Perception and Recoverability of Modified English L2 Codas



Journal Title

Journal ISSN

Volume Title



Previous studies have shown that when L2 learners are faced with structures that are illegal in their L1, they tend to simplify such structures (Weinberger, 1987; Benson, 1988; Sato, 1984; Wang, 1995; Hansen, 2004; Yavaş, 2011; among others). This dissertation examines two different strategies of syllable structure simplification, namely, deletion and epenthesis, using two perception experiments. Specifically, this dissertation investigates the recoverability principle (Weinberger, 1994), which suggests that epenthesis is functionally superior to deletion since it results in relatively less ambiguous structures. Even though both deletion and epenthesis convert the relatively complex CVC syllables into simple CV syllables, their outcomes differ in terms of the degree of lexical ambiguity. If we examine a word with a CVC syllable structure such as lead, the following are possible simplification outcomes: (1) Target word Deleted form Epenthesized form lead [lid] [li] [lidə] We can see that the deleted form results in more ambiguity since it could be interpreted as Lee, leaf, leave, lean, lead, leak, leash, lease, etc. The epenthesized form [lidə], on the other hand, results in less potential ambiguity because it can only be interpreted as lead or leader – the word can only be interpreted as leader if the listener’s dialect involves final [ɹ] deletion. This study hypothesizes that words modified by epenthesis should be chosen more frequently by listeners since epenthesis is better when it comes to meaning preservation (Weinberger, 1994). Thus far, all research dealing with this notion of recoverability has been done with production data. This study attempts to document the perception of recoverability by native and non-native listeners of English and examine the implications of the recoverability principle which cannot be examined in production studies. In the first experiment, listeners from three different language backgrounds were recruited - English, Japanese and Spanish. The participants were presented with monosyllabic words with codas modified by either deletion or epenthesis accompanied by pictures of what the words denote, and they were instructed to choose the modified form of each word that best matched the picture based on their judgment. A mixed model regression test was conducted to see if the listeners’ L1 and the sonority of coda consonants significantly influenced the choice of repair strategy (deletion vs. insertion). The findings revealed that epenthesis was significantly preferred over deletion regardless of the listeners’ L1, which provides support to the recoverability principle. The results showed that the choice of strategy (epenthesis vs. deletion) was significantly influenced by the participants’ L1. Furthermore, the choice of strategy was significantly influenced by the sonority profile. And finally, the interaction between language and sonority was also statistically significant. In Experiment 2, listeners from three different language backgrounds were recruited: English, Japanese and Spanish. In addition to the same monosyllabic words presented in Experiment 1, the participants were presented with bisyllabic words to test whether the findings in Experiment 1 are based on the recoverability principle or an overall preference for bisyllabic forms (Wang, 1995). Our findings revealed that the choice of strategy was significantly influenced by the sonority profile, number of syllables, age of onset, vowel duration and word frequency. Even though epenthesis was significantly preferred over deletion regardless of the listeners’ L1, the findings of the two experiments show that the preference for a specific modification strategy cannot be explained only by the recoverability principle or a preference for bisyllabicity; rather, there is a constellation of factors that influence the modification strategy.