Modeling probabilistic reduction with NDL
25.01.23 - Diversity in Linguistics
Anna Stein
anna.stein@hhu.de
Modeling probabilistic reduction with NDL
Index
WHAT IS PROBABILISTIC REDUCTION?
Probabilistic
-
More predictable words become shorter(Lohmann 2018)
- time is shorter than thyme
- air is shorter than heir
Reduction
-
Predictability in context affects word duration (Seyfarth 2014)
- 'a cup of tea' vs. 'a barrel of tea'
Probabilistic reduction
- Predictability affects syllable and segment duration (Aylett & Turk 2004)
Research question
syllable
word duration
context
segment
arc
flat
Research question
What is the relative contribution of a word’s segment and syllable domain and its adjacent context to word duration (reduction)?
syllable
word duration
context
segment
naive discriminative learning
naive discriminative learning
- NDL has previously been used to model duration effects (Tucker et al. 2019, Tomaschek et al. 2019)
- Cues included context, syllables, segments
- They did not explicitly compare their prediction for word duration
- NDL could validate whether these domains play a role in speech production and to what degree
- Conversational speech data from the Buckeye corpus
NDL Model
'fat cat'
| Context cues | Syllable cues | Segment cues | Outcome |
|---|---|---|---|
| -, cat | fæt | f, æ, t | fat |
| fat , - | kæt | kæt | cat |
NDL predictors
| arc | cat | |
|---|---|---|
| fat | -- | + |
| big | + | - |
| ɑrk | + | - |
| kæt | - | + |
| k | + | + |
| æ | - | + |
| t | -- | + |
| r | + | - |
| ɑ | + | - |
Specific:
- Context
- Syllables
- Segments
Global:
- Segments
- Syllables
- Context
Statistical analysis
Word duration ~
Independent variables:
- Specific context, segments, syllables
- Global context, segments, syllables
Control variables:
- Segment, syllable count
- Speaker, interviewer gender
- Speaker age
- Part of speech
- Speech rate
Random effects:
- Speaker, Word
statistical analysis
Word duration ~
ActivationContext + ActivationSyllables+ ActivationSegment + PriorContext + PriorSyllables +
(1 | Speaker) + (1 | Word) + control variables
Results
Results
Results
Results
Results
Discussion
What is the relative contribution of a word’s segment and syllable domain and its adjacent context to word duration (reduction)?
Research question:
Not all domains seem to play a role in speech production
- The syllable and context domain are mixed
- Both are stronger than the segment domain
- The segment domain has the least contribution
references
Aylett, M., & Turk, A. (2004). The smooth signal redundancy hypothesis: A functional explanation for relationships between redundancy, prosodic prominence, and duration in spontaneous speech. Language and speech, 47(1), 31-56.
Lohmann, A. (2018). Cut (n) and cut (v) are not homophones: Lemma frequency affects the duration of noun–verb conversion pairs. Journal of Linguistics, 54(4), 753-777.
Seyfarth, S. (2014). Word informativity influences acoustic duration: Effects of contextual predictability on lexical representation. Cognition, 133(1), 140-155.
Tucker, B. V., Sims, M., & Baayen, R. H. (2019). Opposing forces on acoustic duration.
Tomaschek, F., Plag, I., Ernestus, M., & Baayen, R. H. (2021). Phonetic effects of morphology and context: Modeling the duration of word-final S in English with naïve discriminative learning. Journal of Linguistics, 57(1), 123-161.
Picture sources
https://3.bp.blogspot.com/-xPGtKecKSyg/VyHvDcNkRcI/AAAAAAAA6Gw/pMMaFXoXoBgEqsI9PTQPKhl7HhBUjTzeACLcB/s800/animal_rat_black.png
https://1.bp.blogspot.com/-23RzpcbS5R4/XXXOrXeT1kI/AAAAAAABUxQ/KpvTE1WQ0cQ2XZXQTmAfYbruwff7XoI5wCLcBGAs/s1600/pet_cat_oddeye_black.png
Thank you for your attention!
https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcSeS7Ico9ZJP7cMAlBbMtNWDa7M44kSy8uBn0vx69TmVbiaI0U&s
naive discriminative learning
1
2
1
3
1
2
3
--
--
++
--
+
+
+
--
--
DivLing Talk
By ansost22
