The retrievability of L2 English multi-word items in a context of strongly form-focused exposure: What matters? Untitled Item

(Forthcoming in Studies in Second Language Acquisition)

The display here enlarges the earlier 2018 display for the assonance experiment. Most of the column headings are different & more explanatory. Everything is now on one page. But no numerical values have been changed or re-assigned. Several new covariates have been added.

# Binary outcome variables: Score1FreeRecall; Score2CuedRecall;Score3Recog (i.e., recognition)

# The focal predictor is Assonance. The focal moderator is Focus. There are various covariates of which Concreteness was envisaged as the most important.

# Scores are binary scores for performance by individual learners on individual test items.

# There are ID numbers for targeted multiword items (MWIs) and for learners.

# For Assonance, 1 = 'assonant', 0 = Nonassonant control

# 'Focus' is short for direction of attention to assonance in targeted items plus manipulation of intentional learning. 1 = direction of attention (DA) to assonant items; 0 = DA to control items.

# Formatting is for mixed-effects logistic multiple regression. # We used the R package 'lme4'.

# Each learner was exposed to 26 MWIs of which 13 assonant & 13 do not. There were 60 upper-intermediate to advanced Dutch-speaking undergraduate learners for the 15-minute delayed free recall test & 55 for the cued recall test a week later. One research question (RQ) was whether assonant MWIs are comparatively memorable. Another had to do with the relative effects of form variables & semantic ones.

# Since learners were to be tested on recall of forms not meanings, we wanted target MWIs that learners would be familiar with or which have meanings likely to be inferrable from the constituent words (CWs).

# Half of the learners were asked to identify, write down, & remember the MWIs whose CWs have same vowel sound. Half were asked to do the same with respect to the MWIs whose CWs have different vowel sounds. Our instructions deliberately implied that the 13 not-to-be-remembered MWIs were irrelevant. Learners were randomly assigned to MWI lists & to conditions of focus. There were 4 lists of MWIs. MWIs were in random order on a list. Control MWIs were randomly selected from a pool of 637 candidates. Assonant MWIs were semi-randomly selected from a pool of 56 candidates. We used a simple algorithm to prevent any CW from showing up on the same list of 26 MWIs. Candidate MWIs had been screened to exclude ones the learners might not know or which might lead to a ceiling effect on account of emotiveness (sex toy) or personal relevance (right hand). We also excluded MWIs that might be too hard on account of idiomaticity, being old-fashioned, being too technical, or being restricted to just one variety of English. See the paper for further details.

# The results of the test of recognition are not reported for three reasons. (1) The difficulties involved in interpreting results of a test of recognition that followed a test of recall. Some of these difficulties we had not been aware of when we set the recognition test. (2) Lack of space to describe the results of the recognition test & to discuss the difficulties just mentioned. (3) The marginality of recognition with respect to the important themes of our study..