The graded salience hypothesis in interlanguage

1
The graded salience hypothesis in interlanguage

• Advanced SLA Class Research Project
• Spring 2004
• Rebekah Johnson

2
The Graded Salience Hypothesis

• 3 things must be true
• a. salient (conventional) interpretation has
  unconditional priority over less salient (novel)
  interpretations the salient meaning is always
  activated
• b. novel interpretation of a salient meaning
  involves a sequential process (1st the salient
  meaning is processed, then the non-salient)
• c. novel interpretation must be more difficult
  to derive ? it requires more and different
  contextual support (Giora 1997)

3
Definitions of Salience

• Giora 1997
• Salience a function of its conventionality,
  familiarity, frequency, or givenness status in a
  certain context (p. 186)
• Giora 1999
• Salient can be retrieved directly from the
  lexicon

4
The Literature

• Many contemporary cognitive psychologists believe
  literal and non-literal language requires and
  equivalence process (Gibbs, Glucksberg, Keysar)
  
• Traditional theorists assume non-literal language
  requires a sequential process and is more
  difficult to understand (Grice, Searle)
• Combined the graded salience theory

5
Constructivist Theory

• Constructivists hold that language learners
  continually revise their ideas over a language
  each time they process a language chunk or item
• Constructions are recurrent patterns of
  linguistic elements that serve some well-defined
  linguistic function (Ellis, 2003, p. 64)
• Cognitive linguists say that the meaning of words
  depends on the perception of the world around us
  and the way things are organized and related

6
Constructivist Literature

• Chunking and bracketing
• Associative learning
• Generic learning mechanisms (not UG)
• Emergentism (beyond generative grammar)
  systematicity emerges from associations and
  connections
• Implicit learning in natural settings
• Natural Language Processing (NLP) and corpus
  linguistics
• Connectionism
• Data-driven processing
• Prototypical representations
• Emphasis on acquisition
• Incremental, context-sensitive,
  structure-sensitive
• Graded, distributed

7

• Construction grammar Language acquisition is the
  acquisition of constructions (sequence formula?
  low-scope pattern? construction)
• Formulae lexical chunks resulting from
  memorizing the sequence of frequent collocations

8

• Sinclair Idioms semi-preconstructed phrases
  that constitute single choices (appear to be
  analyzable into segments but are not) Sinclair,
  1991
• Collocations even more frequent in spoken than
  written language off the top of ones head
  and are made up of single clauses and are often
  highly predictable in terms of sequence
• Pulling language from old memory and making it
  fit the current context context shaping

9
Rationale

• Idiomatic phrases and formulae acquisition are
  essential for obtaining native-like competence

10
The Original Study Giora Fein 1999

• Question ? to see if irony (a form of figurative
  language) is as easy to understand as non-ironic
  (literal) language
• The study
• 60 4th-graders in Tel Aviv
• used native language (Hebrew)
• 20 target sentences put into literal or
  figurative (ironic) contexts
• on the page following the context story, there
  were 2 fragmented words printed, 1 related to
  literal meaning and 1 related to ironic

11
Texts

• A. After he had finished eating pizza, falafel,
  ice-cream, wafers and half of the ice cream cake
  his mother had baked for his brother Benjamins
  birthday party, Moshe started eating coated
  peanuts. His mother said to him, Moshe, I think
  you should eat something.

12

• B. At two oclock in the afternoon, Moshe
  started doing his homework and getting prepared
  for his Bible test. When his mother came home
  from work at eight p.m., Moshe was still seated
  at his desk, looking pale. His mother said to
  him, Moshe, I think you should eat something.

13
Test Words

• l i __ __ l e (little)
• s __ __ p (stop)

14
Results

• Giora and Fein (1999) found that the
  comprehension of ironic language involved both
  ironic and literal meanings (slightly higher
  literal meanings were more salient in the results
  than figurative for the ironic stories
• However, for the literal stories, mainly only
  literal concepts were activated

15
Conclusion

• Giora and Fein (1999) concluded that the graded
  salience hypothesis was a good predictor of
  activation and understanding of ironic language
  refuting the prior researchs claims that the
  literal meaning does not need to be processed
  when there is a figurative meaning

16
This study

• Replicates Giora and Feins (1999) study
• 8 idiomatic sentences embedded in story
  paragraphs,
• Assessment by controlling reaction time
• Tested on NSs and NNSs (all adults) 8 per group
  (16 total)

17
Changes in the current study

• Instead of using irony, uses idioms ? Rationale
  idioms are more important, more often learned by
  language learners Irony is not the same as
  idiomatic metaphorical language it is very
  difficult and only learned at high levels of
  fluency
• Uses both NSs and NNSs
• Uses English

18
Other considerations

• It is important to avoid the comparative fallacy
  (comparing NSs and NNSs)
• We also have evidence that mono-linguals are
  different from multi-linguals, cognitively

19
Hypotheses

• This study hypothesizes that
• 1. NNSs will not select the figurative meanings,
  even when the stories are using the idioms
  figuratively as often as NSs
• 2. Figurative meanings may be more salient to
  NSs, even when the idioms are placed in a
  non-figurative context

20
The Subjectsn16
21
Non-Native Speakers

• L1s
• 1 Polish
• 1 Portuguese
• 2 Chinese
• 4 Japanese
• Years spent studying English
• A range of 8-22 years

22
Non-Native Speakers

• LOR in US
• Ranged from 8 months to 13 years
• Education
• All had BAs, many had MAs, some were currently MA
  or doctoral students
• Age
• Ranged from 23 to 60

23
Non-Native Speakers

• Other languages spoken
• 5 spoke only English in addition to L1
• 3 spoke 2 or more additional languages
• ? multilingual

24
Method

• Participants were told to read each text quickly,
  only once, and to turn the page and fill in the
  first word that came to mind, not going back to
  the story
• After doing the task, the researcher generally
  asked participants for feedback about the tool

25
Text - Literal

• While painting the walls of his house, an old man
  suddenly felt tired. He climbed down from the
  ladder he was on and tried to avoid stepping on
  the paint bucket and many brushes on the ground
  near the ladder, but he couldnt avoid all of the
  paint supply items. He stumbled over some paint
  brushes. Then he made a real mess. The old man
  kicked the bucket.

26
Text - Figurative

• While painting the walls of his house, an old man
  suddenly felt tired. He climbed down from the
  ladder he was on and stumbled over some paint
  brushes. He clutched at his heart and suddenly
  fell to the ground. The mans neighbors called
  911 and the paramedics soon arrived, but they
  were too late. The old man kicked the bucket.

27
Test Words

• d __ __ d (died)
• s p i ___ l e __ (spilled)

28
Tool Variation

• 8 of each set of idioms was used (each set
  differed on whether it was the figurative of
  literal use of the idiom phrase and these were
  randomly ordered
• The researcher created the tool and briefly
  trialed it with native speakers

29
ResultsNative Speakers

• 1 thrown out (filled in both words)
• Mainly correctly chose figurative word when
  figurative context, but often also chose
  figurative word even with literal context

30
ResultsNative Speakers
31
ResultsNon-Native Speakers

• NNSs, like NSs, matched more of the figurative
  words with the figurative stories and with the
  literal stories
• Results are essentially the same for NNSs as for
  NSs

32
ResultsNon-Native Speakers
33
Revisiting the Hypotheses

• This study hypothesized that
• 1. NNSs will not select the figurative meanings,
  even when the stories are using the idioms
  figuratively as often as NSs
• The results do not support this NNSs performed
  nearly the same as NSs

34
Revisiting the Hypotheses

• The second hypothesis was that
• 2. Figurative meanings may be more salient to
  NSs, even when the idioms are placed in a
  non-figurative context
• This seems to be true for both NSs and NNSs,
  figurative meanings were more salient for not
  only the figurative uses (83 for both groups)
  but also for the non-figurative uses (63 and
  62)

35
Implications

• It seems that the graded salience hypothesis
  works in the same way for both NSs and for NNSs
• Figurative meanings seem to be more salient in
  the case of idiomatic phrases, perhaps due to the
  fact that the figurative uses are more frequent
  (thus more salient) than literal uses for idioms

36
A caveat

• There are limitations to this study, however
• The main limitation is the data collection tool.
  There are many possible explanations for results
  and the tool should be re-formatted and
  re-trialed to overcome these issues
• The NNSs in the study are highly educated, often
  graduate students, with long exposures to English
  and may not be representative of NNSs in general
• A small sample was used and a small number of
  text examples were used

37
Tool Trouble

• Feedback from participants noted
• The shortest word was often easiest to see
  first (results favor shorter words)
• The first word was the most noticeable (also
  reflected as more often filled in)
• The stories sometimes seemed contrived
• Some of the stories were ambiguous (could have
  been a literal or figurative sense)

38
Future Research

• Future research should
• Use more varied NNSs with different educational
  backgrounds and more diverse exposure to English
• Use a larger sample population and more texts in
  the data collection tool
• Re-design the tool to avoid issues discussed in
  feedback sessions
• Do a longitudinal study, testing NNSs at
  different intervals to see if prolonged exposure
  to English or longer LORs in the U.S. changes the
  saliency of idioms

39

• Thats all, folks!
• Rebekah Johnson
• sun22flower_at_yahoo.com