pronunciation training, especially for Chinese learners of English. Our approach is grounded on the theory of language transfer and involves a systematic phonological comparison between the primary language (L1 being Chinese) and secondary language (L2 being English) to predict possible segmental and suprasegmental realizations that constitute mispronunciations in L2 English. The predictions are validated based on a specially designed corpus that consists of several hundred hours of L2 English speech. The speech data supports the development of automatic speech recognition technologies that can detect and diagnose mispronunciations. The diagnosis aims to support the design of pedagogical and remedial instructions, which involves text-to-speech synthesis technologies in audiovisual forms.

Biography

Helen Meng is Professor in the Department of Systems Engineering and Engineering Management at the Chinese University of Hong Kong (CUHK). She also serves as Associate Dean of Research for the Faculty of Engineering. She pursued studies to the PhD level in Electrical Engineering at the Massachusetts Institute of Technology. She joined CUHK in 1998 and established the Human-Computer Communications Laboratory in 1999, as well as the Microsoft-CUHK Joint Laboratory for Human-centric Computing and Interface Technologies in 2005. In 2008, the joint laboratory was recognized by China's Ministry of Education as the CUHK Microsoft-MoE Key Laboratory in Human-centric Computing and Interface Technologies. Helen is Editor-in-Chief of the IEEE Transactions on Audio, Speech and Language Processing. She is the recipient of the CUHK Faculty of Engineering Exemplary Teaching Award 2001, Service Award for establishing the Worldwide Engineering Undergraduate Exchange Program in 2004, as well as the Young Researcher Award in 2006. Helen's research interests cover multilingual spoken dialog systems, spoken content retrieval, multimodal processing of speech-centric input and expressive text-to-audiovisual speech synthesis.

Plenary Speakers

Keh-Jiann Chen (Academia Sinica)

A Step toward Compositional Semantics: E-HowNet a Lexical Semantic Representation System Extended from HowNet

The purpose of designing the lexical semantic representation model E-HowNet is for natural language understanding. E-HowNet is a frame-based entity-relation model extended from HowNet to define lexical senses and achieving compositional semantics. The followings are major extension features of E-HowNet to achieve the goal. a) Word senses (concepts) are defined by either primitives or any well-defined concepts and conceptual relations. b) A uniform sense representation model for content words, function words and phrases. c) Semantic relations are explicitly expressed; and d) Near-canonical representations for lexical senses and phrasal senses. We demonstrate the above features and show how coarse-grained semantic composition can be carried out under the framework of E-HowNet. Possible applications of E-HowNet are also suggested. We hope that the ultimate goal of natural language understanding will be accomplished after future improvement and evolution of the current E-HowNet.

Biography

Keh-Jiann Chen is the research fellow of the Institute of Information Science, Academia Sinica. He has been the deputy director of the Institute from 1991 to 1994. Currently he is the principal investigator of the Chinese Knowledge Information Processing (CKIP) group at Academia Sinica.

His research interests include Chinese language processing, lexical semantics, lexical knowledge representation, and corpus linguistics. He had been and continued in developing the research environments for Chinese natural language processing including CKIP Chinese lexical databases, Sinica corpus, Sinica Treebank, CKIP word segmentation and pos tagging system and CKIP Chinese parsers. His current major research topic is compositional semantics and developing a lexical semantic representation system E-HowNet for the purpose.

Dr. Chen is one of the founding members of the Association for Computational Linguistics and Chinese Language Processing (also known as ROCLING). He had served as 2nd term president of the society from 1991 to 1993. Currently he is the board member of the Chinese Language Computer Society, the Advisory Board Member of the International Journal of Computational Linguistics and Chinese Language Processing, the associate editor of the Journal of Computer Processing of Languages and the associate editor of the International Journal of Advanced Intelligence.

Noriko Kando (National Institute of Informatics)

Lesson Learned for Sentiment Analysis Research from the NTCIR Multilingual Opinion Analysis Task

Ik-Hwan Lee (Sangmyung University)

Resultative Constructions as Causal Relations between Events

This paper investigates some resultative constructions in English, Korean, Japanese, and Chinese. In particular, it first examines the intransitive-based resultative constructions in English and Korean and tries to offer a proper semantic account of the relevant data. In addition, it will show that the proposed account is also useful for explaining the so-called V(erb)-V(erb) compound resultatives of Korean, Chinese and Japanese. Previous syntactic accounts of resultatives have relied on the Direct Object Restriction (DOR, henceforth), which dictates that result phrases are predicated of NPs in the object position. The DOR has been argued to be incorrect. In this paper, an attempt is made to provide a comprehensive semantic explanation of resultative constructions, using the notion of 'event' and some semantic restrictions which incorporate 'causal relation', 'aspectual relation', 'canonical result', etc. It will also examine the relationship between the typical intransitive resultatives and the V-V resultative constructions.

Biography

Ik-Hwan Lee obtained his Ph.D. in linguistics from U of Texas at Austin in 1979 with the thesis entitled "Korean particles, complements, and questions". He was a professor at Yonsei U, Seoul, Korea from 1981 through 2008 and currently a "Chair" Professor at Sangmyung U and a Professor Emeritus at Yonsei U. Prof. Lee served as the President of many academic associations including the Linguistic Society of Korea. Currently he is a member of the advisory committee of the English Linguistics Society of Japan and a member of the Executive Committee of CIPL (Comité International Permanent des Linguistes). In 2008 he served as the organizing chair of the 18th International Congress of Linguists (CIL18). Prof. Lee has published many books (English Semantics, Semantics of Psychological Verbs, etc.) and papers ("Semantics of questions", "Double nominative constructions in Korean", "Licensing conditions and event structure of resultatives", "A semantic/pragmatic interpretation of deverbal -er nominals in English", etc.).

Ming Zhou (Microsoft Research Asia)

Generating Chinese Couplets and Poetry with Statistical Methods

Automatic generation of Chinese couplets and poetry is a difficult task in NLP research community and has not been explored deeply. In this paper, we proposed a novel statistical approach to automatically generate Chinese couplets and Chinese poetry.

For Chinese couplets, we present a phrase-based SMT approach to generate the second sentence. First, the system takes as input the first sentence, and generates as output an N-best list of proposed second sentences, using a phrase-based SMT decoder. Then, a set of filters is used to remove candidates violating linguistic constraints. Finally, a Ranking SVM is applied to rerank the candidates. A comprehensive evaluation, using both human judgments and BLEU scores, has been conducted, and the results demonstrate that this approach is very successful.

We extended this approach to further generate classic Chinese poetry and use quatrain as case study. Given a few keywords describing user's intention, a phrase based statistical machine translation model is used to generate the four sentences one by one. The evaluation on the quality of every line of sentence as well as the quality of the generated poem as a whole demonstrate very promising results.

Biography

Ming Zhou received his B.Eng from Chongqing University in 1985 and M.S. and Ph.D. degrees from the Department of Computer Science and Engineering at Harbin Institute of Technology (HIT) in 1988 and 1991. During his study at HIT, he developed China's first Chinese-English machine translation system. During 1991-1999, he worked at Tsinghua University as a Postdoc Researcher for the first two years and then as associate professor. He visited the Chinese University of Hong Kong and the City University of Hong Kong in 1995 and 1996 respectively. During 1996-1999, he visited Kodensha Ltd. in Japan and developed J-Beijing machine translation software product. This product has become the most popular Chinese-Japanese translation product in Japan and was awarded Makoto Nagao prize in 2008 together with other MT products from the Kodensha. In 1999, he joined Microsoft Research Asia as researcher. In 2001, he became the group manager of Natural Language Group. His group has invented many innovations such as new type of input methods for Chinese and Japanese for MS Windows, English assistant for MS Office, keyword extraction for MS SQL server, and word breaker, query speller for search engine, as well as popular web services of Chinese couplets (www.duilian.msra.cn) and English learning search (www.engkoo.com). He has served as area chairs of ACL, IJCAI and other NLP conferences many times. He currently is the member of editorial boards of Journal of Machine Translation, Journal of Computational Linguistics and ACM Transaction of Asian Language Information Processing.