Loading...

The Phonetics and Phonology of Glottalization in Italian

by Jessica Di Napoli (Author)
Thesis 290 Pages

Summary

This book combines theoretical and experimental approaches to provide a comprehensive account of glottalization in word-final syllables in Italian. The speech production study at the heart of the book sheds light on the source of glottalization, the contextual factors determining its occurrence, and the acoustic correlates which characterize its production.
Acoustic analysis of words presenting evidence of glottalization is carried out through visual inspection of the acoustic signal together with spectral analysis of voice quality. Statistical analysis of the data in the study is performed using mixed effects models, as well as tree-based methods including conditional inference trees and random forests. Results of the study have implications for cross-linguistic studies on voice quality.

Table Of Contents

  • Cover
  • Title
  • Copyright
  • About the author(s)/editor(s)
  • About the book
  • This eBook can be cited
  • List of Figures
  • List of Tables
  • Table of Contents
  • 1 Introduction
  • 1.1 Glottalization in Italian
  • 1.1.1 A glottal stop coda for Italian?
  • 1.1.2 Evidence of glottalization as prosodic boundary marking in Italian
  • 1.1.3 Research questions
  • 1.2 Overview of the present study
  • I Theoretical background
  • 2 Glottalization in Italian
  • 2.1 Glottalization and syllable well-formedness
  • 2.1.1 Word-final stressed vowels
  • 2.1.2 Syllable structure
  • 2.1.3 An empty prosodic position for final stressed syllables?
  • 2.1.4 Vayra (1994) – A glottal stop coda for Italian?
  • 2.2 Challenges to Vayra (1994)
  • 2.2.1 Raddoppiamento as compensatory lengthening (a historic mora for final stressed syllables)
  • 2.2.2 All stressed syllables are heavy
  • 2.2.3 Glottalization – not just creak
  • 2.3 Phrase-final glottalization
  • 2.3.1 Phrase boundaries and glottalization
  • 2.3.2 On the nature of phrase-final glottalization
  • 2.3.3 Final lengthening and phrase-final glottalization
  • 2.4 Glottalization and hiatus
  • 2.5 Summary
  • 3 Voice quality – Classification and quantification
  • 3.1 The production of voice quality
  • 3.1.1 Voicelessness and glottal closure
  • 3.1.2 Modal voice
  • 3.1.3 Non-modal phonation
  • 3.1.4 Summary
  • 3.2 Linguistic uses of voice quality
  • 3.3 Investigating voice quality
  • 3.3.1 Characteristics of voice quality in the voice source
  • 3.3.2 Characteristics of voice quality in the speech signal
  • 3.3.3 Qualitative signal-based approaches to classifying glottalization
  • 3.4 Conclusion
  • II Laboratory speech study
  • 4 Glottalization in Central Italian – A glottal stop coda or prosodic edge marking?
  • 4.1 Introduction
  • 4.1.1 Word-level glottalization
  • 4.1.2 Phrase-level glottalization
  • 4.1.3 The present study
  • 4.2 Methods
  • 4.2.1 Speech materials
  • 4.2.2 Speakers
  • 4.2.3 Recordings
  • 4.2.4 Acoustic analysis
  • 4.2.5 Statistical analysis
  • 4.3 Results
  • 4.3.1 Intended and perceived boundaries
  • 4.3.2 Glottalization
  • 4.3.3 Raddoppiamento sintattico
  • 4.4 Summary and discussion
  • 4.4.1 Glottalization as a phrase boundary marker in Central Italian
  • 4.4.2 Glottalization as a marker of juncture
  • 4.5 Conclusion
  • 5 Glottalization in Northern Italian
  • 5.1 Introduction
  • 5.1.1 The present study
  • 5.2 Methods
  • 5.2.1 Speech materials
  • 5.2.2 Speakers
  • 5.2.3 Recordings
  • 5.2.4 Acoustic analysis
  • 5.2.5 Statistical analysis
  • 5.3 Results
  • 5.3.1 Intended and perceived boundaries
  • 5.3.2 Glottalization
  • 5.3.3 Raddoppiamento sintattico
  • 5.4 Summary and discussion
  • 5.4.1 Glottalization as a marker of phrase boundaries in Northern Italian
  • 5.4.2 Phrase edges and prominence
  • 5.5 Conclusion
  • 6 Regional and individual variation in glottalization
  • 6.1 Introduction
  • 6.1.1 An overview of tree-based methods for classification
  • 6.1.2 The present study
  • 6.2 Methods
  • 6.2.1 Data set
  • 6.2.2 Statistical analysis
  • 6.3 Results
  • 6.3.1 Conditional inference tree
  • 6.3.2 Random forest
  • 6.4 Discussion
  • 6.4.1 Predicting glottalization in Italian – boundary strength first
  • 6.4.2 Individual and regional influences on glottalization
  • 6.5 Conclusion
  • III Acoustic characteristics of glottalization in Italian
  • 7 Acoustic correlates of glottalization – A qualitative approach
  • 7.1 Introduction
  • 7.1.1 Investigating glottalization
  • 7.1.2 Characteristics of glottalization in Italian
  • 7.1.3 The present study
  • 7.2 Methods
  • 7.2.1 Speech materials
  • 7.2.2 Annotation and analysis
  • 7.3 Results
  • 7.3.1 Rates of glottalization and breathiness
  • 7.3.2 Acoustic correlates of glottalization
  • 7.3.3 Glottal stop distribution and duration
  • 7.3.4 Duration of glottalization
  • 7.4 Summary and discussion
  • 7.4.1 Preferred acoustic correlates of glottalization in Italian
  • 7.4.2 Speaker variation in the realization of glottalization
  • 7.4.3 Glottal stops
  • 7.4.4 The domain of phrase-final glottalization
  • 7.5 Conclusion
  • 8 Acoustic correlates of glottalization – A quantitative approach
  • 8.1 Introduction
  • 8.1.1 Quantifying glottalization
  • 8.1.2 Acoustic measures of glottalization in Italian
  • 8.1.3 The present study
  • 8.2 Methods
  • 8.2.1 Speech materials
  • 8.2.2 Acoustic analysis
  • 8.2.3 Statistical analysis
  • 8.3 Results
  • 8.3.1 H1*-H2*
  • 8.3.2 H1*-A3*
  • 8.3.3 CPP
  • 8.3.4 HNR
  • 8.4 Summary and discussion
  • 8.4.1 Acoustic measures of phrase-final glottalization in Italian
  • 8.4.2 Effects of vowel quality and gender
  • 8.4.3 Acoustic correlates of stress
  • 8.5 Conclusion
  • 9 Conclusion
  • 9.1 Summary of results and general discussion
  • 9.1.1 Glottal stop coda or boundary marking?
  • 9.1.2 Regional variation?
  • 9.1.3 What is the nature of glottalization?
  • 9.2 Conclusion
  • A Stimuli
  • B Raw data counts – Production study
  • C Statistical models and pairwise comparisons – Acoustic measures (Chapter 8)
  • C.1 H1*-H2*
  • C.2 H1*-A3*
  • C.3 CPP
  • C.4 HNR
  • References
  • Index
  • Series index

| 13 →

List of Figures

2.1 Moraic representations of (a) cita ‘cite’, (b) città ‘city’ and (c) cinta ‘walls’, illustrating the bimoraic status of each stressed syllable (σ*). Bimoraicity is achieved through (a) vowel lengthening, (b) raddoppiamento and (c) a moraic coda

2.2 Moraic representation of città ‘city’ in isolation, with empty final mora. Adapted with permission from Di Napoli (2015, p. 127)

2.3 Historical development of raddoppiamento triggers ĂD > a ‘to’ and TŪ¯) > tu ‘you’, adapted from Repetti (1991, p. 311)

2.4 Moraic representation of città ‘city’ in isolation, featuring glottal stop insertion. Reprinted with permission from Di Napoli (2015, p. 127)

2.5 Moraic representation of città triste ‘sad city’ with raddoppiamento. Reprinted with permission from Di Napoli (2015, p. 127)

3.1 Schematic view of the glottis from above. The arytenoid cartilages are positioned at the rear. Figure from Hayward (2000, p. 222)

3.2 Ladefoged’s continuum of phonation types, from Gordon and Ladefoged (2001, p. 384)

3.3 Parameters of muscular tension in the larynx. From Gobl and Ní Chasaide (2010, p. 395), based on Laver (1980, p. 109). Reprinted with permission

3.4 Schematic representations of the glottal configurations in (a) breath (voiceless) phonation and (b) whisper phonation. Reprinted with permission from Laver (1994, p. 191)

3.5 Schematic drawing of a flow glottogram for modal voice, obtained by inverse filtering. Peak flow rate is marked as Umax. From Hayward (2000, p. 229)

3.6 Waveforms of glottal volume velocity for one cycle of vocal fold vibration (top row) and source spectra of glottal pulses (bottom row) for (a) modal voice, (b) breathy voice and (c) creaky voice. Arrows indicate the ‘corner’ of the glottal pulse (see text), which influences the relative amount of spectral ← 13 | 14 → energy at high frequencies. Reprinted with permission from Stevens (1977, p. 273)

3.7 Landmarks in the spectrum for measuring spectral tilt, including the first two harmonics (H1 and H2) and the harmonics closest to F1, F2 and F3, denoted A1, A2 and A3, respectively, as they refer to amplitudes

3.8 Cepstrum for a modal vowel showing prominent peaks characteristic of a periodic signal. Linear regression line shown is used in the calculation of Cepstral Peak Prominence (CPP). Figure reprinted with permission from Hillenbrand, Cleveland, and Erickson (1994, p. 772).

4.1 Moraic representation of città ‘city’ in isolation, featuring glottal stop insertion. Reprinted with permission from Di Napoli (2015, p. 127).

4.2 Moraic representation of città triste ‘sad city’ with raddoppiamento. Reprinted with permission from Di Napoli (2015, p. 127)

4.3 Phrase-final target syllable [ro] in Dico la parola faro a Gianni (Speaker ED). Both the speech signal and the EGG waveform show evidence of glottalization. This token was labeled as glottalized. Reprinted with permission from Di Napoli (2015, p. 134)

4.4 Word-final target syllable [ro] in Il faro illuminò tutti gli scogli (Speaker ED), showing no evidence of irregularity in the speech or EGG waveforms. This token was labeled as not glottalized. Reprinted with permission from Di Napoli (2015, p. 135)

4.5 Central Italian – Results for boundary strength (significant at p<0.001) comparing the rate of glottalization of tokens at an intended word versus phrase boundary. Lighter shading at the top of a bar, where present, indicates breathy tokens. Lexical stress and segmental context are pooled

4.6 Central Italian – Rate of glottalization plotted as a function of rate of boundary mismatch for two levels of boundary strength (word and phrase)

4.7 Central Italian – Results for lexical stress (significant at p<0.001) comparing the rate of glottalization of tokens for unstressed versus stressed final vowels. Lighter shading at ← 14 | 15 → the top of a bar, where present, indicates breathy tokens. Boundary strength and segmental context are pooled

4.8 Central Italian – Results for segmental context (significant at p<0.001) comparing the rate of glottalization of tokens where the word-final vowel is followed by a consonant to those followed by a vowel. Lighter shading at the top of a bar, where present, indicates breathy tokens. Boundary strength and lexical stress are pooled

4.9 Central Italian – Results by speaker for mean consonant duration (and standard deviation) following an unstressed (RSNP) as compared to a stressed (RSP) final vowel

4.10 Central Italian – Mean consonant duration (and standard deviation) across five speakers following an unstressed (RSNP) as compared to a stressed (RSP) final vowel at two boundary strengths (word and phrase)

4.11 Central Italian – Results across five speakers for mean consonant duration (and standard deviation) following an unstressed (RSNP) as compared to a stressed (RSP) final vowel at an intended phrase boundary for two phonation types (modal and glottalized)

4.12 Central Italian – Results for all seven speakers according to test condition: following boundary (Phrase/Word), following segment (Vowel/Consonant) and lexical stress (Stressed/Unstressed). Lighter shading at the top of a bar, where present, indicates breathy tokens. Average rate of glottalization across conditions is 38 percent (see dashed line)

5.1 Languages and dialects of Italy. The La Spezia-Rimini line is shown as a dashed line cutting across Italy just south of these two cities. From Harris and Vincent (1988, p. 483). Reprinted by permission of Oxford University Press

5.2 Northern Italian – Results for boundary strength (significant at p<0.001) comparing the rate of glottalization of tokens at an intended word versus phrase boundary. Lighter shading at the top of a bar, where present, indicates breathy tokens. Lexical stress and segmental context are pooled

5.3 Northern and Central Italian – Rate of glottalization plotted as a function of rate of boundary mismatch for two levels of boundary strength (word and phrase) ← 15 | 16 →

5.4 Northern Italian – Results for lexical stress comparing the rate of glottalization of tokens for unstressed versus stressed final vowels. Lighter shading at the top of a bar, where present, indicates breathy tokens. Boundary strength and segmental context are pooled.

5.5 Northern Italian – Results for lexical stress and boundary strength comparing the rate of glottalization of tokens for unstressed versus stressed final vowels at a word versus a phrase boundary. The interaction between lexical stress and boundary strength is significant at p<0.

5.6 Northern Italian – Results for segmental context (significant at p<0.001) comparing the rate of glottalization of tokens where the word-final vowel is followed by a consonant to those followed by a vowel. Lighter shading at the top of a bar, where present, indicates breathy tokens. Boundary strength and lexical stress are pooled

5.7 Northern Italian – Results by speaker for consonant duration (and standard deviation) following an unstressed (RSNP) as compared to a stressed (RSP) final vowel

5.8 Northern Italian – Results for all seven speakers according to test condition: following boundary (Phrase/Word), following segment (Vowel/Consonant) and lexical stress (Stressed/Unstressed). Lighter shading at the top of a bar, where present, indicates breathy tokens. Average rate of glottalization across conditions is 48 percent (see dashed line)

6.1 Classification tree predicting intention to smoke: (a) tree representation in the form of a conditional inference tree; (b) plot of the feature space illustrating the binary recursive partitioning used to build the tree. Dark shaded portions represent “yes” responses. From Strobl, Malley, and Tutz (2009b, p. 326). Copyright 2009 by the American Psychological Association. Reprinted with permission

6.2 Relative frequencies of “yes” (dark shading) and “no” (light shading) responses for the smoking data, illustrating the increase in node purity at each split in the tree. From Strobl, Malley, and Tutz (2009b, p. 327). Copyright 2009 by the American Psychological Association. Reprinted with permission ← 16 | 17 →

6.3 Variable importance measures for the smoking data. From Strobl, Malley, and Tutz (2009b, p. 336). Copyright 2009 by the American Psychological Association. Reprinted with permission

6.4 Classification tree (conditional inference) predicting glottalization. Each node provides the p-value determined for the split. Dark shading in the terminal nodes indicates proportion of “yes” responses. Total number of tokens in each terminal node given by n

6.5 Variable importance measures for seven variables predicting glottalization in Central and Northern Italian

7.1 Examples of six categories of acoustic irregularity in phrase-final vowels in the corpus: (a) breathiness (speaker EB, faro), (b) aperiodicity (speaker SC, meta), (c) creak (speaker ED, metà di), (d) diplophonia (speaker EB, bucò), (e) glottal squeak (speaker AL, farò), (f) glottal stop (speaker ED, farò a)

7.2 Rates of phrase-final non-modal phonation by speaker, for both Central and Northern Italian speakers. Dark shading represents glottalized tokens; light shading represents breathy tokens

Details

Pages
290
ISBN (PDF)
9783631765777
ISBN (ePUB)
9783631765784
ISBN (MOBI)
9783631765791
ISBN (Hardcover)
9783631765760
Language
English
Publication date
2019 (February)
Tags
Voice quality Prosody Syllable structure Lexical stress Prosodic boundaries Acoustic analysis
Published
Berlin, Bern, Bruxelles, New York, Oxford, Warszawa, Wien, 2018. 288 pp., 54 fig. b/w, 45 tables

Biographical notes

Jessica Di Napoli (Author)

Jessica Di Napoli teaches and researches at RWTH Aachen University. She holds a PhD in Phonetics from the University of Cologne and an MA in Romance linguistics from the University of Texas at Austin. Her research interests include linguistic uses of voice quality and prosodic boundary marking.

Previous

Title: The Phonetics and Phonology of Glottalization in Italian