Generalizations are driven by semantics and constrained by - - PowerPoint PPT Presentation

▶
generalizations are driven by semantics and constrained
SMART_READER_LITE
LIVE PREVIEW

Generalizations are driven by semantics and constrained by - - PowerPoint PPT Presentation

Sep 25, 2022 106 likes •363 views

Generalizations are driven by semantics and constrained by statistical preemption New evidence from artificial language experiments Florent Perek & Adele Goldberg University of Birmingham & Princeton University Generalizations Previous

slide-1
SLIDE 1

Generalizations are driven by semantics and constrained by statistical preemption

New evidence from artificial language experiments Florent Perek & Adele Goldberg

University of Birmingham & Princeton University

slide-2
SLIDE 2

Generalizing beyond the input

  • Learning a language = generalizing beyond the input
  • For instance, using verbs in novel ways

It meeked (witnessed form) She meeked it (generalized form) (Naigles 1990; Fisher et al. 1991;

Gertner et al. 2006; Fisher et al. 2010; Yuan et al. 2012; Akhtar 1999; Tomasello 2000)

  • Overgeneralization errors (e.g., Bowerman 1990)

?? Don’t giggle me

2

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-3
SLIDE 3

Generalizing beyond the input

  • When and why do speakers generalize beyond their

input? And when and why do they not?

  • What aspects of the input are relevant?

– Does language learning only consist of gleaning statistical regularities in the input? – What about the role of the function of constructions?

3

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-4
SLIDE 4

Artificial language learning studies

(e.g., Casenhiser & Goldberg 2004; Finley & Badecker 2009; Folia et al. 2010; Fedzechkina et al. 2010; Hudson Kam & Newport 2005; Wonnacott et al. 2008)

  • Participants exposed to novel <utterance, video scene> pairs
  • Statistical structure of input is manipulated
  • To test the role of statistics in language learning

4

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-5
SLIDE 5

Experiment 1

  • Two word order constructions: APV and PAV, a suffix –po on

the patient argument

the panda the pig-po mooped (APV: Agent Patient-po Verb) the pig-po the panda mooped (PAV: Patient-po Agent Verb) ‘the panda mooped the pig’

  • Six novel verbs (e.g., glim, moop, wub) referring to transitive actions

(e.g., ‘punch’, ‘push’, ‘head-butt’)

  • Two test conditions

– Lexicalist: 3 APV-only verbs, 3 PAV-only verbs – Alternating: 2 APV-only, 2 PAV-only, 2 alternating verbs

5

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-6
SLIDE 6
  • Constructions are rarely synonymous in natural

languages (cf. Bolinger 1968; Givon 1979; Goldberg 1995)

  • Our two constructions differ in the intensity of the

effect on the patient

– APV: strong effect: the patient rapidly moves across the screen and out of the scene with dramatic gestures – PAV: weak effect: the patient hardly moves, with similar but less ample gestures

6

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

Experiment 1

slide-7
SLIDE 7

Example of APV exposure pair

7

the monkey the panda-po glimmed

Generalizations Previous work Exp.1: Generalization Exp.2:Preemption Conclusion

slide-8
SLIDE 8

Example of PAV exposure pair

8

the panda-po the monkey glimmed

Generalizations Previous work Exp.1: Generalization Exp.2:Preemption Conclusion

slide-9
SLIDE 9
  • Participants: 24 Princeton undergraduates (18-22, 16 female)
  • Exposure (2 days)

– 36 sentence-scene pairs, each verb used 6 times – Participants were asked to repeat each sentence

  • Sentence production task

– Participants described new scenes; verb was given – Each of the 6 verbs presented 4x, twice each with video showing strong and weak effect – Two new novel verbs, not witnessed in the input

9

Generalizations Previous work Exp.1: Generalization Exp.2:Preemption Conclusion

slide-10
SLIDE 10

Example of production trial (strong effect)

10

what happened here? (pilked) the pig the cat-po pilked

  • r

the cat-po the pig pilked

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-11
SLIDE 11

Example of production trial (weak effect)

11

what happened here? (pilked) the pig the cat-po pilked

  • r

the cat-po the pig pilked

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-12
SLIDE 12

To what extent do speakers generalize constructions to unattested verbs?

  • Hypothetical data: conservative, verb-based behavior

12

0.2 0.4 0.6 0.8 1 Strong effec t Weak effect 0.2 0.4 0.6 0.8 1 Strong effec t Weak effect APV production PAV production APV-only verbs in input PAV-only verbs in input Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-13
SLIDE 13
  • Hypothetical data: full generalization across verbs

13

0.2 0.4 0.6 0.8 1 Strong effec t Weak effect 0.2 0.4 0.6 0.8 1 Strong effec t Weak effect APV production PAV production APV-only verbs in input PAV-only verbs in input Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

To what extent do speakers generalize constructions to unattested verbs?

slide-14
SLIDE 14

Experiment 1: Results

Lexicalist condition: no alternating verbs Verb-based conservativeness Full generalization

14

Strong effect Weak effect

SOV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

OSV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Alternating

Strong effect Weak effect

SOV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

OSV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

SOV OSV

Strong effect Weak effect

novel verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

novel verbs

0.0 0.2 0.4 0.6 0.8 1.0

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

PAV-only verbs APV-only verbs PAV-only verbs APV-only verbs

Alternating condition: two alternating verbs

slide-15
SLIDE 15

Mixed effects logistic regression

(to predict the probability of producing APV)

  • Strong tendency to produce APV when the effect is strong (β =

3.4756, p < 0.0001)

  • APV-only verbs tend to be used (slightly) more often with APV

compared to novel verbs (β = 0.8111, p = 0.0013 )

  • Interaction between Condition and Effect: the effect of the

functional difference is weaker in the lexicalist condition (β = - 1.1113, p = 0.0085)

15

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-16
SLIDE 16

Summary of Experiment 1

  • Tendency for participants to generalize (using verbs in

the contextually appropriate constructions)

  • They may ignore usage of individual verbs
  • Linguistic function can overcome statistical information

in the choice of construction

  • Contrasts with Wonnacott et al.’s (2008) results with

synonymous constructions; see also Perek & Goldberg (2015); Thothathiri & Rattinger (2016)

16

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-17
SLIDE 17

Discussion

  • The meaning of constructions is a source of productivity in

natural language (e.g., Goldberg 1995)

  • But constructional generalizations are typically restricted, e.g.,

*Explain me this. (Explain this to me)

  • Statistical preemption: (Goldberg 1995; Goldberg 2006, Boyd & Goldberg

2011; Robenalt & Goldberg 2015, 2016)

Repeated occurrence of a form A when a different form B is expected provides evidence that only A is acceptable

17

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-18
SLIDE 18

Experiment 2: statistical preemption

  • Similar design to Experiment 1
  • 1 PAV-only verb statistically preempted from APV

i.e., used with both strong and weak effect in PAV in exposure

  • Will speakers only use the verb in PAV contexts,

regardless of strength of effect?

  • Will this affect the way they learn the language?

18

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-19
SLIDE 19

Experiment 2: Results

19

Strong effect Weak effect

SOV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

OSV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

Preempted OSV verb

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

novel verbs

0.0 0.2 0.4 0.6 0.8 1.0

SOV OSV

APV-only verbs tend to be used in APV in both contexts PAV-only verbs tend to be used in PAV in both contexts Preempted PAV verbs tend to be used in PAV Novel verbs tend to be used with the contextually appropriate construction Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

APV PAV

slide-20
SLIDE 20

Experiment 2: Results

20

Strong effect Weak effect

SOV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

OSV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

Preempted OSV verb

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

novel verbs

0.0 0.2 0.4 0.6 0.8 1.0

SOV OSV

APV-only verbs tend to be used in APV in both contexts PAV-only verbs tend to be used in PAV in both contexts Preempted PAV verb tends to be used in PAV Novel verbs tend to be used with the contextually appropriate construction Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

Preempted PAV verb APV PAV

slide-21
SLIDE 21

Experiment 2: Results

21

Strong effect Weak effect

SOV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

OSV-only verbs

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

Preempted OSV verb

0.0 0.2 0.4 0.6 0.8 1.0

Strong effect Weak effect

novel verbs

0.0 0.2 0.4 0.6 0.8 1.0

SOV OSV

APV-only verbs tend to be used in APV in both contexts PAV-only verbs tend to be used in PAV in both contexts Preempted PAV verb tends to be used in PAV Novel verbs tend to be used with the contextually appropriate construction Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

Preempted PAV verb PAV-only verbs APV-only verbs APV PAV

slide-22
SLIDE 22

Mixed effects logistic regression

(to predict the probability of producing APV)

  • Again, tendency to produce APV when the effect is strong (β=

2.0433, p < 0.0001)

  • But mitigated by strong effects of VerbType: participants are

more conservative with all verbs APV-only: β =1.3727, p = 0.0002 PA V-only: β = -1.2858, p = 0.0013 preempted PA V: β = -1.4558, p = 0.0026

22

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-23
SLIDE 23

Summary of Experiment 2

  • Productions with the new novel verbs show speakers did learn

the functional difference between constructions.

  • Speakers are also very sensitive to preemptive information;

they used it to infer the restriction on the preempted verb.

  • They were also more lexically conservative with other verbs:

Unlike Exp. 1, APV-only and PA V-only verbs were mostly used with APV and PA V , respectively.

  • à preemptive exposure for one verb provides evidence that
  • ther verbs, too, are restricted in their distributions

23

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-24
SLIDE 24

Conclusion

  • Adult learners are sensitive to the form and function of

newly learned constructions

Speakers are willing to generalize beyond their input according to the function of constructions

  • They are also sensitive to the distribution of verbs

Statistical preemption provides evidence that verbs are restricted in their distributions

24

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion

slide-25
SLIDE 25

Thanks for your attention!

f.b.perek@bham.ac.uk adele@princeton.edu

Akhtar, N. (1999). Acquiring basic word order: Evidence for data-driven learning of syntactic structure. Journal of child language, 26(02), 339-356. Bolinger, D . (1968). Entailment and the Meaning of Structures. Glossa, 2, 119-127. Boyd, J. K. & Goldberg, A. E. (2011). Learning what not to say: the role of statistical preemption and categorization in “a”-adjective production. Language, 81(1), 1- 29. Casenhiser, D . & Goldberg, A. E. (2005). Fast Mapping of a Phrasal Form and Meaning. Developmental Science, 8, 500- 508. Goldberg, A. E. (1995). Constructions: A construction grammar approach to argument structure. University of Chicago Press. Hudson Kam, C. & E. Newport (2005). Regularizing unpredictable variation: The roles of adult and child learners in language formation and change. Language Learning and Development 1, 151 –195. Perek, F . & A. E. Goldberg (2015). Generalizing beyond the input: he functions of the constructions matter. Journal

  • f Memory and Language 84: 108-127.

Thothathiri, M., & Rattinger, M. G. (2016). Acquiring and Producing Sentences: Whether Learners Use Verb- Specific or Verb-General Information Depends on Cue Validity. Frontiers in Psychology, 7. W

  • nnacott, E., E. Newport & M. Tanenhaus (2008). Acquiring and processing verb argument structure:

Distributional learning in a miniature language. Cognitive Psychology 56: 165-209.

25

Generalizations Previous work Exp.1: Generalization Exp.2: Preemption Conclusion