Outline of a computerized Chinese grammar enabling English and Swedish translation

Bengt Sigurd, Hong Gao

Research output: Working paper/PreprintWorking paper

146 Downloads (Pure)

Abstract

This paper presents a computerized grammar which can analyze and generate a sample of pedagogical Chinese sentences, in particular those mentioned in the textbook Kinesiska är inte svårt (‘Chinese is not difficult’) by Göran Malmqvist 1974. Equivalent grammars for English and Swedish have also been constructed allowing translation between the three languages. The grammar model used is the Swetra grammar developed in the MT-project Swetra at the Department of Linguistics, Lund University and used in various translation and generation projects, including the application which generates weather reports in Swedish and English used by SMHI, the Swedish meteorological agency. The Swetra grammar (see Sigurd (ed.) 1994) is written directly in the DCG (Definite Clause Grammar) format and implemented in Prolog (LPAProlog). It has been used for several languages and adapted to various practical needs. Swetra grammar is characterized by its separate functional (-semantic) representation which is abstract enough to be used as an interlingua in automatic translation. The functional representation of Swetra does not show surface details such as word order, case or agreement. Swetra grammar has ways to account for mode, topic, coordinated clauses and subordinate clauses including relative clauses. The word meanings are represented by a standardized English Machinese according to certain conventions. The construction of a computerized grammar of Chinese can be based on traditional Chinese grammar but it requires reconsideration and taking a number of decisions. The computerized grammars presented pinpoint the differences between Chinese, English and Swedish. The Chinese, English and Swedish grammars and lexicons are constructed to enable the automatic translation into or from these languages using the Swetra functional representation as interlingua. In order to allow writing in both Chinese characters and pinyin the lexical items of the grammar are rended in these two systems. Demo sentences showing the use of the Chinese, English and Swedish grammars are presented.
Original languageEnglish
PublisherDepartment of Linguistics, Lund University
Pages 181–199
Number of pages19
Publication statusPublished - 1999

Publication series

NameWorking papers / Lund University, Department of Linguistics, General Linguistics, Phonetics
PublisherDepartment of Linguistics, Lund University
Volume47
ISSN (Print)0280-526X

Subject classification (UKÄ)

  • Comparative Language Studies and Linguistics

Fingerprint

Dive into the research topics of 'Outline of a computerized Chinese grammar enabling English and Swedish translation'. Together they form a unique fingerprint.

Cite this