Recent technological development has had a deep and positive impact on automatic translation. The increase in terms of size and number of parallel corpora produced in many different languages, in conjunction with considerable improvement in the available algorithms, have seen machine translation enter a new stage. In particular, a shift in the paradigm from statistical to neural network-based machine translation has successfully addressed many shortcomings of traditional machine translation systems, thus improving the quality of the proposed translations. So far, collocations and idiomatic expressions have failed to be properly handled by machine translation services. This study sets out to investigate how such combinations are dealt with by automatic machine translation systems in light of their popularity amongst learners. After analysing the Italian translation of a sample of 282 ‘adjective + noun’ English collocations, we have concluded that, although much progress has been made in recent years, collocations still pose a challenge to automatic translation. In reception tasks, MTSs can certainly help users obtain basic understanding of the collocation in the foreign language. Yet, if users are to translate collocations from their mother tongue into the foreign language, such services cannot be trusted blindly as more work is needed for the production of well-formed, natural stretches of language, particularly with respect to the translation of strong collocations.

Can collocations be translated automatically? An evaluation of available machine translation services on a sample of ‘adjective + noun’ combinations / B. Berti. - In: RIVISTA DI STUDI ITALIANI. - ISSN 1916-5412. - 36:1(2018 Apr), pp. 373-389.

Can collocations be translated automatically? An evaluation of available machine translation services on a sample of ‘adjective + noun’ combinations

B. Berti
2018

Abstract

Recent technological development has had a deep and positive impact on automatic translation. The increase in terms of size and number of parallel corpora produced in many different languages, in conjunction with considerable improvement in the available algorithms, have seen machine translation enter a new stage. In particular, a shift in the paradigm from statistical to neural network-based machine translation has successfully addressed many shortcomings of traditional machine translation systems, thus improving the quality of the proposed translations. So far, collocations and idiomatic expressions have failed to be properly handled by machine translation services. This study sets out to investigate how such combinations are dealt with by automatic machine translation systems in light of their popularity amongst learners. After analysing the Italian translation of a sample of 282 ‘adjective + noun’ English collocations, we have concluded that, although much progress has been made in recent years, collocations still pose a challenge to automatic translation. In reception tasks, MTSs can certainly help users obtain basic understanding of the collocation in the foreign language. Yet, if users are to translate collocations from their mother tongue into the foreign language, such services cannot be trusted blindly as more work is needed for the production of well-formed, natural stretches of language, particularly with respect to the translation of strong collocations.
No
English
Settore L-LIN/12 - Lingua e Traduzione - Lingua Inglese
Articolo
Esperti anonimi
Pubblicazione scientifica
apr-2018
Rivista di Studi Italiani
36
1
373
389
17
Pubblicato
Periodico con rilevanza internazionale
Aderisco
info:eu-repo/semantics/article
Can collocations be translated automatically? An evaluation of available machine translation services on a sample of ‘adjective + noun’ combinations / B. Berti. - In: RIVISTA DI STUDI ITALIANI. - ISSN 1916-5412. - 36:1(2018 Apr), pp. 373-389.
reserved
Prodotti della ricerca::01 - Articolo su periodico
1
262
Article (author)
Periodico senza Impact Factor
B. Berti
File in questo prodotto:
File Dimensione Formato  
2018_RSI_Berti.pdf

accesso riservato

Tipologia: Publisher's version/PDF
Dimensione 200.96 kB
Formato Adobe PDF
200.96 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/571699
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact