A typology of the prosodic expression of focus

(1)

27. November 2019

Post-focal compression in Hindi –

Pitch register as a phonological category?

Frank Kügler

Goethe-University of Frankfurt

Marburg - Linguistisches Kolloquium 1

(2)

27. November 2019

Overview

Background

 A typology of the expression of prosodic focus

 Hindi intonation

Production study

 Characteristic repeated rising contours

 Downstep and post-focal compression (PFC)

Perception study

 Presence and absence of PFC in focus identification

Discussion

 Focus perception, speaker variation, and processing of intonational categories

Conclusion

Pitch register is a phonological category like pitch accents and boundary tones

(3)

27. November 2019

A typology of the prosodic expression of focus

(Kügler & Calhoun to appear)

In many languages prosody plays a key role in encoding focus.

A typology of prosodic focus-marking strategies (Kügler & Calhoun, to appear)

 Stress-based cues

 Phrase-based cues

 Register-based cues

Prosodic cues enhance the prominence of a focused word (e.g. Selkirk 1995; Ladd 2008; Calhoun 2010) and/or align the focus with a prosodic boundary (Féry 2013).

(4)

27. November 2019

A typology of the prosodic expression of focus

Stress-based cues:

 Enhancing the prosodic cues of a focused word by means of - higher fundamental frequency (f0),

- greater f0 movement, - lengthening,

- increased intensity and higher spectral tilt on the focused word

- as well as a drop in f0 after it (cf. Ladd 2008, Fletcher 2010, Turk 2012).

Figure 1. Focus position affects prosodic realisation (from Kügler & Calhoun, to app.).

(5)

27. November 2019

A typology of the prosodic expression of focus

Phrase-based cues

 Languages predominantly do not have lexical prominence (stress).

 Intonation units are phrase tones / phrase boundaries.

 Focus induces additional phrase boundaries or deletes them post-focally.

(1) a. VP focus: What did he do?

(Anaményá nyumbá ndí mwáála)φ he hit the house with a rock

‘He hit the house with a rock.’

b. Object focus: What did he hit with the rock?

(Anaményá nyuúmba)φ (ndí mwáála)φ

c. Verb focus: What did he do to the house with the rock?

(Anaméenya)φ (nyuúmba)φ (ndí mwáála)φ

(Chichewa, adapted from Kanerva 1990: 156)

(6)

27. November 2019

A typology of the prosodic expression of focus

Register-based cues

 Mandarin (Xu 1999; 2011):

- Register expansion under focus (high H’s and lower L’s) - Post-focus compression

 Lexical tones are not affected, no pitch accents (stress), no phrase boundaries.

 Hindi may belong to this group of languages.

(7)

27. November 2019

Background – Hindi intonation

Every prosodic phrase is associated with a rising pitch gesture, except the last one (Moore 1965, Harnsberger & Judge 1994, Patil et al. 2008).

Typical areal feature of South Asian languages (Khan 2016)

(Harnsberger 1999)

billi: mami: ke pitSe gayi:

cat Mommy PP behind went

‘The cat went behind Mommy.’

(8)

27. November 2019

Intonation and Prominence in Hindi

Wide focus / All-new sentence:

(10) A: kyaa huaa?

‘What happened?’

B: [graahak ne davaaii ko khariidaa]_F

‘The customer bought the medicine.’

Narrow focus on the second constituent (object):

(11) A: graahak ne kyaa khariidaa?

‘What did the customer buy?’

B: graahak ne [davaaii]_F ko khariidaa

‘The customer bought the medicine.’

 No prominence marking by means of intonation.

(Patil, Kentner, Gollrad, Kügler, Féry, Vasishth 2008)

(9)

27. November 2019

Intonation and Prominence in Hindi

Wide focus / All-new sentence:

(10) A: kyaa huaa?

‘What happened?’

B: [graahak ne davaaii ko khariidaa]_F

‘The customer bought the medicine.’

Narrow focus on the first constituent (subject):

(12) A: kisa ne davaaii ko khariidaa?

‘Who bought the medicine?’

B: [graahak ne]_F davaaii ko khariidaa

‘The customer bought the medicine.’

 Pitch register compression on the 2^nd constituent (object).

(Patil, Kentner, Gollrad, Kügler, Féry, Vasishth 2008)

(10)

27. November 2019

Results – Hindi intonation (Patil et al. 2008)

10

post-focal compression

graahak ne davaaii ko khariidaa

customer ^ERG medicine ^ACC buy.^PST

“The customer bought the medicine.”

Marburg - Linguistisches Kolloquium

(11)

27. November 2019

Contrastive focus in Hindi (Genzel & Kügler 2010)

Contrast on adjective

• Adjective from one to three syllables

1. Q: Kyaa Romina ne purane gaane ko khojaa? ‘Did Romina search for an old song?’

A: Nahi! Romina ne ek naye gaane ko khojaa. ‘No! Romina searched for a new song.’

no Romina ERG DET new song ACC search.PST

2. Q: Kyaa Mohinder ne miithe biskit ko khyaa? ‘Did Mohinder eat a sweet biscuit?’

A: Nahi! Mohinder ne ek nankim bikit ko khayaa. ‘Mohinder ate a salty biscuit.’

3. Q: Kyaa Naveena ne khurdare lungii ko bechaa? ‘Did Naveena sell a raspy lunghi?’

A: Nahi! Naveena ne ek mulayam lungii ko bechaa. ‘Naveena sold a silky lunghi.’

(12)

27. November 2019

Contrastive focus in Hindi (Genzel & Kügler 2010)

• Significant effect on duration:

– Each syllable of a target word is lengthened compared to wide focus renditions

• Significant effect on F0

– The L tone gets significantly lower under contrastive focus (speaker variation!) – The H tone gets significantly higher

– Hence, pitch span is significantly enhanced under contrastive focus

Similar findings for Urdu (assumed to be structurally similar to Hindi (Jabeen

& Braun 2018)

- Longer duration of first and second syllable of words in contrastive focus - Larger F0-range on words in

corrective focus

- Early alignment of H tone in contrastive focus

(13)

27. November 2019

Hindi intonation

Every prosodic phrase is associated with a rising pitch gesture, except the last (Moore 1965, Harnsberger & Judge 1994, Patil et al. 2008).

Pitch rises within an intonation phrase are downstepped to each other (Patil et al. 2008).

Focus is realized by means of post-focal pitch register compression (PFC) (Moore 1965, Harnsberger & Judge 1994, Patil et al. 2008; but see

Genzel & Kügler 2010 for pitch span expansion in case of contrast).

Typologically, Hindi is classified as a “phrase language” (Féry 2010, 2017) – no pitch accents, but only boundary tones

– rising pitch gestures represent L and H φ–phrase tones

– presumably, intonational function is to demarcate prosodic word boundaries or φ–phrases

Hindi uses register-cues to encode focus (Kügler & Calhoun, to appear)

(14)

27. November 2019

Research question

Some thoughts on typological classification:

Intonation languages use pitch to highlight information (pitch accent, e.g. German: Féry & Kügler 2008, English: Breen at al. 2010)

Hindi – as a phrase language – shows no clear prosodic effect on the focused constituent (Patil et al. 2008)

However, Hindi does show a post–focal prosodic effect, i.e.

compressed pitch register or post–focal compression (PFC) Similar to e.g. Mandarin (Xu 1999)

Question arises whether this post–focal cue serves as a perceptual cue for focus interpretation?

If so, could pitch register be interpreted as a phonological category like pitch accents and boundary tones are in intonation languages?

 A production study and a perception study

(15)

27. November 2019

A production study

Do Hindi speakers realize prosodic cues of focus in contrastive ellipsis?

(1) a. raahul ne mãã ko [davaaii]_Fdii naa ki [gaaDii]_F Raahul ^ERG mother ^DAT medicine give ^NEG that car

“Raahul gave the medicine to the mother and not the car.”

b. raahul ne [mãã ko]_F davaaii dii naa ki [naanii ko]_F Raahul ERG mother DAT medicine give NEG that granny DAT

“Raahul gave the medicine to the mother and not to granny.”

Do speakers differentiate sentences like (1) prosodically in order to resolve the ambiguity?

Predictions:

Focus marking is realized by means of PFC (cf. Patil et al.2008).

(1a/b) contain a contrastive focus (object/indirect object).

PFC is expected to be present after the focused object/indirect object.

(16)

27. November 2019

A production study – Method

Speaker & recordings

30 female native Hindi speakers, age 20-25 Recordings on DAT at Delhi university

Speech materials

Two conditions: (i) indirect object contrast (ii) direct object contrast Six sentence pairs as (1)

Interspersed with filler (material from Patil et al. 2008) Data pre-processing

Annotation and F0 analysis in Praat of 30 x 6 x 2 = 360 sentences Per constituent: F0-minimum before F0-maximum, duration

(17)

27. November 2019

A production study – Example

(18)

27. November 2019

A production study – Results

• Prosodic differentiation of (1):

Contrast on indirect object shows PFC on the following object Contrast on direct object shows no PFC on following verb

• Speaker variation: lesser & higher degree of PFC

(19)

27. November 2019

A production study – Results (F0)

Q1a – Difference on Indirect Object

 lmer(log(IOF0max)~cond+(1+cond|speaker)+(1+cond|item),data)

Q1b – Difference on Direct object, Indirect object is in contrast

 lmer(log(OF0max)~cond+(1+cond|speaker)+(1+cond|item),data)

Q1c - Difference on Verb, Direct object is in contrast

 lmer(log(VF0max)~cond+(1+cond|speaker)+(1+cond|item),data)

Q1b Estimate SE t value Sign.

Intercept 5.51883 0.02491 221.55

Condition=IO -0.03287 0.00694 -4.74 *

Q1a Estimate SE t value Sign.

Intercept 5.603747 0.022571 248.27

Condition=IO 0.015450 0.008016 1.93 n.s.

Q1c Estimate SE t value Sign.

Intercept 5.46828 0.02856 191.47

Condition=IO -0.01611 0.01907 -0.84 n.s.

(20)

27. November 2019

A production study – Results (duration)

Q1a – Difference on Indirect Object

 lmer(IOdur~cond+(1+cond|speaker)+(1+cond|item),data)

Q1b – Difference on Object, Indirect object is in contrast

 lmer(Odur~cond+(1+cond|speaker)+(1+cond|item),data)

Q1c - Difference on Verb, Direct object is in contrast

 lmer(Vdur~cond+(1+cond|speaker)+(1+cond|item),data)

Q1b Estimate SE t value Sign.

Intercept 0.418799 0.043654 9.594

Condition=IO -0.048867 0.005335 -9.160 *

Q1a Estimate SE t value Sign.

Intercept 0.53529 0.04529 11.819

Condition=IO 0.01310 0.01079 1.213 n.s.

Q1c Estimate SE t value Sign.

Intercept 0.326627 0.033614 9.717

Condition=IO -0.010851 0.007492 -1.448 n.s.

(21)

27. November 2019

A production study – Speaker variation

187

199

201

(22)

27. November 2019

Summary

 F0-maximum of each constituent is the acoustic cue of the constituent-final H tone.

 Downstep of each H tone was observed, similar to previous findings (Harnsberger & Judge 1996, Patil et al. 2008, Féry 2017)

 No difference of scaling of the H tone on the indirect object:

 no prosodic marking of focus on the focused constituent

 in line with Patil et al. (2008)

 Difference in scaling of the H tone on the direct object:

 post-focal compression after indirect object contrast

 PFC in line with Patil et al. (2008)

 If there is no prosodic marking of focus on the focused word, would listeners be able to identify a focus perceptually?

 What about speaker variation (speaker with PFC and without PFC)?

(23)

27. November 2019

A perception study

Do listeners use the prosodic cue PFC in sentence processing?

In particular, do listeners identify the focus on the basis of PFC?

Sentence completion task:

(2) raahul ne mãã ko davaaii dii naa ki … Raahul ERG mother DAT medicine give NEG that

“Raahul gave the medicine to the mother and not … ”

Predictions

If PFC is present, listeners will identify the indirect object contrast correctly.

If PFC is absent, listeners will show a chance level identification of the contrast (both direct object contrast and speakers without PFC).

gaaDii

“car”

naanii ko

“to granny”

(24)

27. November 2019

A perception study – Method

Speech materials and experimental design Data from 3 groups of speakers:

(i) 0 Hz PFC (ii) 10 Hz PFC (iii) 30 Hz PFC Sentence fragments from five sentences (cf. (2))

– Original data from production study were cut after the conjunction “ki”

– Random presentation of fragments with PRAAT–MFC

– 18 listeners x 5 sentences x 2 conditions x 6 speakers = 1080 trials

Listeners 18 native Hindi listeners (mean age 27 years) Task

Forced choice

sentence completion

(25)

27. November 2019

A perception study – Results

Sentence

Condition Direct Object

Condition Ind. Object

1 0.287 0.824

2 0.519 0.639

3 0.370 0.769

4 0.472 0.741

5 0.556 0.574

Tab.1: Mean identification of

correct sentence completion split by sentence

Speaker

Condition Direct Object

Condition Ind.

Object

5 0.489 0.611

21 0.422 0.622

8 0.478 0.800

18 0.422 0.678

26 0.456 0.767

30 0.378 0.778

Tab.2: Mean identification of

correct sentence completion split by item (speaker)

0 Hz

10 Hz

30 Hz

<

(26)

27. November 2019

A perception study – Results

Linear mixed effects model with “condition” as fixed factor and “listener”

and “item” as random factors

glmer(rating~cond+(1+cond|subject)+(1|item) …

Conditions differ significantly in terms of correct sentence completion.

Condition OBJECT CONTRAST (no PFC): On average, correct sentence completion in 44.1 % of the cases (chance level).

Condition INDIRECT OBJECT CONTRAST (PFC): On average, correct sentence completion in 70.9 % of the cases.

What about speaker variation? Speakers with and without PFC.

Estimate SE z value Pr(>|z|) Sign.

(Intercept) -0.5933 0.4618 -1.285 0.1988 Condition =

Ind.Obj.Contrast

1.9398 0.8316 2.333 0.0197 *

(27)

27. November 2019

A perception study – Results

A model that includes speaker variation

(2 speakers with no PFC compared to 4 speakers with PFC)

glmer(rating~cond*speakervar+(1+cond|subject)+(1|item)…

 No significant effect of CONDITION and SPEAKER GROUP

 Significant interaction in condition INDIRECT OBJECT CONTRAST (PFC)

 speaker variation matters

• Items (speakers) with PFC appear to lead to more correct sentence completion

• Items (Speakers) without PFC appear to lead to chance performance

Estimate SE z value Pr(>|z|) Sign.

(Intercept) -0.5075 0.4835 -1.050 0.29393 Condition =

Ind.Obj.Contrast

1.3397 0.8648 1.549 0.12135 Speaker=PFC -0.1276 0.2164 -0.590 0.55552

Interaction 0.9584 0.3096 3.096 0.00196 **

(28)

27. November 2019

A perception study – Results

• Speakers with PFC have a higher (75.6 %) correct sentence completion identification than speakers with no PFC (61.6 %).

• Post-hoc paired samples t-test confirms this pattern:

t = -8.39, df = 359, p < 0.001

• Model comparison:

Model with speaker variation shows significant improvement compared to the simple model measured as the difference in the deviance (Bates et al. 2015): ΔD = 14.2, p < 0.001.

Interaction plot: Speakers with no PFC (dotted line) compared to speakers with PFC (solid line) split by condition (Obj. vs. Ind. Obj.

contrast)

(29)

27. November 2019

Summary

The presence of PFC matters for focus perception:

If PFC is present, listeners identify the correct sentence completion.

If PFC is not present, listeners perform on chance level.

Speaker variation found in the production of PFC has an impact on focus perception:

Sentences of speakers with no PFC in the ‘indirect object contrast’

condition were completed by chance.

Sentences of speakers with PFC in the ‘indirect object contrast’

condition were completed correctly (75.6 % on average).

 PFC is thus a functional cue to focus identification in Hindi.

Pitch register information, i.e. PFC, is a cue that resolves local ambiguities.

(30)

27. November 2019

Discussion

In some tone languages, PFC is a robust phonetic cue to focus perception (cf. Xu et al. 2012 for Mandarin).

However, Mandarin also shows highlighting of focus on the focused constituent (cf. Xu 1999).

 Thus, there are more cues to focus perception in Mandarin.

Hindi is a language that does not mark focus by means of F0 on the focused constituent (Patil et al. 2008).

However, PFC is a cue that signals focus and that perceptually was shown to function a cue for focus perception.

 Thus, PFC as the only cue in production serves as a cue in perception.

(31)

27. November 2019

Discussion – Perception of focus

 Only few studies that investigate the perception of focus.

 If a prosodic cue systematically found in speech production is present, listeners identify the focus of a sentence (e.g. Botinis et al. 1999, Liu & Xu 2005, Vainio & Järvikivi 2006, Krahmer & Swerts 2007, Xu et al. 2012, Kügler & Gollrad 2015).

 Reliable focus perception for initial or medial focus, not final focus

(Botinis et al. 1999, Xu et al. 2012, Wang et al. 2012).

 For languages with PFC, cue must be present (sentence-finally it is not!)

 Listener’s language background matters (Wang et al. 2012):

 Listeners from a non-PFC language do not identify focus based on PFC

L2 phonological entities are harder identified (cf. Hume & Johnson 2003)

 PFC might be a phonological category similar to pitch accents or boundary tones.

(32)

27. November 2019

Discussion – Speaker variation

 Many studies found speaker variation with respect to the prosodic expression of focus

− German: speakers use predominantly F0 or duration (Baumann et al. 2007)

 Compensating strategy for F0 cues (Gussenhoven 2002):

− Later aligned F0-peaks ~ higher F0-peaks

− Proven for perception (Ladd & Morton 1997)

 Different strategies reported have a common aim:

marking a focused constituent as prosodically prominent

 Speaker variation in Hindi:

− No cue accumulation as in stress-based languages (F0, duration, intensity)

− Presence or absence of the cue (PFC)

− Only presence of cue leads to functionally decisive focus identification

 PFC as a phonological category similar to pitch accents or boundary tones.

(33)

27. November 2019

Discussion – Intonation categories in sentence processing

 Intonation – pitch accents and boundary tones (intonation contour)

 Pitch accents and processing:

− E.g. Focus Attraction Hypothesis (Schafer et al. 1996):

− Default interpretation (in silent reading) (1c)

− Reading (1b) or (1c) depends on the presence of a pitch accent on either ‘plane’

or ‘propeller’

(1) a. The sun sparkled on the propeller of the plane that the mechanic was so carefully examining.

b. The mechanic carefully examined the propeller (high attachment) c. The mechanic carefully examined the plane (low attachment)

 Phrase boundaries and processing:

− Phrase break marked by longer phrase-final duration (Lehiste 1973) (2) a. The hostess greeted the girl with a smile.

b. ( )( ) ‘The girl is smiling’

c. ( )( ) ‘The hostess is smiling’

(34)

27. November 2019

Discussion – Intonation categories in sentence processing

 Intonation – pitch accents and boundary tones (intonation contour)

 Pitch accents and boundary tones as categories of intonation have functional load.

 Pitch accents and boundary tones are typically found in languages using stress-based cues to express focus.

 If pitch register (PFC) has functional load – as was shown in the

identification of focus structure in Hindi – pitch register might also be interpreted as an intonational category

(at least for languages that use register-based cues to express focus).

(35)

27. November 2019

Conclusion

Pitch register – PFC –

– is a cue to focus perception.

– is a cue to disambiguate local ambiguities – thus carries functional load in Hindi.

Pitch register is thus also an effective prosodic cue for disambiguation – besides pitch accents and phrasing.

Pitch register is a phonological category.

(36)

27. November 2019

Thank you!

Acknowledgements

This work was supported by the DFG, projects “Prosody in Parsing” (DFG- KU 2323/2-1) and “Recursivity in Prosodic Phonology” (KU 2323/4-1).

Special thanks to the Co-PI’s of “Prosody in Parsing” Caroline Féry and Shravan Vasishth, and to the project members Gerrit Kentner, Anja Gollrad and Stella Gryllia as well as our student assistant, at that time, Dinah Baer- Henney. Thanks also to Susanne Genzel.

(37)

27. November 2019

References

Botinis, Antonis, Marios Fourakis & Barbara Gawronska. 1999. Focus identification in English, Greek and Swedish. In John J. Ohala, Y. Hasegawa, Manjari Ohala, D. Granville & A. C. Bailey (eds.), Proceedings of the 14th International Congress of Phonetic Sciences (ICPhS), 1557–1560. San Francisco:

University of California.

Breen, Mara, Evelina Fedorenko, Michael Wagner & Edward Gibson. 2010. Acoustic correlates of information structure. Language and Cognitive Processes25(7-9). 1044–1098.

Calhoun, Sasha. 2010. The centrality of metrical structure in signaling information structure: A probabilistic perspective. Language86(1). 1–42.

Féry, Caroline. 2010. Indian Languages as Intonational 'Phrase Languages'. In Imtiaz Hasnain & Shreesh Chaudhury (eds.), Problematizing language studies. Festschrift for Rama Agnihotri, 288–312. Delhi: Aakar Books.

Féry, Caroline. 2013. Focus as prosodic alignment. Natural Language & Linguistic Theory31(3). 683–734.

Féry, Caroline. 2017. Intonation and Prosodic Structure(Key Topics in Phonology). Cambridge: Cambridge University Press.

Féry, Caroline & Frank Kügler. 2008. Pitch accent scaling on given, new and focused constituents in German. Journal of Phonetics36(4). 680–703.

Fletcher, Janet. 2010. The Prosody of Speech: Timing and Rhythm. In William J. Hardcastle, John Laver & Fiona E. Gibbon (eds.), The Handbook of Phonetic Sciences: Second Edition, 523–602. Oxford: Wiley-Blackwell.

Genzel, Susanne & Frank Kügler. 2010. The prosodic expression of contrast in Hindi. In Proceedings of the Fifth International Conference on Speech Prosody 2010, 100143. Chicago, USA.

Gussenhoven, Carlos. 2002. Intonation and interpretation: Phonetics and Phonology. In Bernard Bel & Isabelle Marlien (eds.), Proceedings of the Speech Prosody 2002 Conference, 47–57. Aix-en-Provence: Laboratoire Parole et Langage.

Harnsberger, James D. 1999. The role of metrical structure in Hindi intonation. Paper at South Asian Language Analysis Roundtable.

Harnsberger, James D. & Jasmeet Judge. 1996. Pitch range and focus in Hindi. The Journal of the Acoustical Society of America99. 2493.

Jabeen, Farhat & Bettina Braun. 2018. Production and Perception of Prosodic Cues in Narrow & Corrective Focus in Urdu/Hindi. Proc 9^thInternational Conference on Speech Prosody 2018, Poznan, 30-34.

Kanerva, Jonni M. 1990. Focusing on phonological phrases in Chichewa. In Sharon Inkelas & Draga Zec (eds.), The Phonology-Syntax Connection, 145–

161. Chicago: University Of Chicago Press.

Khan, Sameer u. D. 2016. The intonation of South Asian languages: towards a comparative analysis. In Mythili Menon & Saurov Syed (eds.), Proceedings of FASAL 6, 23–36.

Krahmer, Emiel & Marc Swerts. 2007. Perceiving Focus. In Chungmin Lee, Matthew K. Gordon & Daniel Büring (eds.), Topic and Focus: Cross-linguistic perspectives on Meaning and Intonation(Studies in Linguistics and Philosophy 82), 121–138. Dordrecht: Springer.

Kügler, Frank & Sasha Calhoun. to appear. Prosodic encoding of information structure: A typological perspective. In Carlos Gussenhoven & Aoju Chen (eds.), The Oxford Handbook of Language Prosody. Oxford: Oxford University Press.

Kügler, Frank & Anja Gollrad. 2015. Production and Perception of Contrast: The case of the rise-fall contour in German. Frontiers in psychology6(1254).

1–18.

(38)

27. November 2019

References

Ladd, D. R. 2008. Intonational Phonology, 2nd edn. Cambridge: Cambridge University Press.

Ladd, D. R. & Rachel Morton. 1997. The perception of intonational emphasis: continuous or categorical? Journal of Phonetics25(3). 313–342.

Lehiste, Ilse. 1973. Phonetic disambiguation of syntactic ambiguity. Glossa7(2). 107–122.

Liu, F. & Yi Xu. 2005. Parallel Encoding of Focus and Interrogative Meaning in Mandarin Intonation. Phonetica62. 70–87.

Moore, Robert R. 1965. A Study of Hindi intonation. Ann Arbor: University of Michigan PhD Thesis.

Patil, Umesh, Gerrit Kentner, Anja Gollrad, Frank Kügler, Caroline Féry & Shravan Vasishth. 2008. Focus, word order and intonation in Hindi.

Journal of South Asian Linguistics1(1). 53–67.

Schafer, Amy, Juli Carter, Charles Clifton, JR & Lyn Frazier. 1996. Focus in relative clause construal. Language and Cognitive Processes11(1/2). 135–

163.

Selkirk, Elisabeth O. 1995. Sentence prosody: Intonation, stress, and phrasing. In John A. Goldsmith (ed.), The Handbook of Phonological Theory, 550–569. Cambridge, MA: Blackwell.

Turk, Alice. 2012. The Temporal Implementation of Prosodic Structure. In Abigail C. Cohn, Cécile Fougeron & Marie Huffman (eds.), The Oxford Handbook of Laboratory Phonology, 242–253. UK: Oxford University Press.

Vainio, Martti & Juhani Järvikivi. 2006. Tonal features, intensity, and word order in the perception of prominence. Journal of Phonetics34(3). 319–

342.

Wang, Bei, Chenxia Li, Qian Wu, Xiaxia Zhang & Yi Xu. 2012. Production and Perception of Focus in Huihui Language: Compared with Beijing Mandarin. In Proceedings of the 13th International Conference on Speech Communication and Technology, 663-666. Portland.

Xu, Yi. 1999. Effects of tone and focus on the formation and alignment of f0 contours. Journal of Phonetics27(1). 55–105.

Xu, Yi. 2011. Post-focus compression: Cross-linguistic distribution and historical origin. In Eric Zee & Wai-Sum Lee (eds.), Proceedings of the 17th International Congress of Phonetic Sciences (ICPhS XVII), Hong Kong, 152–155.

Xu, Yi, Szu-Wei Chen & Bei Wang. 2012. Prosodic focus with and without post-focus compression: A typological divide within the same language family? The Linguistic Review29. 131–147.