Assumptions and Rationale
One of the biggest challenges of determining what an infant knows about language is actually tied to language itself. Unlike Piaget (1926), who famously asked older children to reflect on and discuss their understanding of the meaning of words, we have no such luxury of interviewing a 12âmonthâold regarding their wordâreferent links. Even for developmentally simpler skills, we cannot get a 6âmonthâold to give a simple yes or no answer to the question of whether they discriminate two language sounds. It is somewhat paradoxical that language itself is a barrier to understanding language development in infants. The fact that infants have little or no lexical production requires researchers to often turn to tasks that require no language output from the child. Further, infantsâ limited motor skills restrict the measures that can reveal underlying linguistic abilities. Tasks must take advantage of gross motor abilities, such as full head turns (fine pointing or manual selection are difficult); congenitally organized behaviors that infants have strong control over since birth, such as looking or sucking; or, basic psychophysiological responses, such as heart rate.
One of the most valid and reliable tools we have to examine the perceptual skills related to infant language is the habituation task. Habituation is a decrease in a response to a stimulus after repeated presentations. This produces what is termed the habituation curve, a monotonically decreasing behavior in response to a repeated target stimulus. It is a task with a very long history in our field, stretching back to the nineteenth century (for a review, see Thompson, 2009). Indeed, Thompson highlights that the concept is reflected in antiquity: in Aesopâs fables, a fox is quite frightened of a lion upon first meeting him, but becomes less alarmed upon each subsequent viewing. Perhaps Aesopâs example was prescient. Habituation tasks were primarily used for decades with animals (and continue to be used with these populations), with everything from amoebas to dogs showing habituation responses (Harris, 1943).
Considering the long history of the task and ubiquitous nature of the habituation response across other nonâverbal beings (i.e., animals), it is unsurprising that the method was extended to infants in the early twentieth century (see Humphrey, 1933). However, simply habituating an infant to a stimulus is necessarily a bit limiting with respect to what one can say about learning. If, for example, an 8âmonthâold had a reduced behavioral response to the repeated presentation of a phoneme, one could argue that they have formed a memory of that particular sound. But it could also be that the infant is simply tiring. The key to demonstrating that the infant has formed a representation of or learned something about the presented stimulus is dishabituationâan increase in behavioral response to a novel stimulus.
Sokolovâs (1963) comparator model is the classic formulation of this approach. The infant (or adult) has an orienting response to a novel or unexpected nonâthreatening stimulus (e.g., becoming still, looking at the stimulus, reduced heart rate). As it repeats, the infant builds an internal representation of the stimulus. The increasing strength of the representation leads to a greater match between the internal percept and the repeating external stimulus. The initially large orienting response correspondingly reduces as the internal/external match increases. But, if the external stimulus does not match the established internal representation (i.e., a novel stimulus), the infantâs orienting response should reoccur.
Thus, habituation is one of the optimal tasks for testing preâverbal infants as it does not rely on overt productions, but rather on implicit cognitive measures such as those mentioned earlier (e.g., looking time, sucking, heart rate, among others). Further, based on the comparator model, it allows researchers to determine the nature of infantsâ percepts and concepts by testing differing levels of novelty from the habituated stimulus (e.g., changing a habituated word form by one phoneme, or multiple sound changes). If the infantsâ behavioral responses increase to the novel stimulus, it can be concluded that they have the ability to differentiate the habituated and novel stimuli. In this regard, habituation is fundamentally a method to index discrimination ability.
Fantzâs (1964) article in Science on visual habituation in the human infant broadly introduced using this task with very young participants to psychological researchers. However, it is important to note that previous studies had already used habituation with infants, including studies on auditory habituation. For example, Bartoshuk (1962) demonstrated that newborns habituate to tones and dishabituate to tones of a differing intensity, using heart rate as his dependent measure. Once it was determined that infants could habituate and dishabituate to auditory tones, it was a straight road for researchers to examine language sound (i.e., phoneme) discrimination using similar methods.
In one of the seminal works on infant language perception, Eimas, Siqueland, Jusczyk, and Vigorito (1971) used a habituation task with sucking as their measure to investigate 1â and 4âmonthâold infantsâ discrimination of consonants, specifically a voicing contrast. Consonants produced in the same place and manner can differ in the timing of the vibration of the vocal folds. For example, /b/ and /p/ are both produced from the lips and are stops, but they differ in voicing. Vocal cord vibrations occurring approximately 25 ms after the air burst from the mouth (or later) sound like a /p/to English speakers. Vibrations starting before that mark sound like a /b/. In Eimas et al., infants heard a repeated sound contingent on strong sucks. Once their sucking rate decreased by 20%, a novel stimulus was presented in two experimental conditions. In one condition, the novel sound was from the same phonological category (i.e., a new /b/ sound that differed by 20 ms in voicing from the original stimulus) and in the other condition the novel sound came from a different category (i.e., a 20 ms voicing change that crossed the boundary from /b/ to /p/). In the control condition, the same sound was played after the 20% reduction in behavior. Only infants in the differing category condition had a dishabituation responseâincreased sucking when the sound change occurred.
The above experiment highlights some important aspects of infant habituation. First, habituation allows the researcher to test categorical perception in that we can determine if an acoustically different stimulus will engender a continued habituated response or a dishabituation response. As Thompson and Spencer (1966) highlight in their classic list of the characteristics of habituation, âhabituation of response to a given stimulus exhibits stimulus generalization to other stimuliâ (p. 19). Thus, we can assume that the lack of dishabituation to an acoustically different stimulus means that the infant considered it to fall into the same category, or the distinction is too weak to detect. This second explanation is unlikely if you include a condition where a similar magnitude difference elicits a dishabituation response due its crossing of a category boundary, as in the Eimas et al. work.
Second, the use of criterion equated the processing of the stimuli across infants. Based on Sokolovâs (1963) theory, the reduction in the target behavior is commensurate with the increasing robustness of the infantâs memory trace for the stimulus. But, different infants would potentially, and probably, have differing timing with respect to the building of the stimulus memory trace due to individual differences in cognitive skills, particularly attention. By requiring them to reach the same relative decrease, the researcher can assume that they have reached similar processing levels for the stimulus in question.