1

ABSTRACT

does prosodic word recursion cause phonetic initial strengthening?

The paper is the very first study to address the issue of domain-initial strengthening in recursive prosodic words. Domain-initial strengthening concerns the articulatory strength of segments at the left edges of prosodic domains. The widely-used scale of prosodic positions is the prosodic hierarchy.The current study employs a different scale, i.e., recursive prosodic words in which smaller prosodic words are recursively embedded in a larger word. The length of /s/ is measured when it is placed in the initial positions of 2-, 3- and 4-word Vietnamese noun compounds in order to investigate whether prosodic word recursion causes articulatory strengthening. The findings show that there is no significant duration difference among the domains of recursive prosodic words, which reflects no strengthening effect of prosodic recursion on segmental articulation. Therefore, domain-initial strengthening appears to be sensitive to the categories of prosodic constituents rather than the depth of their embedding.

Key words: domain-initial strengthening, prosodic word recursion

Phuong Hoai Dang

May 2013

1

1

1

1

DOES PROSODIC WORD RECURSION CAUSE PHONETIC INITIAL STRENGTHENING?

by

Phuong Hoai Dang

A thesis

submitted in partial

fulfillment of the requirements for the degree of

Master of Linguistics

in the College of Arts and Humanities

California State University, Fresno

May 2013

1

APPROVED

For the Department of Linguistics:

We, the undersigned, certify that the thesis of the following student meets the required standards of scholarship, format, and style of the university and the student's graduate degree program for the awarding of the master's degree.

Phuong Hoai Dang

Thesis Author

Chris Golston(Chair)Linguistics

Brian AgbayaniLinguistics

Sean FulopLinguistics

For the University Graduate Committee:

Dean, Division of Graduate Studies

1

1

1

AUTHORIZATION FOR REPRODUCTION
OF MASTER’S THESIS

xI grant permission for the reproduction of this thesis in part or in its entirety without further authorization from me, on the condition that the person or agency requesting reproduction absorbs the cost and provides proper acknowledgment of authorship.

Permission to reproduce this thesis in part or in its entirety must be obtained from me.

Signature of thesis author:

1

1

ACKNOWLEDGMENTS

First of all, I would like to express my deepest gratitude to my thesis supervisor Chris Golston and committee members Brian Agbayani and Sean Fulop for their devoted support, invaluable guidance and constructive feedback during the preparation and completion of the thesis.

I would like to send my special thank to Thuong Bui and Duc Dang for their indispensable help in the process of data collection.

I also owe a big debt to the participants in the current thesis. I could not have completed the thesis without their patient and devoted participation.

Last but not least, I am indebted to my family who always give me encouragement and support towards the completion of the thesis.

1

1

Page

TABLE OF CONTENTS

Page

LIST OF TABLES......

LIST OF FIGURES......

CHAPTER 1: INTRODUCTION......

1.1 Domain-initial strengthening......

1.2 Prosodic recursion......

1.3 The current study......

CHAPTER 2: NOUN COMPOUNDS IN VIETNAMESE......

2.1 Description......

2.2 Prosodic organization......

CHAPTER 3: METHODS......

3.1 Prosodic domains......

3.2 Participants......

3.3 Materials......

3.4 Data collection procedures......

3.5 Data analysis procedures......

CHAPTER 4:RESULTS AND DISCUSSION

4.1 Results......

4.2 Discussion......

CHAPTER 5:CONCLUSION

REFERENCES......

APPENDICES

APPENDIX A: VIETNAMESE NOUN COMPOUNDS......

1

1

Page

LIST OF TABLES

Page

Table 1. Speech materials for the study...... 18

Table 2. Variation across compounds and speakers...... 23

1

1

Page

LIST OF FIGURES

Page

Figure 1. Prosodic organization of 2-, 3- and 4-noun compounds...... 15

Figure 2. The waveform and spectrogram of the compound mui sɛ ‘car roof’...20

Figure 3. Mean /s/ duration for all speakers...... 21

1

CHAPTER 1: INTRODUCTION

1.1 Domain-initial strengthening

It has been observed that the phonetic properties of segments vary according to their positions in the prosodic structure of languages thanks to a wide range of research regarding the interaction between prosody and segmental articulation. French consonants and vowels have greater amplitude and duration in stressed syllables than in unstressed syllables. English /s/ seems to have less aspiration noise in the middle of Word than at the beginning of Word and Intonational Phrase. There is an increase in the Voice Onset Time of Korean aspirated consonants from Word-medial positions to Word-initial positions to Accentual Phrase-initial positions (see Fougeron 1999 for a review). The examples above are only a few out of different studies on the effect of prosodic positions on phonetic segments which is called prosodic strengthening. Cho (2005: 3867) defines prosodic strengthening as “temporal and/ or spatial expansion of articulation due to accent and/ or prosodic boundaries” and mentions the three so-called strong prosodic positions are the left edges of prosodic domains, the right edges of prosodic domains and accented syllables.

One line of prosodic strengthening research is domain-initial strengthening which concerns “prosodic strengthening associated with left edges of prosodic domains” (Cho et al. 2007: 211). Trask (1996) defines “strengthening” as “a phonological process in which some segment becomes stronger.” The widely used scale of prosodic positions for domain-initial strengthening is the prosodic hierarchy which is a hierarchically organized structure of prosodic domains such as Utterance, Intonational Phrase, Phonological Phrase, Word, Syllable, etc. (Selkirk 1978; Selkirk 1984; Nespor & Vogel 1986; see Selkirk 1995; Selkirk 2009 for a review). In general, domain-initial strengthening is a phonological process in which segments at the left edges of higher prosodic constituents are articulatorily stronger than those at the left edges of lower prosodic constituents.

The phenomenon of domain-initial strengthening has been investigated in different languages such as English (Fougeron & Keating 1996; Keating et al. 1999), Taiwanese (Hsu & Jun 1998; Hayashi et al. 1999; Keating et al. 1999), Korean (Keating et al. 1999; Cho & Keating 2001), French (Keating et al. 1999; Fougeron 2001), German (Kuzla et al. 2007; Kuzla & Ernestus 2011), and Arabic (Al Taisan 2011). These studies generally explore two research questions. The first concerns how the organization of prosodic constituents affects the articulation of speech segments, and the second examines whether the articulatory variation of segments at the beginning of prosodic domains can help mark the prosodic hierarchy of a language.

The common measurements are linguopalatal contact and segment duration. Some acoustic parameters are also used in the domain-initial strengthening studies; however, the results concerning them and their correlations with linguopalatal contact or segment duration are not consistent to reflect the general picture of domain-initial strengthening. On the one hand, Fougeron and Keating (1996) do not find any strong correlations between linguopalatal contact and acoustic measurements like VOT, vowel duration and stop burst energy for the English nasal /n/; Hsu and Jun (1998) and Hayashi et al. (1999) see that there is no significant effect of prosodic positions on VOT duration of the investigated stops in Taiwanese. On the other hand, Cho and Keating (2001) find that the acoustic measurements like stop closure duration, VOT, Total Voiceless Interval, % voicing during closure, vowel duration, stop burst energy, nasal duration and nasal energy vary according to the prosodic positions of the four tested Korean stops and some of them have correlations with linguopalatal contact; Kuzla and Ernestus (2011) find that /b/, /d/, and /g/ have less glottal vibration and /p/, /t/, /k/ have shorter VOT duration after higher prosodic boundaries in German.

The general finding is that speech segments in higher prosodic domains have more linguopalatal contact and/or longer duration than those in lower domains; and the increase of segmental articulation from the lowest to highest domain is usually cumulative. The found pattern of domain-initial strengthening, on the one hand, reflects the effect of prosodic organization on phonetic articulation, on the other hand, provides articulatory and/or acoustic cues to distinguish prosodic boundaries. Note that the findingsof domain-initial strengthening show certain variation across languages, segments, speakers, and prosodic domains. For instance, Cho and Keating (2001:185) observe that Korean has a clearer and more consistent picture of domain-initial strengthening than English, French, and Taiwanese.Regarding the variation across segments, Fougeron (2001:119) sees that “/s/ is less systematically affected by prosodic position compared to the other consonants studied like stops and /l/.” Furthermore, it is observed that not all speakers distinguish all prosodic domains. Fougeron and Keating (1996) report two speakers make distinction among three levels while the other speaker distinguishes only two. In the study which compares domain-initial strengthening in four languages, Keating et al. (1999) also discuss that all speakers make at least two domains distinct, which is robust in the findings.

1.2 Prosodic recursion

In the discussion of prosodic recursion, it is crucial to mention the Prosodic Hierarchy Theory proposed by Selkirk (1978) and developed by Selkirk (1984) and Nespor and Vogel (1986) (see Selkirk 1995; Selkirk 2009 for a review). The theory claims that a string of speech in a language can be exhaustively parsed into different prosodic constituents in which higher domains contain lower ones.

(1) The Prosodic Hierarchy (Selkirk 1995: 5)

UttUtterance

IPIntonational Phrase

PhPPhonological Phrase

PWdProsodic Word

FtFoot

σSyllable

There is a set of constraints on prosodic domination which characterizes the features of the prosodic hierarchy. Specifically, the two constraints of Layeredness and Headedness claim that speech in every language is hierarchically organized in prosodic constituents; higher-ranked constituents dominate lower-ranked constituents. In a stricter sense, the constraints of Exhaustivity and Non-recursivity require that every higher domain must dominate or must be completely parsed into immediate lower domains, which means that there is no level skipping or repetition at any prosodic domain. The first two constraints are inviolable and undominated thanks to the observation that every language has several prosodic levels. On the other hand, it has been argued that Exhaustivity and Non-recursivity are violable and low-ranked because certain prosodic levels are found to be skipped or recurred in a number of languages (Selkirk 1995; Ito & Mester 2009). In sum, the violability and low ranking of Exhaustivity and Non-recursivity result in prosodic skipping and recursion.

Prosodic recursion refers to the repetition of prosodic domains at a certain level of the prosodic structure. Specifically, prosodic recursion involves the embedding of a prosodic domain of a certain level in another prosodic domain of the same level; the larger domains contain the smaller ones (Inkelas 1990; Selkirk 1995; Ito & Mester 2008; Ito & Mester 2009; Kabak & Revithiadou 2009; Selkirk 2009; Féry 2010). Furthermore, it is observed that compound structures are frequently exemplified as prosodic recursion thanks to the notion that the compound structures as well as their components belong to the same category and the compounds are larger constituents containing their smaller components (Ladd 1990; Ladd 1996; Inkelas 1990; Green 2007; Kabak & Revithiadou 2009).

1.3 The current study

As mentioned above, domain-initial strengthening, a type of prosodic strengthening concerning the articulatory strength of segments at the left edges of prosodic constituents has been explored in a wide range of research in which the currently-employed scale of prosodic positions is the prosodic hierarchy of languages. It is interesting to revisit this phenomenon with a different prosodic scale. Particularly, the current study investigates the articulation of phonetic segments in the initial positions of prosodic domains of the same level called recursive prosodic words. The study is the first to examine the interaction between prosodic recursion and segmental articulation.

More specifically, the paper addresses the issue of domain-initial strengthening in recursive prosodic words. Prosodic recursion involves the containment of smaller constituents inside a larger constituent of the same category; therefore, the formation of recursive prosodic words concerns the embedding of smaller prosodic words in a larger prosodic word. Compounds in languages are argued to be recursive prosodic words (Inkelas 1990; Green 2007; Kabak & Revithiadou 2009).

The studied case is Vietnamese noun compounds. Vietnamese is a tonal monosyllabic language in which syllables are argued to be the smallest units of phonological and morphological analysis (Ngo 1984; Nguyen 2011). Particularly, Ngo (1984) proposes the notion of “syllabeme” is the minimal grammatical unit in Vietnamese; it can function as a syllable, a morpheme and a word. Nguyen (2011) argues that a word in Vietnamese is a minimal meaningful unit whose spoken form is a syllable and written form is a separate group of letters. A compound in Vietnamese can be comprised of two or more elements; each is a one-syllable word; and the compounding of these monosyllabic elements creates a new lexical item. Therefore, Vietnamese compounds are appropriate candidates for the notion of recursive prosodic words thanks to the fact that the whole compounds are larger prosodic words which are constituted by smaller prosodic words.

In order to investigate the articulatory variation across initial positions of recursive prosodic words, the duration of the voiceless alveolar fricative /s/ is measured when it is at the beginning of each monosyllabic word of 2-, 3-, and 4- word noun compounds. The two possibilities are hypothesized towards the findings of the study. First, there is a cumulative increase of segmental length from the smallest to the largest constituent-initial positions, which reflects the significant role of the embedding of smaller prosodic words in larger prosodic words. Second, there is no strong distinction among the domains of recursive prosodic words, which implies the effect of prosodic categories on domain-initial strengthening.

Generally speaking, the study aims at exploring the phenomenon of domain-initial strengthening with the scale of recursive prosodic words. /s/ is placed in the initial positions of 2-, 3-, and 4-word compounds and its duration is measured to examine whether prosodic word recursion causes articulatory strengthening.

1

CHAPTER 2: NOUN COMPOUNDS IN VIETNAMESE

2.1 Description[1]

Compounding is the word formation process which involves the combination of existing words to build new lexical items. Vietnamese compounds are formed under such the process; two or more monosyllabic words are combined to generate a single category.There are three word types of compounds in Vietnamese: compound nouns, verbs and adjectives; each of them is classified into two subtypes including coordinate and subordinate compounds.

Coordinate compounds involve the compounding of two or more words in which “each constituent is a center” and “occurs in juxtaposition” (Nguyen 1997:66). Semantically, they are called generalizing compounds because the meanings of the two centers are combined to form a more general lexical item. These centers belong to the same category or are synonyms or antonyms.

(1)a. bàn ɣé

table chair

‘furniture’

b. muə bán

buy sell

‘buy and sell’

c. cɑ̉i cuót

brush polish

‘be meticulous’

Subordinate compounds concern the combination of the words with the head-complement order. The heads are more general lexical items or concepts and the complements modify and narrow the meanings of the heads. Subordinate compounds refer to more specific lexical items and then are called specializing compounds.

(2) a. nɯɤ́k dɑ́

water ice

‘ice’

b. làm ruọŋ

do rice field

‘do farming’

c. sɛ lɯ̉ə

vehicle fire

‘train’

More specifically,coordinate noun compounds involve the combination of nouns and their meanings reflect the generic category of the constituent nouns. The examples in (3), (4) and (5) exemplify the 2-word, 3-word and 4-word noun coordinate compounds respectively. Note that 3-word coordinate compounds are not as popular as the other two.

(3) a.cim muoŋ

bird beast

‘animals’

b. rau kɔ̉

vegetable grass

‘veggies’

c. ruo͎ŋ nɯɤŋ

wet field dry field

‘cultivated fields’

d. ruòi muõi

fly mosquito

‘flies/ bugs’

e. kwʌ̀n Ɂɑ́u

pantcoat

‘clothes’

(4) a. vɯɤ̀n Ɂɑu cuòŋ

garden pond shed

‘traditional Vietnamese farm’

b. Ɂaɲ cị Ɂɛm

brother sister younger sibling

‘brothers and sisters’

c. ræŋ hàm mæ͎t

tooth jaw face

‘the medical study of teeth, jaws and face’

d. tai mũi hɔ͎ŋ

ear nose throat

‘the medical study of ears, noses and throats’

(5) a. bɑ̀ kɔn ko bák

grandmother child aunt uncle

‘relatives’

b. sɤn hɑ̀ sɑ̃ tǽk

mountainriver village principle

‘country’

c. soŋ núi nɯɤ́k nɔn

river mountain water mountain

‘country’

d. zɯɤ̀ŋ tủ bàn ɣé

bed wardrobe table chair

‘furniture’

e. maj lan kúk ʈúk

apricot orchid chrysanthemum bamboo

‘set of four symbolic flowers and plants’

Subordinate noun compounds involve the combination of heads and complements in which the complements follow and modify the heads. Regarding 2-word subordinate compounds, Nguyen (1997) observes that there are the three combinations; the heads are always nouns whereas the complements can be nouns, verbs and adjectives. The examples in (6) exemplify the 2-word subordinate compounds which involve the three types of noun compounding such as a noun and a noun (6a-b), a noun and a verb (6c-d) as well as a noun and an adjective (6e-f).

(13) a. bɔ̀ kɔn

cow/ox child

‘calf’

b. kɤm ɣɑ̀

rice chicken

‘chicken rice’

c.sɛ dạp

vehicle to pedal

‘bike’

d. kɤm nǽm

rice to wisp

‘rice ball’

e. bɔ̀ dɯ͎k

cow/ox male

‘ox’

f. bɔ̀ kái

cow/ox female

‘cow’

The examples in (7) and (8) are 3-word and 4-word compounds in which the heads are left-edged.

(7) a. dɯɤ̀ŋ cʌn ʈɤ̀i

line foot sky

‘horizon’

b. báɲ sɛ bɔ̀

wheel vehicle cow/ox

‘cow/ox cart wheel’

c. bụi than dɑ́

dust coal stone

‘coal dust’

d. vụn báɲ mì

crust cake wheat

‘bread crust’

e. kỏ Ɂɑ́u vét

neck blouse suit

‘suit collar’

1

1

(8) a. vɔ̉ thʌn kʌi dɑ

bark body tree banyan

‘banyan tree bark’

b. duoi kɑ́ nɯɤ́k mæ͎n

tail fish water salty

‘sea water fish tail’

c. thʌn kʌi kɑ̀ cuə

body tree eggplantsour

‘tomato tree body’

d. vɔ̉ báɲ sɛ bɔ̀

tirewheel vehicle cow/ox

‘cow/ox cart wheel tire’

e. kiẻu kỏ Ɂɑ́u vét

model collar blouse suit

‘suit collar model’

Another type of 4-word subordinate compounds is also found to be popular in Vietnamese; they involve the combination of two 2-word subordinate compounds, as seen in (9).

(9) a. bén sɛ mièndoŋ

stationcarregioneast

‘Eastern station’

b. bɯ́k tɯɤ̀ŋ Ɂʌm thaɲ