Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

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.

DeleteCancel

Linguistics Presentation

No description
by

Nicholas Grimes

on 19 April 2010

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Linguistics Presentation

Training English listeners to perceive phonemic length contrasts in Japanese by Keiichi Tajima et al. Pop Quiz!
Can you tell the difference...? Vowel Pairs
Obstruent Pairs
Nasal Pairs
Palatal Pairs Types of length contrasts: Vowel Pairs Obstruent Pairs Nasal Pairs Palatal Pairs Answers: 1. /ka-do/
2. /ka-a-do/ Answers: 1. /ha-k-ke-n/
2. /ha-ke-n/ Answers: 1. /ta-ni-n/
2. /ta-n-ni-n/ Answers: 1. /kja-ku/
2. /ki-ja-ku/ Experiment #1 Participants: Group ET (English Training)
Group EC (English Control)
Group JC (Japanese Control) Experiment #2 Purpose - To investigate how non-native
listeners' perception of Japanese
length contrasts is affected by
contextual factors that affect
segment duration, including speaking
rate, presentation text and position
within the word.

- To address the degree to which
perceptual training improves English
listeners' perception of Japanese
length contrasts, and to assess the
degree to which training generalizes
to novel stimulus conditions that
affect the temporal context. Stimuli and Procedure: - 3 phases; test 1, five days of training, test 2
- Prior to test, non-native listeners were given a
brief description of Japanese length contrasts, with
audio samples and English transcriptions. Test Format - Test stimuli: 76 real word pairs & 3
nonword triplets.

- Each word was read in either isolation
or in a sentence context.

- Talkers were trained voice actors who
spoke standard Tokyo Japanese comfortably.
- List was read at a slow, normal & fast rates

- Task was a single-stimulus, 2 or 3 three
alternative forced choice identification task.

- The test consisted of 1128 trials (16 blocks
of either 114 real word trials or 27 nonword
trials), and within each block of trials, words
from the four contrast types uttered at three
speaking rates were presented in a random order. Training - Only ET underwent training.

- Five days of perceptual identification
training between test one and test two.

- Training trials were identical to the
test trials, except that immediate feedback
was provided concerning listeners' responses
and that correction trials were performed
such that listeners repeated a given trial
until the correct response was selected.

- Each training day lasted for roughly 35-60
min, with a mild tendency for sessions to
become shorter as listeners accumulated training. Results This figure does not provide strong evidence
that training per se led to significantly greater
improvement in performance and factors such as
repeated exposure to the test materials and increased familiarity with the task. Results cont'd - These results suggest that non-native listeners'
ability to identify Japanese length contrasts vary
depending on the contrast type involve.

- The present results do not provide evidence that
training improved performance on all contrast types
or that training improved performance on specific
contrast types.

- Performance was poorer at a fast speaking rate than
at other rates, in both presentation contexts: words
produced in isolation, and words produced in a carrier
sentence.

- Performance for words embedded in a carrier phrase
produced at a fast speaking rate was better than the
performance for words produced in isolation at a fast
speaking rate. This suggests that presentation context
and speaking rate are not independent of each other. - Experiment 2 investigated whether perceptual training
using words produced by professionally trained talkers
would also generalize to non-professional talkers' productions.
It also examined the extent to which performance would vary
depending on the position of the length contrast within the
word, and the extent to which such positional effects are
modified in training.

- All participants were different from those in exeriment 1.

- As in experiment 1, there were three groups of participants:
- Group ET, group EC, and group JC.
Stimuli and Procedure: - The test consisted of 1032 trials, divided into 24 blocks of
either 34 real word trials or nine nonword trials.

- In each block, stimuli from each combination of the following
factors were presented: presentation context (word, sentence), talker, and word type (word, nonword).

- The training stimuli were the same 60 vowel pairs as those
used in experiment 1. Results: - The perceptual training mehod did not lead to a
substantial improvement in overall performance.

- There were no significant differences in listeners'
performance between the professional and nonprofessional
talkers.

- Non-native listeners show varying degrees of difficulty
with different length contrasts, consistent with the results
in experiment 1.

- A significant group-by-test interaction for vowel pairs
suggests that training significantly improved non-native
listeners' perception of contrast types that listeners were
specifically trained with, even though the specific word
pairs used in the tests were different from those presented
in training.

- The lack of significant group-by-test interactions for
obstruent, nasal, and palatal pairs suggests training did
not reliably generalize to contrast types that listeners
were not trained with.

- Non-native listeners' performance was generally much
poorer when the length contrast appeared in word-final
syllables than when they appeared in word-initial syllables.
Perceptual training, however, significantly improved
performance for both word-initial and word-final length
contrasts. Training using professional talkers'
productions also generalized to nonprofessional talkers'
productions.
General Discussion - The results from the study do not provide strong evidence
that perceptual identification training improves non-native
listeners' overall ability to identify Japanese length
contrasts.

- Given that most of the test words were different from the
training words, this suggests that training generalized to
untrained words of the same contrast type.

- As for whether training generalized to untrained contrasts,
the present data do not provide strong evidence that it does.

- Training involving multiple contrast types might be necessary
to improve non-native listeners' perception of various
contrast types.

- Contextual cues are useful for non-native listeners for fast
rate stimuli when phonemic cues for phonemic length are
relatively weak.

- The findings do not provide evidence that training improves
listeners' ability to cope with variation in speaking rate
and presentation context.

- Training significantly generalized to utterances produced by
ordinary nonprofessional talkers.

GD: Part Two - Fundamental frequency contours in addition to differences in duration can assist in differenciating length contrasts for short and long vowels in word initial position.
- /ká-dò/ vs /ká-à-do/

- In conclusion, non-native listeners that were trained to identify Japanese vowel length contrasts did not show greater overall improvement in performance compared to control listeners who did not receive training.
The End
Full transcript