Type and Quantity of Feedback Use as a Measure of Language Proficiency

ML U as a measure of language proficiency A rough solution to the problem of assessing developing proficiency is to consider the development of the so-called mean length of utterance (MLU) for a certain learner, cycle by cycle. This measure, although it has many problems (see below) has often been used to give a very rough idea both of differences in proficiency and of the extent to which acquisition is taking place. In this chapter, an utterance is defined as a stretch of uninterrupted speech produced by one speaker bounded by silence or by contributions from other speakers. The performances of twenty learners, whose names are given below, are examined in this study. The activities studied are conversations (A), role plays (J) and play scenes (C), which are described in chapter 6. Further discussion of the internal dynamics of these activities can be found in Volume 11:11.2. These performances are compared with those of a small group of native speakers (the 'controls') in identical activities. This group is composed of two working-class native speakers of English (Martin and Sheila), and of Swedish (Adam and Eva). In addition, Marl and Nora performed these activities in their native language as well as in Swedish. In Table 8.1 we reproduce the results concerning mean length of utterances for learners grouped according to pairs of source and target languages. The table contains the means for the individuals, as calculated over two activity occurrences in cycle 1 and cycle 3, and the differences between the means in cycle 3 and cycle 1 in order to give an idea of the change that has taken place between the two first and two last recordings. The table reports on individual learners rather than on SL-TL aggregations since there are too few individuals and too many potential differences between them really to justify anything but a report on the individual level. Table 8.1 shows that there is indeed considerable individual variation both between learners of a given target language and between learners sharing a source language. Compare Berta and Alfonso who are both Latin American Spanish -speaking learners of French, or Lavinia and Andrea who are both Italian-speaking learners of English. It is not unlikely that these differences reflect a difference in target language proficiency at the beginning of the period of observation and that there are different rates of acquisition for different learners. The differences between the learners with regard to the MU' for cycle 1 results in the following ordering, which we provisionally take to be a first, rough indicator of differences in proficiency in cycle 1: Ravinder (2.0), Çevdet (2.4), Ergün (2.9), Fatima (3), Madan (3.2), Leo (3.3), Fernando (3.5), Man (3.7), Berta (3.7), Zahra (3.7), Mahmut (3.8), Ilhami (3.8), Andrea (3.9), Marcello (4), Tino (4.3), Mohamed (4.4), Abdelmalek (4.4), Lavinia (4.9), Nora (7), Alfonso (15). A general point to notice is that all the learners except one Mohamed show an increase in MLU between cycles 1 and 3.