CHAPTER 5: DISCUSSION AND CONCLUSIONS
5.7 Implications of the Present Study
5.7.1 Theoretical implications
acceptable model-fit indices of the revised SEM model support such an assumption. The study also simultaneously compared the contributions of character-recognition ability and strategy use to comprehension, and the results suggest that the importance of efficient character recognition overwhelms that of strategy use. However, mere comparison of these two contributions implies linear relationships between the two factors and comprehension, and overlooks the complicated relationships that exist between reading processes because strategy use interacts directly and indirectly with both character recognition and comprehension. In other words, when comparing the contributions of character recognition to those of strategy use, the nonlinear relationship of strategy use with the other processes is overlooked, and its supporting role in comprehension performance would be better captured via a nonlinear and interactive perspective on L2 Chinese reading.
In this study, strategy use was operationalized as conscious cognitive acts, and as having specific purposes in a particular context. As such, strategy use in the study differs from
language-learning strategies, which are defined as “broad, teachable actions that learners choose from among alternatives and employ for L2 learning purposes” (Oxford, 2011, p. 12). Various scholars (Cohen, 2014; Phakiti, 2007; Tseng, Dörnyei, & Schmitt, 2006) have argued against the use of strategy-use surveys such as the Strategy Inventory for Language Learning (SILL)
(Oxford, 1990) in a wide range of learning contexts. Their objections all relate to the idea that, when used for a decontextualized purpose, what such surveys actually collect is the respondents’ accumulated knowledge regarding strategy use, and they thus overlook the context-specific nature of online strategy-use as it occurs in people’s working memory. To overcome the manifest ambiguities in existing definitions of language-learning strategies, and to emphasize learner autonomy, some scholars have begun switching to the notion of self-regulation (Oxford, 2011; Tseng, Dörnyei, & Schmitt, 2006); and Phakiti (2007) goes even further, arguing that the term “strategies” should be used only for online cognitive activities facilitating a particular L2 task performance. Following this line of reasoning, online strategy use for specific learning tasks might be conceptualized as part of strategic competence, and the meta-knowledge accumulated through online strategy use might further feed into self-regulatory competence for language learning.
In addition, scholars investigating communication strategies tend to adopt a view that strategy use is fundamentally a problem-solving technique (e.g., Cohen, 2014; Kasper & Kellerman, 1997; Macaro, 2001). This approach, by its nature, tends to foreground the
compensatory function of strategy use and to downplay its facilitative functions. A theoretical implication of the negative and positive relationships between strategy use and reading processes identified by the present study could be uniting the compensatory and facilitative perspectives into a single framework that treats them as two sides of a coin. The compensatory function may be activated to repair insufficient comprehension or linguistic knowledge for the purpose of facilitating L2 task performance or linguistic competence. Therefore, both functions serve the same purposes, but with different connotations: the compensatory function focusing on the
distinction of deliberate strategies from automatic language skills; and the facilitative function emphasizing the positive contribution of strategies to L2 performance.
The present study drew on L1, L2, and HL reading development, L2 strategic competence, and L2 assessment research to investigate the relationships between selected cognitive constructs of L2 Chinese reading. In so doing, it first investigated the variance
explained by language background (considered as a variable of higher-level reading processing) in L2 Chinese reading ability, character-recognition ability, and strategy use. An important theoretical implication of the results is that higher-level processing skills could be measured more accurately by tapping into the effect of language background on reading processing. On the other hand, although lower-level processing skills may be more language-specific, such skills can be developed quickly even by readers whose L1 orthography is very different from Chinese, and this may lead to lower-than-expected effects of language background on lower-level
processing skills. Additionally, the significant group differences we observed in reading-test performance may have been linked to the nontechnical, narrative, and literary texts included in the reading test, which could have required greater activation of situational models of
interpretation (Grabe, 2009) than the character-recognition test did. In other words, future investigators should bear in mind that large effects of language background on comprehension may be due to the activation of a situational model of reader interpretation, which is related to topical knowledge of the target culture more than to linguistic knowledge of the target language.
It should also be noted here that language background, as investigated in the present study, was operationalized based on Kondo-Brown’s (2005) research, which suggested that, in terms of the amount of linguistic input, parents are significantly more beneficial for home- language development than grandparents. Based on those findings, the current dissertation
initially classified participants by language background based solely on their parents’ native languages. However, this categorization scheme became problematic during screening of the responses to the language-background survey, in that Singaporean CMTLLs reported more Mandarin-use at home than did CFLLs. A reason for this might be that most Singaporean participants identified their parents as native speakers of English due to the prevailing language ideology in Singapore – to say nothing of the fact that, in a multilingual society like Singapore, the distinction among native language, first language, and dominant language can be less than clear. In addition, the language-background survey ignored the possibility that parents can be native speakers of more than one language. These issues may suggest that the status of being a foreign language learner or a heritage language learner might be better treated as being on a continuum for the target language exposure at home, or as an interval variable, rather than as categories or a nominal variable. Transferring language background categories to linguistic home exposure from different family members, i.e., multiple ordinal variables, to an interval variable could be achieved by generating factor scores through PCA and estimating person parameters through Rasch Analysis. In addition, the effect size of language background on different aspects of reading processing could be examined using multiple regression: by treating the factor scores of language background as the DV; the test scores of the reading subtests as IVs; and the R2 of the multiple regression as the effect size of language background.
5.7.2 Methodological implications. This study operationalized reading acts as nonlinear