On Statistical Machine Translation and Translation Theory

Christian Hardmeier

doi:10.18653/v1/W15-2522

On Statistical Machine Translation and Translation Theory

Christian Hardmeier

Uppsala University

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

Abstract

The translation process in statistical machine translation (SMT) is shaped by technical constraints and engineering considerations. SMT explicitly models translation as search for a target-language equivalent of the input text. This perspective on translation had wide currency in mid-20th century translation studies, but has since been superseded by approaches arguing for a more complex relation between source and target text. In this paper, we show how traditional assumptions of translational equivalence are embodied in SMT through the concepts of word alignment and domain and discuss some limitations arising from the word-level/corpus-level dichotomy inherent in these concepts.

Original language	English
Title of host publication	Proceedings of the Second Workshop on Discourse in Machine Translation
Number of pages	5
Publication date	21 Sept 2015
Pages	168–172
ISBN (Print)	978-1-941643-32-7
DOIs	https://doi.org/10.18653/v1/W15-2522
Publication status	Published - 21 Sept 2015
Externally published	Yes

Keywords

Statistical Machine Translation
Word Alignment
Translational Equivalence
Domain
Corpus-Level Dichotomy

Access to Document

10.18653/v1/W15-2522

Cite this

@inproceedings{187fda8cf9264a2e88276f000124efb1,

title = "On Statistical Machine Translation and Translation Theory",

abstract = "The translation process in statistical machine translation (SMT) is shaped by technical constraints and engineering considerations. SMT explicitly models translation as search for a target-language equivalent of the input text. This perspective on translation had wide currency in mid-20th century translation studies, but has since been superseded by approaches arguing for a more complex relation between source and target text. In this paper, we show how traditional assumptions of translational equivalence are embodied in SMT through the concepts of word alignment and domain and discuss some limitations arising from the word-level/corpus-level dichotomy inherent in these concepts.",

keywords = "Statistical Machine Translation, Word Alignment, Translational Equivalence, Domain, Corpus-Level Dichotomy",

author = "Christian Hardmeier",

year = "2015",

month = sep,

day = "21",

doi = "10.18653/v1/W15-2522",

language = "English",

isbn = "978-1-941643-32-7",

pages = "168–172",

booktitle = "Proceedings of the Second Workshop on Discourse in Machine Translation",

}

TY - GEN

T1 - On Statistical Machine Translation and Translation Theory

AU - Hardmeier, Christian

PY - 2015/9/21

Y1 - 2015/9/21

N2 - The translation process in statistical machine translation (SMT) is shaped by technical constraints and engineering considerations. SMT explicitly models translation as search for a target-language equivalent of the input text. This perspective on translation had wide currency in mid-20th century translation studies, but has since been superseded by approaches arguing for a more complex relation between source and target text. In this paper, we show how traditional assumptions of translational equivalence are embodied in SMT through the concepts of word alignment and domain and discuss some limitations arising from the word-level/corpus-level dichotomy inherent in these concepts.

AB - The translation process in statistical machine translation (SMT) is shaped by technical constraints and engineering considerations. SMT explicitly models translation as search for a target-language equivalent of the input text. This perspective on translation had wide currency in mid-20th century translation studies, but has since been superseded by approaches arguing for a more complex relation between source and target text. In this paper, we show how traditional assumptions of translational equivalence are embodied in SMT through the concepts of word alignment and domain and discuss some limitations arising from the word-level/corpus-level dichotomy inherent in these concepts.

KW - Statistical Machine Translation

KW - Word Alignment

KW - Translational Equivalence

KW - Domain

KW - Corpus-Level Dichotomy

U2 - 10.18653/v1/W15-2522

DO - 10.18653/v1/W15-2522

M3 - Article in proceedings

SN - 978-1-941643-32-7

SP - 168

EP - 172

BT - Proceedings of the Second Workshop on Discourse in Machine Translation

ER -

On Statistical Machine Translation and Translation Theory

Abstract

Keywords

Access to Document

Fingerprint

Cite this