In Chapter 4, we touched on some of the current trends in L2 comprehension research, one of which hypothesises that the automatisation of certain operations of cognitive activity facilitates comprehension (Gaonac'h, 1991). We can also invert this concept and propose that in the absence of this automatisation, the execution of a comprehension task in the TL becomes costly in terms of cognition. In our analysis of the data collected by means of an oral sentence translation test, we will attempt to identify not only strategies used by our learners to accomplish the task at hand, but also constraints that prevent them from succeeding.

In Chapter 3 (cf. 3.3.1), our contrastive analysis of French and Polish predicted that the difference between the two systems of basic constituent word order would probably pose more problems for our learners in perception and comprehension than in production. If in the language being acquired the word order alone provides insufficient information to identify the semantic role relations within an utterance, then knowledge of the TL morphological system becomes crucial for the learner. We can imagine the reverse as well, that French learners who have not yet acquired Polish morphology will resort to the word order strategies of their NL to understand a Polish utterance.

Oral Sentence Translation Test

To test a learner's comprehension of TL sentences, Klein (1986) conducted an oral translation test (Heidelberger Forschungsprojekt 'Pidgin-Deutsch' [HPD], 1979; cf. also Klein, 1986) during which learners heard sentences in their TL and translated them immediately into their NL. This task is considered to be a way to gather evidence about how learners '... analyse utterances as input to the language processor' (Klein, 1986: 74). We adapted HPD's translation test for our own purposes and administered distinct versions of the test to our learners at three different periods: 1h30, 3h30 and 7h00 (see test items in Appendix 3). Each test consisted of 8–10 Polish utterances.