Symposium
Chair: Gerald W. McRoberts
Discussant: Catherine T. Best
Early research on infant perceptual development began in earnest in the1960's with the work of Fantz, Salapatik and others in the area of infantvisual perception. However, the study of infant auditory perception laggedbehind, due largely to a lack of easy to use and robust researchprocedures. Two notable developments in methodology helped propel the studyof infant auditory perception forward: 1) Horowitz, et al (1972)introduction of an auditory discrimination procedure in which a repeatedauditory stimulus was contingent on fixation of an unchanging visualstimulus; and, 2) Fernald's (1985) demonstration of infants' preferencesfor infant-directed (ID) speech over adult-directed (AD) speech using anauditory preference procedure in which infants were presented with twounchanging visual targets, each paired with a different auditory stimulus.The merger of selected aspects of these two procedures, and othermodifications, resulted in the Infant-Controlled Auditory Preference (ICAP)procedure(s) that are widely used today (e.g., Cooper & Aslin, 1990;Jusczyk & Aslin, 1998; Pinto, Fernald, McRoberts & Cole, 1999).Despite its wide adoption (over 130 papers published in this decade basedon PsycLit) and increasingly diverse application, many researchers continueto associate the ICAP primarily with the study of infant preferences for IDprosody. And, many who do not use the procedure are unaware of advances andvariations in the methodology that have developed in the last few years.This symposium will highlight some current issues regarding the applicationand the interpretation of the ICAP procedure in the study of speech andmusic perception in infants.The first presentation will focus on the history and development of theprocedure and its use to study the development of ID speech preferences ininfants older than 6 months of age. The second presentation will focus onprocedural variations that allow the use of the procedure with infantsthroughout the first postnatal year, while detailing infants' emergingpreferences for ID speech in their native/ambient language. The thirdpresentation will discuss infants' preferences for aspects of music whilegiving an example of how the preference procedure can provide data thatcannot be derived using habituation/discrimination procedures. The fourthpresentation, based on data gathered with the head-turn version of the ICAPprocedure, will focus on interpretation issues that arise from the factthat some studies find preferences for novelty while others findpreferences for familiarity. The discussant will provide summary commentsand (hopefully) guidance for future directions in the development of theICAP procedure.
Details of individual items:
paper
The study of infant auditory preferences began with Fernald's(1985) introduction of the head-turn auditory preference procedure, showingthat young infants prefer to listen to infant-directed (ID) speech overadult-directed (AD) speech. Since then, important developments haveoccurred, both with the auditory preference procedure as a methodology, andin understanding infant preferences themselves.This presentation will trace the historical development of infant auditorypreference procedures and review recent data supporting the idea thatinfant speech preferences undergo signficant develop during the second halfof the first postnatal year. These developments may not be unrelated.Various procedural and methodological developments have improved on theoriginal procedure. For example, the development of infant-controlledprocedures, and a shift in the dependent measure from the number of timesthe infant chose to listen to a particular stimulus type (e.g., ID vs ADspeech; Fernald, 1985) to the amount of time the infant listened to eachstimulus (e.g., Hirsh-Pasek, et al, 1987; Cooper & Aslin, 1990) very likelyincreased the sensitivity of the preference procedure.At the same time that the procedure has undergone development, reesearch onthe infant speech preferences has shown that newborn infants and1-month-old infants prefer ID speech (Cooper & Aslin, 1990, as do 7 to9month-old infants (Werker & McLeod, 1989). However, while the findings onyounger infants are now well documented, the preferences of infants olderthan 12 months is in question. Infants' preferences for ID over AD speechhas focused on young infants' responses to speech directed to young infants(YIDYoung ID). The preferences of infants older than 12 months areunclear, either for YID or for speech to older infants (OIDOlder ID).Thus, data will be presented that compares younger and older infants'preferences for AD speech vs. YID or OID speech. A preference for OID overAD should not be assumed at either of these ages. For example, prosodicand some linguistic modifications, such as repetition, reach their maximumat about 4 months of age and approach AD speech by 24 months (Stern, etal., 1983). Thus, OID may not hold the attention of younger infants. Onthe other hand, older and more linguistically sophisticated infants mightfind the linguistic properties of AD speech to be equally as interesting asOID speech.Preliminary results, reported at SRCD 1999, indicated several interestingfindings regarding the development of speech preferences: 1) young infants(6M) have a preference for YID speech, but not for OID speech; 2) olderinfants (14M) have a preference for both YID and OID speech, but have asignificantly greater preference for OID. These preliminary data have beenextended in two ways: 1) by completing the original sample of 6- and14-month-old infants (n96); 2) by adding infants at intermediate ages todetermine when the preference for OID speech emerges. The findings will beused to demonstrate the sensitivity of the procedure to detect infantresponses to subtle differences in naturally produced ID speech.
paper
One of the biggest challenges facing researchers of infant development hasbeen the implementation of procedures that accurately reflect infants'perception of the events around them. Although various methodologicalstrategies for studying infant perception have evolved over time, we feelthat those protocols which measure infants' looking behavior areparticularly well-suited to the study of perceptual development. Morespecifically, we will present evidence that the visual-fixation based,auditory preference procedure is a convenient and reliable method forstudying infants' attention to speech across the first postnatal year,given its flexibility regarding both stimulus factors (e.g., visual andauditory) as well as subject factors.Generally, we have tested infants in a procedure that allows them to listento some speech recording as long as they continue to look at a visualdisplay (e.g., a checkerboard, colored circles, a face). When the infantlooks away, both the display and the recording are terminated. When thedisplay is presented again on the next trial, the infant's fixation of thedisplay triggers a second speech recording. The two recordings arealternated across trials (typically 10) and the average amount of attentionto each recording is then compared. By manipulating various aspects ofthis procedure, we have been able to examine a variety of preferences ininfants across the first postnatal year. For example, we investigated 1- to2-month-olds' preferences for infant-directed (ID) speech segments thatvaried in rate. When given a choice between ID-normal and ID-fastrecordings (using a colored circle display), the infants showed longeraverage looking times to ID-normal speech. However, when tested withID-normal and ID-slow recordings, infants showed longer average lookingtimes to ID-slow speech (Cooper, Cooper, Ostroff, & Aslin, 1999). Werecently replicated this preference for ID-slow speech with 4-month-olds.Thus, holding pitch information constant, these experiments demonstratethat infants are sensitive to the temporal features of ID speech.We also examined infants' preferences for their fathers' and mothers'face+voice information. Testing 4-month-olds, we failed to findpreferences for fathers' voices (either AD or ID), faces, or voices+faceswhen compared to those of an unknown male. However, when usingmother/other face+voice displays, 4-month-olds looked significantly longerto their mothers' faces and voices (Ward & Cooper, 1999a and b). Itappears that infants' develop early, multimodal preferences for theirmothers, but do not develop similar preferences for their fathers untilsome later point in time.We also examined preferences for ID-native and ID-foreign speech recordingsin 10-month-olds. Using the colored circle display, infants preferredID-native over four distinct versions of ID-foreign speech. Ten-month-oldsalso showed preferences for an ID-native recording matched in overall tempoand pitch information over an ID-foreign recording. However, infantspreferred an ID-foreign speech recording over an AD-native speechrecording, suggesting that exaggerated pitch characteristics of ID speechare necessary for infant attention to their native language (Ostroff &Cooper, 1999).
paper
Singing to infants is universal across cultures, but little workhas addressed infants' musical preferences. In recent research we haveshown that infants have musical preferences for both structural andperformance aspects of music. Structure refers to the pattern ofrelative pitches and durations defining a piece of music, whereasperformance includes the actual pitch, tempo, and timbre (instrument orvoice quality) of a particular rendition. To examine whether infants are sensitive to performance aspects,we tested 6-month-olds' preferences for infant- directed (ID) versusinfant-absent (IA) versions of the same song. On one trial infants werepresented with one type of singing (e.g., ID) from a speaker on theirleft and on the next trial with the other type of singing (e.g., IA) froma speaker on their right. Such trials alternated; on each trial thesinging played for as long as the infant looked at an illuminated toyunder the speaker. Infants listened reliably longer to ID over IAversions. The ID versions were slower, higher in pitch, and more lovingin voice quality. Which of these performance characteristics are crucialto infants' preferences? To date our studies have shown that thedegree of infants' preference for an ID over AD version is correlatedwith adults ratings of the relative 'lovingness' of the ID over the ADversion. We have also shown that infants prefer higher- overlower-pitched women's singing. In sum, we have shown that infants havedefinite performance preferences, and we have begun to figure out whichperformance characteristics matter to them. In our first study of infants' structural preferences we examinedtheir reactions to consonance (tones that sound pleasant togetheraccording to adults) versus dissonance (unpleasant). Consonance anddissonance are central to musical structure across musical systems, asthey are related to the ebb and flow of tension and relaxation thatconstitutes musical meaning. There is reason to believe that dissonancearises initially from the critical bandwidth structure of the basilarmembrane, thus making it a good candidate to emerge early in life. Usingthe preference procedure described above, we showed that 6-month-oldsprefer consonance over dissonance both in isolated intervals as well asin a Mozart minuet. With younger infants we monitor eye fixation on acentral bull's-eye. Half of the infants get three consonant trialsfollowed by three dissonant trials and the other half get the reverse.To date, the results show that in the former case, 4-month-olds reducetheir looking time from the consonant to the dissonant trials, whereas inthe latter case there is no difference between the initial dissonant andthe following consonant trials. This pattern likely occurs becausehabituation and preference work together in the former case, but againsteach other in the latter. Using the same Mozart minuet, Masataka (underreview) has recently shown a preference for consonance over dissonance innewborn infants of deaf parents. Together, these studies of consonanceversus dissonance clearly show an early, sustained preference for oneaspect of musical structure.
paper
A decade after Horowitz (1975) demonstrated that infant looking times to asimple, unchanging visual stimulus habituate to a repeating auditorystimulus and recover to a change in that auditory stimulus, Fernald (1985)reported that infants prefer to listen (by looking) longer to some auditorystimuli than to others.Jusczyk and his colleagues (Jusczyk, 1998; Kemler Nelson et al., 1995)modified and extended this headturn preference technique to study variousaspects of language development in 4- to 18-month-old infants. Theirstandard paradigm is very similar to the one used by Fernald (1985). Theinfant's gaze is elicited by a blinking light located 90 deg to the leftor right of midline. Once the infant has made a criterion headturn in thedirection of the blinking light, an auditory stimulus is presentedrepeatedly from that location until the infant looks away from theblinking light for 2 consecutive seconds. At least two auditory stimuli(or classes of stimuli) are presented across several trials, and the meanduration of looking to one stimulus compared to the other stimulus istaken as the measure of auditory preference.A modified paradigm introduced by Jusczyk and Aslin (1995) enabled theheadturn preference technique to assess the effects of short-term exposure(learning). Infants are familiarized with a particular auditory stimulus(e.g., a word or a series of sentences containing that word) immediatelyprior to the assessment of auditory preferences for that word compared toone or more novel words. Many studies have reported that infants preferto listen longer to the auditory stimuli that had been briefly presentedprior to the preference test. Thus, in both the standard (no immediatelypreceding familiarization) and the modified (familiarization) paradigm,the prototypical result is a listening preference for the auditory stimulusthat has been experienced more frequently by the infant.However, a number of studies using the modified (familiarization) paradigmhave reported that infants listen longer to the novel auditory stimulusduring test (Aslin, et al., 1998; Echols et al., 1997; Marcus et al., 1999;Saffran et al., 1996). Four new experiments from my lab, in which a totalof 120 8-month-old infants were tested with the modified paradigm, haverevealed both familiarity and novelty effects with the same stimulusmaterials (nonsense words embedded in nonsense-word carrier utterances).The key hypothesized determinant of the direction of infant listeningpreferences (familiarity vs. novelty) is the amount of exposure to thefamiliarization stimulus prior to test. A secondary factor that ishypothesized to influence the direction of infant listening preferences isthe difficulty of encoding information during familiarization, which isdetermined by the auditory context and the age of the infant.A multi-component model of auditory encoding will be presented to accountfor the diversity of results obtained using the headturn preferencetechnique. This model is similar to models of visual encoding and thedynamics of familiarity and novelty preferences (Caplan et al., 1995;Cohen & Gelber, 1975; Hunter & Ames, 1986; Wagner & Sakovits, 1984).