The current study investigated whether adult native English speakers and English-as-a-second-language (ESL) learners exhibit sensitivity to compositional English multi-word sequences, which have a meaning derivable from word parts (e.g., don't have to worry as opposed to sequences like He left the US for good, where for good cannot be taken apart to derive its meaning). In the current study, a multi-word sequence specifically referred to a word sequence beyond the bigram (two-word) level. The investigation was motivated by usage-based approaches to language acquisition, which predict that first (L1) and second (L2) speakers should process more frequent compositional phrases faster than less frequent ones (e.g., Bybee, 2010; Ellis, 2002; Gries & Ellis, 2015). This prediction differs from the prediction in the mainstream generative linguistics theory, according to which frequency effects should be observed from the processing of items stored in the mental lexicon¬¬ (i.e., bound morphemes, single words, and idioms), but not from compositional phrases (e.g., Prasada & Pinker, 1993; Prasada, Pinker, & Snyder, 1990). The present study constituted the first attempt to investigate frequency effects on multi-word sequences in both language comprehension and production in the same L1 and L2 speakers. The study consisted of two experiments. In the first, participants completed a timed phrasal-decision task, in which they decided whether four-word target phrases were possible English word sequences. This task measured how fast participants receptively process a phrase, with their reaction time being the outcome measure. In the second experiment, the same participants completed an oral elicitation task, in which they saw and orally produced target phrases. The outcome measure was the production durations of the first three words (e.g., don't have to worry) in the same target phrases used in the first experiment. Participants were a sample of native English speakers (N=51) and ESL learners (N=52) who can be characterized as being proficient enough to study in an English academic environment (mean internet-based TOEFL scores = 95.52, SD = 6.63) and who had lived in the US for 2-3 years (mean = 2.61 years, SD = 0.56).The results from the first experiment suggested phrase frequency effects in both participant groups and countered the proposal that L2 learners cannot retain information about L2 word occurrences in their memory (Wray, 2002). These results support the prediction from usage-based approaches and further corroborate previous proposals that frequency data from large native English corpora should be representative of the regularities of English input that speakers in general are exposed to (Hoey, 2005; Wolter & Gyllstad, 2013). Moreover, the results entail a need for future L1 and L2 psycholinguistics model to accommodate phrase frequency effects. On the other hand, in the second experiment, both participant groups did not exhibit phrase frequency effects. In light of previous similar compositional multi-word sequence production studies (e.g., Bannard & Matthew, 2008; Ellis, Simpson–Vlach, & Maynard, 2008), which had yielded mixed results, and the results from the first experiment, the absence of the effects in the second experiment could have stemmed from cross-study methodological differences, including the type of experimental tasks used to investigate multi-word sequence frequency effects.
Read
