introduction and overview: prosody€¦ · specific functions of prosody (including the marking of...
TRANSCRIPT
Introduction and Overview:
Prosody
Stefan Baumann IfL-Phonetik
Universität zu Köln
Structure of the Talk Introduction to prosody: Tasks and parameters
Specific functions of prosody (including the marking of information structure)
Prosodic structure and influencing factors
2 Workshop SynSemPro, Cologne, December 1-2, 2014
Prosody: Tasks and Parameters Two main tasks of prosody
– (1) Highlighting: marking prominences – (2) Phrasing: marking division of speech into chunks
Phonetic parameters to achieve highlighting and phrasing – Pitch (movement) – Loudness – Segmental length – Segmental quality – (Pauses)
3 Workshop SynSemPro, Cologne, December 1-2, 2014
Highlighting West Germanic languages (stress accent languages,
see Beckman 1986)
– Lexical stress: word level, abstract, potential for concrete prominence
Ber ‘lin
,Mas sa ‘chu setts
4 Workshop SynSemPro, Cologne, December 1-2, 2014
Highlighting – Postlexical stress: utterance level, concrete prominence, no pitch
movement (duration, intensity)
Example (Huss 1978):
1) Whereas formerly the GOvernment used to increase benefits, now the emPLOyers increase benefits.
2) Whereas formerly the WORkers‘ increase in deductions used to benefit from inflation, now the emPLOyers‘ increase benefits.
postlexical stress
5 Workshop SynSemPro, Cologne, December 1-2, 2014
Highlighting – (Pitch) Accent: utterance level, concrete prominence, pitch
movement (plus duration, intensity)
I bought a HOUSE.
6 Workshop SynSemPro, Cologne, December 1-2, 2014
Highlighting Stress and Accent – concrete prominence of
syllables at utterance level
prenuclear nuclear Position in utterance (status in prosodic hierarchy)
(adapted from Terken & Hermes 2000)
7 Workshop SynSemPro, Cologne, December 1-2, 2014
Highlighting by Phrasing Example Korean
– A: [satšun-enni] [irimi] [mweni] (lit. cousin name what)
– B: [satšun-enni irimi] [suni-dži] (lit. cousin name Suni)
– Analogy to accenting and deaccenting: name is the center/most important part of the question (A) → realised as own phrase
In the answer (B) name is contextually given → not realised as own phrase (“dephrasing“)
8 Workshop SynSemPro, Cologne, December 1-2, 2014
Phrasing Examples
– Telephone numbers: 928 77 66 or 22 22 22 2 – [The car you see is mine] vs.
[The car] [you see] [is mine]
Cues to phrase breaks:
Pauses
Boundary tones (pitch movement associated with unaccented syllables at phrase boundaries)
Tonal reset (jump in pitch either up or down at beginning of new phrase)
9 Workshop SynSemPro, Cologne, December 1-2, 2014
Phrasing Domain-final lengthening (e.g. Wightman et al.
1992, Turk & Shattuck-Hufnagel 2007)
– Pre-boundary segments significantly longer (but not louder and more strongly articulated ≠ stress), slowing down of articulatory gestures
10 Workshop SynSemPro, Cologne, December 1-2, 2014
Phrasing – The larger the prosodic domain the greater the degree of
lengthening
11 Workshop SynSemPro, Cologne, December 1-2, 2014
(from Grice 2006, Frota 2012)
Phrasing: No Phrase Boundary (word boundary)
256 ms
Examples by S. Shattuck-Hufnagel 12 Workshop SynSemPro, Cologne, December 1-2, 2014
Phrasing: Intermediate (= smaller) Phrase Boundary
386 ms
13 Workshop SynSemPro, Cologne, December 1-2, 2014
Phrasing: Intonation Phrase (= larger) Boundary
687 ms
14 Workshop SynSemPro, Cologne, December 1-2, 2014
Phrasing Decrease of intensity at the end of an IP
Domain-initial strengthening (e.g. Keating et al. 2003)
– Consonants hyperarticulated to indicate beginning of higher prosodic domains (e.g. intonation phrase > intermediate phrase > phonological word > foot > syllable)
15 Workshop SynSemPro, Cologne, December 1-2, 2014
Phrasing – Increasing resistance to assimilations in larger domains (e.g. Kuzla
2009)
Weil er vorha[t, W]älder und Seen zu malen...
→ no devoicing of /v/ at beginning of intermediate phrase (= no voicing assimilation)
Benno ha[t W]älder und Seen gemalt.
→ devoicing of /v/ at beginning of phonological word (= voicing assimilation)
16 Workshop SynSemPro, Cologne, December 1-2, 2014
Phrasing Voice quality
– Creaky voice and glottalisation at beginning and/or end of a domain
Abrupt change in rhythm/ tempo – e.g. anacrusis (fast sequence of syllables at beginning of
a phrase)
17 Workshop SynSemPro, Cologne, December 1-2, 2014
Both Highlighting and Phrasing Rhythm (relation between prominent and non-
prominent syllables) – Principle of alternation: thirTEEN + MEN → THIRteen MEN
Tempo (speech rate) – More assimilations and reductions, less phrases and accents in
faster speech
18 Workshop SynSemPro, Cologne, December 1-2, 2014
It’s probably the same thing
Examples by S. Shattuck-Hufnagel 19 Workshop SynSemPro, Cologne, December 1-2, 2014
probably the (fast speech)
20 Workshop SynSemPro, Cologne, December 1-2, 2014
probably (slow speech)
21 Workshop SynSemPro, Cologne, December 1-2, 2014
Neither Highlighting nor Phrasing: Speech Melody
= Intonation in a narrow sense
Same accent placement, same phrasing, different tunes:
Are you coming to the PARty tonight?
Are you coming to the PARty tonight?
22 Workshop SynSemPro, Cologne, December 1-2, 2014
Specific Functions of Prosody
Specific Functions of Prosody
(Grice & Baumann 2007)
24 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Syntactic Structure
– Differentiation of sentence type by intonation contour (in combination with choice of words and word order), e.g. question vs. statement
They went to Berlin? They went to Berlin.
interrogative declarative
25 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Syntactic Structure
– Syntactic disambiguation by phrasing
old men and women
John murdered the man with a gun.
[old men] [and women]
[old] [men and women]
[John murdered the man ] [ with a gun ]
[John murdered ] [ the man with a gun ]
26 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Syntactic Structure – Syntactic(-semantic) disambiguation by phrasing and
intonation contour Is he drinking?
William is not DRINking because he’s unHAPpy.
William is not DRINking because he’s unHAPpy.
yes
no
27 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Syntactic Structure – Syntactic disambiguation by accent placement
I asked the teacher who left.
(Schafer et al. 2000)
I asked the teacher WHO left or: WHO LEFT
I asked the teacher who LEFT
28 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Speech Acts
– Differentiation of illocutionary acts (speaker intention independent of sentence mode) by intonation, e.g.
Why don‘t you go to GeNEva? Why don‘t you go to GeNEva?
information seeking question suggestion / command
29 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure – Information structure: Three dimensions
= Focus (broad, narrow, contrastive) vs. Background
= Information status (given – accessible – new)
= Theme/Rheme or Topic/Comment
30 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure Important/informative elements are highlighted (in
particular by accentuation), unimportant/uninformative elements are not
= Basic assumption of Focus-to-Accent (FTA) approach
Two variants: – Radical (highlighting-based) – Formal (structure-based)
31 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure Radical FTA: Relation between focus and accent is
bidirectional and universal
Single words are in focus and thus highlighted, no larger constituents
Accent position is non-predictable, only dependent on intentions of the speaker and discourse context (→ strictly pragmatic) (see Bolinger, Schmerling)
32 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure Speakers are perfectly free to highlight word A
rather than word B or word C
(1) This is the MAN I was telling you about. (2) This is the man I was TELLing you about.
(Schmerling 1976)
Schmerling: Difference between (1) and (2) could not be accounted for by any conceivable linguistic theory
33 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure However, bidirectionality of focus and accent is
problematic in broad focus: nuclear accent stands in for larger focus domain (focus projection), i.e. focus without accent
– What's up?
– [ I'm going to BerLIN tomorrow. ]F
Further evidence against 1:1-relation: – Focus without accent: Second Occurrence Focus – Accent without focus: Accents on background material
34 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure Formal/ Indirect FTA: Position of (nuclear) accent
determines the interpretation of an utterance’s information structure (= larger constituents)
– to some extent predictable by semantic-syntactic rules of a language (e.g. Gussenhoven 2007)
35 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure Claim: In English and German (final) arguments are
more likely to receive a (nuclear) pitch accent than predicates (if they occur in the same focus domain)
I bought a HOUSE. predicate argument
Ich habe ein HAUS gekauft. predicate argument predicate
36 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure Three different kinds of givenness which may trigger
deaccentuation of the final argument
1. Givenness of a discourse referent (i.e. coreference)
John has an old cottage. a) Last summer he reconSTRUCted the shed.
(shed = cottage) given information = coreference
b) Last summer he reconstructed the SHED. (shed ≠ cottage) new information = no coreference
37 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure 2. Givenness of a lexical item (no coreference)
After the holidays, John arrived in a new car, and also HARry had a new car.
(Deaccentuation of given elements is language-specific:
no/less deaccentuation e.g. in Indian English, Singapore English, Romance languages…)
38 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure 3. Concept-givenness (van Deemter 1994, 1999) due to
an inferential relation between two (non-identical) entities (lexical superordination of anaphor)
Do you like dogs? I like ALL animals. (hyponym – hypernym)
Why do you spend so much time in Naples? It's my favourite CIty in Italy. (part – whole)
39 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure Asymmetry: lexical subordination of anaphor
Referring expressions are less activated than given information (see Chafe 1976, 1994) more likely to be accented
Do you like animals? I like all DOGS. (hypernym – hyponym)
Why do you spend so much time in Italy? I have a friend in NAples. (whole – part)
40 Workshop SynSemPro, Cologne, December 1-2, 2014
Indicating Information Structure In addition:
Accentuation due to contrastive focus, despite referential and lexical givenness (prerequisite: clearly identifiable alternative set)
A: Was it Mary or John who invited you to the party? B: It was [JOHN]F .
41 Workshop SynSemPro, Cologne, December 1-2, 2014
contrastive focus
referentially given lexically given
Indicating Information Structure Information structural meaning differences also
encoded by different pitch accent types (German)
Givenness
42 Workshop SynSemPro, Cologne, December 1-2, 2014
Falling vs. rising onglide
(Röhr & Baumann 2011)
Indicating Information Structure Focus
43 Workshop SynSemPro, Cologne, December 1-2, 2014
Falling vs. rising onglide
(Krüger 2009, Ritter & Grice 2013)
Indicating Information Structure Direct relation between accent types and perceived
prominence
44 Workshop SynSemPro, Cologne, December 1-2, 2014
low < falling < high < rising
(Baumann & Röhr, in prep.)
Prosodic Structure
Prosodic Structure Certain aspects of prosody are grammatical in nature
and as such represented in a phonological representation, called Prosodic Structure
Prosodic structure subsumes three types of structure:
– Prosodic constituent structure (Prosodic Hierarchy)
– Metrical structure (Relative metrical strength of syllables and words)
– Tonal structure (Location and type of pitch accents and boundary tones)
46 Workshop SynSemPro, Cologne, December 1-2, 2014
Some Factors influencing Prosodic Structure
Pragmatics and semantics (information structure)
Syntactic constraints
Well-formedness conditions on the size and eurhythmicity of prosodic constituents
47 Workshop SynSemPro, Cologne, December 1-2, 2014
Prosodic Structure vs. Syntactic Structure
Prosodic edges often align with syntactic edges BUT prosodic structure ≠ syntactic structure
(Steedman 1991) 48 Workshop SynSemPro, Cologne, December 1-2, 2014
Prosodic Structure vs. Syntactic Structure
Well-formed prosodic choices may violate (traditional) syntactic structure
– Length/weight of constituents: a short syntactic constituent can be joined with a fragment of another preferred metrical structure
• The man in the red hat / went home to eat. NP / VP • He went home / to eat and sleep. violates NP / VP
(Shattuck-Hufnagel & Turk 1996) 49 Workshop SynSemPro, Cologne, December 1-2, 2014
Prosodic Structure vs. Syntactic Structure
Speakers have prosodic options – Location of pitch accents
• It’s al-RIGHT vs. It’s AL-RIGHT vs. IT’S AL-RIGHT
– Phrasing • drink a pint of milk a day → drink a / pint a / milk a / day
– Tunes • Are you coming to the PARty tonight? (rising vs. falling)
(Shattuck-Hufnagel & Turk 1996) 50 Workshop SynSemPro, Cologne, December 1-2, 2014
The End! H+L*