the role of vocabulary and grammar knowledge in second-language oral fluency: a correlational study...

29
The role of vocabulary and grammar knowledge in second-language oral fluency: A correlational study Nel de Jong, Free University Amsterdam Laura Halderman, University of Pittsburgh SLRF 2009, Michigan State University

Upload: clement-berry

Post on 16-Dec-2015

217 views

Category:

Documents


0 download

TRANSCRIPT

The role of vocabulary and grammar knowledge in second-language oral

fluency: A correlational study

Nel de Jong, Free University AmsterdamLaura Halderman, University of Pittsburgh

SLRF 2009, Michigan State University

Oral Fluency in L2 Speakers

• Broad vs. narrow definition (Lennon, 1990)– Broad: general oral proficiency– Narrow: speed and smoothness of oral

delivery • Oral fluency depends on fast and automatic

retrieval of vocabulary and processing of grammar knowledge (e.g., Levelt, 1999; Kormos, 2006; Schmidt, 1992)

• Lexical and grammatical knowledge play a large role in second language oral fluency

Lexical knowledge

• Lexical retrieval in writing– Lexical retrieval training of words lead to greater

use of those items and more essential content elements were expressed. No effect on global text quality. (Snellings et al., 2004)

• Lack of (access to) lexical knowledge is a major cause of dysfluencies (Hilton, 2007)

Aspects of Lexical Knowledge

• Breadth– How many words a person knows– Greater breadth leads to fewer lexical searches

• Depth– How well a person knows a word– Greater depth leads to easier integration into

context• Lexical retrieval speed

– How fast a person retrieves a word– Faster retrieval leads to less dysfluencies

Grammatical knowledge

• Automatic syntactic encoding is fast and requires little attention – (Anderson et al., 2004; Kormos, 2006; Segalowitz &

Hulstijn, 2003)

• Automaticity leads to oral fluency – (De Jong & Perfetti, in preparation; Towell, Hawkins, &

Bazergui, 2006)

• Implicit grammatical knowledge can tell us what structures have been proceduralized

Research Goal

Examine the relationship between lexical and grammatical knowledge and oral fluency in a

sample of English Language Learners

Our Tests• Picture Naming – Immediate & Delayed

– Breadth of Lexical Knowledge– Lexical Retrieval Speed (Immediate)– Articulation Rate (Delayed)

• Vocabulary Knowledge Scale– Depth of Lexical Knowledge

• Elicited Imitation– Grammatical Knowledge

• Two minute recorded monologue– Oral production sample

Measures of Oral Fluency

• Temporal Measures:A.Length of fluent runs

– Number of syllables between pausesB.Length of pausesC.Phonation/time ratio

– % of time filled with speechD.Articulation rate

– Syllables per minute(Kormos & Dénes; Towell et al., 1996)

Hypotheses

1. Greater breadth of vocabulary [PN accuracy] => longer fluent runs; higher phonation/time ratio

2. Faster lexical retrieval [Imm. PN RT] => shorter pauses

3. Greater vocabulary depth [VKS] => longer fluent runs

4. Greater grammatical ability [EI] => longer fluent runs

Participants

• 23 students enrolled in English language courses; Speaking course

• High intermediate (~60-79 on Michigan Test of English Language Proficiency)

• L1s: Arabic (3), Chinese (3), French (1), Italian (2), Japanese (3), Korean (5), Russian (1), Slovak (1), and Spanish (1), Turkish (3)

Picture Naming• Immediate

– Timing began as soon as the picture was shown

• Delayed– Timing began at the onset of a “beep” that was

played 3 seconds after the picture appeared

• 24 pictures of nouns each• Frequency bands sampled

– 1-1000, 1001-2000, 2001-3000 & 3001-10,000

• Measures – Accuracy & Reaction Time

Vocabulary Knowledge Scale• 12 nouns, 12 verbs• Four frequency bands:

– 1-2000, 2001-3000, 3001-5000 & 5001-10,000• Definitions:

– 1-9 words; avg. 4.0 words– Only 2,000 most frequent words– Distracter definitions adapted from Vocabulary Levels Test

Target word

Definitions shown when

3 - 5 is selected

Textbox to type

sentence

Elicited Imitation

• 32 sentences, 16 were grammatically incorrect• 8 grammatical structures:

– third person singular –s – regular plural nouns– embedded questions – regular past tense– indefinite articles – relative clauses– Modals – verb complements

• Sentences: 6-11 words, avg. 8.2• Sampled from Erlam (2006)

Elicited ImitationTarget

structureGrammatically correct Grammatically

incorrect

Embedded questions

It’s not clear when the next election will be

Everyone wants to know what is the president like

Relative clauses

It’s wrong to make a promise that you can’t keep .

Jobs that people like them pay a lot

Verb complements

Some universities allow students to travel abroad for a year

Expensive restaurants ask their customers to wearing nice clothes

Results: Time 1PN PN overall overall accuracyaccuracy

PN PN immediate immediate RTRT††

VKSVKS EI EI accuracyaccuracy

LFRLFR -.119-.119 -.337-.337 .438*.438* .597***.597***

MLPMLP .365.365 -.256-.256 -.338-.338 -.029-.029

PTRPTR -.467*-.467* .155.155 .408.408 .187.187

ARAR -.029-.029 -.382-.382 .423*.423* .678***.678***

* p* p < .05; ** < .05; ** pp < .01; *** < .01; *** pp < .001 < .001† † Frequency 1-1000 & 2001-3,000 onlyFrequency 1-1000 & 2001-3,000 only

Results: Time 2PN PN overall overall accuracyaccuracy

PN PN immediate immediate RTRT††

PN PN immediate immediate accuracy accuracy

EI EI accuracyaccuracy

LFRLFR .149.149 -.306-.306 .267.267 .315.315

MLPMLP -.344-.344 .192.192 -.448*-.448* .032.032

PTRPTR .347.347 -.103-.103 .445*.445* -.058-.058

ARAR -.030-.030 -.538*-.538* .067.067 .508*.508*

* p* p < .05; ** < .05; ** pp < .01; *** < .01; *** pp < .001 < .001† † Frequency 1-1000 & 2001-3,000 onlyFrequency 1-1000 & 2001-3,000 only

Conclusions from Correlations• PN accuracy (breadth) correlates with MLP and PTR,

but inconsistently• Outlier in the Pretest measures?• Easier to find appropriate word (more fluent)

• EI (grammatical ability) and VKS (depth) correlate with LFR– Building sentence structures

• EI (grammatical ability) and VKS (depth) correlate with AR– Do students slow down their articulation rate for

planning?

Gains Time 1 – Time 2:Picture Naming Accuracy

• Main effect of pre/post• Main effect of frequency• Interaction - naming type

by frequency– Frequency effect only in

Immediate Naming:– Naming under time

pressure is less accurate

Gains Time 1 – Time 2:Picture Naming Reaction Time

• Main effect of naming type

• Interaction - naming type by time– Effect of time only in

Delayed Naming:– No improvement in lexical

retrieval, but in initiation of articulatory processes (cf. Barry et al., 2001)

Gains Time 1 – Time 2:Elicited Imitation

Main effects• time, structure, accuracy

Interaction (marginally sign.)

• structure X accuracy X pretest/posttest

Conclusions from Pre/Post-tests

• Improvement in vocabulary breadth• Improvement in initiation of articulatory

processes• Naming under time pressure is less accurate

– Lexical retrieval in speeches also occurs under time pressure

• Improvement in grammatical ability– Mostly noun plurals, relative clauses, verb

complements

Hypotheses

1. Greater breadth of vocabulary [PN accuracy] => longer fluent runs; higher phonation/time ratio• PTR: yes; MLFR: no

2. Faster lexical retrieval [Imm. PN RT] => shorter pauses• No support

3. Greater vocabulary depth [VKS] => longer fluent runs• Yes; and higher articulation rate

4. Greater grammatical ability [EI] => longer fluent runs• Yes; and higher articulation rate

Possible Explanations

• Curriculum focuses on academic vocabulary acquisition– Our Pictures sample more general vocabulary and

highly imageable nouns

• Recorded monologues are very open-ended– It’s hard to predict what vocabulary and

grammatical structures they will use

Many thanks to:

• Co-PIs: Prof. Charles Perfetti, Dr. Laura Halderman• Research assistants: Colleen Davis, Mary Lou Vercellotti

• The students and teachers at the ELI• The Robert Henderson Language Media Center• Pittsburgh Science of Learning Center

• Contact: [email protected], [email protected]

This work was supported in part by the Pittsburgh Science of Learning Center, which is funded by the National Science Foundation award number SBE-0354420.

Correlations Gains – Gains

Gain in PTR with gain in Delayed Picture Naming

r = .477, p = .053n = 17

All other correlations n.s.

• Extra slides …

Elicited ImitationTarget structure Grammatically correct Grammatically incorrect

Third person singular –s

Every good story deserves a long title The temperature change a lot every season

Embedded questions

It’s not clear when the next election will be

Everyone wants to know what is the president like

Indefinite articles Everyone likes to live in a big house Everyone has telephone in their home

Modal verbs You must study for years to speak English well

Everyone should to learn a second language

Regular plural nouns

Policemen arrest many criminals every day

Car companies use many machine to build their products

Regular past tense A long time ago, nobody lived in cities Last month, scientist discover a new moon

Relative clauses It’s wrong to make a promise that you can’t keep

Jobs that people like them pay a lot

Verb complements Some universities allow students to travel abroad for a year

Expensive restaurants ask their customers to wearing nice clothes

Overall Conclusions• Vocabulary breadth predicts fluency

– At single points in time: MLP and PTR– Gain: only Accuracy on Delayed Naming with PTR

• Lexical retrieval speed predicts articulation rate– Articulation rate is not a reflection of proceduralization

• Vocabulary depth predicts fluency– At single points in time: MLFR, AR– Gain: no post-test

• Implicit Grammar Knowledge predicts fluency measures the most– At single points in time: MLFR, AR– Gain: no significant correlations with temporal measures