Perceiving word prosodic contrasts as a function of sentence prosody in two Dutch Limburgian dialects

Abstract

This paper investigates the perception of word prosodic contrasts as a function of focus and position in the intonational phrase in two Dutch Limburgian dialects, Roermond and Weert. While their word prosodic contrasts share a historical source, the two dialects differ in that Weert realizes the prosodic contrast by duration, while Roermond uses f₀. In addition, the Roermond dialect, but not the Weert dialect, appears to neutralize the prosodic distinction outside the focus constituent in phrase-internal syllables. The stimulus materials were naturally elicited word pairs in which the prosodic contrast marks a difference in grammatical number. In two perception experiments, listeners decided in a forced-choice task whether the words represented a singular or a plural form. Listeners with a Roermond Dutch background recognized the members of the opposition in focused contexts and phrase-final contexts, but failed to do so in phrase-internal, nonfocused contexts. By contrast, listeners whose native language was Weert Dutch perceived the grammatical number distinction in all contexts with comparable measures of success. Second, the presentation of stimuli consisting of words excised from their sentences significantly impaired the recognition of grammatical number in the Roermond group, but not in the Weert group. These results suggest that the perception of the tonal contrast, but not that of the duration contrast, depends on the intonational context. The fact that in the Roermond dialect lexical and intonational tones are integrated in the same phonological grammar thus turns out to have significant consequences for the functionality of the word prosodic contrast which can be shown to be absent when this phonological contrast is encoded differently.

Introduction

Linguistic communication can generally be regarded as a function of the way meanings are encoded in the phonological structure of languages. The perceptual distinctiveness of segments partly depends on a variety of phonological and phonetic factors. For instance, while in many languages the distinction between voiced and voiceless consonants carries a high functional load in nonfinal positions, it is frequently neutralized at the end of a word or syllable. Similarly, the place of articulation of nasal stops may be contrastive in the syllable onset, but not in the coda. Such distributional patterns are generally due to the less favorable conditions obtaining in the neutralizing contexts for the contrasts concerned, and to the inclusion of these limiting conditions as categorical restrictions in the phonology of languages (e.g. Maddieson, 1984; Flemming, 1995; Boersma, 1998). A prosodic factor acting on the vulnerability of contrasts is the variable precision in the articulation as a function of information status. When words represent new or contrastive information, they tend to be produced with greater articulatory care, yielding more prototypical realizations of the segments in them, while the pronunciation of contextually redundant words is often reduced, resulting in shorter words and the undershoot of articulatory targets (Eefting & Nooteboom, 1993; van Bergem, 1995).

There is growing evidence that such observations at the segmental level also apply to prosodic contrasts. Studies of various tone languages report that the realization of a lexical tone may be affected by its prosodic context, such as the tones of neighboring words (Xu, 1997), even to the extent that a given configuration loses its canonical shape as realized in isolated words (Kochanski, Shih, & Jing, 2003). Sentence intonation can have a strong impact on the melodic shape of tones as well. In Hausa, for instance, polar question intonation suspends the contrast between lexical H and HL (Inkelas & Leben, 1991), and many languages change the lexically specified tone at the end of a phrase to distinguish between different utterance types (Ladd, 1996, page 149; Yip, 2000, page 273; Luksaneeyanawin, 1993). In addition, analogous to what happens at the segmental level, melodic configurations of tones may be modified by the expression of focus or for emphasis effects (Gussenhoven, 2004, page 15). For instance, when a word is focused in Shanghai Chinese, the phonetic realization of its lexical tone uses an expanded pitch range, such that an L-tone becomes lower and H-tone higher (Selkirk & Shen, 1990); the same is true for Mandarin (Xu, 1999). Similarly, Longacre (1952) reported that under stress, the two highest tones of Trique (Oto-Manguean) are pronounced higher and the three lower tones lower than in nonstressed positions. As a result, tonal contrasts become more salient in such privileged positions (cf. also Beckman, 1997).

The Dutch Limburgian dialects of Roermond and Weert are of particular interest in this regard. They belong to a group of Limburgian dialects spoken in the south-east of the Netherlands which display a great deal of overlap in their respective vocabularies, as illustrated by words like bein ‘leg’, derm ‘intestine’ and erm ‘arm’. Additionally, they share a word prosodic contrast between what has recently been referred to as Accents 1 and 2 (Gussenhoven & van der Vliet, 1999; Gussenhoven (2000a), Gussenhoven (2000b)). (The traditional Dutch terms are stoottoon ‘push tone’ and sleeptoon ‘drag tone’, respectively.) However, while Roermond uses a pitch distinction, Weert relies largely on a duration distinction (Heijmans, 2003). Roermond is tonal in the real sense of the word in that it marks lexical distinctions by melodic means. The exact realization of these tones is quite complex, and varies with the intonation contour (declarative vs. interrogative intonation), the information status of the word (within the focus or outside it) and the position of the word in the intonational phrase (final or nonfinal). Interestingly, nonfocused, nonfinal contexts have been claimed to neutralize the distinction (Gussenhoven, 2000a). The dialect of Weert, by contrast, may well have been tonal in the past, in which case it has replaced the melodic cues with durational ones (Heijmans, 2003; Verhoeven, 2003). Although it has been suggested that the vowel duration difference is accompanied by an alignment difference of a rising-falling pitch configuration (Verhoeven & Connell, 1992), it would appear to be the vowel duration difference which is phonologically relevant. Focus will have a general effect on articulatory precision (e.g. de Jong, 1995), and thus also on the realization of vowel quantity contrasts. Focus is expressed by means of pitch accents in Dutch, and pitch accented syllables are longer than equivalent unaccented syllables (‘accentual lengthening’, Cambier-Langeveld & Turk, 1999). However, as pointed out by Sieb Nooteboom (personal communication), while quantity contrasts are commonly neutralized in unstressed syllables, as in Dutch (Rietveld, Kerkhoff, & Gussenhoven, 2004), there are no examples of languages that neutralize a quantity contrast outside the focus constituent.

A second possible difference between the two phonological contrasts concerns the extent to which acoustic cues may be nonlocal. The prosodic contrast employed for the lexical distinctions in Roermond is integrated with the intonational structure of the sentence, and it is therefore possible that the cues to the perception of the lexical tone contrast are less reliably confined to the speech signal for the word in question, and may in part depend on the correct interpretation of the intonation contour for the sentence as a whole. On the other hand, a quantity contrast is locally detectable, and is unlikely to have acoustic cues across the boundaries of the word in which it occurs.

Given these observations, it is not unreasonable to assume that the difference between the tonal encoding in Roermond and the quantity encoding in Weert will have significant consequences for the perception of contrasts across intonational conditions. The discriminability of the tonal distinctions in Roermond is likely to vary from one intonational context to another, while the quantity opposition in Weert is likely to be less dependent on intonational context and position in the intonational phrase: in this respect, the quantity contrast is similar to segmental contrasts, which will be similarly robust against such variation in the context. Finally, the discriminability of the prosodic contrast in Weert should not be expected to depend on the interpretation of the intonation contour for the sentence, as it may be in Roermond. As a result, the excision of words from their sentence contexts is arguably more harmful to the perception of the contrast in the dialect Roermond than in the dialect of Weert.

In order to provide a better insight into the salience of the prosodic contrasts in these dialects and achieve an understanding of their communicative function, a cross-dialectal perception study was conducted. Following a methodological procedure for functional, crosslinguistic analyses of prosodic distinctions outlined in Swerts, Krahmer, and Avesani (2002), we aimed to find out to what extent the perception of prosodic contrasts is dependent on whether it is signalled by f₀ (Roermond) or duration (Weert). It was expected that this distinction is perceivable in all prosodic contexts in Weert, whereas in Roermond the distinction may be neutralized in the nonfocused, nonfinal context. In addition, it was expected that the prosodic distinction in Weert remains robust when words are excised from their context, whereas for Roermond the contrast will be more vulnerable when listeners do not have access to the wider intonational contour.

Because we were interested in the perception of the word prosodic contrast in natural utterances, we conducted the perception experiment with the help of spoken data rather than with stimuli with artificially manipulated f₀ and duration specifications. In order to be able to relate the perception results to the stimuli we used to obtain them, we report f₀ and duration measurements for the experimental words for the two dialects separately, for each of the prosodic contexts used. This will be done in Section 3. First, however, Section 2 describes the selection of lexical minimal pairs, the prosodic contexts and methodological procedures. Section 4 presents the results of the perception experiment, while Section 5 offers a conclusion.