- Email: [email protected]
Cry analysis in infants with infectious and congenital disorders of the larynx
International
Ekevier
Journal of Pediatric
Biomedical
Otorhinolatyngology,
4 (1982)
151- 169
Press
Cry analysis in infants with infectious and congenital disorders of the larynx Jan l&es, Katarina (J. R.) Rehabilitation Uniuersity Hospital. Depurtment
Centre of Speech ond Hewing Children’s Hosprtul. Umoersi!v Centre
far Stutistics
Dehaen and
Disorders und (F. D.) Depurtment of Pediutrtcx.
iJnioersi@ of Brussels, Luurheekluun
of Pediutrics.
29 (Finlund) und (M.D.)
Michelsson, Freddy Marc Despontin
101. 1090 Brussel.~ (Belgtum) and (K. M.) II
of Helsinkr.
Stenhtickrnkutu
I I. SF-00.?90 Ilelsrrrk,
und Opemtions Reseurch. iJnioersi
1050 Brussels (Belgium)
(Received January (Accepted January
6. 1982) 18. 1982)
Summary Thirty induced pain cries from the same number of infants with infectious or congenital disorders of the larynx (infectious laryngitis, laryngomalacia, paresis of the recurrent nerve and subglottic stricture) have been analyzed by means of sound spectrography. For each cry 21 phonetic features have been evaluated. This cry material was then compared with 120 pain cries from healthy infants. In the pathological group a significant increase could be noted in the occurrence of the following cry attributes: second pause, abnormal melody types (rising, falling-rising, flat and no melody types), instability of the fundamental frequency, bi-phonation, vibrato, half-voiced voice quality, noise concentration, and inspiratory stridor. Furthermore, a significant decrease could be noticed in the occurrence of voiceless cries, falling and rising-falling melody types, and glottal roll. These findings show that such spectrographic features as very high maximum and minimum pitch, bi-phonation, gliding, and abnormal melody type occur more often in cries of infants with central nervous system involvement. Moreover, such parameters as instability of the fundamental frequency and noise concentration as indicators of neurologic disorders need further consideration. Except features of inspiratory stridor, this study revealed no really clear parameters typical of peripheral diseases of the vocal tract.
0 I65-5~76/X2/oooO-oooO/$O2.75
2~ 19X2 Elsevier
Biomedical
Press
Introduction Since the early sixties, a research group led by Wasz-Htikert, Helsinki, has been investigating the induced pain cries of normal and abnormal infants by means of sound spectrography. In the course of these studies, summarized in 1976 by SirviG and Michelsson [ 161 the most important finding has been that the cry characteristics of infants with pathological conditions involving the central nervous system (CNS) deviate from normal cry patterns. Very high maximum and minimum pitch, bi-phonation, glide, abnormal melody types, noise concentration and instability of the fundamental frequency occur frequently in the crying of children with CNS pathologies, but are not, or at least seldom, found in healthy crying, and in the cries from infants with feeding tubes [ 131, with cleft palate [ 121, and with hypothyroidism [9]. These features are therefore believed to have pathognomic value for neurological involvement [ 161. In order to confirm the hypothesis that they really originate from central nervous system dysfunction, they should be absent from the cries from infants with other peripheral diseases. As Lester and Zeskind [6] state ‘continued study of features that are primarily caused by peripheral mechanisms as in infants with laryngeal and oral abnormalities who are not neurologically impaired is crucial’. The purpose of the present study was to investigate whether or not these so-called ‘neuropathological’ cry characteristics emerge in infants with laryngeal diseases such as infectious laryngitis, laryngomalacia, paresis of the recurrent nerve and subglottic stricture.
Material Between 1973 and 1981, induced pain cries from 30 infants with laryngeal disorders were tape-recorded at the Children’s Hospital, University of Helsinki, Finland, and at the Department of Pediatrics, University Hospital, Brussels, Belgium. At the time of recording, the ages of 21 of these children varied between 8 days and 12 months; the remaining 9 were between 13 and 24 months old. In 18 cases, the clinical diagnosis was infectious laryngitis, while 12 babies showed other disorders of the larynx: 8 infants had laryngomalacia, 2 presented with paresis of the recurrent nerve, and 2 suffered from subglottic stricture. As far as the laryngitis cases are concerned, the diagnosis was quite clear, so that no laryngoscopies were made; in all the other cases direct laryngoscopies were performed under narcosis. These examinations confirmed the diagnosis of laryngomalacia in all cases. One of the patients with paresis of the recurrent nerve showed a bilateral paresis, the other a unilateral one. In one of the babies with subglottic stricture, the stenosis was situated at about 1 cm below the vocal cords; it was so severe that, when the child was not ventilated, there was only an opening of 1 mm in diameter. In the other case the stricture was situated at about the same place, but was less pronounced. None of the infants had any other disease besides the laryngeal one, and 28 of them had been born with normal birth weights. Only the two babies with subglottic stricture were premature, who had developed the stenosis as a consequence of
intubation for a long time after birth; the cry recordings, however, were performed when the infants were several months old, and therefore, it is improbable that the prematurity could have influenced the spectrographic cry characteristics. Indeed. Michelsson [8] showed that the cry features of prematures become normal at the time they normally should have been born. One pain cry was analyzed from each of these 30 infants. The control material consisted of 120 pain cries from healthy babies from 0 to 7 months, described in 1968 by Wasz-Hiickert et al. [ 191.
Methods All pain cries were induced by pinching the babies in the upper arm or the ear. The cry signals were recorded on an Uher 4000 L tape recorder or on a Sony TC-55 cassette recorder, with an AGK-58-D-200 microphone. The recordings were analyzed with a Kay Sonagraph 7029-A. For the spectrographic analysis of our pathological material, we have applied exactly the same methods as were used for our reference material. Each of the 30 cries described in Table I was the first following a painful stimulus; the second and third cries after the stimuli were also analyzed in some cases, but these results are not included in the statistical analysis. With a view to making the data also useful in later spectrographic cry studies, the analysis was done as completely as possible: besides the 7 ‘neuropathognomic’ cry features already mentioned in the Introduction, 14 other parameters were evaluated in each cry. The altogether 21 cry characteristics studied are briefly defined below. These definitions are based on the descriptions of the cry parameters in Lind et al. [7], Wasz-Hiickert et al. [19], Sirvio and Michelsson [ 161 and Michelsson et al. [ 151. Only the definition of the last parameter, inspiratory stridor, is original, as this characteristic has not been in Scandinavian cry research up to now. (1) Latency: time between the painful stimulus and the onset of the following first pain cry. (2) Duration: duration of the first cry after the stimulus, measured from the onset of the first cry signal with a duration of at least 0.4s until the end of the last cry signal with a duration of at least 0.4 s. (3) Secondpause: time between the end of the first cry after the stimulus, and the onset of the second cry after the stimulus; if it is less than 1.0 s it is not designated as a second pause. (4) Maximum pitch: highest value the fundamental frequency reaches, outside the shift parts and the glottal roll. (5) Minimum pitch: lowest value the fundamental frequency reaches, outside the shift parts and the glottal roll. (6) Shift: abrupt upward or downward change of the fundamental frequency: the shift part is always the shortest part of the cry. (7) Glide: up or down movement of the fundamental frequency with at least 600 Hz in less than 0.1 s.
160
(8) Melody type: the general up- and/or downward movement of the fundamental frequency in a cry; the melody type can be falling, rising-falling, rising, falling-rising or flat, and sometimes no melody type is detectable. (9) Instability ofthefundumentulfrequency: the fundamental frequency is unstable when it shows frequent, considerable up and down movements. (10) Double harmonic break: parallel series of relatively weak harmonics, occurring simultaneously with the fundamental and its harmonics. (11) Bi-phonation: additional fundamental with its own harmonics, occurring together, but not parallel with the fundamental and its harmonics. ( 12) Furcation: sudden split of the fundamental frequency and/or its harmonics into a series (mostly 2 or 3) of weaker harmonics, each of them with its own frequency contour. (13) Tonal pit: considerable down- and upward movement of the fundamental frequency; the fall in pitch must exceed 30% of the original value, and occur in less than 0.4 s. (14) Deviation of pitch: rare, hardly definable feature, involving, the deviation of the harmonics away from each other. (15) Vibrato: at least 4 successive, fast up- and downward movements of the fundamental frequency. (I 6) Glottal roll: part of a cry with very low pitch and intensity. (17) Voice quality: can be labelled as voiced, half-voiced or voiceless; a cry is voiced when the fundamental frequency is measurable over its entire duration, and voiceless when there are no reliable frequency measurements possible; in all other cases, the cry is half-voiced. (18) Continuity: this characteristic is attributed to cries only consisting of one part, and thus being uninterrupted. (19) Glottal plosiues: very short, separate cry fragments (C 0.4 s). (20) Noise concentration: concentration of noise in a relatively small band occurring in an area between f 1200 and f 2500 Hz. (21) Inspirutory stridor: a loud, more or less voiced inspiration, with a lot of noise components over the entire frequency range. Some of the most important spectrographic features are represented schematically in Fig. 1.
Results Table I shows the results of the cry analysis for the 30 cries from infants with laryngeal disorders compared with those for the group of 120 normal infants. Inspiratory stridor was the main symptom of the laryngeal diseases investigated, and occurred in 77% of the pathological cries. The feature was absent in normal cries. Fig. 3 shows a typical example of the inspiratory stridor, following a cry from a baby with infectious laryngitis. Further, there was a significant increase in the occurrence of the following cry attributes: second pause, abnormal melody types (rising, falling-rising, flat and no
161
TABLE
I
SPECTROGRAPHIC CHARACTERISTICS PARED WITH THOSE FROM INFANTS
OF PAIN CRIES FROM HEALTHY WITH LARYNGEAL DISORDERS Healthy
Number Number I. 2. 3. 4. 5. 6. 7. R.
9. IO. I I. 12. 13. 14. 15. 16. I I.
IX. 19. 20. 21.
of infants of cries analyzed
Latency. mean (s) Standard deviation Duration. mean (s) Standard deviation II Pause(%) Maximum pitch, mean (Hz) Standard deviation Minimum pitch, mean (Hz) Standard deviation Shift (%) Glide (S) Melody type (Ri) Falling Rising-falling Rising Falling-rising Flat None Instability of fundamental frequency (%) Double harmonic break (%,) Bi-phonation (%) Furcation (%) Tonal pit (%) Deviation of pitch (%) Vibrato (%) Glottal roll (S) Voice quality (I%) Voiceless Half-voiced Voiced Continuous signals (%) Glottal plosives (%) Noise concentration (S) Inspiratory stridor (I)
120 120 2.0 I.2 2.7 1.3 39 670 90 390 80 35 0
0 55 0 0 0 0 0 6X
x 24 6X 71 39 0 0
infants
INFANTS
COM-
Infants with laryngeal abnormalities 30 30 2.2 1.X 3.2 2.0 80 *** 6X2 263 422 133 23 0
13 *** 37 IO ** 0 0 0 13 *** 17 *** 0*
53 ** 47 67 23 7* 77 ***
* P (0.05, ** P ~0.005, *** P
162 Khz
I
1.0
2.0
sec.
Fig. I. Schematic drawing of a sonagram displaying the following pitch of fundamental frequency: 3. minimum pitch of fundamental harmonic break; 6. vibrato: 7, glide; 8. furcation.
characteristics: I. shift: 2. maximum frequency: 4. bi-phonation: 5. double
Khz
--
of a pain cry from a healthy
50
2:o
110 Fig. 2. Sonagram
sec.
infant.
110
210
3.0
--T-
410
5.0 0
610 set
Fig. 3. Sonagram of a pain cry from a baby with infectious laryngitis at the age of 8 months. pathologically loud and noisy inspiration following the cry, is called inspiratory stridor.
The
163
*-
noise concentration
_
..---.-
I
1
I
1.0 Fig. 4. Spectrogram
. .
2.0
3.0
sec.
of a pain cry from an infant with laryngomalacia. at the age of 14 days. Note the between about I500 Hz and I700 Hz. almost over the entire duration of the crv.
melody types), instability of the fundamental frequency, bi-phonation, vibrato, half-voiced voice quality, and noise concentration. A significant decrease was noted in ‘the occurrence of voiceless cries, falling and rising-falling melody types. and glottal roll. No significant differences were observed in the other cry characteristics. A cry from laryngomalacia can be seen in Fig. 4; a cry from bilateral paresis of the recurrent nerve in Fig. 5. These cries can be compared with the normal pain cry in Fig. 2. Khz
1.0
sec.
Fig. 5. Sonagram of a pain cry from a baby with paresis of the recurrent nerves at the age of 26 days. Note the very high pitch at the beginning of the cry. and the voicelen?, central part.
164
Discussion The pathological material Four different laryngeal diseases were included in our pathological material, because the main purpose of this study was not to point out the cry characteristics of individual laryngeal disorders, but to investigate whether or not cry attributes which are thought to have neuropathognomic value appear in the cries from infants with different kinds of laryngeal diseases as a general group. Although 9 of these infants were older than 1 year on the day of the cry recording, their cries were included in our material, because, apart from the first 5 days after birth, cry characteristics do not seem to change much in the course of the first 2 years of life. This conclusion is justified by a comparison of the spectrographic analysis of 30 cries from 15 healthy infants between 7 months and 2 years [5], with the corresponding data for 120 cries from infants between 0 and 7 months [19]. The only differences seem to be slight decrements in the values for duration and maximum pitch. The reference material The control series consisted of 120 pain cries from healthy babies from 0 to 7 months, described in 1968 [19]. However, at that time many cry parameters were not yet measured, because they were only discovered in later cry studies of different diseases. Therefore, some reference data are drawn from other sources, as seen in Table I [ 12,15,16]. The value 0 for inspiratory stridor is founded on our experience that the phenomenon is not observed in normal infants. Neuropathognomic cty features To date, the phenomena very high maximum and minimum pitch, glide, biphonation, frequent occurrence of rising, falling-rising, flat and no melody types, noise concentration and instability of the fundamental frequency have been regarded to be common in neurological disorders. Table II shows the values for these features in two neurological disorders (central asphyxia and bacterial meningitis) and in two disorders which may involve cerebral dysfunction (hyperbilirubinemia and marasmus); for practical reasons these 4 pathological conditions will be briefly referred to in the rest of this discussion as ‘neurological’ diseases. Table III shows to what extent these characteristics are also found in the ‘peripheral’ disorders investigated up to now (cleft palate, hypothyroidism and laryngeal disorders), and in healthy infants. From Table II it can be seen that all the values for maximum pitch in the ‘neurological’ diseases were significantly higher than normal. In the ,‘peripheral’ diseases on the other hand, no mean maximum pitch value significantly exceeded the normal level. In hypothyroidism the maximum pitch was even significantly lower than in the control series. The mean maximum pitch in the cries from infants with laryngeal disorders (682 Hz) was not significantly higher than the maximum pitch in normal cries. One infant with paralysis of the recurrent nerve (Fig. 5) however, reached a maximum pitch value of 1500 Hz, which is uncommon in normal cries. In a new control material by Thoden and Koivisto ( 17) consisting of 140 cries
165
TABLE
II
FIVE CRY CHARACTERISTICS DYSFUNCTION
IN FOUR
Diagnosis
Hyperhilirubinemia
source
(18)
No of cries analyzed
PATHOLOGICAL
CONDITIONS
Bacterial meningitis
Central aspyxia (X. II) 70
45
INVOLVIN
(II.
HRAIN
Marsmu\
1I)
(5. 26
15)
I IO
I. Maximum
pitch Mean (Hz) Standard deviation 2. Minimum pitch Mean (Hz) Standard deviation 3. Ri-phonation (8) 4. (ilide (a j 5. Melody type (“F) Rising Falling-rising Flat None * Feature
TABLE
occurs
2 120 I 200
I320 910
I loo x90
I 340 650
960
640 460 26 I4
560 400 40 II
730 370 23 0
650 49 *
13
9
17 ;; 1 50
O7 0I
more frequently
0
than in healthy
j4\27 23 0I
;; i 50 II
cries. but an exact value is not known
III
FIVE CRY INFANTS
CHARACTERISTICS
Diagnosis
Cleft palate
Source No of cries analyzed
(11.12) 52
IN
THREE
PERIPHERAL
Hypothyroidism
(9. II) 40
DISEASES
Laryngeal
diseases
see Table I
30
AND
IN
HEALTHY
Healthv
see Tahlc I’0
I
Maximum pitch Mean (Hz) Standard deviation 2. Minimum pitch Mean (Hz) Standard deviation 3. Ri-phanation (%,) 4. Glide (%,) 5. Melody type (a) Rising Falling-rising Flat None
710 2x0
470 130
6X2 263
670 YO
360 90 9 IO
290 loo 0 0
422 I33 I0 0
3YO
I2
1 i
12 24 0 112
x0 0 0 5
I
‘a 37
y IO
3 17
2
infant\
I
166
from 38 infants, varying in age between 1 day and 6 months, the mean maximum pitch was 620 Hz. In this material the maximum pitch exceeded 1000 Hz in 9% of the cries, and in 1% a pitch > 1500 Hz was noted. Also the minimum pitch was higher in the cries of infants with ‘neurological’ diseases than in the controls and infants with ‘peripheral’ disorders (Tables II and III). Bi-phonation, like very high maximum pitch, is very frequent in ‘neurological diseases (Table II), but to some extent it is also present in normal cries and in peripheral diseases. The characteristic has not been found in the normal material presented in Tables I and III, but it was noted in 1% of the normal cries described by Michelsson et al. [12]. We registered it in 3 of the 30 pathological cries analyzed in this paper, and it was also present in 9% of the ‘cleft palate’ cries; it did not occur in cries from hypothyroidism (Table III). It thus becomes clear that it is not the occurrence of bi-phonation per se, but the high frequency of the phenomenon which is typical of CNS disturbances. In all ‘neurological’ diseases mentioned in Table II, except in marasmus, glides occur more frequently than in normal pain cries. In the cries from healthy infants and from babies with hypothyroidism, they are absent. Glides equally did not occur in the 30 pain cries from laryngeal disorders we included in the statistical analysis performed in this study, and which were all first cries after the painful stimulus; we did find the feature once, however, in the third cry after the stimulus from a baby with subglottic stricture. Further, the phenomenon was noted in 10% of the cries from infants with cleft palate. So, as far as glide is concerned, about the same holds true as for bi-phonation: the high frequency of the characteristic, and not its mere occurrence, must be regarded as indicating neuropathology. Since rising, falling-rising, flat and no melody types only occur in 10% of normal cries, they can be referred to as abnormal melody types. In the cries from babies with ‘neurological’ diseases, except in hyperbilirubinemia (Table II), abnormal melody types occurred significantly more often than in the corresponding control materials. In the normal material by Thoden and Koivisto, abnormal melody types occurred in 20% of the cries [ 171. In the present cry material from laryngeal disorders, there is a statistically significant increase in the occurrence of abnormal melody types. In the cleft palate and hypothyroidism materials an increase was noted too, but it was not statistically significant. Thus abnormal melody types seem to be most frequent in the ‘neurological’ diseases, and their very high occurrence may indicate CNS dysfunction, although this hypothesis needs to be confirmed by further investigation. Noise concentration has only been signalled in herpes virus encephalitis [lo], but it should be remarked that the phenomenon has not been looked for explicitly in the other neuropathological groups. In the present study it was observed in 2 of the 30 pain cries from laryngeal disorders; both infants suffered from laryngomalacia (see Fig. 4). As far as instability of the fundamental frequency is concerned, this feature has not been looked for in the cries from hyperbilirubinemia, central asphyxia, cleft palate, and hypothyroidism. It does not appear in normal cries, but 11% of the pain cries from babies with meningitis were unstable, and also 9% of the cries from another
167
group of babies with asphyxia, studied in 1977 by Michelsson et al. [14]. The characteristic was noted in 13% of the cries from laryngeal disorders and needs, together with noise concentration, further investigation before it can be considered to have neuropathological value. It is a fact, however, that CNS dysfunction does affect cry patterns in several typical ways. Apart from the cries from the 30 infants described hitherto, we additionally analyzed the following 3 pain cries. The first was recorded from a 3-month-old patient who besides laryngomalacia also presented convulsions. In this cry the maximum pitch reached a value of 2750 Hz, and there was also a glide. Another baby, who at the age of 6 months showed laryngitis together with convulsions, produced a pain cry with a maximum pitch of 1800 Hz, containing a glide and a bi-phonation. The third baby was diagnosed as having paresis of the left vocal cord, and showed also hyperbilirubinemia. In his cry, a maximum pitch of 2200 Hz, and a glide were observed. Other cry features Inspiratory stridor is mainly found in laryngeal diseases, but can also be a symptom of supralaryngeal or tracheal pathologies involving a stricture in the vocal tract [4]. This feature has not previously been reported in the Scandanavian cry literature, but Dr. J. Hirschberg (Hungary) analyzed it spectrographically in 1966 [I], in 1972 [2], and in 1980 [3]. It is worth noting in this connection that Hirschberg has pointed out the possibility of differentiating between laryngeal and supralaryngeal stricture by means of spectrographic analysis of the stridor. Hoarseness is also a typical characteristic of the cries of infants with abnormalities at the level of the glottis [3,4]. Although our classification system of cries in voiced, half-voiced and voiceless (see p. 160) does not directly quantify the degree of hoarseness in the cries, we found a significant increase of half-voiced cries, accompanied by a decrease (although not significant) of voiced cries in our pathological material. In spite of the absence of totally voiceless cries in this sample, our findings tend to confirm the general pediatric view that a bad voice quality can be indicative of laryngeal pathology. Tonal pit does not appear in normal cries, and does not seem to be present in neurological disorders either, although it must be stated that researchers probably have not been looking for it explicitly in that group of diseases. It occurred in 22% of the cries from cleft palate. In the 30 cries included in the statistical analysis of the present study, tonal pit did not occur, but we found it in two cries from laryngomalacia, not included in the statistical calculations. These two findings are sufficient to state that the feature has no diagnostic value for cleft palate alone, but can also occur with other diseases in the vocal tract. The occurrence of vibrato varies in different materials of normal babies from 0% [8] to 54% [17]. Therefore, it does not seem to be pathognomic for any specific disease, a fact which concerns also the double harmonic break. Glottal roll has been noted in cries of healthy infants from 26% [ 171 to 68% [ 191. A decrease of glottal roll has been noted in cries of very sick babies [8,14].
168
Conclusion In conclusion, it may be stated that inspiratory stridor is the most typical phenomenon of ‘peripheral’ diseases of the vocal tract, while an apparent bad voice quality of the cries can suggest glottic abnormalities. The significance of other features as tonal pit, vibrato and glottal roll in this respect, however, has to be further investigated. High maximum pitch is the best indicator of neurological disturbances; also high minimum pitch, frequent occurrence of bi-phonation and glide, and most probably of rising, falling-rising, flat and no melody types indicate CNS pathologies. According to the present state of infant cry spectrography, the neuropathology of the characteristics noise concentration and instability of the fundamental frequency, needs further consideration.
Acknowledgements We thank the Finnish and Belgian Ministries of Education, the ‘Nationaal Fonds voor Wetenschappelijk Onderzoek’ (Belgium), and the Sigrid Juselius Foundation (Finland), for their financial support for this study.
References 1 Hirschberg,
J., Aphysiologische Stimmbildungen im Sauglingsalter, F&r phoniat., 18 (1966) 269-279. J., Klinische und akustische Analysen von pathologischen .%iuglingstimmen. In J. Hirschberg, G. Sz.&pe and E. Vass-Kovacs (Eds.), Papers in Interdisciplinary Speech Research, Proceedings of the Speech Symposium, Szeged 1971, Akademiai Kiado, Budapest, 1972, pp. 117- 125. 3 Hirschberg, J., Acoustic analysis of pathological cries, stridors and coughing sounds in infancy, Inr. J. Pediat. Otorhinolarg., 2 (1980) 287-300.
2 Hirschberg,
4 Hollinger,
P.H. and Johnston,
K.C.,
The infant
with respiratory
stridor,
Pediat.
C/in. N. Amer.,
2
(1955) 403-411. 5 Juntunen, Paediat.,
K., Sirvio, P. and Michelsson, 128 (1978)
K., Cry analysis
in infants
with severe malnutrition.
Europ. J.
241-246.
6 Lester,
B.M. and Zeskind, P.S., The organization and assessment of crying in the infant at risk. In T.M. Field, A.M. Sostek, S. Goldberg and H.H. Shuman (Eds.), Infants Born ar Rusk. Behavior and Deoelopment. Medical & Scientific Books. New York, NY, 1979, pp. 121-144. 7 Lind, J., Wasz-Hlickert, 0.. Vuorenkoski. V.. Partanen, T.J., Theorell. K. and Valanne. E., Vocal response to painful stimuli in newborn and young infants, Ann. paediat. few.. 12 (1966) 55-63. 8 Michelsson, K., Cry analyses of symptomless low birth weight neonates and of asphyxiated newborn infants, Actu paediat. wand., Suppl. 216 (1971). 9 Michelsson, K. and Sirvio, P., Cry analysis in congenital hypothyroidism, Folia phomar.. 2X (1976) 40-47.
10 Michelsson,
K. and Sirvio, P., Cry analysis of newborn infants with herpes virus hominis encephalitis. In: Proc. 5th European Congress of Perinatol Medicine, Uppsala, Sweden, June 9-12, 1976, Almqvist & Wiksell, Stockholm, 1976, p. 128 (abstr. 212). 11 Michelsson, K. and Wasz-Hiickert, O., The value of cry analysis in neonatology and early infancy. In T. Murry and J. Murry (Eds.), Znfanr Communication: Cry and Early Speech, College-Hill, Houston, TX. 1980, pp. 152-182.
169
12 Michelsson, K., Sirvili, P., Koivisto, M., Sovijarvi, A. and Was-Hockert. 0.. Spectrographic analysis of pain cry in neonates with cleft palate, Biol. Neonute, 26 (1975) 353-358. 13 Michelsson, K., Sit%, P., Koivisto, M. and Wasz-Hockert, O., Comparison of pain cry in nconatc~ with and without feeding tube, Dew. Med. Child Neural., I6 (1974) 397. 14 Michelsson, K.. Sir% P. and Wasz-Hiickert. O., Pam cry in full-term asphyxiated newborn infants and correlated with late findings, Acrcr pclediur. scund., 66 (I 977) 6 I l-6 16. IS Michelsson, K., Sin%, P. and Wasz-Hiickert. O., Sound spectrographic cry analysis of infant\ uith bacterial meningitis, Deu. Med. Child Neural.. I9 (I 977) 309-3 15. I6 Sirvio. P. and Michelsson. K., Sound spectrographic cry analysis of normal and abnormal ncwhorn infants. A review and a recommendation for standardization of the cry characteri
Ekevier
Journal of Pediatric
Biomedical
Otorhinolatyngology,
4 (1982)
151- 169
Press
Cry analysis in infants with infectious and congenital disorders of the larynx Jan l&es, Katarina (J. R.) Rehabilitation Uniuersity Hospital. Depurtment
Centre of Speech ond Hewing Children’s Hosprtul. Umoersi!v Centre
far Stutistics
Dehaen and
Disorders und (F. D.) Depurtment of Pediutrtcx.
iJnioersi@ of Brussels, Luurheekluun
of Pediutrics.
29 (Finlund) und (M.D.)
Michelsson, Freddy Marc Despontin
101. 1090 Brussel.~ (Belgtum) and (K. M.) II
of Helsinkr.
Stenhtickrnkutu
I I. SF-00.?90 Ilelsrrrk,
und Opemtions Reseurch. iJnioersi
1050 Brussels (Belgium)
(Received January (Accepted January
6. 1982) 18. 1982)
Summary Thirty induced pain cries from the same number of infants with infectious or congenital disorders of the larynx (infectious laryngitis, laryngomalacia, paresis of the recurrent nerve and subglottic stricture) have been analyzed by means of sound spectrography. For each cry 21 phonetic features have been evaluated. This cry material was then compared with 120 pain cries from healthy infants. In the pathological group a significant increase could be noted in the occurrence of the following cry attributes: second pause, abnormal melody types (rising, falling-rising, flat and no melody types), instability of the fundamental frequency, bi-phonation, vibrato, half-voiced voice quality, noise concentration, and inspiratory stridor. Furthermore, a significant decrease could be noticed in the occurrence of voiceless cries, falling and rising-falling melody types, and glottal roll. These findings show that such spectrographic features as very high maximum and minimum pitch, bi-phonation, gliding, and abnormal melody type occur more often in cries of infants with central nervous system involvement. Moreover, such parameters as instability of the fundamental frequency and noise concentration as indicators of neurologic disorders need further consideration. Except features of inspiratory stridor, this study revealed no really clear parameters typical of peripheral diseases of the vocal tract.
0 I65-5~76/X2/oooO-oooO/$O2.75
2~ 19X2 Elsevier
Biomedical
Press
Introduction Since the early sixties, a research group led by Wasz-Htikert, Helsinki, has been investigating the induced pain cries of normal and abnormal infants by means of sound spectrography. In the course of these studies, summarized in 1976 by SirviG and Michelsson [ 161 the most important finding has been that the cry characteristics of infants with pathological conditions involving the central nervous system (CNS) deviate from normal cry patterns. Very high maximum and minimum pitch, bi-phonation, glide, abnormal melody types, noise concentration and instability of the fundamental frequency occur frequently in the crying of children with CNS pathologies, but are not, or at least seldom, found in healthy crying, and in the cries from infants with feeding tubes [ 131, with cleft palate [ 121, and with hypothyroidism [9]. These features are therefore believed to have pathognomic value for neurological involvement [ 161. In order to confirm the hypothesis that they really originate from central nervous system dysfunction, they should be absent from the cries from infants with other peripheral diseases. As Lester and Zeskind [6] state ‘continued study of features that are primarily caused by peripheral mechanisms as in infants with laryngeal and oral abnormalities who are not neurologically impaired is crucial’. The purpose of the present study was to investigate whether or not these so-called ‘neuropathological’ cry characteristics emerge in infants with laryngeal diseases such as infectious laryngitis, laryngomalacia, paresis of the recurrent nerve and subglottic stricture.
Material Between 1973 and 1981, induced pain cries from 30 infants with laryngeal disorders were tape-recorded at the Children’s Hospital, University of Helsinki, Finland, and at the Department of Pediatrics, University Hospital, Brussels, Belgium. At the time of recording, the ages of 21 of these children varied between 8 days and 12 months; the remaining 9 were between 13 and 24 months old. In 18 cases, the clinical diagnosis was infectious laryngitis, while 12 babies showed other disorders of the larynx: 8 infants had laryngomalacia, 2 presented with paresis of the recurrent nerve, and 2 suffered from subglottic stricture. As far as the laryngitis cases are concerned, the diagnosis was quite clear, so that no laryngoscopies were made; in all the other cases direct laryngoscopies were performed under narcosis. These examinations confirmed the diagnosis of laryngomalacia in all cases. One of the patients with paresis of the recurrent nerve showed a bilateral paresis, the other a unilateral one. In one of the babies with subglottic stricture, the stenosis was situated at about 1 cm below the vocal cords; it was so severe that, when the child was not ventilated, there was only an opening of 1 mm in diameter. In the other case the stricture was situated at about the same place, but was less pronounced. None of the infants had any other disease besides the laryngeal one, and 28 of them had been born with normal birth weights. Only the two babies with subglottic stricture were premature, who had developed the stenosis as a consequence of
intubation for a long time after birth; the cry recordings, however, were performed when the infants were several months old, and therefore, it is improbable that the prematurity could have influenced the spectrographic cry characteristics. Indeed. Michelsson [8] showed that the cry features of prematures become normal at the time they normally should have been born. One pain cry was analyzed from each of these 30 infants. The control material consisted of 120 pain cries from healthy babies from 0 to 7 months, described in 1968 by Wasz-Hiickert et al. [ 191.
Methods All pain cries were induced by pinching the babies in the upper arm or the ear. The cry signals were recorded on an Uher 4000 L tape recorder or on a Sony TC-55 cassette recorder, with an AGK-58-D-200 microphone. The recordings were analyzed with a Kay Sonagraph 7029-A. For the spectrographic analysis of our pathological material, we have applied exactly the same methods as were used for our reference material. Each of the 30 cries described in Table I was the first following a painful stimulus; the second and third cries after the stimuli were also analyzed in some cases, but these results are not included in the statistical analysis. With a view to making the data also useful in later spectrographic cry studies, the analysis was done as completely as possible: besides the 7 ‘neuropathognomic’ cry features already mentioned in the Introduction, 14 other parameters were evaluated in each cry. The altogether 21 cry characteristics studied are briefly defined below. These definitions are based on the descriptions of the cry parameters in Lind et al. [7], Wasz-Hiickert et al. [19], Sirvio and Michelsson [ 161 and Michelsson et al. [ 151. Only the definition of the last parameter, inspiratory stridor, is original, as this characteristic has not been in Scandinavian cry research up to now. (1) Latency: time between the painful stimulus and the onset of the following first pain cry. (2) Duration: duration of the first cry after the stimulus, measured from the onset of the first cry signal with a duration of at least 0.4s until the end of the last cry signal with a duration of at least 0.4 s. (3) Secondpause: time between the end of the first cry after the stimulus, and the onset of the second cry after the stimulus; if it is less than 1.0 s it is not designated as a second pause. (4) Maximum pitch: highest value the fundamental frequency reaches, outside the shift parts and the glottal roll. (5) Minimum pitch: lowest value the fundamental frequency reaches, outside the shift parts and the glottal roll. (6) Shift: abrupt upward or downward change of the fundamental frequency: the shift part is always the shortest part of the cry. (7) Glide: up or down movement of the fundamental frequency with at least 600 Hz in less than 0.1 s.
160
(8) Melody type: the general up- and/or downward movement of the fundamental frequency in a cry; the melody type can be falling, rising-falling, rising, falling-rising or flat, and sometimes no melody type is detectable. (9) Instability ofthefundumentulfrequency: the fundamental frequency is unstable when it shows frequent, considerable up and down movements. (10) Double harmonic break: parallel series of relatively weak harmonics, occurring simultaneously with the fundamental and its harmonics. (11) Bi-phonation: additional fundamental with its own harmonics, occurring together, but not parallel with the fundamental and its harmonics. ( 12) Furcation: sudden split of the fundamental frequency and/or its harmonics into a series (mostly 2 or 3) of weaker harmonics, each of them with its own frequency contour. (13) Tonal pit: considerable down- and upward movement of the fundamental frequency; the fall in pitch must exceed 30% of the original value, and occur in less than 0.4 s. (14) Deviation of pitch: rare, hardly definable feature, involving, the deviation of the harmonics away from each other. (15) Vibrato: at least 4 successive, fast up- and downward movements of the fundamental frequency. (I 6) Glottal roll: part of a cry with very low pitch and intensity. (17) Voice quality: can be labelled as voiced, half-voiced or voiceless; a cry is voiced when the fundamental frequency is measurable over its entire duration, and voiceless when there are no reliable frequency measurements possible; in all other cases, the cry is half-voiced. (18) Continuity: this characteristic is attributed to cries only consisting of one part, and thus being uninterrupted. (19) Glottal plosiues: very short, separate cry fragments (C 0.4 s). (20) Noise concentration: concentration of noise in a relatively small band occurring in an area between f 1200 and f 2500 Hz. (21) Inspirutory stridor: a loud, more or less voiced inspiration, with a lot of noise components over the entire frequency range. Some of the most important spectrographic features are represented schematically in Fig. 1.
Results Table I shows the results of the cry analysis for the 30 cries from infants with laryngeal disorders compared with those for the group of 120 normal infants. Inspiratory stridor was the main symptom of the laryngeal diseases investigated, and occurred in 77% of the pathological cries. The feature was absent in normal cries. Fig. 3 shows a typical example of the inspiratory stridor, following a cry from a baby with infectious laryngitis. Further, there was a significant increase in the occurrence of the following cry attributes: second pause, abnormal melody types (rising, falling-rising, flat and no
161
TABLE
I
SPECTROGRAPHIC CHARACTERISTICS PARED WITH THOSE FROM INFANTS
OF PAIN CRIES FROM HEALTHY WITH LARYNGEAL DISORDERS Healthy
Number Number I. 2. 3. 4. 5. 6. 7. R.
9. IO. I I. 12. 13. 14. 15. 16. I I.
IX. 19. 20. 21.
of infants of cries analyzed
Latency. mean (s) Standard deviation Duration. mean (s) Standard deviation II Pause(%) Maximum pitch, mean (Hz) Standard deviation Minimum pitch, mean (Hz) Standard deviation Shift (%) Glide (S) Melody type (Ri) Falling Rising-falling Rising Falling-rising Flat None Instability of fundamental frequency (%) Double harmonic break (%,) Bi-phonation (%) Furcation (%) Tonal pit (%) Deviation of pitch (%) Vibrato (%) Glottal roll (S) Voice quality (I%) Voiceless Half-voiced Voiced Continuous signals (%) Glottal plosives (%) Noise concentration (S) Inspiratory stridor (I)
120 120 2.0 I.2 2.7 1.3 39 670 90 390 80 35 0
0 55 0 0 0 0 0 6X
x 24 6X 71 39 0 0
infants
INFANTS
COM-
Infants with laryngeal abnormalities 30 30 2.2 1.X 3.2 2.0 80 *** 6X2 263 422 133 23 0
13 *** 37 IO ** 0 0 0 13 *** 17 *** 0*
53 ** 47 67 23 7* 77 ***
* P (0.05, ** P ~0.005, *** P
162 Khz
I
1.0
2.0
sec.
Fig. I. Schematic drawing of a sonagram displaying the following pitch of fundamental frequency: 3. minimum pitch of fundamental harmonic break; 6. vibrato: 7, glide; 8. furcation.
characteristics: I. shift: 2. maximum frequency: 4. bi-phonation: 5. double
Khz
--
of a pain cry from a healthy
50
2:o
110 Fig. 2. Sonagram
sec.
infant.
110
210
3.0
--T-
410
5.0 0
610 set
Fig. 3. Sonagram of a pain cry from a baby with infectious laryngitis at the age of 8 months. pathologically loud and noisy inspiration following the cry, is called inspiratory stridor.
The
163
*-
noise concentration
_
..---.-
I
1
I
1.0 Fig. 4. Spectrogram
. .
2.0
3.0
sec.
of a pain cry from an infant with laryngomalacia. at the age of 14 days. Note the between about I500 Hz and I700 Hz. almost over the entire duration of the crv.
melody types), instability of the fundamental frequency, bi-phonation, vibrato, half-voiced voice quality, and noise concentration. A significant decrease was noted in ‘the occurrence of voiceless cries, falling and rising-falling melody types. and glottal roll. No significant differences were observed in the other cry characteristics. A cry from laryngomalacia can be seen in Fig. 4; a cry from bilateral paresis of the recurrent nerve in Fig. 5. These cries can be compared with the normal pain cry in Fig. 2. Khz
1.0
sec.
Fig. 5. Sonagram of a pain cry from a baby with paresis of the recurrent nerves at the age of 26 days. Note the very high pitch at the beginning of the cry. and the voicelen?, central part.
164
Discussion The pathological material Four different laryngeal diseases were included in our pathological material, because the main purpose of this study was not to point out the cry characteristics of individual laryngeal disorders, but to investigate whether or not cry attributes which are thought to have neuropathognomic value appear in the cries from infants with different kinds of laryngeal diseases as a general group. Although 9 of these infants were older than 1 year on the day of the cry recording, their cries were included in our material, because, apart from the first 5 days after birth, cry characteristics do not seem to change much in the course of the first 2 years of life. This conclusion is justified by a comparison of the spectrographic analysis of 30 cries from 15 healthy infants between 7 months and 2 years [5], with the corresponding data for 120 cries from infants between 0 and 7 months [19]. The only differences seem to be slight decrements in the values for duration and maximum pitch. The reference material The control series consisted of 120 pain cries from healthy babies from 0 to 7 months, described in 1968 [19]. However, at that time many cry parameters were not yet measured, because they were only discovered in later cry studies of different diseases. Therefore, some reference data are drawn from other sources, as seen in Table I [ 12,15,16]. The value 0 for inspiratory stridor is founded on our experience that the phenomenon is not observed in normal infants. Neuropathognomic cty features To date, the phenomena very high maximum and minimum pitch, glide, biphonation, frequent occurrence of rising, falling-rising, flat and no melody types, noise concentration and instability of the fundamental frequency have been regarded to be common in neurological disorders. Table II shows the values for these features in two neurological disorders (central asphyxia and bacterial meningitis) and in two disorders which may involve cerebral dysfunction (hyperbilirubinemia and marasmus); for practical reasons these 4 pathological conditions will be briefly referred to in the rest of this discussion as ‘neurological’ diseases. Table III shows to what extent these characteristics are also found in the ‘peripheral’ disorders investigated up to now (cleft palate, hypothyroidism and laryngeal disorders), and in healthy infants. From Table II it can be seen that all the values for maximum pitch in the ‘neurological’ diseases were significantly higher than normal. In the ,‘peripheral’ diseases on the other hand, no mean maximum pitch value significantly exceeded the normal level. In hypothyroidism the maximum pitch was even significantly lower than in the control series. The mean maximum pitch in the cries from infants with laryngeal disorders (682 Hz) was not significantly higher than the maximum pitch in normal cries. One infant with paralysis of the recurrent nerve (Fig. 5) however, reached a maximum pitch value of 1500 Hz, which is uncommon in normal cries. In a new control material by Thoden and Koivisto ( 17) consisting of 140 cries
165
TABLE
II
FIVE CRY CHARACTERISTICS DYSFUNCTION
IN FOUR
Diagnosis
Hyperhilirubinemia
source
(18)
No of cries analyzed
PATHOLOGICAL
CONDITIONS
Bacterial meningitis
Central aspyxia (X. II) 70
45
INVOLVIN
(II.
HRAIN
Marsmu\
1I)
(5. 26
15)
I IO
I. Maximum
pitch Mean (Hz) Standard deviation 2. Minimum pitch Mean (Hz) Standard deviation 3. Ri-phonation (8) 4. (ilide (a j 5. Melody type (“F) Rising Falling-rising Flat None * Feature
TABLE
occurs
2 120 I 200
I320 910
I loo x90
I 340 650
960
640 460 26 I4
560 400 40 II
730 370 23 0
650 49 *
13
9
17 ;; 1 50
O7 0I
more frequently
0
than in healthy
j4\27 23 0I
;; i 50 II
cries. but an exact value is not known
III
FIVE CRY INFANTS
CHARACTERISTICS
Diagnosis
Cleft palate
Source No of cries analyzed
(11.12) 52
IN
THREE
PERIPHERAL
Hypothyroidism
(9. II) 40
DISEASES
Laryngeal
diseases
see Table I
30
AND
IN
HEALTHY
Healthv
see Tahlc I’0
I
Maximum pitch Mean (Hz) Standard deviation 2. Minimum pitch Mean (Hz) Standard deviation 3. Ri-phanation (%,) 4. Glide (%,) 5. Melody type (a) Rising Falling-rising Flat None
710 2x0
470 130
6X2 263
670 YO
360 90 9 IO
290 loo 0 0
422 I33 I0 0
3YO
I2
1 i
12 24 0 112
x0 0 0 5
I
‘a 37
y IO
3 17
2
infant\
I
166
from 38 infants, varying in age between 1 day and 6 months, the mean maximum pitch was 620 Hz. In this material the maximum pitch exceeded 1000 Hz in 9% of the cries, and in 1% a pitch > 1500 Hz was noted. Also the minimum pitch was higher in the cries of infants with ‘neurological’ diseases than in the controls and infants with ‘peripheral’ disorders (Tables II and III). Bi-phonation, like very high maximum pitch, is very frequent in ‘neurological diseases (Table II), but to some extent it is also present in normal cries and in peripheral diseases. The characteristic has not been found in the normal material presented in Tables I and III, but it was noted in 1% of the normal cries described by Michelsson et al. [12]. We registered it in 3 of the 30 pathological cries analyzed in this paper, and it was also present in 9% of the ‘cleft palate’ cries; it did not occur in cries from hypothyroidism (Table III). It thus becomes clear that it is not the occurrence of bi-phonation per se, but the high frequency of the phenomenon which is typical of CNS disturbances. In all ‘neurological’ diseases mentioned in Table II, except in marasmus, glides occur more frequently than in normal pain cries. In the cries from healthy infants and from babies with hypothyroidism, they are absent. Glides equally did not occur in the 30 pain cries from laryngeal disorders we included in the statistical analysis performed in this study, and which were all first cries after the painful stimulus; we did find the feature once, however, in the third cry after the stimulus from a baby with subglottic stricture. Further, the phenomenon was noted in 10% of the cries from infants with cleft palate. So, as far as glide is concerned, about the same holds true as for bi-phonation: the high frequency of the characteristic, and not its mere occurrence, must be regarded as indicating neuropathology. Since rising, falling-rising, flat and no melody types only occur in 10% of normal cries, they can be referred to as abnormal melody types. In the cries from babies with ‘neurological’ diseases, except in hyperbilirubinemia (Table II), abnormal melody types occurred significantly more often than in the corresponding control materials. In the normal material by Thoden and Koivisto, abnormal melody types occurred in 20% of the cries [ 171. In the present cry material from laryngeal disorders, there is a statistically significant increase in the occurrence of abnormal melody types. In the cleft palate and hypothyroidism materials an increase was noted too, but it was not statistically significant. Thus abnormal melody types seem to be most frequent in the ‘neurological’ diseases, and their very high occurrence may indicate CNS dysfunction, although this hypothesis needs to be confirmed by further investigation. Noise concentration has only been signalled in herpes virus encephalitis [lo], but it should be remarked that the phenomenon has not been looked for explicitly in the other neuropathological groups. In the present study it was observed in 2 of the 30 pain cries from laryngeal disorders; both infants suffered from laryngomalacia (see Fig. 4). As far as instability of the fundamental frequency is concerned, this feature has not been looked for in the cries from hyperbilirubinemia, central asphyxia, cleft palate, and hypothyroidism. It does not appear in normal cries, but 11% of the pain cries from babies with meningitis were unstable, and also 9% of the cries from another
167
group of babies with asphyxia, studied in 1977 by Michelsson et al. [14]. The characteristic was noted in 13% of the cries from laryngeal disorders and needs, together with noise concentration, further investigation before it can be considered to have neuropathological value. It is a fact, however, that CNS dysfunction does affect cry patterns in several typical ways. Apart from the cries from the 30 infants described hitherto, we additionally analyzed the following 3 pain cries. The first was recorded from a 3-month-old patient who besides laryngomalacia also presented convulsions. In this cry the maximum pitch reached a value of 2750 Hz, and there was also a glide. Another baby, who at the age of 6 months showed laryngitis together with convulsions, produced a pain cry with a maximum pitch of 1800 Hz, containing a glide and a bi-phonation. The third baby was diagnosed as having paresis of the left vocal cord, and showed also hyperbilirubinemia. In his cry, a maximum pitch of 2200 Hz, and a glide were observed. Other cry features Inspiratory stridor is mainly found in laryngeal diseases, but can also be a symptom of supralaryngeal or tracheal pathologies involving a stricture in the vocal tract [4]. This feature has not previously been reported in the Scandanavian cry literature, but Dr. J. Hirschberg (Hungary) analyzed it spectrographically in 1966 [I], in 1972 [2], and in 1980 [3]. It is worth noting in this connection that Hirschberg has pointed out the possibility of differentiating between laryngeal and supralaryngeal stricture by means of spectrographic analysis of the stridor. Hoarseness is also a typical characteristic of the cries of infants with abnormalities at the level of the glottis [3,4]. Although our classification system of cries in voiced, half-voiced and voiceless (see p. 160) does not directly quantify the degree of hoarseness in the cries, we found a significant increase of half-voiced cries, accompanied by a decrease (although not significant) of voiced cries in our pathological material. In spite of the absence of totally voiceless cries in this sample, our findings tend to confirm the general pediatric view that a bad voice quality can be indicative of laryngeal pathology. Tonal pit does not appear in normal cries, and does not seem to be present in neurological disorders either, although it must be stated that researchers probably have not been looking for it explicitly in that group of diseases. It occurred in 22% of the cries from cleft palate. In the 30 cries included in the statistical analysis of the present study, tonal pit did not occur, but we found it in two cries from laryngomalacia, not included in the statistical calculations. These two findings are sufficient to state that the feature has no diagnostic value for cleft palate alone, but can also occur with other diseases in the vocal tract. The occurrence of vibrato varies in different materials of normal babies from 0% [8] to 54% [17]. Therefore, it does not seem to be pathognomic for any specific disease, a fact which concerns also the double harmonic break. Glottal roll has been noted in cries of healthy infants from 26% [ 171 to 68% [ 191. A decrease of glottal roll has been noted in cries of very sick babies [8,14].
168
Conclusion In conclusion, it may be stated that inspiratory stridor is the most typical phenomenon of ‘peripheral’ diseases of the vocal tract, while an apparent bad voice quality of the cries can suggest glottic abnormalities. The significance of other features as tonal pit, vibrato and glottal roll in this respect, however, has to be further investigated. High maximum pitch is the best indicator of neurological disturbances; also high minimum pitch, frequent occurrence of bi-phonation and glide, and most probably of rising, falling-rising, flat and no melody types indicate CNS pathologies. According to the present state of infant cry spectrography, the neuropathology of the characteristics noise concentration and instability of the fundamental frequency, needs further consideration.
Acknowledgements We thank the Finnish and Belgian Ministries of Education, the ‘Nationaal Fonds voor Wetenschappelijk Onderzoek’ (Belgium), and the Sigrid Juselius Foundation (Finland), for their financial support for this study.
References 1 Hirschberg,
J., Aphysiologische Stimmbildungen im Sauglingsalter, F&r phoniat., 18 (1966) 269-279. J., Klinische und akustische Analysen von pathologischen .%iuglingstimmen. In J. Hirschberg, G. Sz.&pe and E. Vass-Kovacs (Eds.), Papers in Interdisciplinary Speech Research, Proceedings of the Speech Symposium, Szeged 1971, Akademiai Kiado, Budapest, 1972, pp. 117- 125. 3 Hirschberg, J., Acoustic analysis of pathological cries, stridors and coughing sounds in infancy, Inr. J. Pediat. Otorhinolarg., 2 (1980) 287-300.
2 Hirschberg,
4 Hollinger,
P.H. and Johnston,
K.C.,
The infant
with respiratory
stridor,
Pediat.
C/in. N. Amer.,
2
(1955) 403-411. 5 Juntunen, Paediat.,
K., Sirvio, P. and Michelsson, 128 (1978)
K., Cry analysis
in infants
with severe malnutrition.
Europ. J.
241-246.
6 Lester,
B.M. and Zeskind, P.S., The organization and assessment of crying in the infant at risk. In T.M. Field, A.M. Sostek, S. Goldberg and H.H. Shuman (Eds.), Infants Born ar Rusk. Behavior and Deoelopment. Medical & Scientific Books. New York, NY, 1979, pp. 121-144. 7 Lind, J., Wasz-Hlickert, 0.. Vuorenkoski. V.. Partanen, T.J., Theorell. K. and Valanne. E., Vocal response to painful stimuli in newborn and young infants, Ann. paediat. few.. 12 (1966) 55-63. 8 Michelsson, K., Cry analyses of symptomless low birth weight neonates and of asphyxiated newborn infants, Actu paediat. wand., Suppl. 216 (1971). 9 Michelsson, K. and Sirvio, P., Cry analysis in congenital hypothyroidism, Folia phomar.. 2X (1976) 40-47.
10 Michelsson,
K. and Sirvio, P., Cry analysis of newborn infants with herpes virus hominis encephalitis. In: Proc. 5th European Congress of Perinatol Medicine, Uppsala, Sweden, June 9-12, 1976, Almqvist & Wiksell, Stockholm, 1976, p. 128 (abstr. 212). 11 Michelsson, K. and Wasz-Hiickert, O., The value of cry analysis in neonatology and early infancy. In T. Murry and J. Murry (Eds.), Znfanr Communication: Cry and Early Speech, College-Hill, Houston, TX. 1980, pp. 152-182.
169
12 Michelsson, K., Sirvili, P., Koivisto, M., Sovijarvi, A. and Was-Hockert. 0.. Spectrographic analysis of pain cry in neonates with cleft palate, Biol. Neonute, 26 (1975) 353-358. 13 Michelsson, K., Sit%, P., Koivisto, M. and Wasz-Hockert, O., Comparison of pain cry in nconatc~ with and without feeding tube, Dew. Med. Child Neural., I6 (1974) 397. 14 Michelsson, K.. Sir% P. and Wasz-Hiickert. O., Pam cry in full-term asphyxiated newborn infants and correlated with late findings, Acrcr pclediur. scund., 66 (I 977) 6 I l-6 16. IS Michelsson, K., Sin%, P. and Wasz-Hiickert. O., Sound spectrographic cry analysis of infant\ uith bacterial meningitis, Deu. Med. Child Neural.. I9 (I 977) 309-3 15. I6 Sirvio. P. and Michelsson. K., Sound spectrographic cry analysis of normal and abnormal ncwhorn infants. A review and a recommendation for standardization of the cry characteri