Christopher T. Fennell & Janet F. Werker

Dept. of Psychology, University of British Columbia

Chris_Fennell@telus.net

A central controversy in phonological development concerns the question of whether infants represent the full phonetic detail in words from the onset of word-learning or whether the act of learning words is necessary for aiding in the construction of a phonological system. Two recent sets of findings are in apparent contradiction. Stager and Werker (1997) used the ‘Switch’ task to test whether 14-month-old infants are able to use the fine phonetic discrimination capabilities seen at the end of the first year of life when mapping words to objects. In this ‘Switch’ procedure, infants are habituated to two word-object pairings (e.g., Object A – Word A, Object B - Word B), and tested on their ability to detect a switch in the pairing at test (e.g., Object A – Word B). Surprisingly, although infants of 14-months easily learn to map two phonetically dissimilar words (‘lif’ and ‘neem’) on to two different objects, and detect a ‘switch’ when the word-label pairing is violated (Werker, et al., 1998), they failed at this task when the phonetically similar words ‘bih’ and ‘dih’ were used (Stager & Werker, 1997), even though they are demonstrably able to discriminate these two nonce words. It has been proposed that the complexity of mapping sound to meaning for the novice word learner leads to the inattention to fine detail. However, a recent study by Swingley, presented at SRCD 2001, tested 14-month-old infants on their latency to look to an object match when a well-known word vs. a phonetically similar 'mispronounced' word was used (e.g. 'baby' -'vaby'). Looking time to the match was significantly delayed when the mispronounced nonce word was presented in comparison to the correct pronunciation, indicating that infants of this age do have detailed representations of words.

In light of the above findings, we see two possibilities. The first possibility is that the Switch procedure, when compared to the visual latency task, is not sufficiently sensitive to reveal 14-month-olds’ ability to use phonological detail. The second possibility, and the one addressed herein, is that it is infants' a priori knowledge of the stimulus words that is creating this discrepancy. Stager and Werker (1997) habituated the infants to novel words and object combinations, thus making the task one of word learning. Swingley used word and object stimuli that are well known to 14-month-olds, thus making the task one of word recognition. The current study addresses this possibility by using the same task as Stager and Werker (1997), but using well-known words. The study involves habituating 16 infants aged 14 months to two known (as assessed by the CDI) word-object pairings: doll – "Doll" and ball – "Ball". and then testing them on a Switch in the pairing (e.g., object doll – word "Ball"). If the infants notice the switch, it will demonstrate that it is not the task, but the nature of the infants' a priori knowledge, that determines the amount of phonological detail used. If the infants do not notice the switch, further research will be required.

Evidence of success in the Switch task with well known words will also address another debate. One interpretation of the Stager and Werker results is that there is a discontinuity between the phonetic representations used in discrimination tasks and the phonological representations used in lexical tasks. A positive result in the current experiment detracts from this argument by demonstrating that the phonological representations are already present at the earliest ages of lexical use.