Looking up translations

Sketch Engine can handle parallel corpora and multilingual corpora to compare different languages or to look up translation examples.

Why work with parallel and multilingual corpora?

Parallel and multilingual corpora can be used by translators to look up how other translators have already translated a word or phrase but also by linguists to look into the behaviour of the same phrase or word in different languages.

parallel corpus concordance

How to generate a parallel / multilingual concordance?

Log in to Sketch Engine (or click Home), go to the Parallel section to list only parallel corpora and select the first a corpus. . If unsure which to use, OPUS or Europarl are good options.  In the case of a multilingual corpus, the next screen will give you a choice of the second language(s).

The search procedure and the available options are the same as with the standard concordance search. The main difference is that you can optionally specify search criteria for the second language at the same time.

For example: find all sentences in English where  jam  is NOT translated as mermelada into Spanish.

Paralell corpus query

The result will look like this. Sketch Engine will even try to identify and highlight the translation candidates. (click to enlarge)

parallel concordance result

Result screen description

Sketch Engine will display the results next to each other aligned (usually) at sentence level, i.e. the matching sentences will be displayed next to each other. The search word will be highlighted in the first language and translation candidates will be highlighted in the second language.

View options

View options in the left menu are identical to the view options of a standard concordance.  Sort, filter and frequency will only be applied to the first language.

The only new option is


Alignment changes the way the aligned structures (usually sentences) are displayed. It switches between a KWIC display and the whole sentences.