The Role of Lexical Choice in ElicitedImitation Item Difficulty

Elicited imitation (EI) has been used for decades as a means of examining the development of oral language skills in various contexts including normal native language development (Ervin-Tripp, 1964; Keller-Cohen, 1981; Menyuk, 1963;) abnormal language development (Berry, 1976; Lahey, Launer, & Schiff-Myers, 1983; Menyuk, 1964) and second language development (Hamayan, Saegert, & Larudee, 1977; Naiman, 1974). In recent years there has been a resurgence of interest in its use in the examination of oral language skills in second language learners (Chaudron, Prior, & Kozok, 2005; Erlam 2006; Jessop, Suzuki, & Tomita, 2007; Vinther 2002). For a fairly comprehensive review of this literature, see Bley-Vroman and Chaudron (1994), Gallimore and Tharp (1981), Lust, Chien, and Flynn (1987), and Vinther (2002). An examination of the literature on EI suggests that interest in its use centers around two major investigative efforts, (a) psycholinguistic research into the nature of language competence itself (see, for example, Ellis, 2006, and Erlam, 2006), and (b) research into the possibility of finding an indirect and efficient way to estimate the overall oral language proficiency of second language learners (see Chaudron, et al., 2005, and Radloff, 1992). While these two purposes have much in common, it seems to us from the available literature that they may make quite different demands on the design and administration of EI items. If one is to investigate the nature of interlanguage, certain conditions for the elicitation of responses must be met which assure that, on the one hand, the task taps into the implicit linguistic knowledge of the speaker (Ellis, 2006), and, on the other, that the responses are minimally affected by rote memory (Erlam, 2006). It may be critical, for example, to include specific target structures in the EI stimulus sentences. It may be important, as Erlam (2006) claims, to focus the participant’s attention on meaning with each stimulus sentence, to delay the imitation of stimuli in order to reduce the chances of rote repetition, and to assure that stimulus sentences are repeated under time pressure to simulate conditions of unplanned speech and reduce the likelihood of the examinee’s focusing on the form of the sentence. It may also be crucial to show that the ability of subjects to produce particular forms using the elicitation procedure corresponds to their ability to produce these forms in spontaneous speech. But none of these conditions have been shown to contribute to the concurrent validity of EI measures of second language proficiency. Perhaps this is because those who have created such measures have not experimented enough with various forms of stimulus presentation. However, up until now, for the creator of EI language proficiency measures, the major strategy appears to have been to try large numbers of items differing in length, morphological complexity, and syntactic complexity, on large numbers of subjects of varying language backgrounds and proficiency levels, and to empirically test which items elicit consistent responses and ultimately discriminate between learners whose proficiency levels in the language have been verified by other measures (Graham, Lonsdale, Kennington, Johnson, & McGhee, 2008). Thus, concerning our ability to explain how the imitation of utterances provides a reasonable representation of interlanguage, Bley-Vroman and Chaudron (1994) observe, “We regard it as premature to view elicited imitation as a proven method for inferring learner competence, because a considerable amount of research needs to be conducted to understand how performance under