the infant cry: what can it tell us?

Current Problems in

P e d i a t r i ~ Volume 26 Number 9 October 1996

Localized Scleroderma Including Morphea, Linear Scleroderma, and Eosinophilic Fasciitis

Audrey M. Nelson, MD Mayo Clinic Rochester Rochester, Minnesota

The Infant Cry: What Can It Tell Us? Michael J. Corwin, ME), Barry M. Lester, PhD,

and Howard L. Golub, ME), PhD Boston University Schools of Medicine and Public Health,

Brown University School of Medicine, Providence, Rhode Island, and CRI, Chestnut Hill, Massachusetts

]~v~ Mosby ~'~ ATimes Mirror

Company

The Infant Cry: What Can It Tell Us? Michael J. Corwin, MD, Barry M. Lester, PhD,

and Howard L. Golub, MD, PhD

l~ esearch on acoustic analysis of infant crying has had a long history that began with descriptions of cry abnormalities related to medical condi-

tions by using sound spectrographic analysis. The sound spectrograph provided a picture of the cry from which a limited number of acoustic characteristics could be cal- culated and from which a number of descriptive characteristics were described from the visual patterns on the spectrogram. The advent of high-speed computer tech- nology has improved the science of cry analysis with enhanced quantitative methods. The computer-based methods permitted a new terminology to be developed from quantifying acoustic parameters rather than visual pattern descriptions. In this article we summarize the research on the acoustic analysis of infant cry and dis- cuss how this information may be useful for the clinician. Because the terminology used in cry analysis var- ies by analytic strategy and is likely not familiar to the reader, we have provided a glossary of terms as an appendix.

From a conceptual point of view, infant crying is a biosocial phenomenon. ~ Crying provides information about the biological integrity of the infant and is a social signal that affects parenting. We address both the biological and social aspects of infant crying.

Biologic Aspect of Infant Cry Studies relating cry features to medical problems are

in three general categories: (1) significant medical problems that may be identified by standard techniques (e.g., severe brain damage from various causes), (2) medical

Michael J. Corwin, MD, is an associate professor of pediatrics, epidemiology, and biostatistics in Boston University Schools of Medicine and Public Health and Boston City Hospital. Barry M. Lester, PhD, is a professor of pediatrics and psychiatry at the Brown University School of Medicine in Providence, Rhode Island. Howard L. Golub, MD, PhD, is affiliated with CRI in Chestnut Hill, Massachusetts. Curr Probl Pediatr 1996;26:325-34. Copyright �9 1996 by Mosby-Year Book, Inc. 0045 -9380 /96 /$6 .00 + .10 5 3 / 1 / 7 6 9 4 1

problems that are currently undetectable until it is too late for treatment (e.g., sudden infant death syndrome), and (3) medical conditions that may be readily identified but would benefit from an improved ability to define prognosis (e.g., prematurity). This article reviews the studies performed in each of these three areas to provide an overview of how cry analysis may help us understand specific disease processes and improve our ability to determine the diagnosis or prognosis of medical problems.

Infants with Known Severe Medical Problems

Clinicians have long been aware that infants with a variety of severe medical problems have identifiable cries. Consequently cry research initially focused on measuring the acoustic cry characteristics of infants with known medical problems. Studies of infants with known problems have been important in that they have helped to advance our understanding of the relation of cry characteristics to medical status. In addition, although cry analysis may not be of practical diagnostic value in cer- tain already documented conditions, in some of these, it may offer the potential to provide prognostic information (e.g., long-term neurodevelopmental outcome).

Studies of infants with known significant problems, in general, have used the sound spectrogram and have included the following medical problems: neonatal asphyxia, 2-5 brain damage from various c a u s e s , 2'3'6"9

hypoglycemia, 1~ hydrocephalus, 12,13 encephalitis, 14 meningitis, 9 hypothyroidism, ~5 Down syndrome, 2 various genetic defects, 3,8,162~ and oropharyngeal abnormalities.8, 21

The studies listed served to validate the concept that medical problems that affect the cry-production system do indeed result in changes in the acoustic characteristics of the cry. Examples of the types of changes in the cry spectrogram described in infants with known significant pathologic conditions are described below.

Brain Damage. Infants with brain damage--severe birth asphyxia, meningitis, or hydrocephalus--have cries

Curr Probl Pediatr, October 1996 325

that are more high-pitched and have melody-type changes to more rising, falling/rising, and flat types. The fundamental frequency (f0) is more unstable. Both gliding and biphonation occur. The cry duration of the phonation may change, and the cry becomes either very short or unusually long. Follow-up studies showed that infants with the poorest neurologic outcome had the most abnormal cries. 5

Herpes Encephalitis. In cases of herpes encephalitis, noise concentration was found. This was not true in infants with severe birth asphyxia and brain damage.

Hypothyroidism. In hypothyroidism the f0 is lower than normal, but otherwise the spectrogram resembles that seen in the crying of healthy neonates.

Cri-du-chat. In cri-du-chat there is a high f0 and a flat melody pattern.

Trisomy 13, 18, and 21. In cases of trisomy 13, 18, and 21 the f0 is low and monotonous.

Medical Problems Currently Ondetectable

If one can detect cry changes in known significant medical conditions, the next logical step is to determine whether specific cry characteristics may be associated with important medical problems for which currently available methods of detection are inadequate. The area in which cry analysis appears to offer the greatest potential benefit is the identification of infants at high risk for the sudden infant death syndrome.

Sudden Infant Death Syndrome (SIDS). SIDS is the most common cause of death for infants between 1 month and 1 year of age. The cause of SIDS is not known, and there is no method of identifying potential SIDS infants before their death. Theories regarding the cause of SIDS include abnormalities of the infants' air- way and abnormalities in the area of the infants' brain stem that controls breathing, heart rate, and the muscles of the pharynx. Because abnormalities such as these would be expected to alter cry characteristics, cry analysis has been suggested as potential for the identification of infants at risk for SIDS. Studies reporting cry characteristics of infants subsequently dying of SIDS (described below) include a retrospective review of parental perceptions of the cries of infants who later died of SIDS, four case reports in which the cries of a total of 10 SIDS infants were analyzed by using various methods, and one large prospective study.

Naeye et al. 22 asked parents retrospectively to compare the cries of their infants who died of SIDS with those of living siblings. The parents reported that, compared with living siblings, the SIDS victim had cries

with a different pitch and cries that were more difficult for the parents to interpret (i.e., varied less with the perceived needs of the infant).

Stark and Nathanson 23 recorded the spontaneous cry of a 4-day-old infant who died suddenly and unexpectedly at 6 months of age. The pathologist's diagnosis was SIDS; however, the reported case history suggested a foreign-body aspiration. Spectrographic analysis was used to compare the cry of the infant who died with 4 spontaneous cries recorded at age 3 to 7 days in healthy infants. The cry of the infant who died showed shorter cry utterances and lower intensity cries compared with the cries of the comparison infants. In addition, the cry of the infant who died had more nonvoiced inspiratory snorts, spectrographic evidence of a constriction in the vocal tract, more episodes of high (500 to 1000 Hz) and extremely high (> 1000 Hz) fundamental frequency and more frequent and extreme rapid shifts in fundamental frequency.

Colton and Steinschneider 24 performed spectrographic analysis of spontaneous crying recorded at age 4 days from an infant who subsequently died suddenly and unexpectedly at age 63 days. The infant was born at 38.5 weeks' gestation with a somewhat low birth weight of 2495 gm. As part of a research study, at age 4 and 30 days, she had respiratory pattern analysis that showed relatively frequent apneic pauses; however, she was clinically normal and was officially listed as a SIDS. Four- teen expiratory cry utterances of >0.5 seconds were analyzed for nine acoustic variables and then compared with cry analysis obtained from 124 control infants and 22 siblings of SIDS victims. The nine acoustic variables were fundamental frequency, duration of cry utterances, first, second, and third formants, and four measures of sound-pressure level (overall, and three different spec- tral bands). The cry of the SIDS infant tended to have a lower fundamental frequency, a longer utterance duration, lower formant frequencies, and greater sound-pressure levels compared with those of the control infants. Although the formants tended to be lower overall, some SIDS utterances exhibited the unusual finding of a first formant >2000 Hz and a second formant at 4000 Hz.

Golub and Corwin 25 used computer-aided cry analysis to analyze the cries of 87 infants. Sixty-seven were normal term non-SIDS infants, 17 infants were infants born prematurely who had recovered from a variety of illnesses, 1 infant was a sibling of an infant with apnea, and 2 subsequently died of SIDS. The two SIDS infants and the infant with a sibling who had apnea were the only infants who had cries showing a high first formant:

326 Curr Probl Pediatr, October 1996

a characteristic consistent with a constriction in the vocal tract.

The automated computer-aided analysis system developed by Golub was used to compare the cries of six infants who subsequently died of SIDS with 1238 age- matched controls. 26 Cases 1 and 2 were normal-term infants whose cries were recorded during the first week of life and who died of SIDS at 4 and 6 months. Cases 3, 4, and 5 were also normal infants bom at term. These infants had a severe apneic event between age 2 and 4 weeks, after which the cry was recorded. All three infants died of SIDS within 1 month of the cry recording. Case 3 was atypical for SIDS in that the infant had two prior siblings who died suddenly and unexpectedly in the first month of life. Case 6 was the larger and healthier of a set of twins born at 34 weeks' gestational age. The cry was recorded by the infant's father at age 69 days. The infant died 1 week later. The mean first formant for the SIDS victims was 1815 _ 218 compared with 1395 __. 253 in the control infants (p < 0.0001).

As a consequence of the observations described, Corwin et al. 2v conducted a prospective study to test the hypothesis that the occurrence of a neonatal cry exhibiting a high first formant is a risk factor for SIDS and to evaluate the association between SIDS and other acoustic cry variables. Cries were recorded and medical and de- mographic data were obtained for 21,880 apparently healthy term newborns. Two cries were recorded between days 2 and 7 of life, after a painful stimulus at the time of routine blood drawing. Twelve infants died of SIDS. Age at death ranged from 19 days to 6.5 months. Newborns whose first cry exhibited a high first formant were more likely to die of SIDS than infants whose first cry did not have this characteristic (relative risk 3.5; 95% confidence interval, 1.1 to 12; p = 0.04). The relative risk for SIDS increased to 8.8 (95% confidence interval, 2.2 to 35;p = 0.002) for newborns whose second cry showed that this characteristic persisted. Newborns with the combination of a high first formant and a high number of mode changes on both of two cries had a relative risk of 32 (95% confidence interval, 8.7 to 120).

The acoustic differences observed may provide an explanation for the parents' perceived differences in cries reported by Naeye et al . 22 We would not expect, however, that listening alone would allow accurate identification of infant cries with a high first formant or a high number of mode changes or both, and this is supported by our experience.

Although differences in methods do not allow direct

comparison of the acoustic data, the observations of Colton and Steinschneider 24 of a very high first formant in at least some cry utterances is consistent with the observation of a high first formant in one of the first three cry utterances.

Thus in a population of apparently healthy term newborns without known significant SIDS risk factors, an association has been observed between alterations in neonatal cry acoustics and subsequent incidence of SIDS. The relative risk of 32 is much higher than that of any other purported SIDS risk factor:8; however, in the population at low risk studied (with an overall SIDS rate of 0.6 per 1000), the rate of SIDS increased only to 9.43 per 1000 among infants with both cry characteristics on two consecutive cries. In spite of the large relative risk, therefore, at this time we would not advocate cry analysis as a practical SIDS screening test. In addition, given the post hoc evaluation of one of the cry variables, this finding needs cautious interpretation and requires replication. At this time, infant cry analysis is a potentially important research tool that, when studied in relation to other physiologic measures, may lead to an improved understanding of SIDS.

Medical Conditions with Uncertain Prognoses

Central Nervous System Insults. Studies demonstrating cry abnormalities in infants with severe central nervous system insults have led to speculation that cry analysis may be useful in the identification of infants who may have had significant or less obvious central nervous system damage. Traditional methods of neurodevelopmental assessment, when performed during the first months of life, have not been reliable pre- dictors of subsequent outcome. Recent studies 29 suggesting that early-intervention programs result in substantial benefit to affected infants make reliable identification of these infants particularly important.

Prematurity. Infants born prematurely are the largest group of infants who are at high risk for poor developmental outcome. Unfortunately, it is difficult reliably to predict neurodevelopmental outcome of these infants early in life. The studies described subsequently suggest that infant cry analysis has the potential to improve our ability reliably to identify infants at highest risk for poor developmental outcome.

Michelsson 4 spectrographically studied the pain-in- duced cries of 40 healthy infants born at <34 weeks' gestation, 35 healthy infants born between 35 and 37 weeks' gestation, 50 healthy term infants, and 95 asphyxiated preterm (<37 weeks) infants. None of the


healthy infants had evidence of birth asphyxia. Healthy preterm infants had higher minimum fundamental frequencies, which increased with decreasing gestational age, fewer glottal plosives, and more gliding compared with the healthy term infants. Michelsson 4 also defined normal values for nine cry characteristics and observed that, compared with healthy term infants, preterm infants more frequently had cry signals exhibiting one or more cry abnormalities.Asphyxiated preterm infants had higher fundamental frequencies than healthy preterm infants, as well as increased occurrence of biphonation and more cry signals containing abnormal cry characteristics. Long-term follow-up correlating cry characteristics with neurodevelopmental outcome is not available for these preterm infants.

Golub and Corwin 25 used computer-aided cry analysis to compare the cries of 10 preterm infants with those of 55 healthy term infants. The preterm infants all exhibited at least one of eight abnormal cry patterns compared with only 10 of the 55 healthy term infants. No long-term follow-up was performed of these infants.

The first study correlating cry analysis with long-term neurodevelopmental outcome was reported by Lester. 3~ The sample included 13 normal healthy term infants (mean birth weight, 3.3 kg) and 18 appropriate-for-gestational-age preterm infants (mean birth weight, 1.6 kg). The preterm infants had a history of a variety of medical problems, including respiratory problems (6 with hyaline membrane disease alone plus 3 with broncho- pulmonary dysplasia) and intraventricular hemorrhage (n = 2). All infants had cries recorded at term concep- tional age; these cries were analyzed by using the computer-aided analysis system developed by Golub.

Neurodevelopmental follow-up was performed on all 31 infants at 18 months corrected age and on 21 infants at age 60 months. The study evaluated whether cry characteristics could be used correctly to classify infants as those scoring high or low on the Bayley Scales at 18 months corrected age and on the McCarthy scales at age 60 months. Lower scores on the Bayley were associated with higher fundamental frequency (p < 0.0001), more change in fundamental frequency (p < 0.05), less change in the first formant (p < 0.05), and lower amplitude cries (p < 0.05).

The McCarthy scales include the general cognitive index or IQ, as well as subscales for verbal, perceptual, and quantitative performance. Lower general cognitive scores were associated with more change in fundamental frequency (p < 0.007), higher first formant (p <0.02), and lower amplitude cries (p < 0.007). Lower verbal

scores were predicted by more change in fundamental frequency (p < 0.05) and with less change in first formant (p < 0.05). Lower perceptual scores were related to more change in fundamental frequency (p < 0.05), higher first formant (p < 0.02), and lower cry amplitude (p < 0.05). Lower quantitative scores were associated with higher first formant (p < 0.007), lower cry amplitude (p < 0.02), and shorter cry duration (p < 0.02).

Additional analyses verified that it was not possible to predict the developmental outcome of these infants from their prematurity, newborn neurologic scores, medical risk factors, or socioeconomic factors.

Subsequently Lester et al. 31 extended these findings in a sample of 145 term and preterm infants by demonstrating that cry variables obtained at 1 month of corrected age predicted cognitive outcome at age 4 years, even after medical and social factors were taken into account.

From these preliminary data, Corwin et al. 32 performed a two-site longitudinal study in which acoustic analysis of the cry at 40 weeks adjusted age was used to predict 30-month developmental outcome. The 30-month prediction equations were developed from a sample of 174 infants enrolled in Boston, and then tested, as a replication, on a sample of 110 preterm and 36 term infants enrolled in Syracuse. The predictor equation for the 30-month cognitive testing included only race, gender, days of hospitalization, and number of short cry utterances. When applied to the Boston sample, the equation predicted low scores (<84) with a sensitivity of 0.80 and specificity of 0.77. Infants with a positive test had a relative risk for a low score of 8.4 (95% confidence interval, 3.0 to 23.5;p < 0.001). On replication in Syracuse, the equation predicted low scores with a sensitivity of 0.82 and specificity of 0.68. Syracuse infants with a positive test had a relative risk for a low score of 7.5 (95% confidence interval, 1.7 to 32.6; p = 0.002).

Although the sample size for the replication was relatively small, it is a significant achievement in behavioral research to develop a set of predictor equations on a sample in one site and replicate the findings with those same equations on an independent sample. Cry analysis has the potential to be developed as part of a screening test for poor developmental outcome. However, the results described need to be replicated on a larger sample, and in particular, the specificity rate needs to be improved. We know that labels can become a self-fulfill- ing prophecy. Thus there is the danger that by labeling these children as at risk, the expectations of the care-


giving environment would be reduced, resulting in an iatrogenic pathway to poor cognitive outcome.

Maternal DrugAbuse. Variations in the infant cry have been described in infants of mothers who abused a variety of substances during pregnancy. Blinick 33 used spec- trograms to compare the birth cries of 31 infants of opiate-addicted mothers with 338 infants of mothers not using opiates. They classified cries as being either type I (Typi- cal), type II (Common Variants) or type III (Abnormal), and found that 15 of the 31 infants with in utero opiate exposure had type III (Abnormal) cries compared with only 38 of the 338 nonexposed infants (p < 0.005). The most common abnormality they identified was a high fundamental frequency (12 infants). Corwin et al. 34 also analyzed the cries of infants of opiate-addicted mothers. Computer-aided cry analysis was performed on 90 cries obtained from 24 infants of opiate-addicted mothers and compared with the cries of 32 matched controls. Infants of opiate-addicted mothers had increased hyperphonation (p = 0.015), more short cry utterances (p = 0.02), and a trend toward a higher fundamental frequency (p = 0.08). In addition, longer duration of the first cry utterance correlated with higher scores on the clinical abstinence scale (r = 0.30, p = 0.005).

The cries of infants exposed to cocaine also have been shown to be abnormal. Corwin et al . 35 recorded cries during the first week of life of 768 infants (404 cocaine exposed and 364 nonexposed controls). Ten cry characteristics were determined by using the computer-aided analysis system. Multiple regression analysis was performed to examine the independent effect of cocaine exposure on the cry variables when the effects of the following 11 potentially confounding variables were taken into account: maternal age, ethnicity, education, cigarette use, alcohol use, and marijuana use; infant gestational age, gender, birth weight, length, and head circumference. After these variables were controlled, the infants exposed cocaine showed 0.7 fewer cry utterances (p = 0.001), 2.3 more short cry utterances (p = 0.02), and crying in the hyperphonation mode decreased by 4.6% (p = 0.01). The magnitude of effect was greater in the subpopulation of infants who tested positive for cocaine after birth (n = 128), suggesting a dose-related effect of cocaine exposure.

Lester et al . 36 used a structural-equation model on a subgroup of these infants to test the hypothesis that direct effects of cocaine are responsible for changes in acoustic features consistent with excitation and that indirect cocaine effects (associated with growth retardation) are responsible for acoustic features consistent with

depression. Eighty term cocaine-exposed infants were compared with a group of 80 control infants who were similar with regard to gestational age, ethnicity, and exposure to other substances including cigarettes, alcohol, marijuana, and opiates. The structural-equation analysis confirmed that direct effects of cocaine included longer duration, higher fundamental frequency, and higher and more variable first formant. Indirect effects of cocaine included a longer latency, fewer utterances, lower amplitude, and more dysphonation.

Lester and Dreher 37 analyzed the cries of 20 infants exposed to marijuana and 20 control infants and found that the infants exposed to marijuana had cries that were shorter, had more dysphonation, higher and more variable fundamental frequency, and a lower first formant. A dose/response effect was observed for the lower first formant.

Finally, Nugent et al. 38 analyzed cries of 127 3-day- old Irish infants whose mothers had been interviewed during their last trimester to ascertain their history of alcohol and tobacco use. Mothers consumed an average of 0.21 oz of alcohol/day, with 62% classified as moderate drinkers (0.01 to 0.49 average oz/day), and 10.6% as heavy drinkers (0.5 to 0.99 average oz/day). Mul- tiple regression analysis showed that greater alcohol use was related to more dysphonation and a higher first formant. Cigarette use during pregnancy was reported by 62% of women, with 23% smoking 10 to 20 cigarettes/day and 8% smoking 20 to 40 cigarettes per day. Greater use of cigarettes was associated with a higher fundamental frequency, a higher second formant, and more variability in the second formant.

Hyperbilirubinemia. Bilirubin, at high concentrations, is known to have significant central nervous system toxicity. Mild to moderate elevation of bilirubin that is seen in normal newborns has not been shown to result in clinically apparent toxicity; however, high levels of bilirubin, particularly when associated with hemolytic disease of the newborn, can result in a severe neurologic problem (kernicterus). Kernicterus can include seizures, athetoid movement disorders, and deafness. A characteristic high-pitched cry also has been described by caretakers of these infants.

Although kernicterus can usually be avoided, the relation between the level of bilirubin and the risk of toxicity is not well understood. In addition, it has been suggested that mild toxicity from lower levels of bilirubin might also be possible3 9 No method exists for de- tecting early or mild bilirubin toxicity. Because infants with kernicterus are known to have changes in their cries,


and given the evidence that cry features may predict neurodevelopmental outcome, cry analysis may provide a marker for early central nervous toxicity in infants with moderate elevations in serum bilirubin.

Wasz-Hockert et al. 4~ used the spectrograph to analyze pain cries from 45 infants with hyperbilirubinemia divided into three subgroups: 15 cases with Rh immu- nization, 15 cases with ABO incompatibility, and 15 cases with no blood group incompatibility, as compared with 120 control infants. Maximum bilirubin levels of __20 were reported in 28 of the 45 infants. Neurologic symptoms were reported in 26 infants, and 6 had long- term neurologic problems.

The infants with hyperbilirubinemia were shown to have shorter latency, shorter duration, higher fundamental frequency, and greater change in fundamental frequency over a 0.1-second period. In addition, furcation or biphonation or both were frequent in infants with hyperbilirubinemia and not observed at all in the control infants.

Golub and Corwin 25 used computer-aided cry analysis to examine the pain cries of 24 infants with moderate hyperbilirubinemia (<20 mg/dl) and 63 control infants. Cry characteristics consistent with unstable glottic function were observed in the cries of 22 of 24 infants with hyperbilirubinemia compared with only 8 of 63 cries of controls.

Vohr et al. 4~ used the computer-aided analysis system to examine the cries of 25 infants with bilirubin levels between 10 and 20 mg/dl, compared with the cries of 25 infants with bilirubin values of <8 mg/dl. High levels of bilirubin were associated with more variability in first formant values and an increase in the percentage of time in the phonation cry mode. Brain stem auditory evoked potentials also were performed on these infants and were found to be related to both higher bilirubin levels and changes in the cry.

These studies suggest that moderate hyperbilirubinemia is associated with changes in cry features; however, it is not known whether these cry changes persist or are correlated with subsequent outcome.

Prenatal and Perinatal Lead Exposure. As part of the Mexico City Prospective Lead Study, acoustic cries were elicited from apparently healthy babies. 42 Computer- aided cry analysis was used for cries obtained at age 2 days (n = 75), 15 days (n = 176), and 30 days. Maternal lead levels were measured every 8 weeks during pregnancy and at delivery from the umbilical cord blood. Cry characteristics associated with higher maternal lead levels included lower percentage nasalization and lower

number of cries (associated with higher lead levels in the second and third trimester); and higher fundamental frequency (associated with higher lead levels at 12 weeks and in cord blood). It is not known, however, if these findings are associated with subsequent outcome.

Summary A growing body of literature suggests that the infant

cry provides important information about the biologic status of the infant. Although it is clear that many ques- tions remain, cry analysis has the potential to become an important technique for the early identification of infants having medical problems that affect their cry- production system. Few data are available with regard to whether or not cry changes are persistent or transient or the prognostic implications of particular cry changes.

The results of the large longitudinal studies showing an association between infant cry characteristics and risk for SIDS and demonstrating that cry variables enhance prediction of poor developmental outcome in preterm infants support the potential medical value of infant cry analysis. Additional longitudinal studies, however, are required to better define the role of cry analysis in these and other conditions such as in utero drug exposures or hyperbilirubinemia.

Social Aspect of Infant Cry Crying is the infant's earliest form of communica-

tion, and most parents are readily able to distinguish their infant's hunger cry from their infant's pain cry. Communication, however, is a dyadic event, and much of the early parent/child relationship is negotiated around crying. Communicating through cry is an acoustic phenomenon. Features such as immediacy of cry onset, loudness, pitch, and the temporal patterning of the cry distinguish the basic or hunger cry from the pain cry. Some infants are better communicators than others (i.e., the acoustic features that distinguish the basic cry from the pain cry are well defined). For other infants, these features are less clear. These kinds of differences in how infants cry as well as how the infant cries, result in parents saying that some infants are more readable than others or parents who say that they cannot tell what the infant needs or wants. The infant's cry can be thought of as the infant's contribution to the relationship with the parents. The parent's contribution is twofold. First, the parent has to interpret the infant's cry, and second, the parent has to translate that interpretation into appropriate parenting behavior. As infants vary in their ability to communicate through cry, parents vary in how


they interpret the cry and in the parenting behavior they execute.

Research has documented how adults (parents as well as nonparents), react to infant cries. One strategy has been to use rating scales in which the adult listens to cries and rates the cries along dimensions of how the cry sounded (e.g., loudness, pitch), how the infant sounded (e.g., sick, distressed), and how the listener felt (e.g., angry or sad). These studies show that adults clearly distinguish the cries of normal infants from abnormal infant cries in which the cry abnormality is the result of the kinds of medical problems d i s c u s s e d . 43"49

For example, preterm infants have cries that are more difficult to interpret than those of term infants, which may increase the stress of caring for the preterm infant.50-53

A second strategy has been to use adult physiologic responses as a measure of reactivity to infant cries. Boukydis and Burgess 48 showed increased skin potential responses in response to the cries of infants with difficult temperament. Crying has been implicated in child abuse, and studies have shown that mothers who had previously abused a child showed higher heart-rate responses when listening to infant cries than did con- trois, g4 In another study, adults who scored high on a questionnaire that measures the potential for child abuse showed more extreme changes in heart rate and skin conductance than did controls. 54

It has been suggested that teenage mothers may not accurately perceive their infants' signals. In a study of how teenage mothers perceive the cries of their infants, teenage mothers with infants who had higher pitched cries associated with intrauterine growth retardation perceived those cries as more positive in contrast to nonteenage mothers who perceived the cries of their growth-retarded infants as more negative. 55 These teenage mothers misperceived their infants' cry characteristics, which may have implications for the ability of these mothers to interpret their infants' behavior.

Colic is a good example of how biological and social aspects of cry combine and become problematic for the parent and for the infant. Colic is often equated with excessive crying and is diagnosed from the amount of cryingY Features in addition to the amount of crying have also been used to improve the diagnosis of colic with the Colic Symptom Checklist. 57 These features include the sudden onset of the colic episode, physical signs associated with hypertonia, inconsolableness, and changes in the acoustic characteristics of the cry. In- fants diagnosed as having colic by using the Colic Symp-

tom Checklist show increased pitch, variability in pitch, and turbulence in the cry. Their mothers perceive their infants' cries as more aversive and distressing, and the infants were scored as fussy/difficult on the Infant Char- acteristics Questionnaire measure of temperament. In another study, infants diagnosed as having colic on the Colic Symptom Checklist showed a higher mean heart rate during feeding and more difficult temperament than infants with excessive crying or controls? 8 The mothers of these infants reported more parenting stress. 59 In colic we see how a biologically driven cry condition can cause a perturbation in the parent/child relationship. We are used to thinking of colic as self-limiting, and insofar as the actual crying is concerned, this view is correct. How- ever, if this acute perturbation in the parent/child relationship is not repaired, there may be long-term consequences for the child's emotional development. That is: How do characteristics of the infant and characteristics of the parent combine to determine the developmental outcome of the child?

According to the "Goodness of Fit" construct, pro- posed by Chess and Thomas, 6~ a good fit occurs when there is a match between characteristics of the child and characteristics of the parent, with positive developmental outcome as the expected result. When there is a mismatch between characteristics of the child and characteristics of the parents, the prognosis is more negative. This construct was tested by using acoustic characteristic of the infant's cry and mother's ratings on the cry- perception scales. 6j Matches were defined as mother/ infant pairs in which the infant's cry was found to have high or low pitch and turbulence characteristics based on acoustic analysis and the mother correctly perceived the cry as either high pitched and distressing or lower pitched and not distressing. Mismatches were dyads in which the infant's cry was in the high category and the mother rated the cry as positive, or the cry was lower pitched, and the mother rated the cry as high pitched and distressing. The match/mismatch analysis was done when the infants were 1 month old. At 1 8 months, infants in the two matched groups showed higher Bayley mental scales, higher Bayley subscales on language and cognition, and higher language scores on a separate language test. Mothers in the mismatch groups had lower self-esteem, more depression, and less social support than mothers in the matched groups. This study suggests that mothers who are better able to read their infants' signals, in this case cry, can provide the kind of parenting that facilitates cognitive and language outcome. Factors such as low self-esteem, depression, and


iack of social support may impede the ability of some mothers to provide a more optimal care-giving environment.

Clinical Implications There are several ways in which understanding in-

fant cry can inform clinical practice; these include both biologic and social aspects. From the biologic point of view, there is the long-term goal of develop- ing cry analysis as a screening tool, and we reviewed preliminary studies to suggest that there may be diagnostic utility in the cases of SIDS and longer term developmental outcome in preterm infants. In the future, cry analyses (perhaps in conjunction with other factors) may permit the clinician to identify infants whose cries indicate that they are exhibiting imme- diate effects of a particular illness or insult, identify which individual infants with a particular risk factor are truly affected, better predict long-term sequelae of these exposures, or a combination of these. How- ever, this is for the future. For the present, clinicians know that abnormalities in cry are useful in the dif- ferential diagnosis of brain damage and in some more subtle conditions such as narcotic addiction. In addition, the clinician can keep an "ear" out for changes in the cry that may signal the onset of a medical condition. There are examples in which this is already common practice, such as croup, but conditions such as meningitis or fever caused by infection can also be heralded by changes in cry.

From the social point of view, when parents com- plain of having an infant with crying-related problems such as colic, or difficulty in understanding the mean- ing of an infant's cry, it is important to ask whether there are changes in the cry in addition to the amount of crying and to consider whether the parents are having difficulty coping with their colicky infant that may war- rant a mental health referral. The clinician may also want to listen for indications from the parents that they are having difficulty understanding their infant's cry or other behavior or if the parents' perception of the baby seems to be disparate with the infant's actual behavior. This may suggest a clinical approach toward having the parents change how they perceive their infant and reinter- pret their infant's behavior. On the one hand, this could seem to be a daunting task in the midst of a busy clinical practice and could justify an outside referral. On the other hand, short, discrete interventions during exami- nation of the baby or discussion with parents can have a beneficial effect.

References 1. Lester BM. A biosocial model of infant crying, Vol. III. In:

LipseU L, Rovee-Collier C, editors. Advances in infant research. Norwood (NY): Ablex, 1984:167-212.

2. Partanen TJ, Wasz-Hockert O, Vuorenkoski V, Theorell K, Valanne E, Lind J. Auditory identification of pain cry signals of young infants in pathological conditions and in sound spectrographic basis. Ann Paedia t r Fenn 1967;13:56-63.

3. Wasz-Hockert O, Lind J, Vuorenkoski V, Partanen T, Valanne E. The infant cry: a spectrographic and auditory analysis. London: Heinemann, 1968.

4. Michelsson K. Cry analysis of symptomless low birth weight neonates and of asphyxiated newborn infants. Acta Pediatr Scand 1971;suppl:216.

5. Michelsson K, Sirvio P, Wasz-Hockert O. Pain cry in full- term asphyxiated newborn infants correlated with late findings. Acta Pediatr Scand 1977;66:611-6.

6. Karelitz S, Fisichelli V. The cry thresholds of normal infants and those with brain damage. J Pediatr 1962;61:679-85.

7. Fisichelli VR, Coxe M, Rosenfeld R. The phonetic content of the cries of normal infants and those with brain damage. J Psychol 1966;64:119-26.

8. Lind J, Editor. Newborn infant cry. Acta Paediatr Scand 1965 ;suppl: 163.

9. Michelsson K, Sirvio P, Wasz-Hockert O. Sound spectrographic cry analysis of infants with bacterial meningitis. Dev Med Child Neurol 1977;19:309-15.

10. Koivisto M, Blanco-Sequeiros M, Krause U. Neonatal symptomatic and asymptomatic hypoglycemia: a follow- up study of 151 children. Dev Med Child Neurol 1972; 14:603-14.

11. Koivisto M, Michelsson K, Sirvio P, Wasz-Hockert O. Spec- trographic analysis of pain cry of hypoglycemia in newborn infants. XIV International Congress of Pediatrics, 1974;1:250.

12. Sirvio P, Michelsson K. Sound spectrographic cry analysis of normal and abnormal newborn infants. Folia Phoniatr (Basel) 1976;28:161-73.

13. Michelsson K, Kaskinen H, Aulanko R, RinneA. Sound spectrographic cry analysis of infants with hydrocephalus. Acta Paediatr Scand 1984;73:65-8.

14. Michelsson K, Sirvio P. Cry analysis in herpes encephalitis. Stockholm: Almquist & Wiksell, 1975.

15. Michelsson K, Sirvio P. Cry analysis in congenital hypothyroidism. Folia Phoniatr (Basel) 1976;28:40-7.

16. Fisichelli V, Karellitz S. The cry latencies of normal infants and those with brain damage. J Pediatr 1963;62:724-34.

17. Michelsson K, Tuppurainen N, Aula P. Sound spectrographic cry analysis of infants with karyotype abnormality. Neuropediatrics t980;11:365-76.

18. Lind J, Vuorenkoski V, Rosberg G, Partenan T J, Wasz-Hockert O. Spectrograph analysis of vocal response to pain stimuli in infants with Down's syndrome. Dev Med Child Neurol 1970;12:478-86.

19. Ostwald P, Peltzman P, Greenberg M, Meyer J. Cries of a trisomy 13-15 infant. Dev Med Child Neurol 1970;12:472- 7.

20. Vuorenkoski V, Lind J, Partanen TJ, Lejeune J, LaFourcade J, Wasz-Hockert O. Spectrographic analysis of cries from infants with maladie du cri du chat. Ann Paediatr Fenn 1966; 12:174-80.

21. Michelsson K, Sirvio P, Koivisto M, Sovijarvi A, Wasz- Hockert O. Spectrographic analysis of pain cry in infants with cleft palate. Biol Neonate 1975;26:353-8.


22. Naeye RL, Messmer J, Sprecht R, Merritt TA. Sudden infant death syndrome temperament before death. J Pediatr 1976;88:511-5.

23. Stark RE, Nathanson S. Unusual features of cry in an infant dying suddenly and unexpectedly. In: Bosma JF, Showacre J, editors. Development of upper respiratory anatomy and function. US Department of Health, Educa- tion, and Welfare Publication No. (NIH) 75-941, 1975:233- 49.

24. Colton RH, Steinschneider A. The cry characteristics of an infant who died of the sudden infant death syndrome. J Speech Hearing Disord 1981;46:359-63.

25. Golub HL, Corwin MJ. Infant cry: a clue to diagnosis. Pedi- atrics 1982;69:197-201.

26. Corwin MJ, Golub HL, Kelly D, Lester BM, Shannon DC. Cry analysis in SIDS victims. Pediatr Res 1986;20:471A.

27. Corwin MJ, Lester BM, Sepkoski C, McLaughlin S, Kayne H, Golub HL. Newborn acoustical cry characteristics of infants dying of SIDS. Pediatrics 1995;96:73-7.

28. Hoffman HJ, Damus K, Hillman L, Krongrad E. Risk factors for SIDS: results of the National Institute of Child Health and Human Development SIDS Cooperative Epidemiologi- cal Study. Ann N Y Acad Sci 1988;533:13-30.

29. Infant Health Development Program. Enhancing the outcome of low-birthweight, premature infants: a multisite, random- ized trial. JAMA 1990;263:33035-42.

30. Lester BM. Prediction of developmental outcome from acoustic cry analysis in term and preterm infants. Pediatrics 1987;80:529-34.

31. Lester BM, McGrath M, Boukydis CFZ, Kilis E. Acoustic cry analysis at one month predicts four year cognitive outcome in preterm and term infants. Pediatr Res 1995;37:264A.

32. Corwin MJ, Sepkoski C, Gross S, Lester B, Peucker M, Kayne H, et al. Early prediction of low cognitive indices at 30 months in preterm infants. Pediatr Res 1992;31:244A.

33. Blinick G, Travolga WN, Antopoc W. Variations in birth cries of newborn infants from narcotic addicted and normal mothers. Am J Obstet Gynecol 1971;110:948-58.

34. Corwin M J, Golub HL, Potter M. Cry analysis in infants of narcotic addicted mothers. Pediatr Res 1987;21:190A.

35. Corwin MJ, Lester BM, Sepkoski C, McGlaughlin S, Kayne H, Golub HL. Effects of in utero cocaine exposure on newborn acoustical cry characteristics. Pediatrics 1992;89:1199- 203.

36. Lester BM, Corwin MJ, Sepkoski C, Seifer R, Peucker M, McGlaughlin S, et al. Neurobehavioral syndromes in cocaine exposed newborn infants. Child Dev 1991;62:694-705.

37. Lester BM, Dreher M. Effects of marijuana use during pregnancy on newborn cry. Child Dev 1989;60:765-71.

38. Nugent JK, Lester BM, Greene S, Dorit-Wieczorek D, O'Mahony E Maternal alcohol consumption and cigarette smoking during pregnancy and acoustic cry analysis. Child Dev. In press.

39. Boggs T, Hardy J, Frazier T. Correlation of neonatal serum bilirubin concentration and developmental status at age 8 months. J Pediatr 1973;54:319.

40. Wasz-Hockert O, Koivisto M, Vuorenkoski V, Partanen TJ, Lind J. Spectrographic analysis of pain cry in hyperbilirubinemia. Biol Neonatorum 1971;17:260-71.

41. Vohr BR, Lester B, Rapisardi G, O'Dea C, Brown L, Peucker M, et al. Abnormal brainstem function (brainstem auditory evoked response) correlates with acoustic cry features in term infants with hyperbi l i rubinemia. J Pediatr 1989;115:296-302.

42. Rothenberg AJ, Cansino S, Sepkoski C, Torres LM, Medina S, Schnaas L, et al. Prenatal and perinatal lead exposures alter acoustic cry parameters of neonate. Neurotoxicol Teratol 1995;17:151-60.

43. Frodi AM, Lamb ME, Leavitt L, Donavan W. Fathers' and mothers' responses to infant smiles and cries. Infant Behav Dev 1978;1:187-98.

44. Frodi AM, Lamb ME, Leavitt L, Donavan W. Fathers' and mothers' responses to the appearance and cries of premature and normal infants. Dev Psychol 1978;14:490-8.

45. Zeskind PS, Lester BM. Acoustic features and auditory perceptions of the cries of newborns with prenatal and perinatal complications. Child Dev 1978;49:580-9.

46. Freudenberg R, Driscoll J, Stern G. Reactions of adult hu- mans to cries of normal and abnormal infants. Infant Behav Dev 1978;1:224-7.

47. Lounsbury ML, Bates JE. The cries of infants of differing levels of perceived temperamental difficultness: acoustic properties and effects on listeners. Child Dev 1982;53:677- 86.

48. Boukydis CFZ, Burgess RL. Adult physiological responses to infant cries: effects of temperament of infant, parental status and gender. Child Dev 1982;53:1291-8.

49. Lester BM, Boukydis CFZ, Garcia-Coll CT, Hole W, Peucker M. Infantile colic: acoustic cry characteristics, maternal perception of cry, and temperament. Infant Behav Dev 1992;15:15-26,

50. Bakeman R, Brown J. Early interaction: consequences for social and mental development at three years. Child Dev 1980;51:437-47.

51. DeVitto B, Goldberg S. The effects of newborn medical status on early parent-infant interaction. In: Field T, Sostek A, Goldberg S, Shuman HH, editors. Infants born at risk. New York: Spectrum, 1979:311-32.

52. Lester BM, Zeskind PS. Brazelton scale and physical size correlates of neonatal cry features. Infant Behav Dev 1979;2:393-402.

53. Field T. Interaction patterns of preterm and term infants. In: Field T, Sostek A, Goldberg S, Shuman HH, editors. Infants born at risk. New York: Spectrum, 1979:333-56.

54. Crowe HP, Zeskind PS. Psychophysiological and perceptual responses to infant cries varying in pitch: comparison of adults with low and high scores on the Child Abuse Potential Inventory. Child Abuse Neglect 1992;16:19-29.

55. Lester BM, Garcia-Coll CT, Valcarcel M. Perception of infant cries in adolescent and older mothers. J Youth Adolesc 1989; 18:231-43.

56. Wessel AM, Cobb JC, Jackson EB, Harris GS, Detwiller AC. Paroxysmal fussing in infancy sometimes called "colic." Pe- diatrics 1954;14:421-35.

57. Lester BM, Boukydis CFZ, Garcia-Coll CT, Hole WT. Colic for developmentalists. Infant Med Health J 1990;11:321- 33.

58. Lester BM. Definition and diagnosis of colic. In: Colic and excessive crying. The 105th Ross Conference on Pediatric Research, 1996. In press.

59. Boukydis CFZ. Impact of colic and excessive crying on maternal cry perception and caregiving. Pediatr Res. In press.

60. Chess S, Thomas A. Origins and evolution of behavioral disorders. New York: Brunner/Mazel, 1984.

61. Lester BM, Boukydis CFZ, Garcia-Coll CT, Peucker M, McGrath MM, Vohr BR, et al. Developmental outcome as a function of the goodness of fit between the infant's cry characteristics and the mother's perception of her infant's cry. Pediatrics 1995;95:516-21.


Appendix

GLOSSARY

General Acoustical T e r m s

CRY LATENCY (of a pain stimulus): The time between the pain stimulus and the onset of the first expiratory utterance.

CRY MODES: A function of vibrational mode of vocal folds. Specific cry modes include:

Phonation:

Hyperphonation:

Dysphonation:

Inspiratory phonation:

Category of cry sounds resulting from harmonic vibration (usually between 350 and 750 vibrations per second) of the vocal chords during an expiratory utterance Category of cry sounds caused by a change in vocal register resulting in an harmonic vibration (usually between 1000 and 2000 vibrations per second) of the vocal chords during an expiratory utterance Category of cry sounds caused by a change in vocal register resulting in an inharmonic (or noisy) vibrat ion of the vocal chords during an expiratory utterance Category of cry sounds resulting from vibration of the vocal folds during an inspiratory utterance

CRY MODE CHANGES: The number of times the cry modes change within any given utterance.

FUNDAMENTAL FREQUENCY fro): Base frequency, during harmonic vibration (that include the Cry Modes of Phonation and Hyperphonation) of vocal cord vibration. Fundamental Frequency is usually heard as the pitch of the cry.

FORMANT FREQUENCIES (F v F z ... F ) : Cen- ter frequencies of the theoretically infinite number of resonances of the vocal tract system. The center frequency of the first resonance is defined as the first

formant (F1), and the second is defined as the second formant (F2), etc. Only the first two or three forrnants are usually measured.

G L O T T A L STOPS: Short, expiratory bursts of sound created by a sudden opening and sustained clos- ing of the vocal folds.

Terms Unique to the Sound Spectrogram BI-PHONATION: An apparent double series of har-

monics of two fundamental frequencies. Unlike double harmonic break, these two series seem to be independent of each other.

CONTINUITY: A measure of whether the cry was entirely voiced, partly voiced, or voiceless.

D O U B L E H A R M O N I C BREAK: A seemingly simultaneous parallel series of harmonics in between the harmonics of the fundamental frequency.

FURCATION: Term used to denote "split" in the f0 where a relatively strong signal suddenly breaks into a series of weaker ones, each of them with its own f0 contour.

GLIDING: A very rapid up or down movement of fo- GLOTTAL PLOSIVES: Sudden release of pressure

at the vocal folds producing an impulsive expiratory sound.

GLOTTAL ROLL or VOCAL FRY: Sound resulting from inharmonic vibration of the vocal folds, usually occurring at the end of an expiratory utterance when the signal becomes very weak and f0 becomes very low.

MAXIMUM PITCH: The highest measurable point of the fundamental frequency (f0)"

MELODY TYPE: Fundamental frequency variations that are either falling, rising-falling, rising, falling- rising, or fiat.

MIN IMU M PITCH: The lowest measurable point of the fundamental frequency (f0)-

NOISE CONCENTRATION: High-energy peak at 2000 to 2300 Hz, found both in voiced and voiceless signals. This attribute is clearly audible.

P ITCH OF SHIFT: Frequency after a rapid increase in the f0 contour.

VIBRATO: At least four rapid up-and-down move- ments of f0 within one expiratory utterance.


the infant cry: what can it tell us?

Documents