Sentence Production Flashcards
Meyer Ferreira
Do we know how we are going to end a sentence when we start to utter it?
Evidence for INCREMENTALITY
Summary
We plan sentence structure by re-using available information - syntactic priming
we CAN make it up as go along but prefer not to do this
Whether we know how we will end a sentence when we start speaking probably depends on CONTEXT
Repetition in speech (Kuiper, 1996)
NZ auctioneers and sports commentators - speak very fast
This paper analysed the way they used language at rapid speech rates
When in low speech rate - resembled everyday speech
When in high speech rate - CHUNKS OF SPEECH WITH LITTLE SYNTACTIC VARIABILITY
- Very little variation in syntax
- Reused little chunks of speech they keep repeating
What is the benefits to repeating little chunks of speech with little variability of syntax (arrangement of words)?
Easier for people to understand and to use quickly
Kuiper’s study with NZ auctioneers and sports commentators findings can be generalised to everyday language
Turns out this is true for alot of our language - it is repeated
Other evidence (Altenberg, 1990)
70% of words for part of “recurrent” word combinations
The more you use recurrent word combinations, the more ‘native’ one sounds
Unusual word combinations = perceived as less “native-like”
Syntactic priming
Processing one particular syntactic structure influences the processing of subsequent structures
We tend to REUSE syntactic structures
- within speaker
- between speakers (in dialogue)
Syntactic priming example (Levelt & Kelter, 1982)
Reusing same type of structure "at what time does your shop close?" "at 5 o'clock" "what time does your shop close?" "5 o'clock"
Bock (1986)
Got participants to repeat sentences then describe unrelated pictures
- told it was a memory test
- actually investigating structure of participant’s response/descriptions of pictures
- picture description sentence structure matched sentence they were primed with
Ditrasitive sentence structures
- A rock star sold some cocaine to an undercover agent (PREPOSITIONAL OBJECT)
- A rock star sold an undercover agent some cocaine (DOUBLE OBJECT)
Active and passive structures
- One of the fans was punched by the referee (PASSIVE)
4. The referee PUNCHED one of the fans (ACTIVE)
Task in Bock (1986)
Priming sentence with particular structure
The ppts tended to repeat sentence structure they were primed with
Why do we tend to reuse sentence structure?
EFFORT REDUCTION THEORY
Bock (1986) - why reuse sentence structure
Means we don’t have to generate new sentence from scratch - re-using procedures/representations already ACTIVATED
- Ease demands of message formulation
- Contribute to fluency
Bock & Loebell (1990)
Is it really syntax?
What about repetition of “to” or “by” - does that explain effect?
1. Stella brought a book to Susan (PO)
2. Stella brought a book to study
3. The man is reading a story to the boy (PO)
Only 1 primes 3 despite 1 + 2 similarities
Syntactic priming properties
- Long lived - over 10 intervening sentences
- Occurs for variety of structures
- Unaffected by factors such as TENSE
- Demonstrated in many languages
- Also reflects way people genuinely use language
Gried (2005) - real life study - looking into International Corpus of English
This includes spoken and written English from the 90s
analysed and looked for tendency to reuse syntactic constructions
–> similar to lab-based results = support idea that we do reuse syntactic structure in real life the same way as we do in the lab
Branigan, Pickering & Cleland (2003)
shows we reuse syntactic structure in dialogue too
- pick card from box - describe card to pair (confed to ppt)
On ppt’s next turn , would the experimental ppt use the same structure as the confederate?
In this study - THEY DO
example:
confed says “the nun is showing the banana to the monk” (PO)
will ppt say “the artist is selling the gun to the ballerina” (PO) OR the artist is selling the ballerina the gun (DO)
Will mirror same PREPOSITIONAL STRUCTURE
Across languages - Loebell and Bock (2003)
German and English bilinguals
show that we reuse syntactic structure when switching between languages
Whatever information we are using to plan sentence structure - must be able to be shared between languages
Why does syntactic priming happen?
- Various accounts of mechanism
- EASE FLUENCY - talking made faster and easier
- less planning involved
Studies have shown - speech onset is shorter
- Broca’s aphasia show stronger priming effeects
Broca’s aphasia
Defining feature: reduced complexity of syntactic structure
Hartsuiker and Kolk did similar method to Bock
Broca’s group showed more syntactic priming - more influenced by sentence structure shown than controls
When primed - Broca’s group can produce syntactically complex structures that they would not normally produce spontaneously
Findings with Broca’s aphasia
Suggests adaptive function for syntactic priming
Syntactic priming reasons
- Aids fluency
- Syntactic structure consolidation is to some extent independent of meaning and form
- Shared between production and comprehension (occurs in dialogue)
- Shared even between languages
Do we know how we are going to end a sentence before we start it?
Speech error evidence - blake fruid/brake fluid
Distinction between word and sound exchanges
Scope suggests syntactic content is sketched out at least for the current clause
Meyer (1996)
Ppts describe pictures - “the arrow and the bag”
Auditory distractors
- semantic (e.g. purse) mean the same thing
- phonological (e.g narrow) sounds similar .
Measured reaction time to start sentence
- ppts SLOWER when distractor SEMANTICALLY related to either noun (2 Ss)
- ppts FASTER when distractor FONOLOGICALLY related to FIRST (3 Fs) but not second noun
Reaction times: time to begin a sentence
Meyer (1996) results show
Ppts had prepared the meaning and selected both nouns for the sentence
Sentence already been planned
Ferreira (1996) was interested in HOW we choose which sentence structure to use
How do we choose which word order to use?
Do alternative possibilities COMPETE? - Competition model
Do we decide as the sentence unfolds? INCREMENTAL model
Competition model
More choice should cause difficulty as structures compete
Incremental model
More choice should help - if there is more choice about how we order the words, it will be easier to slot them into the sentence - just talking and see what happens
This study (Ferreira, 1996) - is it better to give than to donate? shows some verbs are more flexible than others
Ppts presented with two words e.g. gave/donated
These words disappeared and 2/3 more words appeared in random order e.g. children toys
Task: produce a sentence beginning with “I gave” or “I donated” and using all of the remaining words
GIVE: can have both double object or prepositional object order
DONATE: only works in PO order - I donated the toys to the children
In the case of give –> more choice than donate
Results: ppts produced fewer errors for “give” than donate
People were quicker to initiate sentences containing “give” not donate
give = an alternator verb Nonalternator verbs (like donate) - don't occur with DO structure - fewer structure options - less syntactic flexibility - only PO structure node active - less competition - a structure node should be chosen more easily if COMPETITION model
Ferreira study SUPPORTS INCREMENTAL MODEL
More choice of word order = more flexibility - higher accuracy with verbs that have more choice of structure and are easier to produce
With donate - you have to stop and think about what verb to use next - inflexible and slows you down
Verbs that allowed syntactic choices were easier to use than verbs that did not
We make sentences as we go along
- incremental model
Differences between Meyer and Ferreira
Meyer used short sentences, Ferreira used longer sentences
Simple and complex constructions may operate differently
In reality, sentence planning
Probably depends on the kind of utterance and situation whether we plan ahead or not
Ferreira & Swets (2002)
Ppts given sums of varying difficulty (e.g. 21+22, 23+68)
43 vs 43 is the answer vs answer is 43
When no time limit, onset for sentence utterance for all 3 and utterance length unaffected by sum difficulty
- when there was a time limit, utterance length was affected by sum difficulty
–> INCREMENTALITY EVIDENCE for utterance under strategic control
Differences between incremental and competitive
- Incremental doesn’t require competition syntactic structure options
- structure nodes are not inhibitorily connected
- both structure nodes can be highly active simultaneously - Incrementality implies syntactic processing CANNOT occur on later phases before earlier phases
- Incremental theories usually claim that syntactic structures are slots that are available to be filled, rather than active plans that influence non-syntactic planning
Results show incremental model predicts that MORE syntactic flexibility
means faster selection times
because structural decisions are not accomplished through flexibility
Competitive vs incremental make opposite predictions about effect of syntactic flexibility on production
If syntactic decisions are accomplished via competition - show a difficulty with syntactic flexibility
What does Ferreira test?
Effect of syntactic flexibility on the ease of language production?
Ferreira measures
Number of errors
Production latency/time to begin sentence
Study compared differences in flexibility in 4 conditions
Does it make production easier or harder?
Manipulating flexibility by use of alternator or non-alternator verbs
Pattern of errors SUPPORTS INCREMENTAL
Ex1: fewest errors occurred in flexible alternator/un-constraining condition
Pattern of production latencies SUPPORTS INCREMENTAL
If supports incremental, should result in shorter production latencies -these were fastest in the flexible alternator condition
Results NOT compatible with competitive theory predictions
Issue of SYNTACTIC PERSISTENCE (Bock, Loebell & Morey, 1992) - dynamical structural preference
Explains low use of DO structure
When language system creates a sentence with a particular structure. the system is inclined to produce a sentence with that structure in the future - ppts exploit flexibility
75% critical sentences in experiment could only be spoken as PO therefore the likelihood of PO being uttered for all sentences is ALOT HIGHER %
Experiment 2 manipulated flexibility differently but with same flexibility conditions
Production latencies were lower and to an extent to were production errors
Participants could create sentences more quickly in flexible alternator/un-constraining condition
Experiment 3: showed over important factors that affect language production
ISSUES WITH FERREIRA STUDY
Production error analysis showed syntactic non-flexibility only resulted in difficult production when participants needed to produce PASSIVE SENTENCES
Results do show beneficial effects of flexibility under different types of structure - acknowledge that the production task used was very different from real life natural production
–> CANNOT GENERALISE
It would be thus strengthened by supporting evidence in more natural situation
IMPORTANT - Conclusion of Ferreira study
People produce sentences more easily under conditions of syntactic flexibility
Having more ways to express a message permits the speaker to choose a sentence that accommodates variation in the way that a message evolves
Syntactic persistence
When syntactic structure can be repeated later on in conversation - occurs in dialogue
- 1 of the more general phenomenas of syntactic priming
Syntactic priming is observed when
Participants are asked to repeat a prime sentence that contains syntactic structure of interest then asked to describe a picture
What do syntactic priming studies show?
Speakers use particular word order if prime sentence used that order
Bock (1986)
Bock & Loebell (1990)
Branigan et al (1995)
Meyer (1996) study (evidence for competitive model)
presented ppts with pictures of pairs of objects which they then had to name or place in short sentences - used SIMPLE SHORT SENTENCES
At the same time, auditory distractor that was semantically or phonologically similar to the first or second noun
If SEMANTICALLY SIMILAR to BOTH nouns - slowed down sentence production
If PHONOLOGICALLY SIMILAR to FIRST noun - was faster at uttering sentence - facilitated initiation of speech
Meyer (1996) conclusion
We prepare meaning of short phrases and select appropriate words before we start speaking but only retrieve the sound of the FIRST word
Discrepancies between models
- Advance planning evidence comes from study of short and simple sentences
Incremental deals with/explains more complex sentence structure constructions
2. Task demands affect how much participants plan in detail before they start speaking
Smith and Wheeldon (1999)
Ppts describe moving pictures - increased production onset latency for single clause sentences belonging with a complex noun phrase
vs a simple phrase
Also takes longer to initiate double clause/compound sentences vs single clause
These results suggest people DON’T plan entire syntactic structure in advance