The effects of different learning conditions on the development of collocational knowledge in a second language

Previous research has shown that incidental exposure to second-language collocations in reading texts can produce gains in declarative collocational knowledge. However, there is little evidence that incidental exposure leads to the acquisition of procedural collocational knowledge. One key study that investigated these two areas was Sonbul and Schmitt (2013). They found that advanced non-native speakers gained substantial declarative knowledge of low-frequency technical collocations after three exposures in two incidental reading conditions in one treatment session; typographic enhancement of the collocations produced more correct answers than no enhancement. However, the researchers found no evidence of procedural collocational knowledge in a primed lexical decision task.  Experiment 1 in this thesis was a conceptual replication and extension of Sonbul and Schmitt (2013). In a counter-balanced learning and experimental condition, 62 advanced adult ESL speakers were exposed to nine occurrences of 15 low-frequency technical collocations in 500-word texts in three sessions on two consecutive days. (input flooding). Three incidental learning treatments were implemented: reading-only (typographically unenhanced), bolding-only and bolding-plus-glossing. Collocational knowledge was assessed in two tests of declarative knowledge: a cued-recall (gapfill) test and a form-recognition (multiple-choice) test. Procedural collocational knowledge was operationalised as a priming effect in a primed lexical decision task. The results of the immediate cued-recall and form-recognition post-tests corroborate Sonbul and Schmitt’s findings: multiple encounters with the collocations produced substantial declarative collocational knowledge, and more declarative knowledge was produced through exposure to typographically-enhanced collocations (with and without glosses) than to typographically-unenhanced collocations. Procedural knowledge was found, but, unexpectedly, only in the reading-only treatment.  Experiment 2 focused on non-technical lexical (verb + noun) collocations and grammatical (preposition + noun) collocations. Two incidental learning conditions were used: bolding and no-bolding. Seventy-eight intermediate-to-upper-intermediate-level adult native speakers of Chinese were exposed to six occurrences of each of 48 non-technical English collocations as they read twelve 170-word treatment texts on two consecutive days. The immediate post-test session comprised a gapfill (cued-recall) test, for measuring declarative knowledge, and a self-paced reading task, for measuring procedural knowledge. The results show an increase in declarative knowledge in both learning conditions. Bolding produced more declarative knowledge of preposition + noun collocations than no bolding; however, bolding was no more effective than no bolding for verb + noun collocations. In the self-paced reading task, the absence of bolding of verb + noun collocations led to a tendency towards the development of procedural knowledge, but this was not the case for the typographically enhanced verb + noun collocations. For preposition + noun collocations (both unbolded and bolded) no evidence of procedural knowledge was found.  The findings of both experiments indicate that input flooding of collocations in incidental learning conditions can produce declarative collocational knowledge, and that typographic enhancement may lead to more declarative knowledge than lack of typographic enhancement. Repeated exposure to typographically-unenhanced collocations in reading contexts may produce procedural knowledge of collocations more quickly than exposure to typographically-enhanced collocations. These findings indicate that declarative and procedural knowledge of MWUs are dissociated in the sense that they follow different developmental paths. In a teaching context, I predict that the knowledge of collocations may be acquired incidentally through the use of texts such as graded readers which contain multiple instances of collocations.