64
On recursive prosodic structure – Case studies from Akan and German Frank Kügler & IfL Phonetik Universität zu Köln Institut für Linguistik Goethe-Universität Frankfurt

On recursive prosodic structure – Case studies from Akan ...kuegler/docs/2017.CCLS-Lecture-Koeln... · On recursive prosodic structure – Case studies from Akan and German

Embed Size (px)

Citation preview

On recursive prosodic structure –Case studies from Akan and German

Frank Kügler&IfL Phonetik

Universität zu KölnInstitut für Linguistik

Goethe-Universität Frankfurt

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Recursivity in prosodic phonology Debate in prosodic phonology on the interpretation of the effects

of syntax in phonology– Direct vs. indirect reference approach

Debate in prosodic phonology on the constituency of prosodic structure– Universal set of prosodic constituents with recursive structure or not

Impacts on prosodic phrasing – A number of phonetic cues that signal edges of prosodic constituents

Case study 1: ATR harmony across words (Kügler 2015) Case study 2: German post-focal downstep (Kügler & Féry 2016)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Syntax-phonology interfaceTwo theoretical views –

The direct approach and the indirect approach.(cf. Truckenbrodt 2007, Selkirk 2011, Downing 2013)

The direct approach: Phonological processes apply directly in a given syntactic structure (e.g.

Kaisse 1985, Odden 1987).

The indirect approach: Phonological processes apply within a given prosodic structure that derives

from syntactic structure (e.g. Selkirk 1986, 2011). Prosodic structure thus mediates between syntactic structure and phonological processes.

Within the indirect approach, different theories emerged – most prominently, Align (Selkirk 2000), Wrap (Truckenbrodt 1999), and Match (Selkirk 2011).

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Prosodic constituents – The prosodic hierarchy

Intonation phrase

Phonological phrase

Prosodic word

Foot (word stress; trochee vs. Iamb)( x . )/(s w) ( . x )/(w s)

Syllable (tone bearing unit (TBU)

Mora (syllable weight)

ι|φ|ω|F|σ|µ

(Nespor & Vogel 1986, Selkirk 1986, 2011)

• Components of the interface between syntax and phonology

• Phonological properties• Reflects syntactic structure Recursivity

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Prosodic constituency Representation of the strict layer hypothesis –well-formed prosodic structure

Violation of the strict layer hypothesis

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Syntax-phonology interfaceAll theories treat recursivity differently. One argument of the proponents of the direct approach is

concerned with recursivity:─ Strict layer hypothesis (Nespor & Vogel 1986):

A prosodic constituent of level α is not allowed to dominate a constituent of the same level α. No recursivity at the level of prosodic constituents.(cf. Odden 1987)

─ Match theory allows for recursive prosodic structure (cf. Itô & Mester 2013).

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Recursive prosodic structure in Match Theory

(Itô & Mester 2007, 2012, 2013)

Minimal constituent

Non-minimal constituent

Maximal constituent

max = is not dominated by min = does not dominate non-min = dominates and is not

dominated by

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Prosodic constituency Universal set of prosodic constituents: Minimal, non-minimal and maximal constituents

of the same category allow for recursivity Relationship between syntactic and prosodic

constituents:Match constraints (Selkirk 2011: 439):

Syntactic constituency Phonological constituentWord (function words, lexical words) ≈ prosodic wordXP (maximal projections) ≈ phonological phraseClause (standard clause, illocutionary clause) ≈ intonation phrase

ι intonation phrase|φ phonological phrase|ω prosodic word

Akan – Kügler (2015)Kügler, Frank (2015) ATR vowel harmony and phonological phrasing in Akan. Phonology, 32(1), 177-204.

AkanKwa language, Niger-Congo family, 8.3 m speakers in Ghana and parts of the Ivory Coast (Lewis, 2009).Data from Asante Twi – one of three major dialects of Akan with its centre in Kumasi (Schachter & Fromkin 1968; Nkansa-Kyerenateng 2010).Data come from field research in Ghana (2012 / 2014) and informant consultation in Berlin (2012-2015).

Akan Phonology – The vowel system

17.09.2012, Library, Dept. of Modern Languages, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Akan Phonology – The vowel systemNine vowel system, two sets of vowels:

[+ATR] [–ATR]

• Advanced tongue root, volume increase of pharyngeal cavity, almost no difference of degree of primary constriction (Lindau 1975, Tiede 1996)

• Raised F1 for [+ATR] vowels (Lindau 1979, Hess 1992, Tiede 1996)

[i]

[e]

[u]

[o]

([æ])[]

[]

[]

[]

[a]

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Vowel harmony within the prosodic word• PronominalandTAMmarkersareaffixed to theverb(cf. (4), (5); Paster 2010).• Word stems are specified for [±ATR] while pronominal and TAM-markers are

underspecified for [ATR] (Clements 1985).• ATR vowel harmony occurs within the prosodic word

(Stewart 1967, Schachter & Fromkin 1968, Dolphyne 1988, Manyah 2006).(4) a. di ‘to eat’ (5) a. de ‘to be called’

b. mi-di [midi] b. me-de [mɪdɪ]1.SG-eat 1.SG-be called

c. wu-di [ʋudi] c. wo-de [ʋʊdɪ]2.SG-eat 2.SG-be called

d. ye-di [jedi] d. yɛ-de [jɛdɪ]1.PL-eat 1.PL-be called

e. mi-i-di [mi:di] e. me-e-de [mɪ:dɪ]1.SG-PROG-eat 1.SG-PROG-be called

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Vowel harmony within the prosodic wordOT constraints HARMONY: All vowels within a prosodic word must agree in the feature [ATR] IDENT[ATR]Root: Every root vowel in the input with the value [αATR] must have a

corresponding root vowel in the output with the value [αATR]. (6)

Footnote:In Akan disharmonic stems exist (e.g. sika‘money’). Therefore, word-level vowel harmony might be more complicated than HARMONYsuggests. Yet, this is beyond the scope of the present paper.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Akan syntax – word orderSVO (e.g. Boadi 2005, Saah 1988, 1994).(6) Kòfi ́ ba ́ há.

K. come.PRES here‘Kofi comes here.’

Head initial characteristics (e.g. Boadi 2005), e.g. N+Adj in (7).(7) n.namfo hunu

pl.friend useless ‘Useless friends’

Serial verb constructions (Baker 1989, Osam 2003) in (8).(8) a. me-yɛɛ adwuma mee-maa Amma b. ma-yɛ adwuma ma-ma Amma

1SG-do work 1SG-give Amma 1SG.PRF-do work 1SG.PRF-give A.‘I work for Amma.’ I have worked for Amma.’

(Baker 1989: 523f)

What does vowel harmony tell us about prosodic phrasing?! Regressive [+ATR] vowel harmony between words

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Vowel harmony across the prosodic word• ATR may spread across a word boundary (Dolphyne 1988:24)(9) Dolphyne (1988:24)

a. ɔpɛ + sika [ɔpesika] ‘s/he likes’ ‘money’ ‘S/he likes money.’

b. frɛ + Kofi [fre kofi] ‘call’ ‘Kofi’ ‘call Kofi.’

• Regressive ATR harmony affects last syllable of the verb.(10) ɔpɛ+ sika ɔpesika

[-ATR] [+ATR] [-ATR][+ATR]

• Only [+ATR] spreads regressively.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

VP-internal regressive vowel harmony• Regressive ATR harmony across words occurs in general.

S V O(11) /adamfʊ + tʊ + kube/ [adamfʊ tu kube]

‘friend’ ‘throw’ ‘coconut’ ‘A friend throws a coconut.’

S V O Mod(12) /kofi + tɔ + kɛtɛ + bebere:/ [kofi tɔ kɛte bebere:]

‘Kofi’ ‘buy’ ‘mat’ ‘many’ ‘Kofi buys many mats.’

S V O O(13) /kofi + aʨɛ + mɪyɪrɪ + sika/ [kofi aʨɛ mɪyɪri sika]

‘Kofi’ ‘give’ ‘my.wife’ ‘money’ ‘Kofi gives my wife money.’

S V O O(14) /kofi + aʨɛ + eno + m:ire/ [kofi aʨe eno m:ire]

‘Kofi’ ‘give’ ‘mother’ ‘mushroom’ ‘Kofi gives his mother mushrooms.’

(Kügler 2015: 191ff)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

VP-internal regressive vowel harmony• Regressive ATR harmony across words occurs in general.

S V O O Mod(15) /kofi aʨɛ madamfʊ kube bebre:/ [kofi aʨɛ madamfu kube bebre:]

kofi give friend coconut many ‘Kofi has given my friend many coconuts as a gift.’

S V O Mod O Mod(16)/kofi aʨɛ suku:ni fɛfɛ:fɛ kukuo kɛsɪ/ [kofi aʨe suku:ni fɛfɛ:fe kukuo kɛsɪ]

Kofi givestudent beautiful pot big ‘Kofi has given a very beautifulstudent a big pot.’

S V O Adv(17) /kofi yɛ: sɛbɛ defedefe/ [kofi yɛ: Sɛbe defedefe]

Kofi do Sɛbɛ completely ‘Kofi destroyed Sɛbɛ completely.’

(Kügler 2015: 191ff)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

NP-internal regressive vowel harmony Different postnominal modifier

Between modifier

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Vowel harmony across the prosodic word Prosodic markedness constraint drives regressive vowel harmony

Markedness constraint restricts the spread

(Note: *[+ATR] interacts with word-level vowel harmony.)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Vowel harmony across the prosodic word Violation of word-level harmony constraints. Prosodic markedness constraint is higher ranked than word-level harmony

constraints, and drives harmony across word boundaries.

– Word-level vowel harmony –

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Blocking of vowel harmony across words However, regressive ATR harmony across words is blocked.

S // V O(15) /akʊkɔ + di + ӕburo/ [akʊkɔ di ӕburo] *(akʊk[o])

‘chicken’ ‘eat’ ‘maize’ ‘A chicken eats maize.’

S V O // Adv(16) /kofi + tɔ + kɛtɛ +ʋukuada/ [kofi tɔ kɛtɛ ʋukuada] *(kɛt[e])

‘Kofi’ ‘buy’ ‘mat’ ‘Wednesday’ ‘Kofi buys a mat on Wednesday.’

S V O // VO (serial verb construction)(17) /kofi + de + kɔtɔ +di + agorɔ/ [kofi de kɔtɔ di agʊrɔ] *(kɔt[o] di)

‘Kofi’ ‘take’ ‘crab’ ‘eat’ ‘game’ ‘Kofi plays with a crab.’

(Kügler 2015: 191ff)

Syntax-Phonology Match in Akan

CP

TP

DP FP

N VP

V' DP

V DP N Adj

N AdjS V O O

ɩ

φ

N φ

φ φ

V φ N Adj

N AdjS V O O

MATCHCLAUSE – Standard clause is the complement of CP. Syntax of simple declarative clause in Akan (Boadi 2005; Saah 1994)

Syntax-Phonology Match in Akan

CP

TP

DP FP

N VP

V' DP

V DP N Adj

N AdjS V O O

ɩ

φ

N φ

φ φ

V φ N Adj

N AdjS V O O

MATCHPHRASE – Left and right edge of a lexical projection in the input syntactic structure matches with left and right edge of φ in phonological representation.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Syntax-Phonology Match in Akan MATCHPHRASE accounts for NP and VP internal RVH.(18) a. [NP Noun Modifier ]

b. φ( )φ

(19) a. [vP [VP Verb [NP Object ] ] [NP Object ] ]b. φ( φ( φ( )φ)φ φ( )φ)φ

Crucial difference between (18) and (19) – φ-phrase boundary. What about MATCHPHRASE and the blocking contexts?

(20) a. [NP Subject ] [VP Verb [NP Object ] ]b. φ( )φ φ( φ( )φ)φ

MATCHPHRASE predicts φ-phrase boundaries between S and V –However, this φ-phrase boundary appears to have different propertiescompared to the VP-internal φ-phrase boundaries.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Recursion-based subcategories of φ Proposal:

The blocking contexts involve maximal φ-phrase edges, while phrase edges in the non-blocking contexts are non-maximal.

A φmax is not dominated by any further φ-phrase

A φnon-max is dominated by a φ-phrase.

ɩ

φ

N φ

φ φ

V φ N Adj

N AdjS V O O

max

max

non-maxnon-max

non-max

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Recursion-based subcategories of φ A φmax between S and V –

what about the remaining blocking contexts?

1. Time adverbials – a φmax between VP and AdvP(21) a. [ kofi tɔ kɛtɛ ʋukuada ] ‘Kofi buys a mat on Wednesday.’

b. [TP [vP [VP Verb [NP Obj.] ] ] [AdvP Adv] ] c. φ( Verb φ( Obj.)φnon-max)φmax φ(Adv)φmax

2. Serial verb constructions – a φmax between VPs(22) a. [kofi de kɔtɔ di agʊrɔ] ‘Kofi plays with a crab.’

b. [vP [VP Verb [NP Obj.] ] [VP Verb [NP Obj.] ] ]c. φ( Verb φ( Obj.)φnon-max)φmax φ( Verb φ( Obj.)φnon-max)φmax

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Vowel harmony across the prosodic wordCrispEdge constraint prevents spreading across prosodic boundaries

(Itô & Mester 1999)Specific CrispEdge

constraint dominatesgeneral version(CripsEdgeφ)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Summary – Vowel harmony and phrasing Data show a general process of regressive [+ATR] vowel harmony

across word boundaries (RVH). RVH is a domain sensitive process – RVH is blocked at certain

prosodic boundaries, i.e. edges of φmax. RVH provides evidence for recursion-based subcategories of φ. [ATR] harmony as a diagnostic for phonological phrase formation is

new!Autosegmental analysis of RVH is comparable to cross-word H-tone

spreading in Copperbelt Bemba(cf.Kula & Bickmore 2015). Typologically, Akan ranks syntax-phonology faithfulness

constraints higher than prosodic markedness constraints (contrary to Xitsonga, cf. Selkirk, 2011).

German – Phrasing of the post-nuclear area(Kügler & Féry 2016)Kügler, Frank, Féry, Caroline (2016) Post-focal downstep in German. Language and Speech, DOI:10.1177/0023830916647204.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Background & Overview Post-focal realization of constituents and prosodic phrasing in relationship

with syntax and information structure:

– Production experiment – Results– Discussion and conclusion

Information structural terminology follows Krifka (2008).Focus elicits a set of alternatives relevant for the discourseGivenness qualifies material that has been mentioned in previous utterancesTopic qualifies the referent talked about in the sentence

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Background Syntactic constituents are mapped to prosodic phrases (Krifka 1984,

Gussenhoven 1983, 1992, Selkirk 1986, 1995, Truckenbrodt 1995, 2007, Féry 1989, 2011, Féry & Kügler 2008).

Predicate + argument form a prosodic phrase (Φ-phrase) together; every additional argument, as well as every adjunct forms its own Φ-phrase (Féry 2011).

In an all-new sentence, every Φ-phrase has its own pitch accent. Downstep is present and there is an optional upstep on the last constituent (Féry & Kügler 2008).

(1) {Was ist passiert?}[Der Hummer]Φ [hat den Reiher]Φ [dem HammelF vorgestellt ]Φthe lobster has the heron the sheep introduced

‘The lobster introduced the heron to the sheep.’

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Background A focused constituent tends to be aligned with the right-hand edge of a

prosodic phrase (Féry 2013)(1) Align-Focus-R, ι-phrase-R (Align-Foc-ι-R):

Align a focus with the right boundary of an intonation phrase. The pitch accent of the focused constituent forms the head of a

phonological phrase. Focus constituent carries the nuclear accent Deaccentuation of post-focal constituents

Effects for phrasing?

From Féry & Kügler (2008)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Background Sentences with a fronted participle to study post-focal constituents. Regular phrasing in marked word order as in unmarked word order:(2) {Was ist passiert?}

[VorgestelltThat der Hummer]Φ [den Reiher]Φ [dem Hammel]Φ Contexts elicitation sentence-initial focus on the participle.(3) {Did the lobster show the heron to the sheep?}

Vorgestellt hat der Hummer den Reiher dem Hammel

Phrasing – two options (a: e.g. Truckenbrodt 1995)

(3) a. [VorgestelltF hat der Hummer den Reiher dem Hammel]Φb. [VorgestelltF]Φ [hat der Hummer]Φ [den Reiher]Φ [dem Hammel]Φ

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

BackgroundFirst option: delete all Φ-phrase boundaries in the post-focal area. Given constituents have no phrase structure above ω. Pitch accent represents the head of the entire φ-phrase.

(3) ( x ) ι-phrase( x ) φ-phrase( x ) ( x )( x )( x ) ω-word

a. [VorgestelltF hat der Hummer den Reiher dem Hammel]Φ

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

BackgroundSecond option: a Φ-phrase boundary is inserted after the focused constituent to fulfil Align-Focus-R. Given constituents are phrased independently (syntax-phonology match). Pitch accent represents the head of the entire ι-phrase.

(3) ( x ) ι-phrase( x ) ( x )( x )( x ) φ-phrase( x ) ( x )( x )( x ) ω-word

a. [VorgestelltF hat der Hummer den Reiher dem Hammel]Φ

(Note that Align-Focus-R, ι-phrase-R is fulfilled at the ι-phrase level)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Background & PredictionsOption 1( ) (ι-phrase)( ) (Φ-phrase)( ) ( ) ( ) ( ) (ω-word)[vorgestelltF]Φ [hat der Hummer]Φ [den Reiher]Φ [dem Hammel]Φ Flat post-focal F0 and declination, no pitch accents.

Option 2( ) (ι-phrase)( ) ( ) ( ) ( ) (Φ-phrase)( ) ( ) ( ) ( ) (ω-word)[vorgestelltF]Φ [hat der Hummer]Φ [den Reiher]Φ [dem Hammel]Φ Post-focal pitch accents as head of Φ-phrase,

downstep (as in pre-focal constituents, cf. Féry & Kügler 2008).

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Speech materials(1a) Hat der Hummer den Streit beendet?

Nein, angefangen hat der Hummer.

(2a) Hat der Hummer den Reiher bestraft?Nein, eingeladen hat der Hummer den Reiher.

(3a) Hat der Hummer den Reiher dem Hammel gezeigt? Nein, vorgestellt hat der Hummer den Reiher dem Hammel.

Condition: Number of arguments

Items – Six repetitions in total: (a) Two different verbs per sentence length:

intransitive: anfangen ‘to begin’ anrufen ‘to call’transitive: einladen ‘to invite’ besuchen ‘to visit’ditransitive: vorstellen ‘to introduce’ zeigen ‘to show’

(b) Three tokens per verb in terms of variation of the animals:Hummer ‘lobster’ Hammel ‘sheep’ Reiher ‘heron’

Nein, angefangen hat der Hummer.Nein, angefangen hat der Hammel.Nein, angefangen hat der Reiher.Nein, angerufen hat der Hummer.Nein, angerufen hat der Hammel.Nein, angerufen hat der Reiher.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Recording Eleven female speakers

- undergraduate students in their twenties at Potsdam University with no speaking or hearing deficits

Sound proof both, digital recording at 16 kHz Presentation software with pre-recorded questions eliciting different focus-

given structures, different word orders, and different number of arguments(Only sentences with initial focus with increasing number of post-focal arguments are used for analysis; remaining conditions resembles patterns of Féry & Kügler 2008.)

Target sentences were to be read as answers to previous questions. Pseudo-randomized list of tokens to avoid monotony

(caused however some obvious misplaced pitch accents)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – HypothesesH1: Post-focal Φ-phrase deletion. An early focus triggers deletion of all

prosodic phrases. Phrasing is primary the consequence of information structure, syntax is only default.

(6) [Angerufen hat der Hummel den Reiher]Φ

H2: Post-focal Φ-phrase preservation. Syntax triggers phrasing, focus affects prominence relationships. Post-focal prominences are preserved in a compressed register.

(7) [Angerufen]Φ [hat der Hummel]Φ [den Reiher]Φ

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Hypothesis (a)H1a. The final low value of the sentence is reached on or immediately after the focused word,

and the remaining constituents are flat and low.

(6) [Angerufen hat der Hummel den Reiher]Φ

H2a. The final low point is reached only at the end of every declarative sentence, regardless of the number of arguments following the focused word.

(7) [Angerufen]Φ [hat der Hummel]Φ [den Reiher]Φ

L1 = final Low

L1 ≠ final Low

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Hypothesis (b)H1b. No pitch accents are present on the given constituents, and only intensity and duration,

compatible with correlates of lexical stress are preserved on the lexically stressed syllable of each argument.

(6) [Angerufen hat der Hummel den Reiher]Φ

H2b. The individual arguments are pitch-accented, albeit in a narrow pitch register.

(7) [Angerufen]Φ [hat der Hummel]Φ [den Reiher]Φ

F0DYN1 = F0DYN2

F0DYN1 ≠ F0DYN2

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Hypothesis (c)H1c. There is no difference between two- and three-arguments sentences: the arguments are

at the same F0 level reached on or after the participle.

(6) [Angerufen hat der Hummel den Reiher]Φ

H2c. There is a difference as a function of number of post-focal constituents: the more arguments are to follow, the higher the arguments start; downstep is expected between post-focal arguments.

(7) [Angerufen]Φ [hat der Hummel]Φ [den Reiher]Φ

F01 = F02

F01 ≠ F02

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Hypothesis (d) Post-focal declination with (b) or without (a) downstep

(a)

(b)

• Declination is a continuous effect over time• Downstep is an intended lowering of F0 independent of the time between

the steps Positive correlation between downward F0 step and temporal distance

between two peaks would show declination. No correlation would show downstep.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Some data

H H L

Nein angefangen hat der Hammel

Pitc

h (H

z)

120120

150150

200200

250250

300300350350

01-1112

0.9 1.2 1.5 1.8 2.1 2.4 2.7 3time (s)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Some data

H H H L

Nein eingeladen hat der Reiher den Hammel

Pitc

h (H

z)

150150

200200

250250

300300350350

02-1213

0.9 1.2 1.5 1.8 2.1 2.4 2.7 3time (s)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Some data

H H H H L

Nein vorgestellt hatderHummerden Reiher dem Hammel

Pitc

h (H

z)

150150

200200

250250

300300350350

06-1311

0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 3 3.3time (s)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

The Experiment – Some data

H H H H L

Nein vorgestellt hatderHummer den Reiher demHammel

Pitc

h (H

z)

150150

200200

250250

300300350350

09-1311

0.9 1.2 1.5 1.8 2.1 2.4 2.7 3 3.3 3.6 3.9 4.2time (s)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

F0 measurementsMeasurements: F0-maximum in each verb / noun F0-minimum left and right to the F0-max

(3) # Eingeladen # hat # der # Reiher # den # Hammel # Calculations:

(1) Position of final low l2_verb vs. l2_final argument

(2) F0 dynamics(i) F0 rise = H – l1(ii) F0 fall = H – l2

(3) Scaling of F0-maximacross-sentence: HA1 / HA2

within sentence: HA1 – HA2

l1 H l2 l1 H l2 l1H l2L

Nein eingeladen hat der Reiher den Hammel

Pitc

h (H

z)

150150

200200

250250

300300350350

02-1213

0.9 1.2 1.5 1.8 2.1 2.4 2.7 3time (s)

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Results – overview of all dataFigure 1. All data

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Results–Hyp(a): final low value of the sentence

No. ofarguments

L1 (Hz)

Final Low (Hz)

1 181 1662 247 1683 235 166

Table 1. F0 minimum in Hz (i) at the end of the verb (L1), and (ii) the end of the sentence (final Low), aggregated over speakers, items, and repetitions.

Table 2. Linear mixed effect model with speaker, item and repetition as random factors, and position of F0 minimum and number of arguments as fixed factors.Contrasts SE t sign.L1 ~ FL 0.02584 -3.38 *Noa1 ~ Noa2 0.02739 9.85 **Noa1 ~ Noa3 0.02584 9.12 **Interaction L1/FL ~ Noa1/Noa2

0.03833 -6.47 **

InteractionL1/FL ~ Noa1/Noa3

0.03655 -6.32 **

Figure 2. Interaction plot.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Results–Hyp(a): final low value of the sentence

No. ofarguments

L1 (Hz)

Final Low (Hz)

1 181 1662 247 1683 235 166

Table 1. F0 minimum in Hz (i) at the end of the verb (L1), and (ii) the end of the sentence (final Low), aggregated over speakers, items, and repetitions.

The final low point is reached only at the end of every declarative sentence, disregarding the number of post-focal arguments.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Results–Hyp(b): Presence/absence of post-focal accents

Measurement of F0 dynamics within each post-focal argument:

Arguments Verb A1 A2 A31 62.7 5.02 100.3 7.7 6.23 64.1 18.2 12.5 7.3

Table 3. F0 dynamics – average F0 rise in Hz to F0 peak within the corresponding domain.

Arguments Verb A1 A2 A31 138.0 13.02 82.9 13.3 12.03 79.2 14.5 12.8 20.6

Table 4. F0 dynamics – average F0 fall in Hz from F0 peak within the corresponding domain.

The individual arguments are pitch-accented, albeit in a narrow pitch register.

• F0 rise increases with increasing number of arguments.

• Relative small rise of approx. 6 Hz in each final argument.

• Overall larger F0 fall in each argument.

• Each rise and each fall, and each rise and fall together are well above JNDs (Kollmeier et al. 2008).

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Results–Hyp(c): F0 scaling of argumentsFigure 3. F0-maximum in Hz for all data on first, second, or third argument (x-axis) for one-argument (dotted line), two-argument (dashed line), and three-argument sentences (solid line).

Arg’s Verb A1 A2 A3 n1 319 (43) 179 (16) 662 330 (53) 190 (17) 179 (18) 553 314 (34) 200 (29) 193 (22) 187 (21) 66

Table 5. Means of F0-maximum per argument and verb in Hz with SD in parentheses.

Cross-sentence comparison

Table 6: Linear mixed effects model with speaker, item and repetition as random factors, and number of arguments as fixed factor.Contrasts SE t sign.Noa1 ~ Noa2 0.01160 4.74 **Noa1 ~ Noa3 0.01088 10.43 ***

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Results–Hyp(c): F0 scaling of argumentsFigure 3. F0-maximum in Hz for all data on first, second, or third argument (x-axis) for one-argument (dotted line), two-argument (dashed line), and three-argument sentences (solid line).

Arg’s Verb A1 A2 A3 n1 319 (43) 179 (16) 662 330 (53) 190 (17) 179 (18) 553 314 (34) 200 (29) 193 (22) 187 (21) 66

Table 5. Means of F0-maximum per argument and verb in Hz with SD in parentheses.

Within-sentence comparison

Table 7: LMEM with speaker, item and repetition as random factors, and position of arguments as fixed factor.

Contrasts SE t sign.2Arg‘s: A1-A2 0.007736 -7.18 **3Arg‘s: A1-A2 0.01190 -3.03 *3Arg‘s: A1-A3 0.01190 -5.90 **3Arg‘s: A2-A3 0.01020 -3.34 *

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Results–Hyp(c): F0 scaling of argumentsFigure 3. F0-maximum in Hz for all data on first, second, or third argument (x-axis) for one-argument (dotted line), two-argument (dashed line), and three-argument sentences (solid line).

Cross-sentence comparison• F0-scaling of initial (and

later) post-focal argument differs as a function of number of post-focal arguments: the more the higher

Within-sentence comparison• F0-scaling differs between

post-focal arguments, i.e. they are in a downstep relation.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Results–Hyp(d): declination vs. downstep

No correlation between F0-maxima(y-axis) and distance in timebetween F0-maxima (x-axis).

Black line = two-argument sentencesgrey line = three-argument sentences

Downstep in post-focal constituents

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

SummaryThis experiment has shown:

1. Final low of the sentence is realized at the very end of each sentence, independent of sentence length.

2. Post-focal arguments bear prominences (pitch accents), though in a very compressed pitch register.

3. The scaling of F0 depends on the number of post-focal arguments; the more post-focal arguments, the higher the scaling.

4. Post-focal prominences are in a downstep relationship.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Discussion and conclusionThe detailed analysis of the realization of the given arguments determine a clear winner in the comparison between the two possible options in prosodic structure: (8b).

(8) {Hat der Hummer den Reiher dem Hammel gezeigt?}

a. Nein, [vorgestelltF hat der Hummer den Reiher dem Hammel]Φb. Nein, [vorgestelltF]Φ [hat der Hummer]Φ [den Reiher]Φ [dem Hammel]Φ

Prosodic structure is not lost in the post-focal part of the sentence. It is realized in a compressed register.

Kügler: Recursive prosodic structure – CCLS Cologne, 08.05.2017

Conclusion Data from regressive vowel harmony across word boundaries in Akan feeds

the debate on constituency in prosodic phonology.– A theory of recursive prosodic structure (cf. Itô & Mester 2013) accounts for the

phrasing patterns in Akan, i.e. the blocking of RVH at maximal φ-phrases

Detailed phonetic analysis of the presence or absence of pitch accents provides insights in the phrasing properties of e.g. post-focal constituents– The post-focal facts may help to understand the presence of optional pitch accents in

certain environments (e.g. embedded PPs, adjuncts and so on).

A further study of phrasing phenomena may eventually result in a better understanding of recursivity in prosodic constituency.– To do: Comparison between theories My project plan of the Heisenberg project in the near future

Medaase!Thank you!

Vielen Dank!

Acknowledgements:Thanks to Kofi Saah, Kofi Dorflo, Charles Marfo, Reginald Akuoko Duah, Monica Amoah Appenteng, Susanne Genzel, Lisa Selkirk, Seunghun Lee Funded by DFG: SFB 632, Project D5 (http://www.sfb632.uni-potsdam.de/projects_d5eng.html), and KU 2323/3-1 / KU 2323/4-1.

63

Selected ReferencesAmeka, F. 1992. Focus constructions in Ewe und Akan: A comparative perspective. In C. Collins & V. Manfredi (Eds), Proc. Kwa

comparative syntax workshop. Cambridge: MIT, 1–25.Baumann, S., & Grice, M. (2006). The intonation of accessibility. Journal of Pragmatics, 38, 1636–1657.Boadi, L. A. 1974. Focus-marking in Akan. Linguistics 140, 5–57.Boadi, L. A. 2005. Three major syntactic structures in Akan. Accra: BlackMaskLimited.Casali, R. F. 2003. [ATR] value asymmetries and underlying vowel inventory structure in Niger-Congo and Nilo-Saharan. Linguistic

Typology 7, 307–382.Clements, N. G. 1985. Akan vowel harmony – a non-linear analysis. In Goyvaerts, D. (ed) Esseys in memory of M.W.K. Semikenke.

Amsterdam: John Benjamins.Dolphyne, F. A.1988. The Akan (Twi-Fante) Language. Its Sound System and Tonal Structure. Accra: Ghana University Press.Fery, C. (2011). German sentence accents and embedded prosodic phrases. Lingua, 12, 1906–1922.Fery, C. (2013). Focus as alignment. Natural Language and Linguistic Theory, 31, 683–734.Fery, C., & Kugler, F. (2008). Pitch accent scaling on given, new and focused constituents in German. Journal of Phonetics, 36, 680–703.Hess, S. 1992. Assimilatory effects in a vowel harmony system: an acoustic analysis of advanced tongue root in Akan. Journal of

Phonetics, 20, 475-492.Ito, J. & Mester, A. 1999. Realignment. In R. Kager, H. v.d. Hulst, and W. Zonneveld (eds.) The Prosody-Morphology Interface. Cambridge

University Press, 118-217.Itô, J., & Mester, A. (2012). Recursive prosodic phrasing in Japanese. In T. Borowsky, S. Kawahara, T. Shinya, & M. Sugahara (Eds.),

Prosody Matters. Essays in Honor of Elisabeth Selkirk (pp. 280–303). London: Equinox Press.Kager, R. 1999. Optimality Theory. Cambridge: CUP.Kobele G. & Torrence H. 2004. The Syntax of Complement Clauses in Asante Twi. MS. U Chicago.Kobele G. & Torrence H. 2006. Intervention and focus in Asante Twi. ZAS Papers in Linguistics 46:161–184. Kula, N., & Bickmore, L. S. (to app.). Phrasal Phonology in Copperbelt Bemba. Phonology, 32,1.Krifka, M. (2008). Basic notions of information structure. Acta Linguistica Hungarica, 55, 243–276.Lewis, M. P. (ed.) 2009. Ethnologue: Languages of the World, Sixteenth edition. Dallas, Tex.: SIL International. (Online version:

http://www.ethnologue.com/).Lindau, M. 1975. [Features] for Vowels. Working Papers in Phonetics, Department of Linguistics, UCLA, UC Los Angeles, No. 30:

http://escholarship.org/uc/item/7gv6z0vq

64

Selected ReferencesLindau, M. 1979. The feature expanded. Journal of Phonetics, 7, 163-176.Kobele G. & Torrence H. 2006. Intervention and focus in Asante Twi. ZAS Papers in Linguistics 46:161–184. Krifka, M. (2008). Basic notions of information structure. Acta Linguistica Hungarica, 55, 243–276.Lewis, M. P. (ed.) 2009. Ethnologue: Languages of the World, Sixteenth edition. Dallas, Tex.: SIL International. (Online version:

http://www.ethnologue.com/).Lindau, M. 1975. [Features] for Vowels. Working Papers in Phonetics, Department of Linguistics, UCLA, UC Los Angeles, No. 30:

http://escholarship.org/uc/item/7gv6z0vqLindau, M. 1979. The feature expanded. Journal of Phonetics, 7, 163-176.Manyah, K. A. 2004. Harmonie vocalique et ATR en twi : aperçu phonologique et étude acoustique préliminaire, XXVèmes Journées

d’Étude sur la Parole (JEP) de l’AFCP, Fès.Marfo, C. O. 2004. The prosody-syntax interface in Akan focus and topic constructions. Proceedings of the 4th Postgraduate Research

Forum on Linguistics, 125-133.Nkansa-Kyerenateng, K. 2010. The Akans of Ghana. Their customs, history and institutions. Kumasi: Sebewie.Saah, Kofi K. 1988. Wh-questions in Akan. Journal of West African Languages, 18, 17-28.Schachter, P. & Fromkin, V. 1968. Phonology of Akan. Working Papers in Phonetics, No.9, Dept. Linguistics, UCLA.Selkirk, E. 2009. On Clause and Intonational Phrase in Japanese: The Syntactic Grounding of Prosodic Constituent Structure. Gengo

Kenkyu 136: 35–73.Selkirk, E. 2011. The Syntax-Phonology Interface. In: Goldsmith, J., Riggle, J. & Yu, A. (eds) The Handbook of phonological Theory.

Oxford: Blackwell, 435-484.Stewart, J. M. 1965. The typology of the Twi tone system. In Bulletin of the Institute of African Studies 1, Institute of African Studies,

University of Ghana, Legon, 1-27.Tiede, M. 1996. An MRI-based study of pharyngeal volume contrast in Akan and English. Journal of Phonetics, 24, 399-421.Truckenbrodt, H. (2002). Upstep and embedded register levels. Phonology, 19, 77–120.Truckenbrodt, H. (2006). Phrasal stress. In K. Brown (Ed.), The encyclopedia of languages and linguistics, 2nd Edition, Vol. 9 (pp. 572–

579). Oxford, UK: Elsevier.Truckenbrodt, H. (2007). Upstep on edge tones and on nuclear accents. In C. Gussenhoven & T. Riad (Eds.), Tones and tunes: Studies in

word and sentence prosody. Vol 2 (pp. 349–386). Berlin, Germany: Mouton de Gruyter.