(Nov. 18th, 2009) Babies cry like mum, but when? Apparently, they learn the typical melody patterns of their mother tongue already in the womb. A report by our corresponding author, Horst Lautsprecher.
Speculation about the capacity of human babies to learn in the womb has seen various fads, with pregnant women playing instructive music at their bellies in the hope of fostering a new Mozart, or sending in coded signals in accord with the ‘Pat-pat rub-rub’ school of prenatal education. However, a Franco-German research alliance has now found evidence that those last months in the noisy womb may indeed enhance baby’s perceptions – by instilling the melodic essence of their mother’s national language!
Led by Kathleen Wermke, from the Centre for Pre-speech Development at Würzburg’s University Hospital, the group announced in Current Biology that “Newborns’ Cry Melody is shaped by their Native Language” (in print December 15, 2009).
Wermke has previously been more interested in studying baby cries as an early sign of “language development disorders”. Now, she has listened more closely to the underlying “melody patterns” and claims that, as early as 2-5 days after birth, babies are already imitating their mother’s speech.
As raw material, the cries of 30 German newborn babies in Berlin (15 boys, 15 girls, mean age 3.8 days) were compared with those of 30 French babies in Paris (11 girls, 19 boys, mean age 3.1 days). The babies were all healthy and full-term (born at a gestational age of 39.5 weeks). Most importantly, they had “normal hearing from a strictly monolingual (French or German) family background”.
“Cry recordings” were made in “pain-free situations” by sticking an “Earthworks TC20” microphone 15 centimeters from the crying baby’s mouth. Despite the microphone’s presence, we are assured the recordings were all made during normal mother-child interactions (feeding time, nappy time, etc. etc.), varying from 3 to 10 minutes “depending on the spontaneous crying behaviour of the neonate”, a cry being defined here as the “vocal output on a single expiration”. Some 2500 cries were recorded in total.
Well away from all those noisy babies, Wermke analysed their digitalised traces, carrying out “high-resolution melody (fundamental frequency contour) computations” with her “Computerised Speech Laboratory 4400” and “Cry-Data-Analysis Program”. This gave her frequency spectrograms of all cries “in order to identify voiced, harmonic cries and to exclude noisy, voiceless cries.” She excluded the latter as “low-dimensional chaos” due to “strong nonlinearities in the restoring forces resulting from an extremely large amplitude-to-length ratio of the vocal folds in newborns”, i.e. as parents of newborn babies can confirm, their crying can be a bit erratic to start with.
For the final melody analysis, precisely 1254 voiced cries were used, representing an average of 21 cries per French “neonate” and 18 for each of the German babies. These were simple cries containing single “rising-and-falling melody arcs” – the fundamental frequency of the pitch typically rising from around 400 Hertz to 550 Hertz then falling again. For Wermke, such melody arcs come in four basic types:
# “falling contour” – a quick rising, slowly decreasing melody;
# “rising contour”, slow rise, fast fall;
# a “symmetrical”, same rise and fall, contour;
# and a “plateau type”, with a “stable fundamental frequency”.
Her final analysis consisted in standardising every cry to a length of 1 second and then determining the time at which the maximum frequency pitch occurred, e.g. after 0.3 or 0.6 seconds.
And, lo and behold – the German “neonates” cried at their highest pitch after a normalized average of 0.44 seconds while those French babies clocked in at a more languid 0.58 seconds. That is, the French displayed “rising contours” while the Germans had “falling contours”. Well, for Wermke, the conclusion is evident: “Newborns prefer exactly the same melody patterns that are typical of their respective mother tongues.”
Her argument for this is that it is “well-known” that the French and German languages have different “intonation patterns”. In French, intonation is characterised by “a pitch rise toward the end of several kinds of prosodic units” (i.e. words) “except for the very last unit of an utterance, which represents a falling contour”. This is a “crucial difference” from German intonation, which “typically” exhibits a falling melody contour, “e.g. from the accented high-tone syllable to the end of the intonational phrase.”
Unfortunately, Wermke makes no mention of ever having recorded any French or German adults and presents no corresponding melody contour analysis of their typically melodic utterances (for which her main citations include a 1961 reference to the “intonation model of Simone de Beauvoir”). This might not matter when she says that the melodic variations in the babies’ cries challenges “the prevailing opinion that newborns could not actively influence their production of sound, that the cry of newborns is determined simply by rises and falls in respiratory pressure and is not influenced by the brain”. However, Wermke goes further: “the dramatic finding of this study is that not only are human neonates capable of producing different cry melodies, but they prefer to produce those melody patterns that are typical for the ambient language they have heard during their foetal life.” Indeed, she proclaims: “we are the first ones to have provided evidence that language begins with the very first cry melodies!”
According to Wermke, the variations in the observed Franco-German newborns’ melody contours show that they have “not only memorized the main intonation patterns of their respective surrounding language” but are also able to reproduce these patterns “in their own production”. Babies “learn the melody of the language in utero”, although she admits “it’s impossible to rule out that they’re showing the results of very rapid learning since birth.” What, in a hospital setting? Because, let us not forget, the “neonates” were all recorded 2-5 days after birth in large Berlin and Paris hospitals, with all the usual questions of single or shared rooms, family visits, how long the baby slept with the mother (mothers worn out from the birth are often happy to see their babies sleep in nearby baby dormitories out of earshot). And what about ambient hospital sounds? Does the Cochin Hospital in Paris have the same background melodic sound as the Charité Hospital in Berlin? Do German babies in Paris produce different melody contours to those in Berlin?
Alright, let’s say that the babies were in fact learning about sound during the last three months “in utero” – What were they hearing? Wermke provides no information about the mothers, beyond that they were “strictly monolingual”. However, although languages often have a standardised profile, there are variations – the German ‘Hochdeutsch’, often associated with Hannover, has to compete with strong dialectical variations elsewhere, including that heard in Berlin. Did the mother speak dialect or standard German/French? What is her typical melody contour? Does she live in a noisy home (next to the Metro or Boulevard Peripherique in Paris), watch television a lot, listen to music (of whatever melodic preference)?
In Wermke’s report, there is no indication that any of these questions have been asked, let alone answered. Instead, she insists that “these data support the importance of human infants’ crying for seeding language development.” Taken as proven that newborns are “highly motivated to imitate their mother’s behaviour in order to attract her and hence to foster bonding”, her next step is “to compare cries from other language backgrounds”, like Chinese and Japanese.
But will the recordings of these “neonates” finally see any attempt to directly pair “voiced, harmonic cries” of individual babies with the specific “melody contours” of their own mothers, rather than vague references to the “typical” patterns of native languages?