From early on, infants are sensitive to and even prefer to listen to their native language. Using the High-Amplitude Sucking (HAS) paradigm, a habituation paradigm to test (speech) perception and discrimination abilities in newborns and young infants, Mehler and colleagues (Mehler et al., 1988) tested 4-day-old French infants on their ability to discriminate their native language (syllable-timed) from a language with different rhythmical properties, namely Russian (stress-timed). In a series of experiments they provided evidence that the infants were able to discriminate both languages, that is, the infants showed a dishabituation reaction in correlation to the change of language, even when the stimuli were low-pass filtered. In low-pass filtered speech, only prosodic information like rhythm and intonation is preserved, while most of the phonetic information is removed.
Furthermore, Mehler et al. (1988) found that the dishabituation reaction was stronger when the stimulus changed from Russian to French. Therefore, they concluded that the infants not only discriminated both languages but also preferred their native language over the non-native one. This preference for the native language, though it could not be replicated by the research group around Mehler themselves (cf., Mehler et al., 1996), was also found in a study by Moon and colleagues (Moon, Cooper & Fifer, 1993) in which they tested the preferences of Spanish and English 2-day-olds using a different methodology designed to test specific preferences. In their study Spanish and English newborns sucking behaviour was contingent upon hearing stimuli from either the native or the non-native language. Moon et al. (1993) found that the infants preferred to listen to their native language over the foreign language.
These studies were supplemented by findings from Bosch and Sebastián-Gallés (1997), who tested 4-month-old infants on their discrimination of two languages coming from the same rhythmic class, namely Spanish and Catalan, both syllable-timed languages (cf., Ramus, Dupoux & Mehler, 2003). They found that the infants, who were raised either in monolingual Spanish or in monolingual Catalan families, were able to distinguish between these two languages, again even when the stimuli were presented low-pass
filtered. This result shows that 4-month-old infants are able to discriminate between their native and a second language, even when both languages are from the same rhythmical class. In a subsequent study, Bosch and Sebastián-Gallés (2001) also provided evidence that 4-month-old Spanish-Catalan bilinguals were able to discriminate between these two languages.
Thus, the studies on the early language perception of infants provide evidence of a sensitivity to and preference for the native language from birth on. Especially, the data from Bosch and Sebastián-Gallés (1997, 2001) seem to suggest that the ability to discriminate two languages found by Mehler et al. (1988) and Moon et al. (1993) was not initiated by the differences in the rhythmical class of these languages as such, but rather by the familiarity with the native language.
Discriminating two non-native languages
However, infants were also found to be able to discriminate between two non-native languages. In their first study, Mehler et al. (1988) also tested the ability of infants to discriminate between two non-native languages belonging to different rhythmical classes, for the French group between English (stress-timed) vs. Italian (syllable-timed) and for the American group French (syllable-timed) vs. Russian (stress-timed). Both groups did not seem to be able to discriminate between two non-native languages. This finding was later subject to reanalysis (Mehler & Christophe, 1995; Mehler et al., 1996). Looking at a dishabituation reaction across test conditions, that is, irrespective of which language was presented during the habituation phase, they found that the French 4-day-olds, but not the American 2-month-olds, showed a dishabituation reaction after the language change, indicating that they were able to discriminate languages from two different rhythmical classes. They assumed that a developmental change in infants’ sensitivities was responsible for that pattern, in such a way that older infants already have more specific knowledge about their native language and treat both unknown languages as not fitting that model, and therefore fail to show discrimination between these languages (Mehler et al., 1996, p. 105f.)
Nazzi and colleagues (Nazzi, Bertonici & Mehler, 1998) also tested newborns on their abilities to discriminate between two languages belonging to the same rhythmical class, like English vs. Dutch. The French newborns tested with the HAS paradigm did not seem to be able to discriminate between English and Dutch, whereas they succeeded in discriminating English from Japanese (mora-timed). Furthermore, the French newborns were able to discriminate languages based on the rhythmic class, that is, they discriminated a test set consisting of English and Dutch utterances from another test set consisting of Spanish and Italian utterances. As all the stimuli presented in the study were low-pass filtered, Nazzi et al. (1998) concluded that newborns are sensitive to the differences in the rhythmic classes of languages (R-Hypothesis; ibid., p. 757), which then might help them to learn about the organisation of the rhythmic class their language belongs to.
A second study by Nazzi and colleagues (Nazzi, Jusczyk & Johnson, 2000a) tested the ability of older infants to discriminate languages across or within rhythmic classes. Using the HTP paradigm (for a detailed description of the paradigm see chapter 3), they found that 5-month-old American infants still were able to discriminate between languages from two different rhythmic classes, even when both were non-native (Italian vs. Japanese).13 Furthermore, the 5-month-olds were not able to discriminate between languages that belong to the same rhythmic class when both languages were non-native (Italian vs. Spanish; Dutch vs. German). Only in the condition in which one of the languages from within the same class was the native language (British English vs. Dutch; and even American English vs. British English) discrimination could be seen. Nazzi et al. (2000a) concluded from the results of the series of experiments that infants did have a sensitivity to the rhythmic class their native language belongs to as opposed to other rhythmic classes. Furthermore, the familiarity with the native language seemed to be an important factor for the discrimination of languages belonging to the same rhythmic class, because the infants were able to discriminate between English and Dutch but not between Dutch and German (native-language acquisition hypothesis; ibid., p. 3f). However, Nazzi et al. (2000a) used natural speech and not low-pass filtered speech.
13 This result partly contradicts the findings by Mehler et al. (1996), who found no discrimination of two
non-native languages by 2-month-old infants. Nazzi et al. (2000a) assume that these different findings are due to the different experimental paradigms used in the two studies.
Therefore, the languages presented differed in more dimensions than just the differences or similarities in rhythm.
In a recent study, Ramus (2002) tried to investigate more closely the influence of rhythm and intonation on infants’ discrimination of different languages. He tested newborns on their discrimination of Dutch and Japanese sentence stimuli, in a condition where the segmental information was manipulated14 and also an identical intonation contour was superimposed on all sentences, leaving alone differences in the rhythm between the two “languages”. In this experiment the newborns showed tentative discrimination between the two languages. As the effects found in this experiment were weaker than the effects found in studies using low-pass filtered speech, in which both rhythm and intonation were preserved, Ramus (2002) suggested that this might indicate that the integration of rhythmic and intonational cues might enhance infants’ language discrimination abilities. The conclusion that not only rhythm but also intonation might play a role in infants’ language discrimination could also be drawn from results of a study from Dehaene-Lambertz and Houston (1998). They found discrimination between English and French in 2-month-old English- or French-learning infants even when presenting rather short utterances (1.2 s), but not when the intonational organisation of the utterances was eliminated by scrambling the words within the utterances.
Taken together, the studies on the perception of rhythmic properties in speech provide evidence for an initial preference for the native language and also for infants’ abilities to discriminate two non-native languages from different rhythmic classes. At the age of 4- to 5-months, the familiarity with prosodic characteristics of the native language seems to have evolved rather precise, because in addition to the ability to discriminate the native language from a different language from the same rhythmic class (Bosch & Sebastián-Gallés, 1997), infants were also able to discriminate between the American and British dialect of their native language (Nazzi et al., 2000a).
This sensitivity to the rhythmic, and also the intonational characteristics of the native language is a prerequisite to notice changes in the rhythmic and intonational pattern
14 All consonants were replaced by a single consonant according to the manner of articulation. All vowels
that may indicate prosodic boundaries in the speech. As described in chapter 2.1 of the present study the boundaries of phrases and clauses are marked by changes in duration, pitch, and the occurrence of pauses. Furthermore, the sensitivity to the rhythmic units in a language, for instance, interstress intervals in stress-timed languages, may allow more sophisticated analysis of the language, as has been proposed by the bootstrapping
hypotheses. These hypotheses claim that infants are able to use knowledge of one level of
the language to infer about other levels of the language. In relation to the topic of the present study, an early sensitivity to the rhythmic properties of the native language might allow infants to establish the relations of larger prosodic units and pauses which then, once established, could be used to infer about the syntactic structure of the language (see Figure 2.1).