Between Legality and Legitimacy
3. Legalidad y legitimidad en administración
This study substantiates the effectiveness of instructed second language acquisition, and more specifically, corroborates previous empirical findings on the efficacy of instruction on the acquisition of FSs (Bishop, 2004; Boers et al., 2006; Boers & Lindstromberg, 2005; Boers et al., 2004; Hsu, 2010; Jones & Haywood, 2004; Lindstromberg & Boers, 2008a; Rott, 2009; Wood, 2009). From the findings discussed in section 4.1, there was strong evidence that the three types of instruction successfully help learners acquire levels of productive knowledge of FSs. Overall the IEEI, Gap-‐fill and SpotDif group outperformed the Control group at a statistically significant level. In the immediate post-‐test, all treatment groups’ mean scores were significantly higher than that of the Control group. In the delayed post-‐test, however, only the mean score of the Gap-‐fill group remained significantly higher. Statistical significance testing through mixed between-‐within subjects ANOVA shows that with regards to the effectiveness of different treatment types, it seems that Gap-‐fill is most beneficial to the acquisition of the productive
knowledge of these FSs. The Gap-‐fill group consistently scored the highest on both post-‐tests, and only the difference between its members’ mean score and that of the Control group remained statistically significant in the delayed post-‐test.
Despite the lack of statistical significance for the SpotDif and IEEI groups in the Productive Knowledge delayed post-‐test, the fact that Cohen’s d effect sizes are moderate and large for all treatment groups as compared to the Control group is evidence for educational significance, especially for experimental studies with a small-‐sized sample. Neill (2008) claims that statistical significance can be misleading when examining effects using small sample size because such a procedure is subject to Type II errors. According to Neill, “statistical significance is not a direct indicator of size of effect, but rather it is a function of sample size, effect size and p level. In these situations, effect sizes and confidence intervals are more informative than significance testing” (Significance testing, para. 8). This means interventions on a large sample with very small effect sizes can have statistical significance, while large effect sizes of treatment on smaller samples are statistically non-‐significant. Thus in this study it is useful to look at Cohen’s
d effect sizes in addition to statistical significance test results in order to determine whether instructional effectiveness is present and to evaluate the magnitude of instruction effect expected in a real setting. As Norris and Ortega (2000) recommend in their meta-‐analysis, researchers who want to investigate instructional effectiveness in second language teaching should calculate effect sizes because statistical significance tests will not provide answers about the presence, the size and the importance of an effect.
As far as effect sizes are concerned, the two communicative tasks (Gap-‐fill and SpotDif) were more successful than the IEEI group in helping learners acquire a higher level of productive knowledge of the target FSs, though not at a statistically significant level. Effect sizes using
Cohen’s d in both the immediate and delayed post-‐tests show that d was largest for the Gap-‐fill group, followed by SpotDif and then IEEI. In both immediate and delayed post-‐tests, the effect sizes for the Gap-‐fill and SpotDif groups as compared to the Control group exceeded Cohen’s convention for a large effect. The effect size for the IEEI group as compared to the Control group was large in the immediate post-‐test but decreased to the moderate level in the delayed post-‐ test. This finding attests to the effectiveness of interactive tasks in the L2 classroom, and is in line with findings from previous studies (R. Ellis & He, 1999; R. Ellis et al., 1994; Gass & Torres, 2005; Kim, 2008; Kowal & Swain, 1994; Loewen & Philp, 2006), although these studies focus on individual words rather than FSs. This study proves that tasks lead to better productive knowledge acquisition of L2 vocabulary, similar to what R. Ellis and He (1999) and Gass and Torres (2005) found in their studies.
As discussed previously in chapter 2, the Involvement Hypothesis (Laufer & Hustijn, 2001) postulates that the effectiveness of a classroom task is contingent upon the levels of involvement factors, i.e. need, search, and evaluation. Higher levels of need, search and
evaluation are conducive to better acquisition of vocabulary. In accordance with the anticipation made based on the Involvement Load Hypothesis prior to the experiment, the two task-‐based treatment types were more effective than the IEEI group in helping learners acquire higher levels of productive knowledge. The IEEI group did not interact with each other to solve problems created by the Gap-‐fill or SpotDif group. Instead of actively working on the input, they received instructor-‐led explanation of the meaning of readily typologically-‐enhanced FSs in the input text, and a brief introduction of the common mistakes they would likely make with these FSs, as well some strategies to better notice the FSs in any given input. This kind of instruction is only capable of activating low to moderate levels of need, search and evaluate, while the other
two types of treatment activated higher levels of these cognitive processes due to the implementational procedures of the tasks. (see Table 3, Chapter 3) It should be noted that as for receptive knowledge, tasks do not seem to have such an advantage over IEEI. However, the effectiveness of tasks in facilitating learners’ acquisition of Productive Knowledge, which is always much harder to achieve (Milton, 2009; Nation, 2001; Schmitt, 2008), has important implications for the classroom. I will return to this point in a section presented later in this chapter.
The higher levels of productive knowledge acquired by learners in the Gap-‐fill group compared to the SpotDif group could be explained by the way the two tasks were actually carried out by the learners. Although I did not have a systematic analysis of qualitative data obtained from classroom recordings, for the SpotDif group, observations in the classroom suggest that the levels of need, search and evaluate were compromised by the fact that some learners seemed to rely on their short-‐term memory to determine which version of the FSs should be chosen. The kind of interaction occurring among learners was relatively cursory, largely limited to inquiring of each other if they remembered what the FSs in the original passage were. Most learners thus completed the task very quickly without evaluating the meaning of the FSs or understanding the surrounding text. The mechanical focus on form during the SpotDif task with apparently little sign of attempting to understand the immediate context of the FSs, nor their meaning, might have caused some learners to acquire the FSs less effectively due to lower involvement load. An explanation of this unintended effect of SpotDif could be the nature of FSs, which are multiword lexical units. This structural composition of FSs could have induced learners’ need to mechanically recall the form of the whole unit, rather than searching for the meaning and understanding the context. In addition, in previous studies where SpotDif was used, it was of a
longer format, with a gap-‐fill component after learners finished comparing the two versions of the original passage. The shortened format of SpotDif in my study, intended to maintain the same time on task across treatment groups, could be another factor leading to less impressive effects compared to previous research. In comparison, the Gap-‐fill group seems to carry out the assigned task in a way exactly as planned: they attempted to know the meaning of the FSs, many times through asking the instructor, as well as decoded the meaning of the surrounding text in order to complete the task collaboratively.
Learners’ attention to the context of new FSs has been proven to play an important role in the learning of FSs at the productive level. Pickering (1982) conducted an experiment on the effectiveness of contextualized versus de-‐contextualized (i.e. translation into L1) presentation of new words and found that context leads to better learning. Furthermore, Cohen and Aphek (1980), cited in Carter (2002), conclude from their studies that learners of higher proficiency levels are more likely to benefit from learning words in context. In the treatment conditions of the current study, target FSs were introduced in context to learners in the reading text, it seems that Gap-‐fill, the follow-‐up task that requires learners to attend more closely to the context around new FSs for the second time, was most successful in helping learners to recall productively these FSs in the immediate and delayed post-‐tests. It is likely that as learners’ attentional resources are drawn to the task of understanding the context, this intensifies the processes of need, search and evaluation in vocabulary acquisition.
5.2. Do IEEI, Gap-‐fill and SpotDif help learners acquire higher levels of receptive knowledge of