Poster group
Details of individual items:
poster
Infants' preferences for ID (Infant-Directed) over AD(Adult-Directed) speech has focused on young infants' responses to speechdirected to young infants (YIDYoung ID). The preferences of infantsolder than 12 months are unclear, either for YID or for speech to olderinfants (OIDOlder ID). This study 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. Using an infant-controlled auditory-preference procedure based onCooper & Aslin (1990), McRoberts (1999) assessed infants speech preferencesat 6M & 14M, by comparing AD speech with either YID or OID speech toinvestigate the effects of age and age-of-addressee on ID speechpreferences. Stimuli were excerpts from recordings of mothers interactingwith their 4.5- (YID) or 12- to 18-month-old (OID) infants, and speaking toa female research assistant (AD). Two ID & 2 AD samples were includedfrom 3 mothers at each age. The prosodic characteristics of the ID sampleswere similar to published data at each age (M F0 382 Hz for OID, 350 Hzfor YID, and about 200-210 Hz for both AD samples). At each age, one groupof infants heard OID and another group heard YID, each paired with theirrespective AD samples. Each infant received 12 trials (6 ID/6 AD inalternating order); half began with an ID trial. Preliminary results,reported at SRCD 1999, indicated several interesting findings regarding thedevelopment of speech preferences: 1) 6M infants have a preference for YIDspeech, but not for OID speech; 2) 14M infants have a preference for bothYID and OID speech, but have a significantly greater preference for OID.These preliminary data have been extended in two ways: 1) by completingthe original sample of 6- and 14-month-old infants (n96); 2) by addinginfants at intermediate ages to determine when the preference for OIDspeech emerges. Because preferences for ID speech with prosodic characteristicssimilar to our OID have been reported previously for young infants,non-prosodic factors were suspected to be responsible for the lack ofpreference for OID at 6M. In recent research, 6M infants preferred IDspeech with lexical repetition over linguistically matched ID speech withprosodic repetition but no lexical repetition (Pinto, 1998). Thus, afollow-up study is underway in which the repetition in OID speech wasincreased by repeating utterances presented on each trial. Preliminaryresults show that 6M infants have a preference for this repetition-enhancedOID speech, supporting the idea that 6M infants' speech preferences are atleast partially a function of their attention to the linguistic structurein speech.
poster
Across the first postnatal year infants prefer to listen to speech directedto infants (ID speech) over speech between adults (AD speech). Because IDspeech differs from AD speech on many prosodic and linguistic dimensions,much research has focused on which factor(s) is responsible for increasedattention to ID speech. These studies have found that vocal pitch is amajor influence on infant attention. However, recent studies from our labhave begun to focus on the temporal characteristics of ID speech,specifically speaking rate. We found that 1- to 2-month-olds preferredID-normal over ID-fast speech (ID speech at a rate twice that of normal),but preferred ID-slow speech (ID speech at a rate half that of normal) overID-normal. Because the pitch information in all of these contrasts wasconstant, infants' preferences for the slower speech in both experimentswas most likely affected by speaking rate. Although the preference forID-normal over ID-fast speech could be attributed to experience, infants'preferences for ID-slow over ID-normal speech is more puzzling. It seemsunlikely that the latter preference was based on experience because whenadults slow their speech, absolute pitch and pitch variability typicallydecrease. In other words, it is unlikely that infants have heard speech atsuch a slow rate with exaggerated pitch. To determine whether extensiveexperience with natural ID speech affects infants rate preferences, wetested 4-month-olds with the same ID-normal and ID-slow utterances used totest 1- to 2-month-olds.A preliminary sample of 12 four-month-olds were tested with two sets of IDspeech recordings (normal and slow), which were sequentially paired with avisual display (colored circles). Infants' looking times to the displaywere recorded. Sessions were 12 trials in duration (6 trials of each speechtype), with order counterbalanced. A 2 X 2 mixed ANOVA was performed onthe mean looking times as a function of speech type. The between subjectsvariable was order (ID-normal first, ID-slow first), and speech type(ID-normal, ID-slow) was the within subjects variable. Infants lookedsignificantly longer on the ID-slow speech trials (M 20.72 sec., SD 6.0) compared to the ID-normal speech trials (M 15.6 sec., SD 5.6; F(1,10)8.14, p < .02). No other factors were significant. Thus, both 1- and 4-month-olds attend more to ID-slow over ID-normalspeech. It seems unlikely that this preference is the result ofexperience, since infants would not have had contact with ID speechcontaining the pitch and rate characteristics of the ID-slow samples. Onepossibility is that slower speaking rates may facilitate infants'achievement of states that are maximally conducive to sustained attention.Additionally, given the moderate complexity of our visual display, infantsmay have been able to attend longer when concurrently presented with lesscomplex speech. That is, as the complexity of the speech increases (e.g.,ID-normal), visual fixations to more complex visual events may beattenuated. We plan to test this hypothesis by decreasing the complexityof the visual display, and measuring its impact on infants' speechpreferences.
poster
The language input to which infants attend has important consequences forthe development of speech perception. For instance, infants preferentiallylisten to infant-directed (ID) speech over adult-directed (AD) speechacross the first postnatal year, in both native and nonnative languages.Attention to this specialized speech register seems to ease the task ofphonetic perception. In one study infants could only discriminate theword-embedded phonemes /ra/ and /la/ when they were exaggerated withinfant-directed intonation. Since infants perceptually attune to only thosephonemes that are contrasted in their native language by 10- to 12-monthsof age, the information that ID speech affords infants during this timeperiod is of special interest. In other words, it is important to determinewhether ID speech highlights native linguistic information, therebycontributing to the native language tuning process. In a recent experimentfrom our lab, 10-month-olds displayed preferential attention tonative-language ID speech over nonnative ID speech in various tonal andnon-tonal languages. The present experiment was designed to take a closerlook at this preference. Specifically, 20 native-English 10-month-old infants were tested in asequential auditory preference procedure for their attention to ID speechin their native language (English) versus ID speech in a foreign language(French), with the melodic structure of the utterances held constant. Thatis, the speech stimuli were recorded from a single bilingual mother readingnursery rhymes in each language. This produced speech samples with nearlyidentical intonation patterns (e.g., alike in absolute pitch, pitchvariability and duration). Infant looking times corresponding to each typeof speech were analyzed by a mixed ANOVA with order (ID-English first vs.ID-French first) as the between-subjects factor and speech type (ID-Englishvs. ID-French) as the within subjects factor. The results of the ANOVAindicated a significant order by speech type interaction, F (1,18) 4.88,p < .05. Analyses of the simple main effects revealed that among thoseinfants who heard English first, there was a significant difference betweenattention to ID-English speech (M 9.27 sec, SD 3.19 sec) and ID-Frenchspeech (M 6.14 sec, SD 1.92 sec), t (9) 2.94, p < .05 (2-tailed).Among the infants who heard French first, there was no significantdifference between attention to ID-English speech (M 8.12 sec, SD 4.16sec) and ID-French speech (M 7.9 sec, SD 2.3 sec), t (9) -.278, p >.05 (2-tailed). Overall, the 10- month-old infants in this experimentshowed a significant preference for the ID-English nursery rhymes overID-French nursery rhymes when the English rhymes were presented first.Although there was a trend in this same direction for the other order(French rhymes first), this mean difference was not significant unless thefirst looks of the session were eliminated from the analysis. Nevertheless,it appears that the10-month-olds were attending to linguistic-levelinformation in the speech, as the intonation patterning across the speechtypes was equated. This finding suggests that ID speech may in facthighlight linguistic information for infants during the time of perceptualtuning to native-language phonetic characteristics.
poster
When addressing infants, adults systematically alter their speech. Asearly as four weeks after birth, infants show a preference for this speechregister, ID speech or 'baby talk' (Cooper & Aslin, 1990). BT ischaracterized by features including elevated F0, elongated vowels, andexpanded F0 contours (Fernald & Kuhl, 1987; Pegg, Werker & McLeod, 1992).However, in certain circumstances, these acoustic cues may also becharacteristic of ADS, particularly in utterances communicating positiveaffect. Because previous studies have failed to control for affectivecontent of BT and ADS, it is possible that the observed preference for BTis due to the appeal of positive affective cues typical of the BT signal.Thus, infants' well-known preference for baby talk may actually be apreference for happy talk.The present research investigated the relationship between affective toneand register in infant listening preference. Identical sentences wereembedded near the end of brief stories designed to establish happy, sad,or neutral affective contexts. A speaker read these in BT or ADS manner.This permitted production of stimuli controlled for syntactic complexityand lexical content across registers. Adult subjects rated the sentences,using 7-point Likert scales, for affective tone and register. Accurateratings were obtained for both affect and register using a forced-choiceprocedure. Acoustic analyses revealed distinct prosodic characteristicsfor BT and ADS. As expected, BT and positive affect were somewhatconflated, and it proved difficult to identify instrumentallydisambiguating acoustic cues. However, these factors were reliablydistinguished perceptually, evidenced by accurate ratings from adultsubjects.Infants' preferences were tested using a standard headturn preferenceprocedure. In an initial control study, infants showed a reliablepreference for positive BT over neutral ADS, replicating the classicfinding in this area. In Experiment 1, preference for speech register wasmeasured while affective tone was held constant. Results showed noevidence of BT preference for any type of affect. Experiment 2 measuredaffective preference within register. In this study, infants preferredhappy talk to neutral talk within each of the BT and ADS registers. Mostimportant, Experiment 3 altered the usual correlation of affect andregister. In this study, infants displayed a significant preference forhappy ADS over neutral BT, a reversal of the classic finding. Experiment 4measured affective preference for neutral BT over sad AD, revealing apreference for neutral BT. This suggests that infants' preference areguided specifically by positive affective tones, rather than by affectiveintensity alone. These results suggest that the so-called BT preference is largely apreference for particular affective properties of speech. Mammals havebeen using vocal means for expressing affect to their offspring since longbefore human language appeared; evolutionary studies have documented theadaptive significance of selective attention to positively-toned stimuliin infancy (Scherer, 1986). Ongoing studies are investigating the extentto which speech affect may modulate infant attention, in addition topreference, and the manner in which positive affect may influence infantspeech processing.
poster
Previous research has shown that 6-month-olds categorized 650 Hz low-pass filtered infant-directed (ID) utterances, but that 4-month-olds did not. These results suggest that prosodic information without segmental content provides sufficient support for ID categorization by 6-month-olds. These results, however, do not allow us to determine if 4-month-olds' failure to categorize resulted because a) low-pass filtered stimuli lack the additional (higher) spectral cues that infants typically hear in natural ID speech and that may facilitate their categorization of ID utterances, or b) there is a developmental progression from 4 to 6 months in infants' categorization of ID stimuli. The present study examined categorization of naturalistic, unfiltered ID utterances by both 4- and 6-month-old infants. The unfiltered versions of the approving and comforting ID stimuli used in our previous research were presented to infants in the current study. In Experiment 1, 4-month-olds (n 32) heard 7 different unfiltered tokens from one class of ID utterance (approving or comforting), one in each of 7 familiarization trials; this was followed by either a novel and discriminably different test token from the same category (control group) or a novel test token from the unfamiliar category (experimental group). ID stimuli were presented contingent on infant fixation of either of two computer monitors on which identical visual stimuli were presented. Because it indexed attention to the auditory stimuli, the primary dependent variable of interest was total duration of looking (TDL) at the visual displays during three trial blocks. Block 1 consisted of averaged TDL data from trials 1 and 2, block 2 consisted of averaged TDL data from trials 6 and 7 (final two familiarization trials), and block 3 consisted of TDL data from trial 8 (test trial). Both age groups decreased TDL across familiarization trials (ps < .01). Recovery of responding on the test trial was examined with a group (experimental vs. control) X familiarization stimulus (approval vs. comfort) X trial block (last familiarization block vs. test block) repeated measures ANOVA. Four-month-olds did not categorize; they did not increase looking to a test token from an unfamiliar class. Six-month-olds (n 41) presented the same unfiltered ID stimuli in the same procedure did provide evidence of categorization. These infants recovered responding to the unfamiliar but not to the familiar category token, revealing that they categorized ID utterances which conveyed both prosodic and segmental information (F (1, 38) 4.55, p < .04). The results of these studies, combined with those of Moore et al. (1997), reveal a developmental progression in infants' categorization of ID speech. These findings also provide evidence that 6-month-old infants can treat as similar physically and discriminably different ID utterances that are produced in similar contexts and that are intended to communicate similar messages. The development of this ability, a prerequisite for responding to utterances as meaningful communicative signals, has implications for infants' developing social and communicative competence.
poster
It has been found that infant-directed speech (IDS) differs from adult-directedspeech (ADS) in its higher overall pitch and pitch variation, longer durationof syllables, hyperarticulation of vowels, and greater affective salience.Three functions of IDS have been suggested: to attract and maintain attention;to convey and elicit positive affect; and to teach language. It could behypothesised that pet-directed speech (PDS) would share, with IDS, theattentional and affective functions, but not the linguistic/didactic function.The purpose of the study reported here is to delineate more clearly these threefunctions of IDS and the measures used to identify these, especially incomparison with PDS. Twelve mothers of 6-month-old infants were studied. Eachof the mothers also had a pet, a cat or dog. Mothers were left with a goodquality tape recorder which could be worn over the shoulder, and a lapelmicrophone, for several days, and asked to record IDS, PDS, and ADS at theirleisure in their home. They were also given three toys, a sheep, shoe, and ashark, and asked to play with these with their infant and their pet in theseparate IDS and PDS sessions. ADS with the experimenter was recorded inanother session, and included some discussion of the toys used in theexperiment. The toys were employed to ensure that the vowels /i/, /u/, and /a/were present in the three styles of speech. The data were subjected toacoustic, phonetic, and affective analyses. There were no substantialdifferences between the speech of mothers to boys versus girls, or cats versusdogs. Acoustic measures were conducted on the words 'sheep', 'shoe', and'shark', extracted from the 12 mothers' IDS, PDS, and ADS. Mean F0 ( pitch),pitch variation, and syllable duration, were all much higher in IDS and PDSthan in ADS, and for F0 and duration IDS was higher than PDS. Low-pass filteredspeech samples (25-sec duration) were extracted from each of the 12 mothers'IDS, PDS and ADS and rated by 20 undergraduates blind to the speech style, onfive scales. A factor analysis of the rating scale scores, produced twofactors an Affective Salience and a Directing Behaviour factor. Analysis ofvariance of factor scores revealed statistically equivalent Affective Salienceand Directing Behaviour in PDS, higher Affect than Directing in IDS, and higherDirecting Behaviour than Affective Salience in ADS. The first three formants,F1, F2, F3, of the vowels, /i/, /u/, /a/, in mothers' productions of the targetwords were measured to investigate the degree to which the speech revealeddidactic hyperarticulation. These analyses revealed that individual mothers hadparticular patterns of hyperarticulation, but no overall consistent pattern wasevident. The results of the three sets of analyses are discussed in relation tothe three possible functions of IDS, and the possible functions of PDS, bothwith regard to the expected effect on the recipient, and the facilitativeeffect that this style of speech may have for the speaker's expression ofemotion.
poster
Motherese speech produces the highest listening discrimination in an operant auditory preference procedure among infants compared to adult directed-speech (Fernald,1985). Experiments by PelE1ez et al. (1995, 1998) demonstrate that contingent maternal vocal imitation produces the highest rates of infant vocalizations compared to noncontingent vocal stimulation and compared to motherese speech. The present experiment expands on theses findings by using a single-package treatment which combines motherese speech and contingent maternal vocal imitation in an alternating pattern. 10 infants, 4-to-10-mos old, with delayed receptive and expressive language skills participated in daily experimental sessions for approximately one week each. Daily sessions lasted 30 minutes and consisted of between 10 to 20 trials of one minute each. The infant receptive and expressive language skills were assessed through the Receptive-Expressive Emergent Language Scale for the Measurement of Language Skills in Infancy (REEL Scale 2). Procedure. A single subject A1B1A2B2 reversal design was implemented across four phases: Baseline (A1), motherese/imitation treatment (B1), reversal/extinction (A2), and motherese/imitation treatment (B2). The baseline (A1) measured the number of infant vocalizations in the absence of any motherese or vocal stimulation. During the treatment phase (B1), the experimenter produced motherese speech with a high pitched voice and exaggerated intonation to elicit infant vocal responses. Vocal imitation made by an experimenter immediately followed any of the infant vocalizations. When an infant did not vocalized within 10 sec, the experimenter began motherese speech until the next infant vocalization was elicited. During the reversal/extinction (A2) phase experimenter no vocal stimulation followed infant vocalizations and continued until an extinction effect was observed. In the last treatment phase (B2), the motherese/imitation stimulation was reinstated. Results suggest that infant vocalization rate doubled, and in some subjects tripled, that of baseline and reversal/extinction phase. Few studies have included a package treatment that combines several effective sources of contingent adult stimulation (e.g., PelE1ez, et. al. 1996), however, this is the first study to find that both motherese speech and maternal vocal imitation combined can serve as a powerful technique to increase infant vocalizations rate, especially the vocalizations of those infants showing language delays.