Morphologically Governed Accent in Optimality Theory

John D. Alderete

• Publisher: Taylor & Francis
• Year: 2013
• eBook ISBN: 9781135727093
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

Alderete examines the influences of morphological factors on stress and pitch accent within Optimality Theory.

CHAPTER 1

Introduction

1.1 Morphologically Governed Accent in Optimality Theory

This dissertation is about accent systems in which word structure has an important role in the characterization of accentual processes. These processes affect phonological categories like stress and tone, but importantly, they cannot be described with reference to sound structure alone. Morphological factors like the accentual properties of roots and certain diacritically marked affixes must also be recognized. As many accent systems studied here have a strict limitation of one accent per word, one major accentual process examined in this thesis is Accent Resolution, the deletion of accent in words with more than one inherently accented morpheme. This pattern of deletion often shows a preference for retention of accent in the root, which underscores one important function of morphological structure. Other morpho-accentual phenomena examined here include morphologically triggered de-accentuation (or ‘dominance effects’), accent insertion (often known as pre- and post-accentuation), and certain accentual shifts. The occurrence of these processes is, in many cases, directly tied to affixation, and so they too are inherently morphological. The focus here is almost exclusively on word accent, as the accentual processes under examination are mostly word-level phenomena, but some parallels with other levels of structure are made throughout this thesis.

1.1.1 Goals

This thesis has two basic goals. The first goal is more or less a descriptive one and involves arguing for the existence of two distinct types of morpho-accentual processes (1). The first type is an analogue to a well-known type of vowel harmony system where the features of root vowels have the effect of ‘overriding’ the featural specifications in prefixes and suffixes. In many accent systems, the presence of an accent in the root likewise overrides accent in affixes, hence the term ‘root-controlled accent’. The second type is fundamentally different from the first. Generally linked to affixation, ‘affixcontrolled’ accentual processes require a change in the accentuation of the base, which is usually the root or the stem.

(1) a. Root-Controlled Accent (RCA): inherent accent in the root precludes the realization of accent elsewhere in the word.

b. Affix-Controlled Accent (ACA): the attachment of an affix correlates with a mutation of the accent in the base of affixation.

The second goal, related to the first, is to argue for a specific theoretical model that accounts for these two types of morpho-accentual processes in an explanatory way.

To my knowledge, the first characterization of an accent system in terms of root-control is given in Hill & Hill 1968, which describes stress-accent in the Uto-Aztecan language Cupeño. In this language, both roots and affixes have an accented/unaccented contrast, but when an inherently accented root combines with an accented affix, the root accent ‘wins’, as exemplified below with some characteristic examples (the roots are underlined).

(2) Root-Controlled Accent in Cupeño (Hill & Hill 1968)

a.	/pé + yax/	→	pé-yax	‘He says’
	/pé + yax + qál/	→	pe-ya-qál	‘He was saying’
b.	/pé + ʔávu + qál	→	pe-ʔávu-qal	‘S/he was wanting’
	/pé + pulín + qal/	→	pe-pulín-qal	‘She gave birth’

Part of the descriptive aim of this thesis is to extend this idea to other, better-known accent systems like Russian and Japanese. In the analyses of these systems, root-controlled accent has a very similar role, causing the deletion of an accent outside of the root. The parallels between these systems run deeper than this, however, which can be seen by examining a second pattern of RCA.

RCA has an equally important role in blocking the application of other accentual processes. For example, Cupeño has a set of suffixes which place an accent on the root-final syllable (3a), but this process is blocked in words with accented roots because such an insertion would preclude the realization of a root accent (3b).

(3) Blocking Effect of RCA in Cupeno

a.	/wena + nukpre/	→	wená -nuk	‘having put in’
	/pé + ma + čipre/	→	pe-má-či	‘with my hand(s)’
b.	/méme + ykepre/	→	méme-yke	‘to the ocean’
	/tiviiʔe + maapre + le/	→	tívi?-me-l	‘small round basket’

Likewise, in Russian and Tokyo Japanese, certain patterns of pre-accentuation are only found in words with unaccented roots, an observation which extends to other morpho-accentual processes. In sum, RCA has the effect of privileging roots in the realization of inherent accent, both in the concatenation of more than one inherently accented morpheme and in the application of morpho-accentual processes.

In the second type of morpho-accentual process, affix-controlled accent, affixes are in charge accentually. However, this process is not simply the symmetric counterpart to RCA; it shows a different kind of behavior altogether. The main characteristic of ACA that sets it apart from RCA is that ACA demands a change in the accentuation of the base to which the affix is attached. Thus, affix-controlled processes run counter to the underlying force behind RCA because they demand a change in the prosody of the base, which typically contains the root. Three affix-controlled processes examined in detail here are illustrated below with some examples from Tokyo Japanese.

(4) Affix-Controlled Accent in Tokyo Japanese (Poser 1984, McCawley 1968)

a. Dominance effects require a deletion of base prosody

/edo + kko/	→	edo-kko	‘native of Tokyo’
/kóobe + kko/	→	koobe-kko	‘native of Kobe’

b. Pre-accentuation requires an insertion of accent into the base

/yosida + ke/	→	yosidá-ke	‘the Yoshida family’
/nisímura + ke/	→	nisimurá-ke	‘the Nishimura family’

c. Accent shifts require a shift of base prosody

/kúzu + ya/	→	kuzú-ya	‘junkman’
/toma + ya/	→	toma-ya	‘mat seller’

A common type of affix-controlled process, exemplified in (4a) with the suffix -kko, is deletion of the prosody of the base to which the affix is attached. It is clear why this type of deletion, sometimes called a ‘dominance effect’, is antagonistic to RCA: RCA strives to preserve the accent of the root, while -kko demands deletion of the root accent. Another type of affix-controlled process is pre-accentuation (or post-accentuation for prefixes), which causes insertion of an accent somewhere in the base, as shown in (4b). Pre-accentuation may also run counter to the imperative to realize root accent in this system because, as is typical, there is one accent per word in Japanese; the insertion of an accent thus entails the deletion of the base accent, as in nisimurá-ke. A final type of ACA involves accent shift or ‘flop’, exemplified in (4c) with the suffix -ya. While not in direct conflict with RCA, accent shifts of this kind resemble the other types of ACA in that they mutate the accent of the base. The accentual change demanded by -ya is a shift of the prosody of the base, as observed in words with accented stems like kuzu-ya. To summarize, affix-controlled accent is an obligatory change in the accentuation of the base of affixation; because of this change of base accent, ACA may be in conflict with root-controlled accent.

The identification of these two morpho-accentual processes raises the following two questions for the theory of morphologically governed accent.

1. How are the differences between RCA and ACA to be described and explained?

2. How is the conflict between RCA and ACA to be resolved?

The observation that RCA and ACA are fundamentally different poses the formal question of how to distinguish them as separate classes of morpho-accentual phenomena. Furthermore, the inherent difference between RCA and ACA leads to situations of conflict where RCA demands preservation of a root accent, while ACA requires a change of the root accent, even the deletion of it. Moreover, the negotiation of this conflict is a subtle matter. In some contexts RCA wins, as found in pre-accentuation in Cupeño. In the case of Japanese, however, ACA wins, as the pre-accenting suffix -ke induces an insertion of accent across the board, even in words with accented roots. How then is the conflict between these morpho-accentual phenomena to be modelled in a way that accounts for the observed cross-linguistic differences?

There is also a larger theoretical issue to be addressed in the analysis of these morpho-accentual processes. Both RCA and ACA have clear parallels in other morpho-phonological alternations and these parallels require explanation. As alluded to above, RCA is like root-controlled vowel harmony in that the phonological patterns observed in words are determined by the roots contained in these words. Affix-controlled accentual processes also have well-established parallels with non-accentual processes. Pre-accentuation, for example, can be compared to the length alternations induced by certain affixes, like pre-lengthening suffixes found in many languages (e.g., Yidiñ, Dixon 1977): both pre-accenting and pre-lengthening suffixes trigger a change of the phonological make-up of their bases. These considerations lead to a third important question, which has both descriptive and theoretical implications.

3. What accounts for the parallels to RCA and ACA found in other areas of phonology?

If the parallels pointed to above are valid, then it would be a significant liability if the ultimate theory of RCA and ACA did not make these connections. A major goal of this thesis, therefore, is to account for the properties of ACA and RCA with principles that are generally available in linguistic theory. The correct theory, by this desideratum, should account for the range of accentual processes with theoretical constructs that are not specific to a theory of accent, but rather, sufficiently general to extend to other morphologically governed phonological patterns.

1.1.2 Synopsis of Theoretical Arguments

In this dissertation, I argue that RCA and ACA, despite important differences, are united under the rubric of Faithfulness in Optimality Theory (Prince & Smolensky 1991, 1993, McCarthy & Prince 1993b). In particular, RCA is a consequence of the privileged faithfulness status for roots, as has been shown in other empirical domains, such as vowel harmony and dissimilation (see especially Beckman 1997 [1998]). The patterns of root retention in Cupeño and other languages are a consequence of root-specific Faithfulness constraints, a result that relates RCA to other root-controlled phenomena.

Affix-controlled accent, on the other hand, is modelled as an obligatory violation of Faithfulness, compelled by a new type of constraint, Anti-Faithfulness. Anti-Faithfulness constraints evaluate a pair of morphologically related words and trigger an alternation by forcing a violation of Faithfulness. Consistent with the desideratum 3 given above, the second half of this dissertation argues that Anti-Faithfuln...

Table of contents

• Cover
• Title Page
• Copyright Page
• Preface
• Acknowledgements
• Abstract
• Table of Contents
• 1. Introduction
• 2. Root-Controlled Accent in Cupeño
• 3. Restricted Edge Effects in Root-Controlled Accent Systems
• 4. Transderivational Faithfulness and Anti-Faithfulness
• 5. The Role of Anti-Faithfulness in Morpho-Accentual Phenomena
• 6. Conclusion
• Bibliography
• Index