minimal pairs in sof
TRANSCRIPT
When Minimal Pairs
Meet SOFChelsea Fink
Overview
• Motivation: Previous work with focus shows that in its
second-occurrence, a word generally lacks the pitch
accent found in the first-occurrence. However, other
phonetic correlates (i.e. duration) can indicate focus on
the SO.
• Question: What affect (if any) does SO have on
minimal pairs?
• Method: Measure the VOT of minimal pairs in FO and
SO
• Results: VOT actually got shorter in SOF
• Conclusion: Test this out with more subjects, and see if
this is a consistent result
Background
• Focus in Phonology
• Hayes and Lahiri (1991): Looked at Bengali and found
that focus plays a role in determining the structure of
phonological phrases.
• Kenesei & Vogel (1995): Expanded on H&L work with
Bengali, and also looked at several other languages
(English, Italian, etc.).
• If an element in a sentence is focused, it either interrupts
the phonological rules, or causes them to extend beyond
their usual limits.
• Confirmed what H&L stated: Phonological Phrases are
the domain for focus.
Background
• The Semantics
• What is second-occurrence focus? “An expression
that is in the scope of a focus-sensitive operator, is the
semantic focus of that operator, and which is a repeat of
an earlier focused occurrence” (Beaver et al. 2007)
• Focus-sensitive operators, such as only or even, need an
element that is prosodically prominent within their scope
(Partee 1999)
• If that is the case, then the SO of an element should
have the same prominence as its FO. (Which is doesn’t.)
• Why are semanticists so interested in the phonology of
SOF?
Background• The Semantics: Continued…
• According to Partee, if these second-occurrences are
able to lack some sort of prosodic prominence (such as
pitch accent), it would force semanticists to accept one of
two options:
• 1. Words like only can sometimes be associated with
focus, and sometimes not. This is the worst scenario
• 2. Take a pragmatic approach, and deal with items in
their context. Sometimes, items lack pitch accent
because of their surroundings. (Ex: “There is no
sunshine” doesn’t mean that the entire planet is blacked
out.)
Motivation
• Beaver et al. (2007)
• Studied SOF and found that phrasal stress is key in
marking focus.
• Pitch accent is optional on items in SOF
• Phrasal stress is still there when pitch accent is not.
• When pitch accent is suppressed in SO, other elements,
such as duration and vowel quality can help determine
where the focus is.
Hypothesis
• All of this talk about SOF lacking a pitch accent has me
wondering…
• How does VOT compare for minimal pairs in FO and
SO?
• Because focus is generally determined by pitch accent, it
is common to look at peak F0, F0 range, duration, and
intensity. In the SO, it’s common to rely on longer
duration and higher intensity.
• That being said, I anticipate that VOT would lengthen on
minimal pairs in their SO.
Method
• Subject: 25 y/o male with English as his L1
• Stimuli: 4 sets of minimal pairs – zoo/sue, time/dime,
pun/bun, beach/peach
• The 8 words were presented randomly in a list of 25
words. Subject was instructed to say each word two
times before moving on to the next one.
• This list was presented first, in order to get an isolated
recording of the minimal pairs.
Methodtime
peach
bun
sue
pun
dime
Method• After reading the list, the subject read through a mixed
list of sentences and discourse sets containing the min pairs. Each sentence or set was read twice before moving on to the next one.
• Example:
• 1a. Where did John go?
• 1b. John went to the beach party.
• 2a. Julie said she wanted to see the bun book, but James only brought a pun book.
• 2b. Even Josh only brought a pun book.
• Discourse like 2a & 2b were set up in a way that the operator in the subject NP of 2b “leads the sentence to be uttered with a nuclear pitch accent in the initial NP, while the fact that the post-nuclear material is repeated seems to lead to an unaccented realization” (Beaver 2007).
Analysis
• Used waveforms and spectrograms to determine the
VOT for each of the words.
• Environments: Isolated, Embedded (Q&A), FOF, and
SOF
• Took the averages of the 1st and 2nd recording from all of
the environments.
Results: Isolated
-190
216
83
-15
93
-11
74
-11
-250
-200
-150
-100
-50
0
50
100
150
200
250
zoo
sue
time
dime
pun
bun
peach
beach
Results: FOF vs SOF
-124
190
82
-14
-100
149
80
-15
-150
-100
-50
0
50
100
150
200
250
FOF
SOFzoo
sue time
dime
VOT got shorter on the SO!!!
Results: FOF vs SOF
97
-11
75
-12
69
-13
70
-11-20
0
20
40
60
80
100
120
pun bun peach beach
FOF
SOF
Results:
FOF, SOF, Iso, Embed
92
-19
68
-10
97
-11
75
-12
69
-13
70
-11
93
-11
74
-11
-30
-10
10
30
50
70
90
110
EMB
FOF
SOF
ISO
pun bun peach beach
Results:
FOF, SOF, Iso, Embed
-131
184
85
-17
-124
190
82
-14
-100
149
80
-15
-190
216
83
-15
-200-175-150-125-100
-75-50-25
0255075
100125150175200225
EMB
FOF
SOF
ISO
zoo
sue time
dime
Results
• After seeing that the VOT got shorter, I went back and
looked at the absolute duration for the minimal pairs in
first and second-occurrence…
• peach: FO = 418 ms SO = 292 ms
• beach: FO = 310 ms SO = 338 ms
• time: FO = 339 ms SO = 325 ms
• dime: FO = 256 ms SO = 294 ms
• pun: FO = 339 ms SO = 278 ms
• bun: FO = 308 ms SO = 262 ms
• sue: FO = 345 ms SO = 292 ms
• zoo: FO = 291 ms SO = 266 ms
Conclusions
• As mentioned earlier, phrasal stress is often determined
with F0 peak and range, duration, and intensity. In the
lack of a pitch accent, duration and intensity often
increase…
• Because of that, I wouldn’t anticipate that VOT alone
could determine whether or not an item is focused, but I
figured we could at least draw some conclusions on what
would happen to VOT in SOF. (Clearly, I missed the
mark on that one)
• What to do differently next time: get more subjects! I
would be interested in seeing if the shorter VOT is a
common thing, or if this one subject just happened to be
consistently faster.