CHAPTER 6. Chunks in writing 101 SB programs. On the other hand, longer, more lexically based, idiomatic chunks seem to contribute more to L2 uency and authenticity but need more exposure to be mastered. Our ndings are also in line with Piggott (2019), who found that the students who had had relatively more L2 exposure used signi cantly more chunks overall. e second aim of this study was to explore the relationship between chunks and di erent CAF measures. Chunks are thought to be processed and stored as single units, thus reducing cognitive load (Forsberg, 2010). Chunks may compensate for limited L2 automation (Gunnarsson, 2012), allowing learners to produce more language in less time, as retrieval of prefabricated sequences is claimed to be easier than the linguistic encoding required to produce grammatical sentences from its components. Chunks are thus expected to reduce processing e orts (Pawley & Syder,1983; Wray & Perkins, 2000) and, as a result, may enable uent production, particularly during cognitively demanding tasks such as L2 writing (Forsberg, 2010). is may, therefore, explain the signi cant, positive correlation that was attested between chunk coverage and uency (notably text length). In our study, there was a signi cant correlation, albeit with a relatively low e ect size, and we assume that the use of whole, longer chunks - which can be produced in one go - have contributed somewhat to the total number of words produced. Piggott (2019) used chunk use as a uency measure, and this seems to be warranted by the present ndings. ere is also a signi cant correlation between chunks and complexity (average sentence length) but with a relatively low e ect size. erefore, we cannot assume that the relationship of chunks and sentence length is a direct one. For example, we would expect sentence length to be a ected by the number of clauses, and there were no di erences between the groups in complement use (which include dependent clauses) and discourse chunks, which connect especially main clauses. It seems that sentence length correlates strongly with the use of longer, more lexically based chunks. Finally, there was no link between Guiraud, a lexical diversity measure, and chunks. is can be explained by the fact that chunks are o en combinations of frequently used words, which themselves will not contribute to diversity. Because the e ect sizes between chunk coverage and the CAF measures were relatively low, the question is if chunks should be considered subcomponents of the traditional CAF triad or be considered a separate pro ciency measure in its own right. It is likely that chunks contribute to uency and longer sentences, but rather indirectly. e argument to treat chunks as a measure in its own right might be that students can write very long texts with very long sentences without many errors, but they may not sound idiomatic at all. is does not necessarily have to be a problem, but if sounding target-like is considered an objective, then the use of native-like chunks should be considered. is would be in line with the pro ciency rubric used in Hou et al (2018), as they evaluated Idiomaticity separately from Complexity, Accuracy and Fluency.
RkJQdWJsaXNoZXIy MTk4NDMw