Kathleen Wermke

Relation of melody complexity in infants' cries to language outcome in the second year of life: a longitudinal study.

This study is part of the German Language Development Studys prospective longitudinal research programme on infants from birth until the age of 3 years. Thirty‐four infants were retrospectively classified into two groups (normal/delayed) by their language skills at 2.5 years of age. Frequency spectrograms and melodies of about 11,000 cries from the first 16 weeks of life were analysed using a CSL 4400. A Melody Complexity Index was calculated at monthly intervals. Infants with less than 45% complex melodies in their cries during the second month were found to be almost five times more likely to develop a language delay as infants with a higher proportion. For infants above the cut‐off of .45, development of a language delay condition could be ruled out with a probability of 89%. Although the results need to be interpreted cautiously, the data indicate a possible relation between early melody development and later language outcome.

Musical elements in human infants’ cries: In the beginning is the melody

Crying is the earliest sound production of human infants on their long way toward language. Here, we argue that infants’ early crying contains melodic constituents for both musical and prosodic structures. This view is based on our findings that cry melodies become increasingly complex during the first months of life and, that complex cry melodies are composed of shape-specific melody arcs. We found that cry melodies contain frequency ratios that show a certain preference of musical intervals. We also observed that young infants are capable of uttering shape-similar melody arcs at different frequency levels, that means they have an aptitude for frequency transposition from birth on. Moreover, we observed that the production of phonatory breaks within single expiratory sounds generates rhythmical elements and points to a flexible time organization. Our data support the view that in crying elementary constituents of both musicality and language faculty are unfolding. The results may elucidate the relation between emotionally charged sounds and music respectively language and suggest direction for further research.

Classification of cries of infants with cleft-palate using parallel hidden Markov models

This paper addresses the problem of classification of infants with cleft palate. A hidden Markov model (HMM)-based cry classification algorithm is presented. A parallel HMM (PHMM) for coping with age masking, based on a maximum-likelihood decision rule, is introduced. The performance of the proposed algorithm under different model parameters and different feature sets is studied using a database of cries of infants with cleft palate (CLP). The proposed algorithm yields an average of 91% correct classification rate in a subject- and age-dependent experiment. In addition, it is shown that the PHMM significantly outperforms the HMM performance in classification of cries of CLP infants of different ages.

Variability of the cry melody and the melody spectrum as indicators for certain CNS disorders

Pain cries of 42 infants during the first week of life were investigated by signal analysis methods, extracting the time variations of the fundamental frequency [melody] with high accuracy in frequency and time. Several methods are proposed to define measures for the variability of this melody. We tested the suitability of these measures as indicators for certain brain disorders. Indicators which measure fine and fast frequency modulations [microvariability] have been proven to be possible candidates for pathology indicators. Our melody spectrum is an effective tool for overviewing the several components of variability. The pathology indicators derived from the melody spectrum yield a statistically significant difference between infants with severe cerebral disorders and those in the control group. Due to the existence of discontinuities during prespeech development, pathology indicators are strongly dependent on age, maturity and individual peculiarities and have to be closely tuned to developmental stages.

Melody as a primordial legacy from early roots of language

The stormy development of vocal production during the first postnatal weeks is generally underestimated. Our longitudinal studies revealed an amazingly fast unfolding and combinatorial complexification of pre-speech melodies. We argue that relying on “melody” could provide for the immature brain a kind of filter to extract life-relevant information from the complex speech stream.

Tuning phenomena of melodies and resonance frequencies (formants) in infants pre‐speech utterances

In a former study [Wermke et al., Med. Eng. Phys. 24, 501–514 (2002)], an increasing tuning between laryngeal (melody) and pharyngeal (resonance frequencies) activity was demonstrated during pre‐speech development. This tuning was observed unexpectedly early during development in mitigated cries and earliest non‐cry utterances and prepares probably articulation in speech‐like vocalizations of older infants. A new retrospective study supported this assumption by comparing tuning processes in cries and early non‐cry utterances (N=2500) in two groups (low versus high word production performance at 18 months) of term‐born healthy infants (N=20). Additionally, age‐matched comparisons were made in 4 cleft‐lip‐palate‐infants treated with a palatal plate. In order to demonstrate the interaction between melody and resonance control we designed a special graphical representation. The tracking function of the resonance frequencies is displayed synchronously to the melody and its harmonics. Resonance frequencies in p...

Changes of voice parameters and melody patterns during the first year of life in human twins

The study analyzed changes of cry characteristics of twins using spectrograms, fundamental frequency contours (melody), and several indicators of microvariability of the fundamental frequency (F0) calculated with the aid of a KAY Elemetrics CSL Model 4305. Subjects were 15 identical and 15 fraternal healthy twin pairs, whose spontaneous utterances were recorded every 6 weeks over the first year of life. Developmental changes of F0, jitter, perturbation quotient, degree of voiceless, degree of subharmonic components as well as the evolution of complex melody patterns were analyzed. Correlations of the voice data with anthropometric data (weight, length, head circumference) and obstetrical factors (time of gestation, presentation, birth modus) were calculated to consider influences by growth discordance or different birth histories. As identical twins have a much greater congenital synchronization of development, universals of prespeech evolution are more convincingly expressed in these pairs (lower intrapa...

Special Session on Infant Cry Analysis - resonance Development and Formant Tuning Phenomena in Infant's Crying

The tracking of resonance frequencies and the analysis of their interaction with the fundamental frequency (F0) allows a description of (pre-) articulatory activity in very young infants. Subjects are six healthy infants. Spontaneous cries were recorded weekly from the 4 until the 20 week. For resonance frequency estimation a spectral parametric technique was applied, which was based on autoregressive models whose order is adaptively estimated on subsequent signal frames [1]. Cry melodies exhibiting different degrees of complexity (e.g. single-arc-melodies, multiple-arc-melodies) were selected for analysis. We found that resonance (formant) tuning occurs much earlier than expected. Here we demonstrate the early occurrence of a tuning between resonance frequencies and the cry melody in infants from 8 weeks onward. A more intense tuning between the melody and the lower resonance frequencies was found beginning about the 2 / 3 month. This tuning is interpreted as an early articulatory activity in infant’s crying. In a broader perspective it is attributed to a language-related behaviour preparing formant tuning in speech. Medical applications are seen for infants with disturbances of the vocal tract transfer function, e.g. infants with cleft-lip-palate.

Special Session on Infant Cry Analysis - Plenary Lecture : Time Variations of the Fundamental Frequency (Melody) and Resonance Frequencies in Infant's Crying - Key Parameters for Pre-Speech Development

The ability to perceive and to produce the time varying fundamental frequency (melody) is an extremely important component of auditory information and a fundamental aspect of language. The fundamental frequency is an essential parameter of prosody. In adult language perception, prosody can guide the syntactic analysis of spoken sentences [1]. Concerning infant language perception it was shown that young infants recognize utterances in their language based on prosodic cues before they become sensitive to its segmental characteristics [see review in 2]. Speakers use the F0 modulation to stress particular elements in an utterance or to indicate the beginning or end of a syntactic phrase. Recently, Drayna et al. [3] demonstrated in a twin study the influence of genes on the ability to recognize correct pitch and melodies. They could show that the perception of pitch is highly heritable. Research examining patients with brain damage has indicated that melodic information may be processed primarily by a cortical system in the right hemisphere. A close link between the processing of melodies and the processing of language has been demonstrated in a recent study by Maess et al. [4] who found that music processing involves a neural network normally seen to be active during language processing. This finding strongly supports a direct relationship between the processing of language and music from a functional and neuroanatomical view. The importance of F0 and related parameters is also well described for infant’s and children’s sound production. The importance is not only given by research results in the framework of “cry-diagnosis”, but also by findings within the field of pre-speech development and language acquisition [5-10]. Moreover, the interaction between laryngeal (melody) and pharyngeal (resonance frequencies) activity is one of the key aspects for prespeech research [e.g. 11]. Tuning processes between the cry melody and resonance frequencies are preparatory activities for an intentional articulation in speech. BRIDGING FROM CRYING OVER BABBLING TO SPEECH BY USING A MATHEMATICAL MODEL