Relationships between early infant speech segmentation skills and later narrative ability

1
Relationships between early infant speech
segmentation skills and later narrative ability

• Laura Gutowski, Melanie Hill, Rochelle Newman,
  and Nan Bernstein Ratner
• The University of Maryland,
• College Park

2
Background

• A large amount of research has examined infants
  ability to segment the running stream of speech.

3
Segmentation skills seen in most infants

• Infants can reliably locate words within the
  fluent speech signal by 7.5 months,
• Jusczyk Aslin, 1995
• can store this information over long stretches of
  time
• Jusczyk Hohne, 1997
• and can do so in the presence of competing noise
  and talker variability
• Houston Jusczyk, 2000 Jusczyk, Pisoni
  Mullennix, 1992 Newman Jusczyk, 1996

4
The question what happens to infants who cant
segment?

• In all of these studies, MOST infants performed
  the task at above chance-levels, but SOME did
  not.
• Most studies of infant speech processing report
  on abilities evidenced by the majority of infant
  participants, and only report group results.
• Few studies link data from individual children
  across a range of ages.

5
Is there a relationship between segmentation and
later language ability?

• Do individual differences among infants speech
  processing abilities relate to those infants
  later language acquisition?
• A handful of studies have attempted to answer
  this question
• Molfese and colleagues (speech sound
  lateralization)
• Trehub Henderson (1996) (stop-gap detection)
• Benasich colleagues (2002a,b) (Rapid Auditory
  Processing)
• Tsao, Liu Kuhl (2004) (vowel discrimination)

6
Participants in the current study

• Participants were 26 children who had
  participated in speech segmentation studies at
  7-12 months of age. These studies were originally
  conducted by the late Peter Jusczyk of Johns
  Hopkins University and his colleagues.

7
Participants breakdown between segmenters and
nonsegmenters

• Our particular focus was later language
  development for 12 children who had shown speech
  segmentation ability (segmenters) and for 14
  who had not (nonsegmenters).
• Statistical analyses were run to examine whether
  infant perceptual performance affected any
  later-assessed abilities including the structural
  characteristics of the childrens narratives,
  produced in response to the picture book, Frog,
  where are you?

8
Participants (continued)

• The group of children showing evidence of
  segmentation ability had an average age of 54.3
  months (range 48.5 67.9), while children not
  evidencing segmentation ability (nonsegmenters)
  had an average age of 56.6 months (range 49.7
  69.0).This distribution is not significantly
  different (t(25) .98, p gt.30).
• Maternal level of education for segmenters was
  17.2 years that for mothers of non-segmenters
  was 16.8 years. This distribution does not differ
  significantly (t(25) 0.89, p gt.38).

9
Narrative analyses

• We used a narrative framework proposed by
  Trabasso Rodkin (1994) to compare the Frog
  narratives of the two groups.
• This rubric characterizes stories according to
  the following 5 categories
• Setting
• Initiating Event (s)
• Goal
• Attempt(s)
• Outcome(s)
• Each childs story was scored for these features
  on a CHAT transcript (MacWhinney, 2000).

10
Sample of CHAT narrative coding
11
First a note about formal language test results

• Overall spoken language quotient on TOLD-3
• Segmenters mean quotient of 125
• Nonsegmenters mean quotient of 111
• Significant difference (t(25)2.44, plt.05,
  eta-squared .192, indicating a large effect)

12
Results, language (continued)

• Semantics subscale on TOLD-3
• Segmenters120, non-segmenters 108
• Significant difference (t (25)2.21, plt.05,
  eta-squared 0.163, large effect).
• Syntax subscale on TOLD-3
• Segmenters 126, non-segmenters 112
• Significant difference (t(25)2.26, plt.05,
  eta-squared 0.170, large effect

13
Results Narratives

• Overall, children who had been segmenters as
  infants produced higher quality narratives than
  those who had not demonstrated such skills.
• Specifically,
• for story setting, segmenters were significantly
  more likely to include information about
  possession (that the frog belonged to the boy),
  than were nonsegmenters (means of 1.6 vs. 1.2 out
  of a possible 2.0 points, p lt .05, one-tailed).

14
Results (continued)

• There were no differences between the groups for
  numbers of initiating events contained in the
  stories however, there was a nonsignificant
  trend for segmenters to relate more attempts to
  find the frog than were seen in stories produced
  by nonsegmenters (means of 3.8 vs. 3.4
  respectively).
• Utterances describing outcomes were slightly more
  frequent in the stories of segmenters as well,
  but the difference between groups did not reach
  significance (means of .92 vs. .71,
  respectively).

15
Results (continued)

• Finally, significant differences were seen
  between the groups for numbers of stories with
  stated goals, with the segmenters much more
  likely to make the goal of the boy's search
  explicit (means of .58 vs. .14, respectively, p lt
  .05, one-tailed).

16
Infant speech perception, later language, and
later IQ

• While our results could be attributable to
  differences in cognitive ability in the original
  sample of infants, follow-up using the nonverbal
  portion of the Kaufman Brief Intelligence Test
  suggest that language abilities, but not
  intelligence, are related to earlier infant
  speech processing abilities.

17
Results K-BIT non-verbal scores

• K-BIT Matrices
• Segmenters 111
• nonsegmenters 112
• a nonsignificant difference (in addition to being
  in the opposite direction we might have
  predicted t(25)-0.38, p.71, eta-squared
  0.006)

18
Conclusions

• How infants perform on speech segmentation tasks
  at ages 7-12 months is strongly related to their
  later language performance at ages 4 - 6 years.
• This does not appear to be the result of simple
  intelligence differences, as segmenters did not
  perform better on a nonverbal intelligence
  measure.

19
Discussion

• This study was part of a larger study to examine
  language outcomes of children who had
  participated in speech processing studies as
  infants.
• Some later outcomes were more dramatic than
  others.
• For example, standardized test scores and lexical
  styles (Ter Avest et al., this conference) seemed
  to differentiate the groups more than narrative
  styles.
• This makes some sense, as narrative style may be
  less dependent upon specific syntactic and
  lexical knowledge. As noted, the two groups were
  very similar in overall cognitive outcome, which
  may influence narrative structure more strongly.

20
Future work

• Future work should examine these relationships
  longitudinally and prospectively, examining
  infant language skills beyond segmentation.
• Thus, we should follow a single group of children
  from early laboratory performance through to
  later stages of language development to identify
  relationships among these skills.
• This may lead to greater ability to identify
  children at increased risk for language delay and
  disorder.

21
Acknowledgments

• Funding for this study was provided by the
  Bamford-Lahey Childrens Foundation, which also
  graciously supported my attendance at this
  meeting.
• The original laboratory work with the infants
  that is reported here was supported by NICHD
  (15795) to the late Peter Jusczyk, and a Senior
  Scientist Award from NIMH (01490) to Professor
  Jusczyk. We would like to take a moment to
  remember his many important contributions to the
  field.

22
References

• Benasich, A., Tallal, P. (2002). Infant
  discrimination of rapid auditory cues predicts
  later language impairment. Behavioral Brain
  Research, 136, 31-49.
• Benasich, A., Thomas, J., Choudhury, N.,
  Lappänen, P. (2002). The importance of rapid
  auditory procesing abilities to early language
  development Evidence from converging
  methodologies. Developmental Psychobiology, 40,
  278-292.
• Houston, D. M., Jusczyk, P. W. (2000). The role
  of talker-specific information in word
  segmentation by infants. Journal of Experimental
  Psychology Human Perception Performance,
  26(5), 1570-1582.
• Jusczyk, P. W. (1997). The discovery of spoken
  language. Cambridge, MA MIT Press.
• Jusczyk, P. W., Aslin, R. N. (1995). Infants
  detection of the sound patterns of words in
  fluent speech. Cognitive Psychology, 28, 1-23.
• Jusczyk, P. W., Hohne, E. A. (1997). Infants'
  memory for spoken words. Science, 277, 1984-1986.

23

• Jusczyk, P. W., Houston, D. M., Newsome, M.
  (1999). The beginnings of word segmentation in
  English-learning infants. Cognitive Psychology,
  39, 159-207.
• Jusczyk, P. W., Pisoni, D. B., Mullennix, J.
  (1992). Some consequences of talker variability
  on speech perception by 2-month-old infants.
  Cognition, 43, 253-291.
• Kaufman, A., Kaufman, N. (1990). Kaufman
  Brief Intelligence Test. Circle Pines, MN AGS.
• Molfese, D. L. (1989). The use of auditory evoken
  responses recorded from newborns to predict later
  language skills. In N. Paul (Ed.), Research in
  infant assessment (Vol. 25, No. 6, pp. 68-81).
  White Plains, NY March of Dimes.
• Molfese, D. L. (1992). The use of
  neuropsychological measures to predict long-term
  outcomes in new-born human infants. In M.
  Tramontana S. Hooper (Eds.), Advances in child
  neuropsychology, Volume I. New York Springer
  Verlag.
• Molfese, D. L., Molfese, V. J. (1985).
  Electrophysiological indices of auditory
  discrimination in newborn infants The bases for
  predicting later language development. Infant
  Behavior and Development, 8, 197-211.

24

• Molfese, D. L., Molfese, V. J. (1997).
  Discrimination of language skills at five years
  of age using event related potentials recorded at
  birth. Developmental Neuropsychology, 13,
  135-156.
• Molfese, D. L., Molfese, V. J., Espy, K. A.
  (1999). The predictive use of event related
  potentials in language development and the
  treatment of language disorders. Developmental
  Neuropsychology, 16, 373-377.
• Molfese, D. L., Searock, K. (1986). The use of
  auditory evoked responses at 1 year of age to
  predict language skills at 3 years. Australian
  Journal of Communication Disorders, 14, 35-46.
• Newcomer, P., Hammill, D. (1997). Test of
  Language Development - Primary (TOLD-P3). Austin,
  TX Pro-Ed.
• Newman, R. S., Jusczyk, P. W. (1996). The
  cocktail party effect in infants. Perception
  Psychophysics, 58(8), 1145-1156.
• Trabasso, T. Rodkin, P. (1994). Knowledge of
  goal/plans A conceptual basis for narrating
  Frog, where are you?. In Berman, R. A., Slobin,
  D.I. (Ed.) Relating events in narrative A
  crosslinguistic developmental study. (pp 85-108).
  Hillandale, NJ Lawrence Erlbaum Assoc. Press.

25

• Trehub, S. E., Henderson, J. (1996). Temporal
  resolution in infancy and subsequent language
  development. Journal of Speech, Language,
  Hearing Research, 39, 1315-1320.
• Tsao, F.-M., Liu, H.-M., Kuhl, P. K. (2004).
  Speech perception in infancy predicts language
  development in the second year of life A
  longitudinal study. Child Development, 75,
  1067-1084.