Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


How do I choose which measure of fluency to use?

Presentation held at EUROSLA Conference, August 30, 2013, Amsterdam. http://aclc.uva.nl/conferences/eurosla-23/eurosla-23.html

Nivja De Jong

on 3 September 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of How do I choose which measure of fluency to use?

How do I choose which measure of fluency to use?
Mathematical Considerations
Speech rate
Articulation rate
Pruned speech rate
Mean syllable duration
Phonation-time ratio
Mean length of runs
Number of silent pauses per minute total time
Mean pause duration
Number of filled pauses per minute
Number of repairs per minute
per minute total time or speaking time

Theoretical Considerations
Fluency: the extend to which the speech production processes are running smoothly, resulting in relatively fast speech without unnatural hesitations

Do I want the measures to reflect L2-specific fluency?

Because: L2 fluency measures can be predicted to a large extent by L1 fluency behavior (i.e. reflect personal speaking style!)
Practical considerations
Breakdown fluency: counting filled pauses by listening / counting and measuring silent pauses in PRAAT
Speed fluency: counting syllables (phonemes!) by hand is very time consuming. For relatively noise-free sound files, automatic measurement is possible (De Jong & Wempe, 2009)
Repair fluency: by hand (e.g. using CLAN software/notations)
It depends on the purpose and the research questions
Aspects of fluency (Tavakoli & Skehan, 2005):
breakdown, speed, and repair

Do I want confounded or unraveled measures?

Measures of breakdown fluency
number of silent pauses per speaking time
number of filled pauses per speaking time
mean duration of silent/filled pauses
Measures of speed fluency
articulation rate =
# syllables / phonation time
mean syllable duration =
phonation time / # syllables
Measures of repair fluency
number of repetitions per speaking time
number of repairs per speaking time
Unraveling measures of fluency
Confounding measures of fluency
phonation time ratio =
total speaking time / total time

mean length of utterance = speaking time / (number of pauses + 1)

pruned speech rate = (number of syllables - filled pauses, repeated syllables, repairs) / total time
What is a silent pause?
> 100 ms?
> 1000 ms?

> 250 ms
Goldman-Eisler (1968):
"uh dealing with the problem"
De Jong & Bosker, 2013
mean syllable duration is more normally distributed
Segalowitz' notions of fluency
cognitive fluency
perceived fluency
utterance fluency
Evidence for L2-specific fluency measures (2)
mean syllable duration: strongly related to L2 proficiency/skills
mean pause duration: low (if any) relation to L2 proficiency/skills
De Jong et al., 2012;
De Jong et al., 2013
R = .3 - .5
R = .01 - .05
Levelt et al., 1999
Segalowitz: 7 fluency vulnerability points
combining number and duration of pauses
combining breakdown and speed
combining breakdown, speed, and repair fluency
Evidence for L2-specific fluency measures (1)
Fluency gain over time within speakers:
Towell et al. (1996): mean length of runs, speaking rate, but NOT mean duration of silences
Segalowitz & Freed (2004): mean length or runs, speaking rate
O'Brien et al. (2007): replicated Segalowitz & Freed
Cross-sectional approach:
Riazantseva (2001): proficient speakers outperform lower proficiency speakers on breakdown measures
Tavakoli (2011), Skehan & Foster (2007), Riazantseva (2001): L2 speakers pause more often within AS-Units, clauses, constituents than L1 speakers do
Predicting L2 fluency measures from L1 fluency behavior:
De Jong et al. (2013):
mean syllable duration R = .21 (lowest)
mean pause duration R = .57 (highest)
Derwing et al. (2009):
(pruned) speech rate R up to .64
number of pauses per second R up to .41
Pearson r with proficiency
Thank you!

Pearson Language Tests
Utrecht University

L2-specific or general measures of fluency?

Separate measures for separate aspects or one measure that encompasses all?

De Jong et al. (2012): http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=8514278
De Jong et al. (2013): http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=8859231
De Jong & Bosker (2013): http://www.diss2013.org/Proceedings_DiSS_2013.pdf
De Jong & Wempe (2009): http://link.springer.com/article/10.3758/BRM.41.2.385
Derwing et al. (2009): http://journals.cambridge.org/production/action/cjoGetFulltext?fulltextid=6385632
Levelt et al. (1999): http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=30915
O'Brien et al. (2007): http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=1392672
Segalowitz (2010): http://www.routledge.com/books/details/9780805856620/
Segalowitz & Freed (2004): http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=214862
Skehan & Foster (2007): Skehan, P. and Foster, P. (2007) ‘Complexity, accuracy, fluency and lexis in task-based performance: a meta-analysis of the Ealing research’ in P Van Daele,,S., Housen, A ., Kuiken, F., Pierrard, M. and Vedder, I. (eds) Complexity, Fluency and Accuracy and Fluency in Second Language Use, Learning and Teaching Brussells: VWK pp207-226
Riazantseva (2001): http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=1427744
Tavakoli (2011): http://eltj.oxfordjournals.org/content/65/1/71.short
Tavakoli, P., & Skehan, P. (2005): http://benjamins.com/#catalog/books/lllt.11.15tav/details
Towell et al. (1996): http://applij.oxfordjournals.org/content/17/1/84.abstract
Full transcript