1 of 46

The Role of pragmatic markers in

L2 Interactive Fluency

Dan Brown

Grand Valley State University

brownda1@gvsu.edu

Publication Co-Authors:

Julieta Fernández

University of Arizona

Amanda Huensch

University of Pittsburgh

Aquinas College

Tiny URL for article (in press; Pre-proof):

https://tinyurl.com/3y9wubdu

Tiny URL for slides:

https://tinyurl.com/bdxwc2sk

2 of 46

Background Overview

Fluency defined

Three problematic areas in fluency research

What are Pragmatic Markers?

How might Pragmatic markers relate to L2 fluency?

2

Background

3 of 46

What is oral fluency?

What is fluency?

(1) Broadly (layman’s perspective): overall language proficiency�

(2) Narrowly (L2 researchers’ perspective): smoothness, flow, automaticity and ease of delivery (Lennon, 1990; Segalowitz, 2010)

3

Background

4 of 46

Defining fluency → Viewed in two ways:

(1) Broadly (layman’s perspective): overall language proficiency�

(2) Narrowly (L2 researchers’ perspective): smoothness, flow,

automaticity and ease of delivery (Lennon, 1990; Segalowitz, 2010)

4

Background

5 of 46

What is oral fluency (from an L2 research perspective)?

Utterance fluency

Objective temporal measures (e.g., speed, pausing, breakdown)

Ability to plan and execute speech efficiently

Cognitive fluency

Perceived fluency

Impressions or inference of listeners about a speaker’s fluency based on a speech sample

In practice fluency is often operationalized, taught, and measured as a listener construct

Segalowitz, N. (2010). The cognitive bases of second language fluency. New York: NY: Routledge.

Background

6 of 46

What influences our perceptions of fluency?

Number of (unfilled) pauses�per second
Total pause time
Speech rate
Mean length of utterance
Phonation-time ratio
Number of stressed words / 1 min
Pace
Grammatical Accuracy
Lexical diversity
Accentedness

No mention of interactional features

Background

Predictors of fluency:

7 of 46

Three problematic areas in how fluency is operationalized:

1

Monologic bias in tasks & rating.

2

Attested (authentic) proficient utterances

vs.

Idealized models of oral production

3

Lack of connection between “stakeholders” (teachers, students, assessment, and L2 research).

7

Background

8 of 46

What are Pragmatic Markers (PMs)?

“You know,...” “Well,...” “I mean,….” “....I see….” “Oh, yeah!” “So,” “Right,...”

Also: ‘Utterance particles’ / ‘Discourse markers’ / ‘Continuatives’

→ conflicting views on what is considered a PM; in flux

They signal transitions in the evolving process of a conversation

highlight the force of the message
signal the interactive relationship between speaker, hearer, and message
sequentially dependent to what happens before and after in the conversation.

Represent roughly 10% of highly interactive speech

8

Background

9 of 46

Types of PMs

9

Fung & Carter, 2007:

Interpersonal and cognitive PMs occur with less frequency and diversity in L2 speech compared to L1 interaction.

Referential and structural PMs have been found to occur more frequently in L2 speech, closer to rates found in L1 discourse.

10 of 46

10

Interpersonal	Cognitive	Referential	Structural
Marking shared knowledge: you know, you see, See	Denoting thinking process: Well, I think, I see, and	Cause/Contrast: Because, cuz, but, and, yet, however, nevertheless	Opening and closing of topics: Now, OK, right, alright, well
Indicating attitudes: Well, really, I think, obviously, absolutely, basically, actually, exactly, sort of, kind of, to be honest, just	Reformulation / Self-correction: I mean, that is, in other words, What I mean is, to put it another way	Coordination: And Disjunction: Or	Sequence: First, firstly, second, next, then, finally
Showing responses: Ok, right, alright, yeah, yes, I see, great, oh great, sure, yeah	Elaboration: Like, I mean	Consequence: So	Topic shifts: So, now, well, and what about, how about
	Hesitation: Well, sort of	Digression: Anyway	Summarizing opinions: So
	Assessment of the listener’s knowledge: You know	Comparison: Likewise, similarly	Continuation of topics: Yeah, and, cuz, so

Which types of pragmatic/discourse markers are typically taught in TESOL?

And in which contexts?

11 of 46

11

Interpersonal	Cognitive	Referential	Structural
Marking shared knowledge: you know, you see, See	Denoting thinking process: Well, I think, I see, and	Cause/Contrast: Because, cuz, but, and, yet, however, nevertheless	Opening and closing of topics: Now, OK, right, alright, well
Indicating attitudes: Well, really, I think, obviously, absolutely, basically, actually, exactly, sort of, kind of, to be honest, just	Reformulation / Self-correction: I mean, that is, in other words, What I mean is, to put it another way	Coordination: And Disjunction: Or	Sequence: First, firstly, second, next, then, finally
Showing responses: Ok, right, alright, yeah, yes, I see, great, oh great, sure, yeah	Elaboration: Like, I mean	Consequence: So	Topic shifts: So, now, well, and what about, how about
	Hesitation: Well, sort of	Digression: Anyway	Summarizing opinions: So
	Assessment of the listener’s knowledge: You know	Comparison: Likewise, similarly	Continuation of topics: Yeah, and, cuz, so

Less common in L2 learner speech

More likely taught in academic writing?

12 of 46

Why focus on pragmatic markers?

Their role in evaluations of oral performance:�

“You have problems with fluency. You use lots of pauses and too many fillers”

→ but so do expert speakers (??)

Potential positive role in perceptions of oral performance (e.g., Fernández et al., 2014; Wei, 2011)
Higher levels of perceived fluency found to be associated with more diverse use of PMs (Hasselgren, 2002)
Grammatical errors are easily singled out, but non-expert uses of pragmatic markers are not so easily identifiable
ESL/EFL students have been found to use PMs less frequently and with less diversity → pedagogical opportunity??

12

Background

13 of 46

Current Study’s Research Questions

To better understand the construct of fluency in interaction, and in turn, the potential functional value of PMs:

1. To what extent does expert-like use of interpersonal and cognitive

pragmatic markers (PMs) influence perceived fluency of English L2

interactive speech?

2. Does the L1-L2 identity of raters affect these ratings?

13

14 of 46

Research Design

Matched-guise experiment → allows comparison of listeners’ perceptions of two different samples of speech produced by the same speakers (without being aware it’s the same speaker)
Comparison of two audio clips of the same speakers in conversation; one with (+)PMs and one without (-)PMs
Two conversations/dyads of focus → 4 speakers for comparison
Two versions of a survey (one focus dyad in each), with a counterbalanced design = 4 surveys

14

Dyad #1; Order A

+PM

-PM

Dyad #1; Order B

(reversed)

-PM

+PM

Dyad #2; Order A

+PM

-PM

Dyad #2; Order B

(reversed)

-PM

+PM

Methods

Conversation between L2 speakers 1 & 2

Conversation between L2 speakers 3 & 4

Survey 1

Survey 2

Survey 3

Survey 4

15 of 46

Four advanced L2 English speakers (graduate students in the U.S.)
Male / female pairings to help for individual ratings
Unscripted interactions with a classmate
Topic: high-stakes test they’ve both recently taken (shared knowledge, deep personal/emotional significance)→ + interpersonal and cognitive PMs

15

Speech Samples

Methods

16 of 46

16

75 seconds
Digital Editing using Audition

(audio editing software)

Controlled for speakers use of language features known to influence fluency ratings (e.g., speed, pausing, pace, false starts, grammar errors, etc.)
Guided by baseline data (8 native speakers’ expert-like use in the same task conditions) for frequency, variety, and placement of PMs
Cutting, splicing, rearranging of phrases, sentences, and PM use.
Piloting for ‘naturalness’

Methods

Speech Samples

17 of 46

+Pragmatic Marker Sample

17

+PM

-PM

Survey link

18 of 46

-Pragmatic Marker Sample

18

+PM

-PM

Survey link

19 of 46

19

20 of 46

Survey versions / order

VERSION 1

1. Advanced L2 (- PM use): Arabic L1 male; Korean L1 female

2. Distractor #1: advanced L2; Spanish L1 female; Arabic L1 male

3. Distractor #2: L1 speakers

4. Distractor #3: low proficiency L2; Portuguese L1 female; Chinese L1 male

5. Advanced L2 (+ PM use): same speakers as sample 1

VERSION 2

1. Advanced L2 (+ PM use): Japanese L1 male; Russian L1 female

2. Distractor #1: advanced L2; Spanish L1 female; Arabic L1 male

3. Distractor #2: L1 speakers

4. Distractor #3: low proficiency L2; Portuguese L1 female; Chinese L1 male

5. Advanced L2 (- PM use): same speakers as sample 1)

20

Methods

21 of 46

243 raters

185 English L1 speakers
58 English L2 speakers

Convenience/snowball sampling → email, social media, friends of friends, spread domestically and internationally

→ over-representation of university students and young adults

Excluded linguists (current students or degrees related to linguistics, TESOL, etc.)
Excluded language teachers

21

Participant Respondents

Methods

22 of 46

Quantitative data:

Two-way ANOVA: Mean ratings between +PM and –PM segments

Qualitative data: Coded open ended comments to identify features of fluency and disfluency

Comments coded “+/- Explicit fluency”--> Explicit mention of PMs (e.g., “She used too many filler words like ‘you know ‘ ”)
Comments coded “+/- Implicit fluency”--> favored one performance (+PM or -PM) over the other and attributed (dis)favorable features related to fluency in one performance (e.g., “She spoke without hesitation….”)

22

Analysis

Methods

23 of 46

Two-way ANOVA Results → Significant but modest effect sizes in rating differences between +/- PM segments (for 3 out of the 4 speakers)
PMs accounted for 2.7%, 17.5%, 8.5%, & 6.5% of the total variability in ratings for each speaker, respectively
No significant interaction effect between listeners L1/L2 status → The advantages for +PM use were found exclusively in the L1 English listeners’ ratings (L2 raters did not rate them significantly differently)

23

Quantitative Results

Results

24 of 46

Ratings by L1/L2 listener status

24

Results

25 of 46

25

Qualitative coding of raters’ justifications

Methods

Feature included	Examples
Speed	“Speech rate is fast enough to appear natural”
Pauses	“...with few breaks”
Flow	“...stumbles a few times…”
Clarity	“...was clear in expressing his ideas”

Features excluded	Examples
Vocabulary	“Did not use a wide range of vocabulary”
Accent	“He’s clearly foreign; noticeable accent”
Pronunciation	“Pronounced in an odd way”
Prosodic features	“Good intonation"
Accuracy/grammar	“Some small articles missing”
Comprehensibility	“Hard to understand”
Sentence structure	“Formed proper sentences”
Attitude	“Was relaxed; exuded confidence”
General assessments	“Has a good handle on the language; he’s so good!”

26 of 46

26

c

Results

27 of 46

Occurred in 8% of respondents’ comments
13 raters had negative comments explicitly about PMs for L2 speakers only
1 rater had negative comments explicitly about PMs for both L1 and L2 speakers

27

Qualitative Results: Explicit (dis)favorable mention of PMs

Results

Same rater, describing Native Speakers with similar PM use (frequency, variety, placement)…

Rater description of L2 Speakers…

“pauses or uses filler words such as ‘like’ or ‘I mean’ indicates he is pausing to translate thoughts into English.”

“…flowed easily”

“He said ‘so’ and ‘yeah’ and ‘um’ a lot and was clearly trying to find her words”

”...very fluent.”

“She always said ‘you know,’ she maybe can not express something that she wants to express..”

“He used the right social makers appropriately.”

28 of 46

Quantitative Results: Some evidence in favor of PM use supporting perceptions of L2 interactive fluency (esp. for native listeners)
Qualitative: About twice as many raters described the same speakers when using PMs as more fluent in comments
Despite apparent advantages for fluency, evidence of bias in favor of native speakers’ use of PMs

28

To sum up…

Results

29 of 46

Implications for fluency research

Fluency can’t just be seen as a (relatively stable) trait of the individual, but rather as collaborative interactional achievement:

Learners ability to co-construct fluency

Interactive fluency should be seen as a performance tied specific speech situations/purposes

29

Discussion

30 of 46

Implications for teachers

“But what about a focus on intelligibility? Abandoning native speaker norms? The growing recognition of World Englishes?”
How do fluent speakers develop and maintaining fluid interaction?
How do we typically teach oral fluency in speaking? Examples?

30

Discussion

Monologue or dialogue?

31 of 46

Level: Intermediate to Advanced; Higher Ed / Adult / Secondary?
Objective: Enhance learners’ perceived interactive fluency through a language awareness-raising approach; Ex: Smallwords (Fernandez& Yuldashev, 2015)
Pedagogical Trajectory:

Analysis of authentic interaction in transcripts → Roticing & explicit teaching of PM functions
Functional transcript analysis (close-ended)
Data (e.g., corp ora) navigation (open-ended)
Free recorded practice with rubric use (e.g., Hasslegren, 2002)

31

Approaches to teaching PMs in TESOL

Pedagogical Implications

Link to sample material s

32 of 46

32

/ Higher Ed.

/ Adult

/ Secondary