• Keine Ergebnisse gefunden

Phonological Representations of Standard German

3.3 Summary and Implications for Acquisition

Although German and Swiss German are closely related languages, differ-ences are found in the phonetics and phonology of the respective stop sys-tems. German uses a two-way laryngeal contrast that distinguishes voiceless unaspirated stops from voiceless aspirated ones. As was shown in chapter 2, the primary phonetic attribute that marks the contrast in production and in perception is an ACT distinction. Swiss German also distinguishes two categories of stops but it does not use a laryngeal contrast. Instead, sin-gleton stops contrast with geminates in length, with CD being the essential phonetic correlate for both production and perception.

Basic differences between the German and the Swiss German stop sys-tems are not only found in the respective phonetic implementations of the tense/lax opposition in the two languages, but also extend to phonological representations. As became obvious in the preceding sections, it is not only a question of different featural specifications. Rather, the German and the Swiss German stop contrasts are assumed to be represented on different phonological levels. The German ACT contrast is represented by means of a laryngeal feature on a segmental level, whereas the Swiss German CD con-trast is represented prosodically on a suprasegmental level. For the German laryngeal contrast it was shown that the debate on which feature specifica-tion is most appropriate to represent the tense/lax opposispecifica-tion is far from settled. The Swiss German contrast between singletons and geminates is represented by association lines to timing units within the syllable. Sin-gletons occupy one timing slot, geminates are linked to two timing units, thus directly influencing syllable structure. A comparative depiction of a

3.3 Summary and Implications for Acquisition 59 potential phonological representation of the German laryngeal contrast and of the Swiss German length contrast is given in figure 3.8.

Figure 3.8: Phonological representation of German and Swiss German stops.

/b, p, p:/stand for root nodes. Note that the laryngeal nodes without featural marking are redundant and should be omitted completely (see Lombardi 1991, 1995a,b). They are added here to emphasise the representational differences.

Some remarks are due considering the implications that the two different types of stop contrasts pose for language acquisition and in particular for the present research. As will become evident in more detail in chapter 4, early language acquisition is determined by a developmental change from a universal phonetic perception to a language-specific phonological/phonemic perception. Thus, although early phonological representations are not fo-cussed on in the experiments presented in the following chapters, phonology comes in in terms of the respective phoneme inventories of the languages under investigation.

With regard to phonetics, chapter 2 showed that both the German and the Swiss German stop contrast are grounded on temporal parameters, namely the duration of a voicing lag (ACT) in the case of German and the duration of the closure phase (CD) in the case of Swiss German. From an acoustic point of view, the laryngeal contrast as it is realised in aspira-tion languages is fairly salient (e.g., Stevens & Klatt 1974). Therefore, it is not surprising that young infants succeed in discriminating short lag/long lag stop contrasts (e.g., Eimas et al. 1971; see chapter 4 for details). The laryngeal contrast in thevoice languages, where (the presence and absence of) closure voicing is the primary phonetic correlate to mark the opposition, is less salient than the aspiration contrast. Consequently, it is also more difficult to distinguish for young infants (e.g., Eimaset al.1971; Laskyet al.

1975). Given that the Swiss German length contrast is not one of different noise (i.e., aspiration) intervals but of different intervals of silence, it is less salient than the German phoneme distinction and, thus, is assumed to be harder to discriminate for very young children. The perception of both ACT and CD contrasts by infants in their first year of life will be discussed at length in chapter 4.

In order to understand and differentiate the words of a language, children need not only discriminate phonetic speech contrasts but they must also be-come aware of which contrasts are used to convey meaning in the language they are about to acquire. In other words, they have to become aware of the phoneme inventory and of the phonology of their mother tongue. In-fants begin to comprehend their first words already in the second half of the first year of life (see Jusczyk 1997:89) and they utter their first own produc-tions around the age of 12 months. Nonetheless, a fine-grained phonological system seems to emerge only somewhat later in the second year of life. Per-ceptual studies suggest that the ability to distinguish minimal pairs is only about to evolve at the age of 14 months. While familiar words are dis-tinguished successfully, 14-month-olds exhibit difficulties in distinguishing unfamiliar words that differ in one phoneme only (e.g., Stager & Werker 1997; Fennell & Werker 2003). Only at an age of 17 months and older, more reliable distinction is observed in word-learning tasks – as opposed to the less demanding word-comprehension tasks (e.g., Werkeret al. 2002).

A crucial factor in the acquisition of a phoneme inventory and of an abstract representation of contrasts seems to be perceptual salience. Fikkert (2010; also cf. Altvater-Mackensen & Fikkert 2010) argues that perceptually salient elements (e.g., vowels and onset consonants) are the first elements chosen for production and also the first elements to be featurally specified (see also chapter 4). The Dutch laryngeal contrast between prevoiced and voiceless unaspirated stops is less salient than the English short lag/long lag aspiration contrast. Supporting the assumption that more salient contrasts are acquired before less salient contrasts, Kager et al. (2007) show that, in production, the laryngeal contrast in Dutch is acquired later (completed beyond age 2;6) than the English laryngeal contrast (completed by the age of 2;0; cf. Kager et al. 2007 and references there).With regard to speech perception, van der Feest (2007) demonstrated that PoA contrasts in Dutch seem to be specified earlier in infants’ phonological systems than the less salient Dutch laryngeal contrast.

In analogy to these findings, the phonetic bases of the German and the Swiss German phoneme contrasts suggest that the German laryngeal con-trast will be easier to acquire phonologically than the Swiss German length

3.3 Summary and Implications for Acquisition 61 contrast. As said before, the different representational systems (segmental level/features vs. suprasegmental level/timing units) provide an interesting research topic per se, which cannot be dealt with in detail in the present work. The experiments presented in chapters 6, 7 and 9 are designed to assess infants’ speech sound discrimination skills only, without testing their phonological knowledge required for word recognition and word learning.

Nonetheless, the role of an abstract phonological system in the process of first language acquisition will be addressed in chapter 4. A possible explana-tion for how infants could come to find abstract labels or representaexplana-tions for the phonetic categories deducible from the acoustic signal will be discussed in connection with statistical learning and an exemplar-theoretic approach to phonological acquisition (also cf. chapter 7). Yet, before turning to the acquisition of phonology, chapter 4 will start with a detailed account of infants’ perception skills in the first year of life.

Chapter 4

Acquiring Phoneme