Accuracy on the Forced-Choice Task and Cues Used During Think-Aloud Protocols
As in Zyzik and Azevedo (2009), the FCT was scored in a binary fashion: 1 point for choosing the correct form and 0 points for choosing the wrong form, both forms, or checking neither. The mean score for the group was 8.71 (maximum of 16) and the standard deviation (SD) was 2.3. The mean scores for each condition are presented in Table 1. Note that mean scores are out of 4 (there were four items in each condition).
Table 1 reveals that condition 4, which contrasted adjective and noun forms, yielded the lowest mean score among the four conditions. A repeated measure ANOVA confirms that the differences between conditions are statistically significant (3, 102) = 9.13, p < 0.001). The partial eta squared (rfy value (0.212) indicates that this was a relatively small effect. Next, a paired samples’ /-test was performed to determine where the differences have occurred. The Alpha level was set at 0.008 to avoid a type 1 error. The results of this test indicate significant differences between conditions 1 and 3, conditions 1 and 4, and conditions 2 and 4 (p < 0.008 for all comparisons). An item analysis was conducted to determine if certain items on the FCT were more difficult than others. Mean scores on individual items ranged from 0.23 for the noun/adjective contrast in corte/corto (‘cut/short’) to 0.83 for the verb/noun contrast in sientes/sentimientos (‘feel/feelings’). Additional items that yielded scores at or below chance level were pasar/pasan (to spend/they spend), carinosa/carino (‘affectionate/affection’), tienen/tener (‘they have/to have’), and fuerte/fuerza (‘strong/strength’). The learners’ verbal reports, discussed later, will clarify why certain items were particularly problematic.
Thirty-five participants completed a 16-item task, thus yielding an initial data set of 560 items to be coded for cue use. However, some of the data had to be discarded from further analysis (the cases of ‘neither’ and those that were not identifiable). The remaining data set consists of 526 tokens of cue use, which were grouped into three main categories: semantic cues, grammatical cues, and external cues. Table 2 presents the raw tokens and the relative frequency of each cue. Subsequently, I describe each cue and provide examples from both concurrent (CTA) and retrospective (RTA) think-aloud protocols. Semantic cues. This category reflects learners’ reliance on word meaning to make a decision between the two contrasting forms. As expected, word meaning is accessed through the participants’ native language, English. In their comments, participants often used the phrases ‘X means Y’ or ‘that’s the word for X’.
I guess I’ve just heard people refer to affection as carino. (PI 3, RTA) In the above examples the participant chooses a form based on its assumed equivalent in English. Note that participant is attending to word meaning rather than sentence-level syntactic or morphological information. This led to the wrong choice in the case of fuerte (strong) and the correct response in the case of carino (affection).Syntactic cues. This category reflects learners’ attention to sentence-level distributional patterns. For example, some learners focused on the distribution of finite forms or sequences of words that they judged to be possible or not. Note that the above example involves using derivation as a cue in conjunction with semantic cues: the participant accessed the meaning of the word via the LI.With years of experience in making delicate Cheap Louis Vuitton Handbags, we are waiting to provide the crowning touch to your style. Just feel free to shop Louis Vuitton solar PM shoulderbag M93124 in grey here.
Processing your request, Please wait....
