Prior Knowledge, Comprehension and the L2 User

1
Prior Knowledge, Comprehension and the L2 User

• Ernesto Macaro
• University of Oxford

2
Research evidence

• The latest research shows that drinking red wine
  results in a significant reduction in
  cardio-vascular related diseases (Bourgogne et.
  al. 2004)
• Therefore everyone should drink lots of red
  wine.
• Research shows that high consumptions of alcohol
  lead to alcohol dependency, cirrhosis of the
  liver, significantly high levels of marriage
  break-downs.

3
Research evidence

• A meta-analysis of SLA research shows that Focus
  on form and focus on forms both lead to higher
  levels of language acquisition compared to focus
  on meaning (Norris and Ortega 2000
  Effectiveness of L2 instruction)
• Therefore we should have focus on form designed
  second language learning.
• Focus on forms (and even consistent focus on
  form) lead to neglect of skills development
  poor vocabulary growth rates de-motivation
  significantly high levels of marriage
  break-downs.

4
Please listen to the following news item
5
News Report Lotticks in Hotel!

• A Reading man who found lotticks and izzids in
  his supposedly furbustuous Caribbean hotel was
  awarded 459 in damages yesterday by a local
  magistrate.
• Paul Batters paid 1300 to Atlantic Pacific Tours
  in March 2000 for a furbusty holiday on the
  island of Martinique.
• The firms ancaps promised a furbusty hotel, free
  happaps from the airport, free use of the hotels
  gabonmang and beaches.
• However, on his arrival, Mr Batters found there
  was no one to meet him at the airport, the hotel
  room was infested with lotticks and izzids, and
  the gabonmang was completely flooded from the
  ninth hole onwards.
• Mr Batters successfully sued Atlantic Pacific
  Tours, who claimed that they had been organizing
  holidays on the island for 20 years and had never
  received any uptips.

6
Research evidence

• A systematic review shows that Prior Knowledge
  of a topic helps listeners with comprehension.
  (Macaro et al 2005)
• Therefore language teachers should use texts
  which the students have prior knowledge of.
• Or should they?
• Lecturers should provide L2 users with the text
  of their lecture beforehand
• Or should they?

7
Research Evidence

• is like a second-hand car
• You really need it to get around
• But you should really have a good look under the
  bonnet before buying it!

8
Prior Knowledge (schemata)

• Knowledge of the topic/area
• Rhetorical knowledge structure organization of
  the text/discourse
• Context knowledge a lecture an interactive
  seminar

9
example

• Types of prior knowledge we might have of the
  topic
• floods
• schema

Global Specific event Personal
Geographical lecture News report Personal account
10
Top-Down Processing Application of PK
Understanding Spoken Text
parsing
perception
Bottom-up processing
11
Why the interest in Prior Knowledge?

• the most efficient comprehension is one where the
  listener uses the least amount of surface
  information from the text to achieve the maximum
  amount of meaning

12
Studies testing the facilitating effect of
familiar topic

• Generally, if listener knew topic (personal or
  specific knowledge) they understood it better
  than if they did not know the topic
• However some caveats
• PK sometimes led to wild guessing (or
  overextending)
• PK effect only strong in open-ended
  comprehension, not specific items
• PK only accounted for a small of the variance
  in comprehension
• Markham and Latham (1987) Long (1990) Jensen
  Hansen (1995)

13
Studies testing the facilitating effect of
stimulating prior knowledge

• Topic not necessarily very familiar
• global knowledge stimulated
• Teacher presents students with advance organizer
  activities
• Mind mapping statements or questions
• Generally, comprehension was facilitated
• No evidence of long-term effect on skill of
  listening on strategy use
• Teichert (1996) Herron et al. (1998)

14
Studies strategy use and successful listening

• often described as investigating successful
  listeners versus unsuccessful listeners
• Two hypotheses are tested here,
• Learners of unequal listening proficiency use
  different strategies.
• some learners of equal general proficiency might
  be adopting more effective listening strategies
  than others.

15
Testing Hypothesis 1

• more effective listeners use PK to infer meaning
  rather than working it out from the text itself
• Less effective listeners use strategies such as
  listening out for single words, translating into
  L1,
• Hence claims for the superiority of top-down
  approaches
• OMalley et al (1989) Vandergrift (1998) Chien
  and Wei (1998).

16
Studies testing hypothesis 2

• No studies really testing this hypothesis but
  these two come near it.
• Chiang and Dunkel 1992
• Tsui and Fullilove 1998

17
Chiang and Dunkel (1992)

• the importance of two factors in learners'
  ability to understanding spoken English language
  texts in lectures
• prior knowledge, operationalized as familiar or
  unfamiliar topic,
• passage-dependent and passage-independent test
  items.

18
Chiang and Dunkel (1992)

• 360 students took the Comprehensive English
  Language Test and on the basis of their scores
  were divided into low and high listening
  proficiency groups.
• Students at each level were then randomly
  assigned to one of the four experimental
  conditions familiar or unfamiliar topic of text
  passage-dependent or passage-independent test
  items.
• Subjects in each group listened to one of the
  lectures recorded in one of the four conditions.

19
Chiang and Dunkel (1992)

• Results
• Prior knowledge
• Generally subjects scored higher when they
  listened to the familiar topic than the
  unfamiliar topic (predicted).
• There was no significant difference between HILP
  and LILP in this respect.
• However, the significant effect of prior
  knowledge only appeared on the subjects'
  performance on the text-independent items. This
  suggests that it did not help them with main
  ideas contained in the text nor with specific
  details.

20
Tsui and Fullilove (1998)Bottom-up or top-down
processing as a discriminator of L2 listening
performance. Applied Linguistics 19/4

• Topic
• Is bottom-up processing (focusing on words and
  phrases in the text) more important than top-down
  processing (using the listener's prior knowledge
  and inferencing) in discriminating the listening
  performance of L2 learners.

21
Tsui and Fullilove (1998)

• Background
• Previous research has suggested that poor
  listeners spend too much time in bottom-up
  processes (local) rather than top-down (global)
• However, some researchers have suggested that
  what makes poor readers is their inability to
  recognize words rapidly and construct an accurate
  representation. Local skills have to be mastered
  to take much of the guesswork out of reading.
  Same for listening?

22
Tsui and Fullilove (1998)

• Method
• Investigated the performance of candidates in one
  section of a listening paper in large-scale
  public examinations in Hong Kong media items
  short texts simulations of news items or adverts.
  (English L2)
• Independent Variables
• Question type global or local (see examples)
• Schema type (1) initial input is congruent with
  subsequent input
• (2) initial input is incongruent
  (refuted by) with the subsequent input (see
  examples)

23
examples

• Local and Global questions
• A. Butterfly catching
• B. Bird Watching
• C. Travelling
• D. Kite Flying
• A. Tom Everly
• B. Bobby Walker
• C. Mike Harman
• D. Isabella OGrady

• Matching and non-matching schema type
• A. A kitchen knife
• B. An oven timer
• C. A cleaning cloth
• D. A special dish
• A. the direction the wind was blowing
• B. the strong jets of water from the fire hoses
• C. the prompt call by residents to the fire
  services
• The quick action of the firemen

24
Tsui and Fullilove (1998)

• Sample 177 test items taken by 20,000
  candidates. 98 matching global questions 20
  non-matching 49 matching local questions 10 not
  matching
• Analysis mean criterion the mean scores in
  the entire paper of the candidates who chose
  those multiple choice options.
• An option with a high mean criterion was chosen
  by candidates who scored higher in the entire
  paper.

25
Tsui and Fullilove (1998)

• Results
• Consistently showed that (correct) items of
  non-matching schema type (i.e harder) yielded the
  higher mean criterion scores (i.e. were chosen by
  the most successful students) (predicted)
• No significant differences between local and
  global questions. (not predicted!)
• Mean criterion scores of non-matching schema type
  items among the global questions were
  significantly higher than those of the matching
  schema type among the global questions.
  (predicted)
• Mean criterion scores of non-matching schema type
  items among the local questions were
  significantly higher than those of matching
  schema type among the local questions. (not
  predicted!)

26
Tsui and Fullilove (1998)

• Conclusions and implications
• The biggest problem occurred with non-matching
  schema listeners unable to process subsequent
  input which contradicted their initial schema.
  Either they were weak at bottom-up processing or
  not combining strategies
• Learners need to be taught how to use prior
  knowledge to help understand but also they need
  to be reliant on rapid and accurate decoding.
• Most effective listeners combine top-down and
  bottom up strategies

27
Lectures and the L2 user

• Lectures will activate prior knowledge of some
  sort. Which kind?
• What kind of lectures traditional interactive
  semi-interactive?
• Individual a variable?
• We need to understand why and when prior
  knowledge is leading to misunderstanding of
  lectures
• For a review on academic listening see
  Flowerdew (1994)

28
Ruhe 1996

• Enhanced lecture comprehension through the
  provision of an organizational graphic a mind
  map similar to advance organizers.
• A sample of 103 students with mixed L1s were
  matched graphic provided versus no graphic
  provided and vocabulary provided in lecture
  order versus vocabulary provided in non-lecture
  order.
• graphic provided group scored higher than the
  control whilst there were no significant
  differences between the control and the two
  vocabulary provided conditions.
• In other words, all conditions except the control
  would have activated schemata but only the
  graphic revealed the organizational patterns of
  the lecture.

29
Use of metaphor in lectures (Littlemore 2001)

• Metaphor science is witchcraft
• science is the topic of the metaphor
• witchcraft is the vehicle of the metaphor
• the common ground is what is shared by
  participants in the metaphor
• The common ground of metaphor is often culturally
  specific

30
Lectures include metaphor because

• Metaphors are evaluative (usually negative)
• Metaphors label new concepts being introduced
• Metaphors allow the lecturer to be deliberately
  vague
• Metaphors provide frameworks for ideas
• Metaphors make language entertaining and memorable

31
Littlemore 2001

• Method
• Bangladeshi students of civil service reform
• Researcher followed their lectures
• Students asked to note down difficult language in
  lectures
• 20 Students given 10 metaphors to interpret and
  to say how they had derived the meaning

32
Littlemore 2001

• Findings
• Although lecturers varied in their use of
  metaphor it was always present somewhere
• Of 180 words judged difficult, 145 were
  metaphorical
• Most of the participants misinterpreted at least
  one of the metaphors in a way that seriously
  affected their understanding of the lecturers
  position/opinion.
• Participants wrongly used both schematic
  knowledge (PK cultural background) and
  contextual knowledge about their course, to
  interpret the metaphor

33
Main findings of the review on PK

• There is a positive association between Prior
  Knowledge and listening comprehension
• Studies where Prior Knowledge was deliberately
  stimulated by the teacher (i.e. advanced
  organizer type studies) found that students
  short term listening comprehension performance
  was greater

34
PK review findings

• Prior Knowledge can be misused if it is not
  supported by later in-text information or if the
  listener is not listening out for possible
  contradicting information.
• The way in which Prior Knowledge is used as a
  comprehension strategy may vary depending on the
  learners L2 language proficiency. Lower
  proficiency learners likely to misuse prior
  knowledge more. (Previous research concluded it
  was a question of either use or non-use)

35
Implications (for teachers/test-constructors)

• Texts should be selected carefully by teachers to
  take into account both the facilitating and
  potential pitfalls of prior knowledge.
• Facilitating comprehension may engender
  motivation.
• Limiting exposure to texts where the topic is
  familiar to the listener, may lead to
  under-developing bottom-up processes crucial for
  confirming hypotheses generated.
• Tests should include questions which require
  understanding of information which may contradict
  a listeners general knowledge of a topic.

36
Implications for lecturers/teachers of L2 users

• Lecturers should exercise caution in their use of
  metaphor
• Raise students awareness of metaphor use
• Perhaps provide mind-mapping activities at the
  beginning of a lecture
• Find out PK of students (global/specific/personal)
  
• Provide key words on slides to guide the
  understanding.
• Check understanding on schema-non-matching
  information

37
Implications for researchers

• More research on the different types of PK
• How does PK interact with different lecture
  types/styles
• Disentangle hypothesis 1 from hypothesis 2
• To identify successful listening strategies, need
  to control for general proficiency and PK.

38
Prior Knowledge (schemata)
How much is given beforehand
Lecture comprehension L2 user
L2 proficiency
Mode of delivery