speakers and hearers use prosody to disambiguate scopally...
TRANSCRIPT
Speakers and hearers use prosody to disambiguate scopally ambiguous sentences
Kristen Syrett, Georgia Simon, Kirsten Nisula
Background
Production Experiment
Perception Experiments
Selected References Baltazani 2002 The prosodic structure of quan4fica4onal sentences in Greek; Fodor 2002 Psycholinguis4cs cannot escape prosody; Hirschberg & Avesani 2000 Prosodic disambigua4on in English and Italian; Kadmon & Roberts 1986 Prosody and scope: The role of discourse structure; Jackendoff 1972 Seman&c interpreta&on in genera&ve grammar; Koizumi 2009 Processing the not-‐because ambiguity in English: The role of pragma&cs and prosody; Ladd 1980 Intona4on and Grammar; McMahon, Lidz, & Pierrehumbert 2004 Suprasegmental cues to meaning in child-‐directed speech
Conclusions and Implications 1. Speakers can distinguish between interpretations using surface cues, but these cues (a) are
highly variable within and across speakers & contexts, (b) depend on the quantifier and its position, (c) are not restricted to sentence-final contour, (d) are not linked to scope alone.
2. Hearers reliably recruit such cues to arrive at the speaker’s intended interpretation, in most, but not all cases. Success depends on the quantifier, speaker, and scope.
3. Psycholinguistic (including acquisition) studies that present participants with scopally ambiguous sentences should control for prosody, taking into account these patterns. It is not enough to say that prosody was “kept neutral” or was delivered “naturally.”
NELS 2012
Purpose Determine whether certain scopal rela&ons elicit a par&cular prosodic contour, and whether any surface-‐level signature varies with the context.
Theoretical Jackendoff (1972): prosody is 4ed to scopal rela4ons ‘A accent’ (fall): ¬ is part of the presupposi&on, not focus; ‘B accent’ (non-‐fall): ¬ is associated with the focus (asser&on) Objec4ons: Liberman & Sag 1974; Ladd 1980; Ward & Hirschberg 1985; Kadmon & Roberts 1986 Emerging picture: While it is possible that a fall-‐rise contour may (strongly) favor a ‘¬ wide scope’ reading, this contour is neither necessary nor sufficient, and should vary with the context (informa4on structure, QUD, speaker knowledge of salient scalar alterna4ves).
ALL the men didn’t go.
Participants 26 undergraduates (data from 19 used)
Procedure Par4cipants were recorded reading a series of short passages:
Stimuli Target sentences were all ambiguous, with controls clearly able to be disambiguated prosodically.
Sentences were presented in a brief discourse context favoring one possible interpreta4on.
Purpose Determine whether in the best case scenario, hearers can match a prosodic rendi&on of a sentence to its intended interpreta&on.
Experiment 1 Participants
44 undergraduates
Stimuli and Procedure 24 minimal pairs selected from 4 speakers (3 from produc4on experiment, 1 experimenter) 6 pairs each:
2 all, ¬; 1 ¬, many/most; 2 ¬, because; 1 only/even
context 1 (M > ¬) context 2 (¬ > M)
Experiment 2 Participants
37 undergraduates
Stimuli and Procedure 18 minimal pairs selected from 3 speakers (2 F from previous experiment, 1 experimenter) 6 pairs each:
2 all, ¬; 1 ¬, many/most; 2 ¬, because; 1 only/even
context 1 (M > ¬) context 2 (¬ > M)
Acknowledgments This work benefitted from a Rutgers startup grant to K. Syrett and funding from the Aresty Research Center at Rutgers University, as well as technical assistance from Paul de Lacy and the Rutgers Phonology lab, and discussions with Shigeto Kawahara.
Corresponding author: Kristen Syrett [email protected]
Experimental Speakers and hearers use prosodic cues to disambiguate a wide variety of syntac4c structures and presupposi4on/focus dis4nc4ons (Speer, et al, 1993; Price et al. 1991), as well as non-‐quan4fica4onal ¬/because ambiguity (Cooper & Paccia-‐Cooper 1986; Hirschberg & Avesani 1997, 2000; Koizumi 2009), although there is considerable varia4on in produc4ve strategies. Greek speakers and hearers do recruit prosodic cues for scopal disambigua4on (Baltazani 2002/3), but previous studies in English leave the ques4on open (McMahon et al. 2004; Jackson 2006).
Abstract Since Jackendoff (1972) first claimed that sentences such as (1) can be reliably disambiguated via a sentence-‐final contour, (2), various researchers have ques4oned this strength of this rela4onship and the nature of the link between prosody and sentence meaning, arguing for a pragma4c account. (1) All the men didn’t go. (2) a. ∀ > ¬ (none) ‘A accent’
b. ¬ > ∀ (not all) ‘B accent’ To date, systema4c evidence bearing on this ques4on has been lacking. We present a set of produc4on/percep4on experiments designed to inves4gate whether speakers and hearers recruit auditory cues (including, but not limited to sentence-‐final contour) to disambiguate such sentences. While there is considerable variability in speaker produc4on, there are surface-‐level cues to interpreta4on. Moreover, hearers successfully recruit these cues to arrive at the correct interpreta4on. We argue that psycholinguis4c studies inves4ga4ng par4cipants’ ability to access mul4ple interpreta4ons of scopally ambiguous sentences should carefully control for prosody.
Sample sentences test items (n=28) type All the magnolias won’t bloom. ∀, ¬ Liam doesn’t know many alumni. ¬, many Neil doesn’t enjoy most musicals. ¬, most
control items (n=28) type They’re not late because of his driving. ¬, because Warren only likes the Orioles. focus She even painted the garage. focus Alan punched Owen and then he kicked him. pronominal ref.
read passage silently answered
ques4on re-‐read passage aloud, recorded Sentences blocked, pseudorandomized
Sonorance and presence of con&nua&on sentence controlled for Analysis: comprehension scores of 1; targets excised in Praat
A few years ago, the township decided to plant magnolia saplings to line a path through the park. The saplings on the north side were planted mainly in sand and haven’t been geLng nearly enough nutrients. However, the soil near the south side is rich, and the magnolias are thriving there. All the magnolias won’t bloom. But I bet the ones on the south side will.
Results Analysis 1: Sentence-‐Final Contour
Analysis 2: AcousLc Analysis 5 acousLc measures -‐ quanLfier, sentence-‐final word max F0, max F0 loca4on, F0 st. dev., word dura4on all: most delayed max F0 on quan4fier and shortest sentence-‐final word in context 4, longest sentence-‐final word in context 3; many/most: shortest quan4fier AND longest sentence-‐final word in context 1
quantifier! context- scope-- scalar-alternative- %-falling-contour--
all- 1- all->-¬- quantity- 93.4-- 2- ¬->-all- which- 89.1-- 3- all->-¬- which- 71.1-- 4- ¬->-all! quantity! 95.5-many/most- 1- M->-¬- quantity- 91.3-- 2- ¬->-M! which- 65.1-- 3- ¬->-M- which-- 63.0-
!
Which sentence should follow? a. He really has to make more connec4ons. b. But the ones he knows are well
established in the community.
Speakers blocked, minimal pairs in sub-‐blocks Sentences presented via headphones in lab using Superlab Sentence presenta&on within blocks randomized Analysis: comprehension scores of 1, RTs <15s
Blocking and presenta&on via Superlab as before Analysis: comprehension scores of 1, RTs <3.5s
Results Experiment 1
Experiment 2
Neil is a fan of serious theatre. I bought 4ckets to the musical "Oklahoma!" for him as a gio. My friend Adam was concerned about the choice. I realized he was right. <<Neil doesn’t enjoy most musicals.>> He thinks they are very cheesy.
A B
Liam doesn’t know many alumni.
All % significantly above chance 4 x 2 x 3 ANOVA (speaker, scope of ¬, lexical item): ME of speaker, ME of scope of ¬, significant interac&ons
x3
All % significantly above chance, with excep&on of ¬ > all 3 x 2 x 3 ANOVA (speaker, scope of ¬, lexical item): ME of scope of ¬, significant ¬ x lexical item interac&on