| old_uid | 235 |
|---|
| title | The Segmental Anchoring Hypothesis Revisited |
|---|
| start_date | 2005/11/24 |
|---|
| schedule | 13h30 |
|---|
| online | no |
|---|
| summary | In recent years there has been more and more evidence that LH points in the tonal space behave as phonological targets. Indeed, H peaks of both nuclear and prenuclear pitch accents are produced with an amazing degree of stability in tonal scaling and alignment across languages (see Bruce 1977, Arvaniti et al. 1998, 2000; Ladd 1986; Liberman and Pierrehumbert 1984; Prieto et al. 1995, 1996; Silverman and Pierrehumbert 1990; Xu 1999, 2002, among others). Previous investigations on the alignment behavior of prenuclear accents in Spanish and in other languages have observed that while L values are consistently 'anchored' with the onset of the accented syllable, H positions are more variable. These studies have generally emphasized prosodic factors, and, specifically, the right-hand prosodic context on the location of accent peaks (upcoming accent or boundary tone, tonal crowding, vicinity to a prosodic boundary, etc.) -see Silverman & Pierrehumbert, 1990 for English; Prieto, van Santen & Hirschberg, 1995. However, a more recent line of work suggests that when such prosodic effects are excluded alignment of F0 targets are consistently governed by segmental anchoring and that strict alignment effects are pervasive under changes of syllabic/segmental structure and speech rate (Arvaniti & Ladd 1995 for Greek, Ladd et al. 1999 for English and Xu 1998 for Chinese; see also Schepman et al. submitted, for Dutch, Atterer & Ladd, to appear, for German). The idea behind all this evidence is that both the beginning and the end of a rising pitch accent are anchored to specific points in the segmental structure, regardless of segmental or syllable structure composition.
In the first part of the talk, I will present 2 experiments which have the aim of testing the so-called 'anchoring hypothesis' for Castilian Spanish and examine whether LH points behave as segmental anchors in prenuclear position. Experiment 1 focuses on segmental and syllabic composition, while Experiment 2 examined peak behavior under changes in speech rate. For Experiment 1, 4 speakers of Castilian Spanish read three times a corpus of 96 sentences exhibiting the phonological variety needed to test our hypothesis (open vs closed; complex vs. simple onset, as well as different consonant and vowel types), for a total of 1152 repetitions (96 utterances x 4 speakers x 3 repetitions). For experiment 2, the same speakers were told to read twice 32 target syllables in proparoxytone words, first at a normal speech rate, then at a fast rate and finally at a slow rate, for a total of 768 utterances (32 utterances x 3 speech rates x 4 speakers x 2 repetitions). Results of this study reveal that the strict anchoring hypothesis cannot be maintained, as Hs do not strictly align with a given segmental position (for example, the end of the vowel). Crucially, syllable type and speech rate has a significant and consistent effect of F0 peak location.
The second part of the talk will present a production and a pilot perception experiment with prenuclear accents in Central Catalan. In order to test the the presence of word-edge tones in Catalan and to investigate the effects of within-word position on F0 peak location we designed a corpus of potentially ambiguous utterances which are only distinguished by word boundary location (e.g., Catalan Mirà batalles 'he watched battles' vs Mirava talles 'he used to watch carvings'; Spanish Compraré mostazas 'I will buy pots of mustard' vs. Compraremos tazas 'We will buy cups'). Notice that the utterances have the same exact segmental and accentual composition. For experiment 1, 4 speakers read a total of 40 ambiguous utterances four times (160 sentences per speaker, a total of 480 utterances per language). For experiment 2, 20 listeners heard 40 ambiguous utterances extracted from experiment 1.
The results of the production experiment reveal that prenuclear H tones are not consistently aligned to the end of the word. Yet, a clear allphonic difference was found between the two utterance types, namely, a significant effect of within-word position. As in other languages, as syllables get closer to the word-boundary H peaks get more retracted. The results again do not support the strict anchoring hypothesis.
In sum, the segmental anchoring hypothesis cannot be maintained for Catalan and Spanish, as clear effects of syllable structure (coda type, presence of a coda) and word position are found on H location in these languages. I finish the talk claiming that H placement is conditioned not only by tonal pressure, but also by prosodic domain adjustments (ie, syllables and prosodic words). Finally, H placement might also be constrained by perceptual mechanisms (ie, the coda effect found in a variety of languages). |
|---|
| responsibles | Lécuyer |
|---|
| |