Segmental Structure and Tone Linguistische Arbeiten Edited by Klaus von Heusinger, Gereon Müller, Ingo Plag, Beatrice Primus, Elisabeth Stark and Richard Wiese Volume 552 Segmental Structure and Tone Edited by Wolfgang Kehrein, Björn Köhnlein, Paul Boersma and Marc van Oostendorp ISBN 978-3-11-034109-6 e-ISBN (PDF) 978-3-11-034126-3 e-ISBN (EPUB) 978-3-11-037749-1 ISSN 0344-6727 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2018 Walter de Gruyter GmbH, Berlin/Boston Printing and binding: CPI books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com Contents Björn Köhnlein, Marc van Oostendorp Introduction 1 Michael Becker, Peter Jurgec Interactions of tone and ATR in Slovenian 11 Paul Boersma The history of the Franconian tone contrast 27 Cathryn Donohue Tones and vowels in Fuzhou revisited 99 Laura J. Downing Grounding Nguni depressor effects 109 Wolfgang Kehrein There’s no tone in Cologne: against tone-segment interactions in Franconian 147 Paul Kiparsky Livonian stød 195 Björn Köhnlein Synchronic alternations between monophthongs and diphthongs in Franconian tone accent dialects: a metrical approach 211 Marc van Oostendorp Tone, final devoicing, and assimilation in Moresnet 237 Keywords index 253 Languages index 257 Björn Köhnlein, Marc van Oostendorp Introduction Tonal and segmental phonology are core areas in the study of sound patterns. Roughly, segmental phonology deals with phenomena that affect the pronun- ciation of vowels and consonants. Tonal phonology, on the other hand, inves- tigates the use of fundamental frequency for linguistic purposes, at the lexical level (protoypically tone languages, such as Mandarin) or at the post-lexical level (prototypically intonation languages, such as Dutch or German). Particu- larly after the rise of autosegmental phonology (Leben 1973, Goldsmith 1976), many similarities between tone and segmental structure have been uncovered: for instance, both tones and segmental features spread, and both interact with metrical structure, such as with word stress, in similar ways. In most models of phonology, they therefore receive identical treatment. For instance, both are considered to be composed of features, possibly living in some feature-geometric arrangement on a set of autosegmental tiers. Such a model makes a number of predictions, however, that do not always seem to be sufficiently verified. First, it predicts that tones and segmental features such as [Labial] or [Voiced] have a similar kind of behaviour. Secondly, it suggests that tonal features interact as freely with segmental features as the latter do among each others. These predictions seem to be partly true; but there are also certain problems. This volume seeks to reevaluate the nature of tone–segment interactions in phonology. The contributions address, among other things, the following basic questions: what tone–segment interactions exist, and how can the facts be incor- porated into a phonological account of the language? Is interaction between tones and vowel quality really universally absent? What types of tone–consonant interaction do we find across languages? What is the relation between diachrony and synchrony in relevant processes? The papers in this volume deal with the relation between segmental features and tone from a variety of angles. There is an empirical emphasis on Franconian dialects in the Netherlands, Belgium and Germany, but other languages and lan- guage families are also discussed, such as Fouzhou (Chinese), Livonian (Finnic), Nguni (Bantu), and Slovenian (Slavic). It is shown how some of the predictions of the standard model seem more problematic, and solutions to some of these problems are also mentioned. Björn Köhnlein, The Ohio State University, [email protected] Marc van Oostendorp, Meertens Institute, [email protected] https://doi.org/10.1515/9783110341263-001 2 Björn Köhnlein, Marc van Oostendorp In this introduction, we briefly sketch some of the theoretical background: which of the predictions are verified? Which ones are problematic? 1 Tones and segments: similar behaviour Tones and segments sometimes behave in a similar way. Consider for example the influence of word stress on the distribution of tone and vowel quality. The interaction between vowel quality and stress is a driving factor in phonology. For instance, it has been repeatedly observed that vowels in stressed syllables tend to be more sonorous than those in unstressed syllables. In a nutshell, this means that the lower a (peripheral) vowel is, the likelier it is that this vowel will be stressed (Kenstowicz 1997, 2004; de Lacy 2006, 2007). One example of sonority-driven stress can be found in Takia, a North New Guinea language: in the language, word stress falls on the syllable that contains the most sonorous vowel (Ross 2002, de Lacy 2007):1 (1) Sonority-driven stress in Takia a. [ŋi-ˈsaŋes] ‘hawk’ [ˈŋa-sol] ‘1sg-flee’ b. [kirˈŋen] ‘her/his finger, toe’ [ŋi-ˈemi] ‘your (pl.) legs/feet’ c. [ifuˈno] ‘s/he hit you’ [mulˈmol] ‘a kind of a tree’ (1a) demonstrates that in words containing a low and mid vowel, the low vowel will receive stress, as it is of higher sonority. Likewise, in (1b, c), the mid vowels are preferred over the high vowels. These facts can be expressed in a sonority scale; for purposes of illustration, the sonority scale for tense vowels is provided in (2). The higher a vowel is ranked in the sonority, the likelier it is that this vowel will be a potential stress-bearing unit. (2) Sonority scale for tense vowels in descending order a > e, o > i, u Crucially, interactions between prominence and stress are not restricted to segments but can also be found in the relation between tone and metrical structure. As de Lacy (2002) argues, the interaction between tone and stress is governed by a tonal prominence scale, comparable to that for vowels: high tones (H) and metrically strong positions attract each other, as do low tones (L) 1 If vowels are of equal sonority, the rightmost one will be stressed. Introduction 3 and metrically weak positions; mid tones (M) are located in between H and L. The resulting prominence scale, which closely resembles the sonority scale in (2), is provided in (3). (3) Tonal prominence scale in descending order H > M > L One relevant example comes from Ayutla, a Mixtec language where tonal prom- inence is relevant for determining the position of stress. For our purposes, it serves to state that stress is realized on a syllable carrying the most prominent tone in the word (the full pattern is somewhat more complex, see Pankratz & Pike 1967, de Lacy 2002 for discussion). In disyllabic words, stress is therefore on H, if the word contains a high tone (4a, b, c). In the absence of H, stress will fall on M (4d). (4) Some stress patterns in Ayutla disyllabic words a. ˈHL [ˈʃínìʔ] ‘hat’ b. LˈH [páˈlà] ‘brown sugar’ c. MˈH [[jāˈkwaʔ] ‘it is crooked’ d. ˈML [ˈnāmà] ‘wall’ As these examples demonstrate, tone and vowel quality interact with metrical structure in a similar way. Metrically strong positions correspond to relatively prominent tones/vowels, and metrically weak positions correspond to less prom- inent tones/vowels. 2 Non-interaction between segmental features and tone Given that tonal features and the segmental features responsible for vowel height interact with the stress system in a similar way, one would expect them to also interact with each other. Interestingly, however, there seems to be much less clear-cut evidence for such interactions. In fact, it has even been assumed that direct interactions between tones and vowels are absent in the world’s languages. Consider the following universal that has been proposed in the literature: (5) An alleged universal of tone systems: no interactions of tones and vowel quality Vowels and tones cannot determine each other’s quality (Hombert 1977, 1978; Hombert, Ohala & Ewan 1979). 4 Björn Köhnlein, Marc van Oostendorp At a phonetic level, correlations between pitch and vowel quality have been established: higher vowels tend to be produced with higher pitch than lower vowels, which is referred to as intrinsic pitch height (e.g. Lehiste 1970). Therefore, one might expect that this could potentially result in phonological contrasts; yet it is commonly assumed that this interaction does not lead to cognitively controlled phonological patterns.2 There are two main arguments underlying this assumption. First, the effect of intrinsic pitch height is rather small and may not be universal (Kingston 2007). Furthermore, little evidence has been put forward so far that would indicate a direct relation between tone and vowel quality, which is why many phonologists still hold (5) to be correct (see e.g. de Lacy 2007). Notably, there is more evidence for interactions between consonants and tone, particularly with respect to those between laryngeal features and tone; but still, there seems to be an asymmetry concerning the direction of attested interactions. The basic facts are as follows: cross-linguistically, two types of interactions between voicing and tone are particularly frequent. First, voiceless consonants tend to have a raising effect on the tone of the next vowel whereas voiced consonants tend to have a lowering effect on the vowel to their right. In Suma (Bradshaw 2000), for instance, imperfective verbs start with a High tone (6a), except when they begin with a voiced obstruent, in which case the first tone is rising (6b). Diachronically, the effect is also known from other tone languages, where voicing contrasts on onset consonants turn into tone contrasts on follow- ing vowels (see Yip 2002 for summary and discussion).