blank space

UCLA Working Papers in Linguistics, no.4

Papers in Phonology 4

edited by Adam Albright and Taehong Cho

Table of Contents

Katherine Crosswhite The analysis of extreme vowel reduction 1–12
Alhaji Maina Gimba Downdrift in Bole 13–30
Henk Harkema Dutch schwa is a regular vowel 31–58
Shannon Madsen Stress patterns in Spanish blends 59–86
Mira Oh Suffix faithfulness in Korean 87–120
Jie Zhang Non–contrastive features and categorical patterning in Chinese suffixation Max[f] or Ident[f]? 121–178
Adam Albright and Bruce Hayes Distributional encroachment and its consequences for morphological learning 179–190

Pricing and ordering info

Back to UCLA WPL publication list


Katherine Crosswhite – The analysis of extreme vowel reduction

In this paper, I examine systems of vowel reduction that employ two "levels" of reduction: moderate and extreme. Phonetically, the occurrence of extreme reduction seems to be linked to (non-phonemic) duration, and typically causes reduction in vowel sonority. Moderate reduction, on the other hand, can cause increases as well as decreases in vowel sonority. Based on this type of observation, I hypothesize that two-pattern vowel reduction systems utilize two distinct forms of reduction. That is, extreme reduction and moderate reduction are distinct phenomena, not "stronger" and "weaker" versions of a single process.

Up to table of contents

Alhaji Maina Gimba – Downdrift in Bole

Bole, a language of the west-Chadic family spoken in Northern Nigeria, has a downdrift phenomenon where the actual pitches of words descend such that a high following a low is lower than the preceding high. It has been shown across languages that a domain of downdrift (or downstep) is smaller than a whole sentence (e.g. Silverstein 1976, Pierrehumbert and Beckman 1988, Leben 1989). In this paper, I attempt to characterize the domain of downdrift in Bole based on pitch tracks of utterances which vary in the syntactic structure, the length of phrase, focus, and speech rate. The results show that the downdrift tends to break before a Prepositional Phrase or a heavy noun phrase, though not always. Results show that the downdrift is sensitive to the phonological length, focus, and speech rate; downdrift can be blocked in a sentence of the same syntactic structure when the constituent is long or when uttered at fast rate. The domain of downdrift is often longer than a Phonological Phrase and smaller than an Intonation Phrase (=IP). Thus, I call this domain 'an Intermediate Phrase (=ip).' However, interestingly, the data do not support the Strict Layer Hypothesis. That is, the domain of downdrift is sometimes smaller than the domain of a process of Low Tone Raising (LTR) (Gimba 1998), which applies within a Phonological Phrase. This suggests that the domains of these two rules are constrained by different components of a grammar. The domain of LTR is highly sensitive to the morphosyntactic information while the domain of downdrift (i.e. ip) or declination (i.e. IP) is sensitive to the semantic, syntactic and pragmatic information as well as performance driven conditions such as speech rate.

Up to table of contents

Henk Harkema – Dutch schwa is a regular vowel

The goal of this paper is to show that schwa in Dutch is a regular vowel. First, it will be established that there is no need to treat schwa, the shortest Dutch vowel, as a long vowel, contra Trommelen (1984). Secondly, it will be demonstrated that there is no need to posit two levels of syllabification to capture the properties of schwa, contra Kager (1990).

Up to table of contents

Shannon Madsen – Stress patterns in Spanish blends

Blends in Spanish reveal important and interesting properties of correspondence. Blend formation is a nonconcatenative process, in that to express a blend word (BW) certain parts of each source form (SF) are merged rather than concatenated one after the other, as in compound words (Pineros 1999). Blends reveal three meanings: the meaning of each of the two source forms, plus the meaning of the blend word. The three meanings that are revealed correspond to the three morphological words present in the output structure. Since there are three morphological words in a blend word but only one prosodic word that licenses them, the source forms need to arrange themselves so as to maximize their usage of the left and right edges of the one prosodic word that is available to them. This is why in blend words the two source forms either begin or end at the same point. That is to say, in blend words, SF1 and SF2 are either both anchored at the left edge of the blend word they compose, or the right edge. The analysis presented here concentrates on stress patterns in the formation of blends. I introduce some constraints not before discussed in the literature. One, which I call PRESERVE STRESS, deals with the fact that Spanish blends strive to maintain the whole stress pattern of one of the source forms, that is, it is preferred that the blend word preserve the number of syllables and the same sequence of stressed and unstressed syllables as one of the source forms. Another, *CV REPRESENTATIVE, prevents the source forms from contributing too little to the blend word. Furthermore, the constraints IDENT (Stressed syllable)(SF-BW) and MAX (Stressed syllable)(SF-BW) ensure that the CV patterns of syllables carrying stress in each of the source forms necessarily surface at some point in the blend word. This means that if CiVj is a stressed syllable in a SF, and CaVb is a stressed syllable in SF2, both CiVj and CaVb need to surface in the blend word. With the appropriate ranking we can explain how Spanish blends pattern, including aspects that until now were left unexplained.

Up to table of contents

Mira Oh – Suffix faithfulness in Korean

It has been argued that languages make more effort to preserve elements of lexical than nonlexical items (McCarthy and Prince 1995, Casali 1997). This paper, however, contends that suffixes should be recognized as an independent morphological category to be faithful and that suffix faithfulness outranks root faithfulness in Korean verbal morphology. The fact that suffix faithfulness and root faithfulness are separately rankable from other faithfulness and markedness constraints also accounts for l-final verbs and nominal suffixation. Furthermore, suffix faithfulness will be shown to play a role in derivation and elision in hiatus context as well.

Up to table of contents

Jie Zhang – Non-contrastive features and categorical patterning in Chinese suffixation Max[f] or Ident[f]?

This paper investigates the interaction between the diminutive suffix // and the stem coda nasal n or N across Chinese dialects. The most common data pattern is: /CVn+r/->[CVr], /CVN+r/->[CV~r]. Two theoretical claims are made in its analysis. First, non-contrastive phonetic differences can induce categorical phonological patterning, as I show experimentally that the vowel in the CVN stem is more nasalised than the vowel in the CVn stem, and argue that it is this difference that leads to the difference in phonological patterning between CVn and CVN upon /r/-suffixation. Second, MAX[F] constraints are needed in phonological analyses, since for the data in question, they make fewer assumptions on categorization and better prediction in their factorial typology than IDENT[F] constraints.

Up to table of contents

Adam Albright And Bruce Hayes – Distributional encroachment and its consequences for morphological learning

We describe a common but neglected pattern of linguistic exceptions, which involve "distributional encroachment." This occurs when the distribution of allomorphs is determined by phonological context, but a few exceptional forms take the "wrong" allomorph. For learning algorithms, this can complicate the task of identifying distributions. We present an algorithm for learning allomorph distributions, then show how it can be modified to handle distributional encroachment.

Up to table of contents