Upload
moses-washington
View
218
Download
1
Tags:
Embed Size (px)
Citation preview
Incrementality in Production
Binding Study update…
• Fiorentino/Minai conjecture…
– On Principle B studies, “we observed the following. The results from the experiments using proper names showed effects of binding-theory incompatible antecedents (Badecker & Straub, 2002; Kennison, 2003). However, with full NPs (Nicol & Swinney, 1989; Clifton et al., 1997) and when the accessible proper name is introduced in a lead-in context sentence (Kennison, 2003; Runner, 2003) the apparent violations of Principle B were not evident.”
– “If we assume that the early filter works on discourse representations, then it might be possible to propose a split among full NPs and proper names in terms of the richness of their discourse representation.”
Binding Study update…
• Proposed designs…
Incrementality in Comprehension
• How much of what you know can be deployed immediately?
– Numerous linguistic constraints involving• Complement structure
• Long-distance movement
• Binding, etc.
– ‘Immediately’ here has only word-level granularity
– Next step: how is this achieved?
Incrementality in Production
Incrementality in Production
• Different domains
– Speech errors
– Flexibility and incrementality
– Look-ahead in planning
Message
FunctionalProcessing
PositionalProcessing
PhonologicalEncoding
a. Lexical selectionb. Function assignment
a. Constituent assemblyb. Inflection
Adapted from Bock & Levelt (1994)
Speech Errors
“…the most slippable units are the most basic units in languageproduction […] each of these - the word, the morpheme, and thephoneme - is the basic building block for a particular linguisticlevel.” (Dell 1995, p. 190)
Word Errors - Category Constraint
Incrementality in Production
Binding Study
• Materials from Badecker & Straub are available
• 6 sets of 4, 12 fillers, comprehension questions
– A: Clare, Diogo, Valérie
– B: Matt, Ivan, Heather
– C: Lisa, Takuya, Hajime, Kaori
– D: Rob, Chunyuan, Utako
Message
FunctionalProcessing
PositionalProcessing
PhonologicalEncoding
a. Lexical selectionb. Function assignment
a. Constituent assemblyb. Inflection
Adapted from Bock & Levelt (1994)
Wilhelm Wundt (1832-1920)
“…the main job will be to do as muchas can be done with strictly incrementalproduction. This is a time-honoredprinciple in psycholinguistics. Wundt(1900) said that word order follows thesuccessive apperception of the parts ofa total conception [Gesamtvorstellung].Of course, Wundt added that this canonly hold to the degree that word orderis free in a language.” (Levelt 1989, p. 26).
V. Ferreira 1996
• Incremental models predict easier production with syntactic flexibility for two reasons– All structures are freely available to be filled
– Strict incremental construction permits the most active lexical representation (rather than syntactic competition) to determine structural decisions.
+
I gave
toyschildren
to
250ms
500ms
1500ms
250ms
Until button press
I gave toys to the children.I gave the children toys.I donated toys to the children.*I donated the children toys.
She gave it to the child.*She gave the child it.She gave the box to him.She gave him the box.
confused
himstory
250ms
1500ms
1000ms
Until button press
The story confused John. John rejected the story.John was confused by the story. The story was rejected by John.The story confused him. He rejected the story.*Him was confused by the story. *The story was rejected by he.
confused
rejected
Ferreira & Dell 2000
• “Production proceeds more efficiently if syntactic structures are used that permit quickly selected lemmas to be mentioned as soon as possible. We call this the principle of immediate mention.” (Ferreira & Dell, 2000, p. 299)
• Availability effects
– The coach knew (that) you missed practice.
– ‘that’ omitted more frequently the more accessible the embedded subject is.
Ferreira & Dell 2000
• Repetition
– I knew (that) I had booked a flight for tomorrow.
– You knew (that) I had booked a flight for tomorrow.
– I knew (that) you had booked a flight for tomorrow.
– You knew (that) you had booked a flight for tomorrow.
Ferreira & Dell 2000
Ferreira & Dell 2000
Ferreira & Dell 2000
Picture-Word Interference
Picture-Word Interference
Picture-Word Interference
cat cat
cat
Picture-Word Interference
duck duck
duck
Picture-Word Interference
rice rice
rice
Levels of Encoding
• Task: picture description
– Conjunctions: “the arrow and the bag”– Simple sentences “the arrow is next to the bag”
• Auditory distractor: semantic, phonological, unrelated
• Interference effects - delay in utterance onset latencies
– Semantic NP1 NP2– Phonological NP1
(Meyer 1996)
Look-Ahead
• “Is the verb an obligatory part of the advance planning unit?”
• Task - simple scene description + distractors– Intransitive: verb + subject– Transitive: verb + subject + object
• Prompts– Auf dem nächsten Bild sieht man wie… S (O) V
in the next pic. sees one how…– Und auf dem nächsten Bild… V S (O)
and on the next pic. …
• Distractors: SEM, UNREL, SYN, IDENT, NONE
(Schriefers et al., 1998)
Look-Ahead
(Schriefers et al., 1998)
Look-Ahead
• “The production system does not have to wait for successful retrieval of the verb lemma when it occurs late in the utterance.”
(Schriefers et al., 1998)
Look-Ahead
(Schriefers et al., 1998)Distractor onsets after 200ms
Look-Ahead
• “It appears that speakers can assign syntactic functions without knowing the verb lemma and its subcategorization frame and argument structure, and they do so if the verb does not occur in utterance initial position.”
(Schriefers et al., 1998)
Look-Ahead
(Schriefers et al., 1998)
Expt. 5 - simple main clauses, no prompts“…the verb is only part of the grammatical advance planningunit if it occurs in utterance initial position.”
More Production
Summary so far…
• Evidence for incremental grammatical encoding
– V. Ferreira (1996): give/donate alternation - opportunistic choice of word order
– Schriefers et al. (1998): lack of picture-word interference effects on verb in S (O) V structures
– V. Ferreira & Dell (2000): modulation of use of that.
Arithmetic
• Add the following
– 21 + 4 =– 4 + 21 =
(Dutch, French speakers)
(Brysbaert et al., 1998)
Arithmetic
• Language contrast
– Dutch: 51ms advantage for 4 + 21 order– French: 56ms advantage for 21 + 4 order
• Radically incremental account
– “the Dutch speakers try to get access to the unit of the response first, because they can start programming the pronunciation of the answer as soon as the value of the unit is known. In contrast, the French speakers have to capitalise on the value of the ten, which they must know before the response execution can be started” (p. 67)
(Brysbaert et al., 1998)
Arithmetic
• Say the following
– 24 + 31• “[…]”
• “[…] is the sum”
• “the sum is […]”
(Ferreira & Swets, 2002)
Expt 1: utterance duration unaffected by difficultyExpt 2: utterance duration affected by difficulty
Look-Ahead in Grammatical Encoding
Swets & F. Ferreira (2003)
(Swets & Ferreira 2003)
(Swets & Ferreira 2003)
(Swets & Ferreira 2003)
How Incremental is Production?
• This question can mean different things
– What are the units of grammatical encoding?
– What are the units of phonological encoding?
– How closely time-locked are grammatical and phonological encoding?
Comprehension-Production
• Same or different?
Syntactic Priming
1. Lexical prime (e.g. worship)2. Structural prime (e.g. the dog
was chased by the cat)
Priming Effects
• Production - Production
• Comprehension - Comprehension
• Comprehension - Productionetc.
Mirror Neurons
• Cells respond for observation and generation of actions
• Single unit recordings in macaques, fMRI in humans
• Suggests a tight connection between perception and generation
(Rizzolatti et al. 1996)
(Iacoboni et al. 1996)
‘…we understand others through an ‘internal act’ that recaptures the sense of their action…’ (Rizzolatti et al. 2001)
Mirror Neurons
• Mirror neurons are typically selective for a specific action (i.e. finite capacity system)
• Long term encoding of link between perception and generation of the same action
• Various ways in which mirror neurons might connect to input and output systems
• Well-known analog in language…
(Rizzolatti et al. 1996)
(Iacoboni et al. 1996)
Words
• Words encode specific concepts (i.e. finite capacity system)
• Long-term representations, shared across perception and generation
• Various ways in which lexical items might connect to input and output systems
• Does not entail an ‘ideomotor’ theory of words
• Same for phonemes, features
/døg/
DOG
Surely the enthusiasm about mirror-neurons is supposed to be about more than that?
V. Ferreira - optionality vs. ambiguity
“…comprehension and production share high level processing goals, but differ in terms of processing implementation.”