Corpora of maternal and second language pairs are obvious language learning materials; they offer a wealth of insights into comparative lexical, syntactic and frequency properties of the paired languages. Parallel concordances have indeed been making their way into the classroom since the 90s.

(See Joseph Réseau's site at http://www.uhb.fr/campus/joseph.rezeau/concord.htm for a practical example).

The most common use is however limited to simple parallel concordances on the sentence level, focused on a given word or syntactic configuration. In the language learning perspective, along with the quality (grammatical and semantic tagging), representivity (discourse type and language level), and size of the corpus, the degree of alignment is another important factor that determines what materials can be usefully extracted from a parallel corpus. All corpora are not aligned to the same degree.

The online Canadian Hansard (the French and English record of the Senate and Parliament of Canada at http://www.parl.gc.ca) is aligned on the macro structural level, i.e. pages and sometimes paragraphs are aligned.

The derivative Hansard Collocation Database (HCD; http://edziza.arts.ubc.ca/winder/hansard) is aligned on smaller units, sentences, at the meso structural level. Presently, no online French-English corpora is aligned by words and expressions, the micro structural level.

We will consider in this talk how language learning materials can be extracted at each of these three levels, with a focus on the difficulties, properties, and promises of the micro alignment of the Hansard Collocation Database.

(For more information on parallel corpora and language learning, see Tim Johns Data-Driven Language Learning Page:

http://www.eisu.bham.ac.uk/johnstf/timconc.htm.

For a Flash preface to bilingual terminology research using the online Hansard and the HCD see the "Video introductions to the Canadian Hansard and to the and to the Hansard Collocation Database" at the HDC Help Page:

http://edziza.arts.ubc.ca/winder/hansard/faq.cfm.)