Semantic-based transfer

Dorna, Michael; Emele, Martin C.

doi:10.3115/992628.992684

Cited by 25 publications

(15 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Concerning stratification, we could hypothesize that translators will seek a nonambiguous relation between semantics and lexico-grammatical expression (this could be another facet of simplification); or, they may try to leave ambiguities intact when they can, i.e., when it is possible to recast the same ambiguity in the target language. The latter is an approach that has been recently advocated in machine translation (e.g., Dorna & Emele 1996). Concerning delicacy, we can hypothesize that if the source language is more specific in a particular grammatical system than the target language, then translators will tend to choose a less specific option in the target language (e.g., imperative in English for polite imperative in German) with the possible effect of losing some of the meaning expressed in the source language text.…”

Section: Outlook: Research On the Specific Properties Of Translationmentioning

confidence: 99%

System-oriented and text-oriented comparative linguistic research

Teich

1999

LiC

View full text Add to dashboard Cite

The main concern of this paper is to develop a model of cross-linguistic variation that is applicable to various kinds of comparative linguistic research. The motivation for this lies in the observation that there is little interaction among the major areas of comparative linguistic investigationlanguage typology, contrastive linguistics, translation studies, and the computational modeling of multilingual processes as implemented in machine translation or multilingual text generation. The divide between them can be characterized by a general orientation towards describing the relation between language systems (as in language typology) vs. describing the relation between texts (as in translation studies). It will be suggested that with a model of cross-linguistic comparison that accommodates both the system view and the text view on cross-linguistic variation, language typology, contrastive linguistics, translation studies and multilingual computational linguistics can be shown to have mutually compatible concerns rather than being entirely disjunct endeavors. The model proposed is based on Systemic Functional Linguistics (SFL), using the representational categories SFL sets up as parameters along which cross-linguistic variation can be described. The fundamental assumption brought forward by SFL that acts as a unifier of concerns is that texts are ultimately instantiations of the language system under certain specifiable contexts of use. A model of cross-linguistic variation based on SFL thus bears the promise of opening up the text view for the system-oriented branch of cross-linguistic study, and the system view for the text-oriented branch. I illustrate the model with data from several European languages, concentrating on the register of instructional text.

show abstract

Section: Outlook: Research On the Specific Properties Of Translationmentioning

confidence: 99%

System-oriented and text-oriented comparative linguistic research

Teich

1999

LiC

View full text Add to dashboard Cite

show abstract

“…Within the modular system architecture, the dialogue and discourse processing is situated in between the components for semantic construction (Gamb~ck et al, 1996) and semantic-based transfer (Dorna and Emele, 1996). It uses context knowledge to resolve semantic representations possibly underspecified with respect to syntactic or semantic ambiguities.…”

Section: Methodsmentioning

confidence: 99%

Definiteness predictions for Japanese noun phrases

Heine

1998

Proceedings of the 17th International Conference on Computational Linguistics -

View full text Add to dashboard Cite

One of the major problems when translating from Japanese into a European language such as German or English is to determine definiteness of noun phrases in order to choose the correct determiner in the target language. Even though in Japanese, noun phrase reference is said to depend in large parts on the discourse context, we show that in many cases there also exist linguistic markers for definiteness. We use these to build a rule hierarchy that predicts 79,5% of the articles with an accuracy of 98,9% from syntactic-semantic properties alone, yielding an efficient pre-processing tool for the computationally expensive context checking.

show abstract

“…This form of representation also has the advantage, as [10] claim, that additional constraints which are important for generation in the target language, e.g. topic/focus in sign languages, may be made explicit.…”

Section: Semantic Representationmentioning

confidence: 99%

“…This has been extended by introducing labels for different kinds of semantic predicates. As in Verbmobil's VIT representation the labeling of all semantic entities allows a flat representation of the hierarchical structure of arguments and operator embeddings [10,11]. In contrast to Vermobil's uniform labeling, an ontology for all DRS propositons has been introduced to facilitate the mapping between the flat semantic structure of the DRS to the nested input structure of the target language specific HPSG, as required by the generation algorithm in ALE [5].…”

Section: Semantic Representationmentioning

confidence: 99%