|Year : 2022 | Volume
| Issue : 1 | Page : 18-24
Articulation rate and speech rhythm in child-directed speech and adult-directed speech
Jyothi Shivaswamy1, Santosh Maruthy2, Animesh Barman1
1 Department of Audiology, All India Institute of Speech and Hearing, Mysore, Karnataka, India
2 Department of Speech-Language Sciences, All India Institute of Speech and Hearing, Mysore, Karnataka, India
|Date of Submission||24-Dec-2021|
|Date of Decision||31-Jan-2022|
|Date of Acceptance||26-Feb-2022|
|Date of Web Publication||27-Jun-2022|
Department of Audiology, All India Institute of Speech and Hearing, Manasagangothri, Mysore - 570 006, Karnataka
Source of Support: None, Conflict of Interest: None
Introduction: Studies have demonstrated that mothers exaggerate linguistic, segmental and supra-segmental properties of Child-Directed Speech. However, these studies have majorly focused on acoustic characteristics of pitch and its related measures. There has been relatively little research on speech rhythm. Though many studies report slower articulation rate relative to Adult-Directed Speech, still there is no conclusive evidence across different languages. Aims: This study aims to examine articulation rate and rhythm in Kannada speaking mothers, a Dravidian language which is less explored. Methods and Material: Twenty-five dyads of mothers and their children were recruited from the local community through random sampling for the study. The mean age of these children was 2.08 years (SD = 0.61, range= 1.89). The mean age of mothers was 30.96 years (SD= 3.44, range= 13). Articulation rate was calculated by dividing the number of syllables per second by the total duration of fluent speech in each two-minute sample. Speech rhythm was measured using an automatized approach, i.e., Envelope Modulation Spectra (EMS). Results: Results demonstrated that mothers spoke slower to their children when compared to speaking with the adult supporting the universality nature of Child-Directed Speech. Conclusions: However, results showed no conclusive evidence for the analysis of speech rhythm and hence gives a future direction to explore on the use of EMS in the normal population is mandated.
Keywords: Adult-directed speech, articulation rate, child-directed speech, speech rhythm
|How to cite this article:|
Shivaswamy J, Maruthy S, Barman A. Articulation rate and speech rhythm in child-directed speech and adult-directed speech. J Indian Speech Language Hearing Assoc 2022;36:18-24
|How to cite this URL:|
Shivaswamy J, Maruthy S, Barman A. Articulation rate and speech rhythm in child-directed speech and adult-directed speech. J Indian Speech Language Hearing Assoc [serial online] 2022 [cited 2022 Aug 12];36:18-24. Available from: https://www.jisha.org/text.asp?2022/36/1/18/348428
| Introduction|| |
Adults often change their speaking style when interacting with young infants or children commonly referred to as infant-directed speech (IDS) or child-directed speech (CDS), respectively.,, Compared to adult-directed speech (ADS), CDS is majorly characterized by the use of higher pitch and exaggerated pitch contours and slower tempo and hyperarticulated vowels and is more rhythmic in nature., Researches in the last few decades have shown that CDS fulfills three main functions: first, it helps to hold children's attention and arousal levels, second, to communicate maternal affect, and third, in facilitating language acquisition., The speech adjustments which adults make while communicating with children or infants can be summarized under linguistic and suprasegmental characteristics.
Linguistic characteristics of child-directed speech
Studies have demonstrated that adults adjust the linguistic complexity, especially the semantic and pragmatic features, according to the child's age and stage of language development., Studies report that in CDS, mothers produced sentences that were grammatically simplified or incomplete. And their speech consisted more repetitions, questions and commands.
Decreased speech rate is one of the unique characteristics of CDS. It is reported that slower speaking rate can enhance word recognition, and a better expressive vocabulary in children. Speaking rate is defined as the speed at which an individual produces articulatory movements for the production of speech. It is measured by either speech rate or articulation rate (AR), both defined as “the number of output units per unit of time.” Speech rate includes pause intervals, whereas AR does not include them. Studies have reported that AR is lower in CDS than in ADS.,, However, few studies did not find an increase in AR with the age of the child and also demonstrated cross-linguistic differences.,, These results suggest that there is still no conclusive evidence that the AR of CDS is slower than ADS across languages. Hence, more cross-linguistic studies are warranted.
Suprasegmental characteristics of child-directed speech
The prosodic aspect of mothers' speech to children is less understood. Although few studies report that CDS is more rhythmic, higher in pitch, and contains slower, more exaggerated pitch contours than ADS,, only increase in pitch has been documented in cross-linguistic studies. Payne et al. used interval-based rhythm metrics to analyze the speech, and the results showed that all three languages (English, Catalan, and Spanish) were more “vocalic” (higher %V) than adult speech.
The results of these studies may suggest a biological emphasis or “universality” in CDS across languages that may serve a common function., However, few studies have reported language-specific variation as well. It can be seen that most of the research work carried out are on “stress-timed” languages and dialects. However, there is a dearth of literature in Indian languages where majority of the languages are “syllable timed.” Particularly, there are no studies in Kannada (a Dravidian language spoken in Karnataka).
The standard rhythm metrics used to measure the rhythm of speech require manual marking of vocalic and consonantal intervals on a spectrographic display which can be challenging and time consuming.,,, To overcome these problems in the present study, an attempt has been made to use automated measurement of the temporal regularities in the amplitude envelope of the speech waveform. Envelope modulation spectra is a spectral analysis of the low-rate amplitude modulations of the envelope for the entire speech signal and within select frequency bands. This measurement requires no segmentation, no special procedures for silent pauses, and no linguistic assumptions. The present study aims to compare and explore on CDS and ADS with respect to the AR and rhythm in Kannada.
| Materials and Methods|| |
Twenty-five dyads of mothers and their children participated in the study. They were recruited from the local community through random sampling. A sample size of 25 mothers was proposed initially. Thus, the initially proposed sample size of 25 participants was well above the reported G*power sample size. with α level as 0.50, β level as 0.84, and the effect size of existing investigations. Investigators calculated the effect size for AR and speech rhythm reported in the previous studies based on the given mean and standard deviation (SD) values. The sample size calculated using G*power ranged from 21 to 23 participants. A self-reported questionnaire was administered on mothers to rule out any history of audiological, psychological, neurological, or communication disorders [Appendix 1].
[Table 1] shows the demographic details across individual children and their mother. The mean age of these children (10 males and 15 females) was 2.08 years (SD = 0.61, range = 1.89). The mean age of mothers was 30.96 years (SD = 3.44, range = 13). All the mothers were native speakers of Kannada (a Dravidian language spoken in the state of Karnataka, India). Mothers' educational levels ranged from a higher primary education to a master's degree. Informed written consent was obtained from the participants. This research and the recruitment of the participants were approved by the Ethical Committee.
|Table 1: Demographic details of individual participants and their children|
Click here to view
Digital recording of mother's speech was carried out in a sound-treated laboratory using head-mounted wireless microphone (Sennheiser PC 3) linked to an amplifier (ProElite 6 Channel External Sound Card 5.1 Surround Sound USB 2.0) with 16-bit resolution and a laptop (Sony Model PCG 71811W) with a sampling rate of 44 kHz. Sony Cyber-Shot Digital Camera (model-15DCR-SR88E) was used to video recording. The camera was positioned in such a way that it was not visible to the mother and the child, allowing a natural interaction between the child and the mother.
The study comprised two conditions, CDS and ADS. In CDS condition, mothers were asked to sit with their child comfortably on a chair. Quiet animal models were given to the mother and instructed the mother was instructed to play and interact with the child as they normally would do at home. In ADS condition, an adult experimenter conducted a semi-structured interview with each mother asking questions related to the interaction that they had with their child in CDS condition. Each recording lasted approximately 7–8 min for both the conditions. Both the conditions employed spontaneous speech in order to generate naturalistic sample as possible. In total, there were 50 recordings (CDS condition: 25 recordings; ADS condition: 25 recordings).
For the analyses, the CDS and ADS recordings were divided into segments. A segment is defined as a period of mothers' speech not interrupted by vocalizations of the child or noises from the environment. PRAAT software (version 6.0.43) was used to identify and excise these segments. All the speech samples were transcribed manually using orthographic transcription.
Measurement of articulation rate
The analysis was carried out in PRAAT software (version 6.0.43). As initial few utterances had significant number of hesitations and pauses, initial 60 second utterances of mothers were not considered for the study. AR was calculated by dividing the number of syllables per second by the total duration of fluent speech. All the disfluent utterances, hesitations, and silent pauses of 250 ms or longer were excluded from the analysis.
Measurement of rhythm
Envelope modulation spectral analysis was employed for the measurement of the rhythm. The study attempted to calculate modulation spectra for amplitude envelopes extracted from the full signal and seven-octave bands (center frequencies of 125, 250, 500, 1000, 2000, 4000, and 8000 Hz). From each of these eight modulation spectra, six variables, namely peak frequency, peak amplitude, energy 3–6 Hz, energy 0–4 Hz, energy 4–10 Hz, and energy ratio, were computed. These yielded a total of 48 variables (8 envelopes ×6 metrics) which were calculated using a fully automated program developed in MATLAB (MathWorks).
Each participant's recordings were processed, wherein the MATLAB code was generated for all 48 variables for both conditions. To obtain these variables, the speech signal was filtered into its constituent octave bands using a pass-band eighth-order Chebyshev digital filters, following which the amplitude envelope of the entire signal was extracted by half-wave rectification, and filtered using a 30-Hz low-pass fourth-order Butterworth filter. It is then down-sampled (to 80 Hz, mean subtracted). The power spectrum of the down-sampled envelope was calculated using a 512-point fast Fourier transform with a Tukey window and converted to decibels for frequencies up to 10 Hz (normalized to maximum autocorrelation). The result was the six EMS metrics which was computed from the resulting spectrum for each band and also the full signal resulting in a total of 8 frequency bands. For each speaker, the average values obtained in both conditions were calculated across eight envelopes for six metrics (totally yielding 48 variables for each participant in CDS and ADS conditions).
Intra-judge and inter-judge reliability of articulation rate
Intra-judge and inter-judge reliability measures were obtained for 5% of the recorded samples. The first examiner reanalyzed the data and measured AR for intra-judge reliability after a week. For the inter-judge reliability another speech language pathologist, who was a native speaker of Kannada language with a minimum experience of 5 years in acoustic analysis and analyzed the samples. Cronbach's alpha value was calculated for both intra-judge and inter-judge reliabilities.
| Results|| |
Statistical analysis was performed using IBM SPSS software version 20.0 (Armonk, NY: IBM Corp). The Shapiro–Wilk test of normality revealed that the data were normally distributed(/p/> 0.05) and hence parametric tests were done. AR was compared between the two speech conditions, CDS and ADS, using independent-sample t-test. For the analysis of rhythm in both the conditions, six dependent variables were taken into consideration, i.e., peak frequency, peak amplitude, energy 3–6 Hz, energy 0–4 Hz, energy 4–10 Hz, and energy ratio across eight-octave bands. Two-way repeated measures ANOVA (with the condition and octave bands as within-subject factors) was done to study the main effects of the condition, octave bands, and interaction between condition and octave bands. The statistical alpha corrected P value was considered to be/P/= 0.05, and the effect sizes were reported as partial eta square for ANOVA.
The AR was calculated by number of syllables per second (excluding the pauses of 250 m or longer). Mean values were obtained for both the CDS and ADS conditions. [Figure 1] shows the comparison of overall mean ARs between CDS (M = 5.09, SD = 0.75) and ADS (M = 5.94, SD = 0.71). Independent t-test was done to determine whether there was a statistically significant difference between the ARs in both conditions. The overall results suggested a statistically significant decrease in the rate of speech during CDS when compared to ADS (t (48) = −4.05, /p/< 0.01). This result was in accordance with the previous studies by Fernald and Simon in German-speaking mothers, Bernstein-Ratner in American-English mothers, Tang and Maidment in Cantonese-speaking mothers, and Van de Weijer in Dutch and German speakers. In [Figure 1], syllables per second are on the Y-axis and the CDS and ADS conditions are on the X-axis. Error bars indicate SD values.
|Figure 1: Comparison of articulation rate (syllables per second) between child-directed speech and adult-directed speech conditions. Error bars indicate standard deviation values|
Click here to view
The descriptive data (mean and SD) for six variables, namely peak frequency, peak amplitude, energy 3–6 Hz, energy 0–4 Hz, energy 4–10 Hz, and energy ratio across eight-octave bands in both speaking conditions, i.e., for CDS and ADS, are given in [Table 2].
|Table 2: Means and standard deviations for all six envelope modulation spectral variables (peak frequency, peak amplitude, energy 0-4 Hz, energy 4-10 Hz, and energy ratio) across eight frequency bands (full band, 125 Hz, 500 Hz, 1 kHz, 2 kHz, 4 kHz, and 8 kHz) for both child-directed speech and adult-directed speech conditions|
Click here to view
[Table 3] shows F values, uncorrected /P/values, and partial eta-squared effect size comparison of inferential statistical analyses for all variables (peak frequency, peak amplitude, energy 3–6 Hz, energy 0–4 Hz, energy 4–10 Hz, and energy ratio) between the two conditions. The mean peak frequency values were found to be higher in the CDS condition when compared to ADS condition. Comparable results were obtained between the two conditions for the remaining five variables (peak amplitude, energy 3–6 Hz, energy 0–4 Hz, energy 4–10 Hz, and energy ratio).
|Table 3: F values, uncorrected/P/values, and partial eta-squared effect sizes (ηp2) for all inferential statistical analyses for all variables (peak frequency, peak amplitude, energy 3-6 Hz, energy 0-4 Hz, energy 4-10 Hz, and energy ratio) between child-directed speech and adult-directed speech|
Click here to view
Repeated measures ANOVA showed a statistically significant (/P/< 0.05) main effect for octave bands for variables, namely peak frequency, peak amplitude, energy 0–4 Hz, and energy 4–10 Hz. However, there was no statistically significant main effect for the two conditions. A significant interaction effect between condition and octave bands (/P/< 0.05) was established only for peak frequency [Table 3]. Hence, a separate rANOVA was run for each condition, i.e., CDS and ADS. The results suggested that there was no main effect across octave frequencies for CDS condition(F [4.15699.74 = 1.726, /p/= 0.148]). However, in ADS condition, there was significant main effect across octave frequencies (F (1.790, 42.951) =21.356,/P/= 0.000). Bonferroni correction was applied and so all the effects are reported at <0.00625 level (0.05/8). Bonferroni post hoc analysis suggested a significant difference between the following octave bands: full band versus 125 Hz, full band versus 8 kHz, 250 Hz versus 8 kHz, 1 kHz versus 125 Hz, 1kHz versus 8 kHz, 2 kHz versus 8 kHz, 4 kHz versus 8 kHz.
Further, independent t tests were done to compare between CDS and ADS conditions across eight octave bands. The results suggested statistical significant difference between conditions for full band, 500 Hz, 1 kHz, and 8 kHz octave band. (full band [t (48) = 4.91,/P/< 0.01], 125 Hz [t (48) = −0.81,/P/= 0.42], 250 Hz [t (48) = 2.77,/P/< 0.008], 500 Hz [t (48) = 4.25,/P/= 0.01], 1 kHz [t (48) = 3.28,/P/< 0.002], 2 kHz [t (48) = 1.51,/P/= 0.13], 4 kHz [t (48) = 2.40,/P/= 0.02], and 8 kHz [t (48) = −3.03,/P/= 0.004].
| Discussion|| |
The objective of the current study was to compare CDS and ADS in Kannada-speaking mothers pertaining to two aspects: (a) AR and (b) speech rhythm. Spontaneous speech of 25 mother–child dyads was analyzed. Mothers were instructed to play with their children to obtain CDS and interact with the adult for ADS.
The results showed a significant difference between the CDS and ADS conditions, i.e., mothers spoke slower to their children (5.09 syllables/s) when compared to speaking with the adults (5.94 syllables/s). It can be inferred from the study that Kannada-speaking mothers tend to slow down their speech with the intent to enable the child to understand and facilitate word learning. Slower speaking rate presumably makes sounds more distinct from one another and gives infants with additional perceptional information when compared to fast speaking rate in ADS. In our study, we also found that mothers tend to simplify and repeat utterances more often to their children. Furthermore, slower speaking rate is generally associated with better speech intelligibility in normal adult listeners.
However, contrasting result was reported in Sri Lankan Tamil speakers, who did not show evidence of slower AR in IDS. However, the study was done on only 5 Sri Lankan Tamil speakers, and the researchers agreed that the results must be tempered by the unbalanced nature of the data (few of the recordings were missing). Whereas in a study by Han et al. Mandarin Chinese IDS did not show evidence of slower AR compared to the Dutch. Here, researchers opined that rhythmic class of the language may have an effect on the temporal modifications and thereby affect speech rate. Being a syllable-timed language (without lexical stress), Mandarin Chinese has nearly equal weight and time in all syllables. Whereas Dutch is a stress-timed language (with lexical stress), the stressed and unstressed syllables are distinguished by syllable weight and duration at the word level.
The present study is in support of the previous findings which also reported slower AR when speaking to children and infants.,, The reason for the slower AR may be because mothers tend to make adjustment to the child's age by prolonging words and placing stress on novel and content words, thereby giving child the time to process and learn. Although decreased speech rate has not been found to directly impact child language outcomes, it has been found to improve word recognition in infants, a skill that is necessary for language acquisition.
The second aim of the study focused on the rhythmic aspect of speech, which has gained less attention with respect to CDS. Speech rhythm was measured using an automatized approach, i.e., Envelope Modulation Spectra (EMS). It is a method of measuring the amplitude variations across multiple frequencies in a speech sample. It depicts the slowly varying amplitude changes that occur within a signal and the distribution of energy in the amplitude fluctuations across designated frequencies are collapsed over time. In our study, we hypothesized that there will be a significant difference between the ADS and CDS conditions. There was no significant main effect of condition for any of the octave frequencies. However, the results of peak frequency showed some interaction effects between the conditions. However, the other variables failed to demonstrate a significant effect of conditions.
Overall, the results suggest that there is no significant change in the underlying rhythm with respect to CDS and ADS. This is the preliminary study conducted on CDS using a novel method to measure rhythm. As this is one of the first studies on normal-hearing mothers, this area needs further exploration. The empirical results reported in this study might have some limitations. The first is the age range of the children considered for the study, which is between 1 and 3 years. Studies have reported that the speech rhythm directed to children becomes more like adult by the age of 1 year. Dominey and Dodane state that “the essential acoustic property of CDS is the exaggeration or modulation of characteristics that are already present in ADS.”. Here, they refer to general prosodic characteristics such as increase in pitch, vowel duration, and pause duration and not to rhythm specifically. Lee et al. on the model-based analyses (rhythmogram) also report that mothers modify language rhythm as a function of infant age, and language rhythm becomes more adult like over the first 12 months.
| Conclusion|| |
The difference between the speech directed toward a child as that of an adult has been investigated from many aspects – lexical, syntactic, phonological, and semantic aspects. This paper attempts to investigate linguistic and prosodic aspects of CDS in Kannada language, a language that has received less attention. The results of our study suggest that mothers tend to reduce the rate of speech when talking to their children supporting the universality of CDS. However, more such studies are required in other Indian languages to support our study. In this study, we did not find a significant difference in speech rhythm between CDS and ADS. As EMS is one of the novel methods used in this study, it is difficult to come to a conclusion about the use of EMS in studying rhythm. This gives us a future direction to explore on the use of EMS in the normal population.
Further studies on linguistic modifications such as syntactic, semantic, and pragmatic aspects of CDS can be explored, and at the segmental level, prosodic aspects such as pitch characteristics, clause boundary cues, vowel space area, and dispersion can be studied in other Indian languages. Many studies report that spoken language input to a developing child has a strong influence on the development of linguistic and cognitive abilities., Hence, there is a need to do more of longitudinal studies to examine the long-term relationship between mother's speech at an early age and toward later ages. This would provide valuable information role of CDS on the development of language in children. There is also a need to do long-term studies and cross-sectional studies to investigate the age-related changes in the acoustic modifications adopted by mothers. Our work also invites future explorations focusing on studying how a speaker adjust their speech when speaking to a child who is not typically developing (e.g., hearing impairment, delay in speech, and language development).
The authors thank the participants for their cooperation. We thank the Director, All India Institute of Speech and Hearing, Mysuru, for permission to carry out the study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Fernald A. Human maternal vocalizations to infants as biologically relevant signals: An evolutionary perspective. In: Barkow JH, Cosmides L, Tooby J, editors. The Adapted Mind: Evolutionary Psychology and the Generation of Culture. New York Oxford University Press; 1992. p. 391-428.
Soderstrom M. Beyond babytalk: Re-evaluating the nature and content of speech input to preverbal infants. Dev Rev 2007;27:501-32.
Song JY, Demuth K, Morgan J. Effects of the acoustic properties of infant-directed speech on infant word recognition. J Acoust Soc Am 2010;128:389-400.
Kuhl PK, Andruski JE, Chistovich IA, Chistovich LA, Kozhevnikova EV, Ryskina VL, et al.
Cross-language analysis of phonetic units in language addressed to infants. Science 1997;277:684-6.
Kitamura C, Burnham D. Pitch and communicative intent in mother's speech: Adjustments for age and sex in the first year. Infancy 2003;4:85-110.
Trehub SE, Unyk AM, Trainor LJ. Maternal singing in cross-cultural perspective. Infant Behav Dev 1993;16:285-95.
Katz GS, Cohn JF, Moore CA. A combination of vocal fo dynamic and summary features discriminates between three pragmatic categories of infant-directed speech. Child Dev 1996;67:205-17.
Liu HM, Kuhl PK, Tsao FM. An association between mothers' speech clarity and infants' speech discrimination skills. Dev Sci 2003;6:1-10.
Thiessen ED, Hill EA, Saffran JR. Infant-directed speech facilitates word segmentation. Infancy 2005;7:53-71.
Snow CE. Beginning from baby talk: Twenty years of research on input in interaction. In: Gallaway C, Richards BJ, editors. Input and Interaction in Language Acquisition. New York: Cambridge University; 1994. p. 3-12.
Brousseau L, Malcuit G, Pomerleau A, Feider H. Relations between lexical-temporal features in mothers' speech and infants' interactive behaviours. First Lang 1996;16:41-59.
O'Grady W. How Children Learn Language (Cambridge Approaches to Linguistics). Cambridge: Cambridge University Press; 2005.
Fernald A, Taeschner T, Dunn J, Papousek M, de Boysson-Bardies B, Fukui I. A cross-language study of prosodic modifications in mothers' and fathers' speech to preverbal infants. J Child Lang 1989;16:477-501.
Zangl R, Klarman L, Thal D, Fernald A, Bates E. Dynamics of word comprehension in infancy: Developments in timing, accuracy, and resistance to acoustic degradation. J Cogn Dev 2005;6:179-208.
Raneri DP. Infant Directed Speech: Maternal Pitch Variability, Rate of Speech, and Child Language Outcomes. Master's Thesis, University of Maryland, College Park; 2015. Available from: https://drum.lib.umd.edu /handle/1903/16736. [Last assessed on 2020 June 01].
Robb MP, Maclagan MA, Chen Y. Speaking rates of American and New Zealand varieties of English. Clin Linguist Phon 2004;18:1-15.
Tsao YC, Weismer G, Iqbal K. Interspeaker variation in habitual speaking rate: Additional evidence. J Speech Lang Hear Res 2006;49:1156-64.
Jacewicz E, Fox RA, O'Neill C, Salmons J. Articulation rate across dialect, age, and gender. Lang Var Change 2009;21:233-56.
Fernald A, Simon T. Expanded intonation contours in mothers' speech to newborns. Dev Psychol 1984;20:104-13.
Tang JS, Maidment JA. Prosodic aspects of child-directed speech in Cantonese. Speech Hear Lang 1996;9:257-76.
Van de Weijer J. Language input to a prelingual infant. In: Sorace A, C. Heycock, & R. Shillcock., editors. Proceedings of the GALA '97 Conference on Language Acquisition: Edinburgh University Press; 1997. p. 290 293.
Thanavisuth C. Acoustic Qualities of IDS and ADS in Thai. Conference on Spoken, (Icslp 98); 1998. P. 1 4. Available from: http://www. Iscaspeech.org/archive/icslp_1998/i98_0912.html. [Last accessed on 2020 Oct 08].
Narayan CR, McDermott LC. Speech rate and pitch characteristics of infant-directed speech: Longitudinal and cross-linguistic observations. J Acoust Soc Am 2016;139:1272-81.
Han M, Jong NH, Kager R. Language specificity of infant-directed speech: Speaking rate and word position in word-learning contexts. Lang Learn Dev 2021;17:221-40.
Blount BG, Padgug EJ. Prosodic, paralinguistic, and interactional features in parent-child speech: English and Spanish. J Child Lang 1977;4:67-86.
Fernald A, Mazzie C. Prosody and focus in speech to infants and adults. Dev Psychol 1991;27:209-21.
Grieser DA, Kuhl PK. Maternal speech to infants in a tonal language: Support for universal prosodic features in motherese. Dev Psychol 1988;24:14-20.
Payne E, Post B, Astruc L, Prieto P, Vanrell M. Rhythmic modification in child directed speech. Oxford University Working Papers in Linguistics. Philol Phon 2009;12:123-44.
Lee CS, Kitamura C, Burnham D, Todd NP. On the rhythm of infant- versus adult-directed speech in Australian English. J Acoust Soc Am 2014;136:357-65.
Dellwo V. Rhythm and speech rate: A variation coefficient for deltaC. In: Karnowski P, Szigeti I, editors. Language and Language-Processing. Frankfurt/Main: Peter Lang; 2006. p. 231-41.
Grabe E, Low EL. Acoustic correlates of rhythm class. In: Gussenhoven C, Warner N., editors. Laboratory Phonology. Berlin: Mouton de Gruyter; 2002. p. 515 46.
Ling LE, Grabe E, Nolan F. Quantitative characterizations of speech rhythm: Syllable-timing in Singapore English. Lang Speech 2000;43:377-401.
Ramus F, Nespor M, Mehler J. Correlates of linguistic rhythm in the speech signal. Cognition 1999;73:265-92.
Liss JM, LeGendre S, Lotto AJ. Discriminating dysarthria type from envelope modulation spectra. J Speech Lang Hear Res 2010;53:1246-55.
Kalashnikova M, Burnham D. Infant-directed speech from seven to nineteen months has similar acoustic properties but different functions. J Child Lang 2018;45:1035-53.
Boersma P, Weenink D. PRAAT: Doing Phonetics by Computer (computer program version 6.0.43); 2005. Amsterdam: Institute of Phonetic Sciences. http://www.praat.org/ .
Chon H, Kraft SJ, Zhang J, Loucks T, Ambrose NG. Individual variability in delayed auditory feedback effects on speech fluency and rate in normally fluent adults. J Speech Lang Hear Res 2013;56:489-504.
Chon H, Sawyer J, Ambrose NG. Differences of articulation rate and utterance length in fluent and disfluent utterances of preschool children who stutter. J Commun Disord 2012;45:455-67.
Bernstein-Ratner N. Dissociations between vowel durations and formant frequency characteristics. J Speech Hear Res 1985;28:255-64.
Bradlow AR, Pisoni DB. Recognition of spoken words by native and non-native listeners: Talker-, listener-, and item-related factors. J Acoust Soc Am 1999;106:2074-85.
Dominey PF, Dodane C. Indeterminacy in language acquisition: The role of child directed speech and joint attention. J Neurolinguistics 2004;17:121-45.
Kaplan PS, Bachorowski JA, Smoski MJ, Hudenko WJ. Infants of depressed mothers, although competent learners, fail to learn in response to their own mothers' infant-directed speech. Psychol Sci 2002;13:268-71.
[Table 1], [Table 2], [Table 3]