![Page 1: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/1.jpg)
Syntax-‐prosody mismatches in Irish and English verb-‐ini9al structures
Emily Elfner McGill University
![Page 2: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/2.jpg)
Introduc9on
• Syntax-‐prosody interface: concerned with the mapping of syntac9c to prosodic structure
• To what extent can prosody be used as a test for syntac9c cons9tuent structure?
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 2
![Page 3: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/3.jpg)
Introduc9on
• Between-‐language comparison is complicated by several factors:
1. Language-‐specific syntac9c differences 2. Language-‐specific prosodic/eurhythmic
preferences 3. Language-‐specific intona9onal tunes
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 3
![Page 4: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/4.jpg)
Introduc9on
• In this talk: I discuss experimental work that compares two languages, (Connemara) Irish and (North American) English
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 4
![Page 5: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/5.jpg)
Introduc9on
• Main idea: control for factor 1 (language-‐specific syntac9c differences) to beXer assess the roles of factor 2 and (to some extent) factor 3 – Compare the prosody of VSO (basic transi9ve) sentences in Irish with VOO (ditransi9ve/double object construc9ons) in English.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 5
![Page 6: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/6.jpg)
Introduc9on
• Three types of prosodic cues were examined: – Dura9on – Loca9on and likelihood of prosodic pauses – F0 contours (pitch tracks) and overall pitch paXerns
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 6
![Page 7: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/7.jpg)
Introduc9on
• Main findings: – Speakers of the two languages behave remarkably similarly on all three prosodic measures in the two structures examined
– … but not in the way that we might expect given our assump9ons about the underlying syntac9c structure, giving rise to apparent syntax-‐prosody mismatches.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 7
![Page 8: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/8.jpg)
Introduc9on
• This suggests that we may need to re-‐evaluate how mismatches are integrated into our theory of syntax-‐prosody mapping, and ask the following ques9ons: – How do we diagnose when we have a mismatch, and when should we use prosodic evidence to revise our assump9ons about the underlying syntac9c structure?
– What types of prosodic paXerns are indica9ve of a universal paXern, and which are language-‐specific?
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 8
![Page 9: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/9.jpg)
Experiments: Design
![Page 10: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/10.jpg)
Experimental design
• 4 produc9on experiments: – Experiments 1 & 2: Connemara Irish – Experiments 3 & 4: North American English
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 10
![Page 11: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/11.jpg)
Experimental design
• Experiments looked a comparable syntac9c structures: VSO (basic transi9ve) sentences in Irish and VOO (double object) construc9ons in English.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 11
![Page 12: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/12.jpg)
Syntax: Irish
• Basic word order in Irish is VSO: Leanann Liam Ó Móráin Niall Ó Mearlaigh. follows Liam Ó Móráin Niall Ó Mearlaigh ‘Liam O’Moran follows Niall O’Marley.’
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 12
![Page 13: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/13.jpg)
Syntax: Irish
• Evidence from syntac9c cons9tuency tests suggest that subject and object form a cons9tuent in Irish, to the exclusion of the verb (McCloskey 1996, 2011). – Follow McCloskey (2011) in the assump9on that V moves to ΣP through head movement, S moves to Spec,TP, and O remains in VP.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 13
![Page 14: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/14.jpg)
Structure of an Irish transi9ve (VSO) sentence (McCloskey 1996, 2011)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 14
![Page 15: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/15.jpg)
• Crucially: S and O form a cons9tuent to the exclusion of V, ignoring category labels.
Syntax: Irish
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 15
![Page 16: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/16.jpg)
Syntax: English
• (Arguably) comparable structures in English: VOO (double object construc9ons) – Assump9on: pronominal subjects are prosodic cli9cs.
He brought Marvin their leXer.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 16
![Page 17: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/17.jpg)
Syntax: English
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 17
![Page 18: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/18.jpg)
Syntax: English
• Larson (1988) analysis: “The situa9on posited here for VP in English is analogous to the situa9on widely assumed for S in VSO languages” (Larson 1988: 344; also Dowty 1982, Jacobson 1987)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 18
![Page 19: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/19.jpg)
Syntax: Comparison
• In both languages: post-‐verbal arguments appear to form a cons9tuent to the exclusion of the verb.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 19
![Page 20: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/20.jpg)
Experiments: Methods and Materials
![Page 21: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/21.jpg)
Materials and design
• 4 experiments: VSO structures (Irish) and VOO structures (English)
Names Branching DPs
Connemara Irish Experiment 1 Experiment 2
N.A. English Experiment 3 Experiment 4
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 21
![Page 22: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/22.jpg)
Materials and design
• Each experiment: 2x2 design varying subject/object complexity
Branching object (B)
Non-‐branching object (N)
Branching subject (B) BB BN
Non-‐branching subject (N) NB NN
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 22
![Page 23: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/23.jpg)
Sample materials: Experiment 1 (Irish names)
‘Liam (O’Moran) follows Niall (O’Marley).’
Condi4on V S O
N1 N2 N3 N4
BB Leanann Liam Ó Móráin Niall Ó Mearlaigh
BN Leanann Liam Ó Móráin Niall
NB Leanann Liam Niall Ó Mearlaigh
NN Leanann Liam Niall
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 23
![Page 24: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/24.jpg)
Sample materials: Experiment 2 (Irish DPs)
‘Neasa/nice children pick (ripe) apples’
Condi4on V S O
N1 A2 N3 A4
BB Piocann pick.pres
páisG children
deasa nice
úlla apples
aibí ripe
BN Piocann páisG deasa úlla
NB Piocann Neasa úlla áibí
NN Piocann Neasa úlla
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 24
![Page 25: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/25.jpg)
Methods: Irish experiments • Experiment 1 (Irish names):
– Recorded at Trinity College Dublin in a sound-‐aXenuated booth – 6 bilingual na9ve speakers of Connemara Irish, living in Dublin at the 9me of recording
– 3 male, 3 female (ages 19-‐46) – 4 condi9ons x 6 items x 4-‐5 repe99ons of the experiment
• Experiment 2(Irish DPs): – Recorded in Carraroe, Ireland in a classroom at the Acadamh na hOllscolaíochta Gaeilge and in Dublin at Trinity College Dublin
– Part of a larger experiment – Analysis for 8 speakers out of 12 recorded (7 female, 1 male); ages 22-‐60
– 4 condi9ons x 12 items
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 25
![Page 26: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/26.jpg)
Sample materials: Experiment 3 (English names)
• Context: Sarah Thompson, the head of the department, is responsible for providing each professor with a research assistant.
Condi4on V O O
N1 N2 N3 N4
BB She alloKed Mary MacEwan Lauren O’Hara
BN She alloKed
Mary MacEwan Lauren
NB She alloKed
Mary Lauren O’Hara
NN She alloKed
Mary Lauren
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 26
![Page 27: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/27.jpg)
Sample materials: Experiment 4 (English DPs)
Condi4on V O O
N1 N2 N3 N4
BB He brought Marvin’s sister Mandy’s leKer
BN He brought
Marvin’s sister their leKer
NB He brought
Marvin Mandy’s leKer
NN He brought
Marvin their leKer
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 27
![Page 28: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/28.jpg)
Methods: English experiments
• Both experiments were recorded in McGill prosody lab – Experiment 3: 18 par9cipants x 4 condi9ons x 8 items
– Experiment 4: 16 par9cipants x 4 condi9ons x 8 items
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 28
![Page 29: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/29.jpg)
Methods
• Experiment 3 (English names): items presented with a neutral context
• Experiments 1, 2, and 4 (Irish names, Irish DPs, English DPs): items presented with no context
• Data were excluded for the following reasons: – Technical problems with the recording – An obvious disfluency – Obvious perceived emphasis on a par9cular word
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 29
![Page 30: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/30.jpg)
Analysis: All experiments
• Data were aligned using the prosodylab-‐aligner (Gorman et al. 2011) and analysed used Praat (Boersma & Weenink 2013) and R.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 30
![Page 31: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/31.jpg)
Background: Boundary Strength
![Page 32: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/32.jpg)
Boundary Strength
• Rela9ve boundary strength: theory that some prosodic boundaries are stronger than others. – To some extent this correlates with prosodic categories (PWd, Phonological Phrase, Intona9onal Phrase)
– But also applies to differences within categories (Ladd 1986, 1988; Kubozono 1989, 1992; Féry & Truckenbrodt 2005; Wagner 2005, 2010)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 32
![Page 33: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/33.jpg)
Background: Boundary Strength
• Dura4onal effects (including lengthening as well as pauses) are used to disambiguate syntac9c structures (Lehiste 1973, and others).
• Pre-‐boundary lengthening: increased dura9on of segments/words before a prosodic boundary.
• Rela4on to boundary strength: dura9onal effects increase with boundary strength (Price et al. 1991, Wightman et al. 1992, Wagner 2005).
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 33
![Page 34: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/34.jpg)
Background: Dura9onal effects • Example (Lehiste 1973): Boundary rank (| vs. ||) is determined by depth of embedding.
• Dura9on is longest before the strongest boundary.
1) [[Steve or Sam] and Bob] will come. Steve | or Sam || and Bob
2) [Steve or [Sam and Bob]] will come. Steve || or Sam | and Bob
• Boundary strength by rank: pipes (| vs. ||) • Pre-‐boundary lengthening: a vs. a (longest)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 34
![Page 35: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/35.jpg)
Predic9ons
• Predic9ons: dura9on and pauses – Words should have a rela9vely longer dura9on before a rela9vely strong prosodic boundary
– Pauses should be more likely at a rela9vely strong prosodic boundary
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 35
![Page 36: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/36.jpg)
Predic9ons: Boundary strength
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 36
Strong Weak Weak Strong
V | S || O V | O || O
V || S | O V || O | O
V [ S O ] V [ O O ]
(Syntax)
(Prosody)
![Page 37: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/37.jpg)
Experiments: Results
![Page 38: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/38.jpg)
Prosodic boundaries
• Three cues to prosodic boundaries will be considered in this talk: – Loca9on and likelihood of prosodic pauses – Rela9ve dura9on (pre-‐boundary lengthening) – Pitch contours and presence of tonal cues at prosodic boundaries
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 38
![Page 39: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/39.jpg)
A note on presenta9on • For ease of presenta9on, results are presented as branching/
non-‐branching condi9ons:
Condi4on V O O
N1 N2 N3 N4
BB She alloKed Mary MacEwan Lauren O’Hara
BN She alloKed
Mary MacEwan Lauren
NB She alloKed
Mary Lauren O’Hara
NN She alloKed
Mary Lauren
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 39
![Page 40: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/40.jpg)
A note on presenta9on • Resul9ng in a comparison of two condi9ons:
Condi4on V O O
N1 N2 N3 N4
Branching subject
She alloKed Mary MacEwan Lauren (O’Hara)
Non-‐branching subject
She alloKed
Mary Lauren (O’Hara)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 40
![Page 41: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/41.jpg)
1. Results: Pauses
• Measurements: – Pauses automa9cally iden9fied by prosodylab-‐aligner (Gorman et al. 2011)
– Measurement of dura9on of pause following words of interest
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 41
![Page 42: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/42.jpg)
Pauses are more common following the first argument
• Example: experiment 3 (English names)
Condi4on V O O
N1 N2 N3 N4
BB She alloKed Mary MacEwan Lauren O’Hara
BN She alloKed
Mary MacEwan Lauren
NB She alloKed
Mary Lauren O’Hara
NN She alloKed
Mary Lauren
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 42
![Page 43: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/43.jpg)
Pauses
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 43
V | N1 (N2) | N3 (N4)
branching subject non-branching subject
0.00
0.25
0.50
0.75
1.00
1-V 2-N1-subj 3-N2-subj 1-V 2-N1-subj 3-N2-subjWord
count pause
no
yes
Experiment 1: Irish namesbranching subject non-branching subject
0.00
0.25
0.50
0.75
1.00
1-V 2-N1-subj 3-N2-subj 1-V 2-N1-subj 3-N2-subjWord
count pause
no
yes
Experiment 2: Irish DPs
branching subject non-branching subject
0.00
0.25
0.50
0.75
1.00
1-V 2-N1-subj 3-N2-subj 1-V 2-N1-subj 3-N2-subjWord
count pause
no
yes
Experiment 3: English namesbranching subject non-branching subject
0.00
0.25
0.50
0.75
1.00
1-V 2-N1-subj 3-N2-subj 1-V 2-N1-subj 3-N2-subjWord
count pause
no
yes
Experiment 4: English DPs
![Page 44: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/44.jpg)
Results: Pauses
• Pauses are more common following the first post-‐verbal argument than following the verb – Par9cularly true of Irish, where pauses are more frequent
– Also observed in BenneX (2008) for Irish (natural storytelling speech in Donegal Irish)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 44
![Page 45: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/45.jpg)
Results: Pauses
• Implica9on: there is a stronger prosodic boundary following the first post-‐verbal argument than following the verb:
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 45
Weak Strong
V | S || O V | O || O
![Page 46: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/46.jpg)
2. Results: Dura9on
• Predic9on: We should see pre-‐boundary lengthening before a strong prosodic boundary
• Dura9on (roughly) normalized by dividing raw word dura9on by number of phonemes
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 46
![Page 47: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/47.jpg)
B. Dura9on of V vs. S/DO
• Example: experiment 3 (English names):
Condi4on V O O
N1 N2 N3 N4
BB She alloKed Mary MacEwan Lauren O’Hara
BN She alloKed
Mary MacEwan Lauren
NB She alloKed
Mary Lauren O’Hara
NN She alloKed
Mary Lauren
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 47
![Page 48: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/48.jpg)
Dura9on: lengthening on right edge of first argument
branching subject non-branching subject
0.000
0.025
0.050
0.075
1-V 2-N1-subj3-N2-subj 1-V 2-N1-subj3-N2-subj
Nor
mal
ized
dur
atio
n (s
)
Subject branchingnessbranching subject
non-branching subject
Experiment 1: Irish namesbranching subject non-branching subject
0.00
0.03
0.06
0.09
1-V 2-N1-subj3-N2-subj 1-V 2-N1-subj3-N2-subj
Nor
mal
ized
dur
atio
n (s
)
Subject branchingnessbranching subject
non-branching subject
Experiment 2: Irish DPs
branching subject non-branching subject
0.000
0.025
0.050
0.075
1-V 2-N1-subj3-N2-subj 1-V 2-N1-subj3-N2-subj
Nor
mal
ized
dur
atio
n (s
)
Subject branchingnessbranching subject
non-branching subject
Experiment 3: English namesbranching subject non-branching subject
0.0
0.1
0.2
0.3
0.4
1-V 2-N1-subj3-N2-subj 1-V 2-N1-subj3-N2-subj
Dur
atio
n (s
)
Subject branchingnessbranching subject
non-branching subject
Experiment 4: English DPs
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 48
V | N1 (N2) | N3 (N4)
![Page 49: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/49.jpg)
Results: dura9on
• Results are similar for both languages in all four experiments: – Lengthening at the right edge of the subject (Irish) or first object (English)
– The verb is not lengthened rela9ve to the subject
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 49
![Page 50: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/50.jpg)
3. Results: F0
• While F0 analysis is ongoing, speakers seem to employ similar techniques to achieve a “neutral” pronuncia9on of these types of sentences.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 50
![Page 51: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/51.jpg)
Results: F0
• Specifically: – Rise in F0 on the verb (“LH” accent) – Fall in F0 on the rightmost word of each argument (“HL” accent)
– (Op9onal) rise in F0 on the leumost word of the first argument (“LH” accent)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 51
![Page 52: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/52.jpg)
Results: F0
• These are comparable to the paXerns described for neutral (all new) sentences Connemara Irish in Elfner (2012), where LH and HL accents provide evidence for prosodic “bracke9ng”: – LH indicate the leu edge of a (non-‐minimal) prosodic phrase
– HL indicates the right edge of a prosodic phrase
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 52
![Page 53: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/53.jpg)
Results: F0
• To illustrate: – Sample pitch tracks from each language – Chart showing mean difference in F0 values (rise in F0 vs. fall in F0) by word in two of four experiments (experiments 1 and 3).
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 53
![Page 54: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/54.jpg)
Sample pitch tracks
• Common paXern: rise-‐fall-‐fall
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 54
V | S | O
LH HL HL
V | O | O
![Page 55: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/55.jpg)
Sample pitch tracks
• Note: not all par9cipants exhibited these paXerns (par9cularly in English), but a subset of speakers in each language and in each experiment employ this paXern.
• This suggests that this is a possible paXern in both languages.
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 55
![Page 56: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/56.jpg)
Experiment 1 (Irish names)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 56
sil BUAILEANN MAIRIN sp ROININ sp NI MHEARA
hit.pres Mairín Roinín Ní Mheára
Mairín hits Roinín Ní Mheara
100
350
150
200
250
300
Pitc
h (H
z)
Time (s)0 2.4
HL LH
HL
![Page 57: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/57.jpg)
Experiment 2 (Irish DP)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 57
sil BRISEANN MUINTEOIR sp PLATAI NUA sp
break.pres teacher plates new
A teacher breaks new plates.
100
350
150
200
250
300
Pitc
h (H
z)
Time (s)0 1.727
HL LH
HL
![Page 58: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/58.jpg)
Experiment 3 (English names)
sil THEY LENT MOIRA NORA OCONNOR sp
100
350
150
200
250
300
Pitc
h (H
z)
Time (s)0 1.646
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 58
HL LH
HL
![Page 59: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/59.jpg)
Experiment 4 (English DPs)
sil HE PROMISED LOUIE NINA’S DETAILS THAT
100
350
150
200
250
300
Pitc
h (H
z)
Time (s)0 1.818
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 59
HL
LH
(HL)
![Page 60: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/60.jpg)
Quan9ta9ve analysis: F0 rises/falls
• Method: – Measurements for mean F0 values in third quadrant and first quadrant of the target word
Third-‐first = mean 3rd quadrant – mean 1st quadrant
Third-‐first pitch accent type
>0 Hz LH (‘rise’) < 0 Hz HL (‘fall’)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 60
![Page 61: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/61.jpg)
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4
sil LEANANN LIAM O MORAIN sp NIALL sp SIN E sp sil
100
350
150
200
250
300
Pitc
h (H
z)
Time (s)0 3.03
Illustra9on: Third-‐first
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 61
![Page 62: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/62.jpg)
Mean third-‐first
branching subject non-branching subject
-10
0
10
1-V 2-N1-subj 3-N2-subj 1-V 2-N1-subj 3-N2-subjSubject branchingness
Third
-firs
t (H
z)
Subject branchingnessbranching subject
non-branching subject
Experiment 1: Irish names
branching subject non-branching subject
-30
-20
-10
0
10
1-V 2-N1-subj 3-N2-subj 1-V 2-N1-subj 3-N2-subjSubject branchingness
Third
-firs
t (H
z)
Subject branchingnessbranching subject
non-branching subject
Experiment 3: English names
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 62
V | N1 (N2) | N3 (N4)
![Page 63: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/63.jpg)
Summary: F0
• While more detailed analysis is necessary, it is intriguing that speakers from both languages have the op9on of employing a similar pitch contour: – Rises (LH) on the verb – Falls (HL) on the rightmost word of subject (and object)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 63
![Page 64: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/64.jpg)
Discussion
![Page 65: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/65.jpg)
Summary
• Par9cipants in all four experiments behaved similarly: – Pre-‐boundary lengthening is observed on the post-‐verbal argument (subject/first object) but not on the verb
– Pauses are more common following the post-‐verbal argument (especially in Irish) and rarely occur following the verb
– (Op9onal) rise-‐fall-‐fall pitch contour involving rise (LH) on verb and falls (HL) on right edge of each post-‐verbal argument
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 65
![Page 66: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/66.jpg)
Summary
• Schema9cally:
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 66
V | S | O
LH HL HL
Pre-‐boundary lengthening Prosodic pauses
V | O | O
![Page 67: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/67.jpg)
Analysis: Prosodic boundary strength
• These results suggest that prosodic boundary following the first post-‐verbal argument is stronger than the boundary following the verb:
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 67
V | S || O V | O || O
![Page 68: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/68.jpg)
A mismatch?
• What does this tell us about the underlying syntac9c structure?
• To what extent can we (or should we) use prosody/boundary strength as a test for cons9tuency?
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 68
![Page 69: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/69.jpg)
A mismatch?
• For both structures, the two post-‐verbal arguments were predicted to form a cons9tuent to the exclusion of the verb:
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 69
Irish VSO English VOO
![Page 70: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/70.jpg)
A mismatch?
• Based on boundary strength alone, it is therefore surprising that there is a stronger prosodic boundary between the two post-‐verbal arguments than following the verb.
• But how do we know when the prosodic parse is truly a mismatch with the syntac9c structure?
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 70
![Page 71: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/71.jpg)
A mismatch?
• HYPOTHESIS A: V forms a cons9tuent with the first argument, an apparent mismatch with the syntax:
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 71
HL HL
HL HL
LH LH
![Page 72: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/72.jpg)
A mismatch?
HYPOTHESIS B: Only the right edges of phrases are marked by lengthening and pauses (there is no right phrase edge following V because V is not phrasal): preserves syntac9c assump9ons
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 72
LH HL HL
LH HL HL
![Page 73: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/73.jpg)
A mismatch?
• How can we decide between these two possible accounts?
• And how do we know whether this confirms our hypotheses about the underlying structure, or goes against them? – Whether or not a prosodic parse is a mismatch depends on our syntac9c assump9ons…
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 73
![Page 74: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/74.jpg)
A mismatch?
• However: If we need to derive the mismatch from the syntax using edge-‐marking, can (or should) we just derive this from the syntax itself? – i.e. Can we use the prosodic evidence as an indica9on that the syntac9c story is more complicated/different than it seems?
– Are apparent mismatches a sign that we should revise our syntac9c assump9ons?
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 74
![Page 75: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/75.jpg)
Implica9on
• Implica9on: VSO/VOO structures are underlyingly [VS]O/[VO]O:
• The pressure would then be on the syntac9c component to derive these structures…
• (to be discussed in Michael Wagner’s talk…)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 75
HL HL
HL HL
LH LH
![Page 76: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/76.jpg)
Conclusion
• The prosodic similarity between the Irish and English structures suggests that there is an important role for structure, even in the case of apparent mismatches. – While further typological research is needed, it seems unlikely that this is due to coincidence, and hence unlikely to be due to language-‐specific factors.
– … it seems that Tagalog may also show some of these paXerns, at least in terms of pitch paXerns (Norvin Richards’ and Joey Sabbagh’s talks tomorrow…)
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 76
![Page 77: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/77.jpg)
Conclusion
• However, this raises the ques9on of how much we want to rely on prosody/boundary strength as a test for cons9tuency– a theore9cal ques9on without a clear answer. – … But maybe we’ll know more at the end of the workshop!
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 77
![Page 78: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/78.jpg)
Thank you!
![Page 79: Syntax’prosody-mismatches-in-Irish- and-English-verb](https://reader034.vdocument.in/reader034/viewer/2022042601/62628b1fa5860e337a0d052d/html5/thumbnails/79.jpg)
Acknowledgements • Many thanks to the Irish and English speakers who par9cipated in the
experiments. • Ailbhe Ní Chasaide and Maria O’Reilly at Trinity College Dublin and Treasa
Uí Lorcáin at the Acadamh na hOllscolaíochta Gaeilge for help recrui9ng par9cipants and for arranging recording space.
• Michael Wagner, Jim McCloskey, the McGill Syntax-‐Phonology Reading Group and the McGill prosody lab for comments and assistance.
• McGill prosody lab RAs for help running and processing the English experiments: David Fleischer, Lauren Garfinkle, Thea Knowles, Elise McClay, Erin Olson, Symon Jory Stevens-‐Guille
• SSHRC postdoctoral fellowship (756-‐2011-‐0285) • NSF grant “Effects of Syntac9c Cons9tuency on the Phonology and
Phone9cs of Tone” (BCS-‐1147083) to Elisabeth Selkirk • SSHRC Standard Research Grant on Rela9ve Boundary Strength to Michael
Wagner, as well as funding from the CFI/CRC program
2014-‐05-‐08 ETI3: Prosody and Cons9tuent Structure 79