Florit-Pons Vila-Gimenez Rohrer Prieto V1 Submitted

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

The development and temporal integration

of co-speech gesture in narrative speech: A longitudinal study


Júlia Florit-Pons1, Ingrid Vilà-Giménez1, Patrick Louis Rohrer1,2, Pilar Prieto3,1
1
Dept. of Translation and Language Sciences, Universitat Pompeu Fabra, Barcelona, Catalonia
2
Laboratoire de Linguistique de Nantes - UMR 6310, Université de Nantes, Nantes, France
3
Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Catalonia
julia.florit04@estudiant.upf.edu, ingrid.vila@upf.edu,
patrick.rohrer@upf.edu, pilar.prieto@upf.edu

[7]). It is well-established that infants start coordinating


Abstract pointing gestures and speech before their first words [8],
Children from 5 to 6 years of age have been shown to start showing that alignment is found from early stages. However,
producing non-referential beat gestures in narrative speech. less is known about gesture-speech temporal integration as
However, it still remains unclear how the use of these non- children begin adding non-referential beat gestures to their
referential gestures along with referential iconic gestures gestural repertoire. To our knowledge, only one study [6] has
evolves over time in children’s narrative discourse, and how the examined the alignment between gesture and children’s speech
temporal integration between gestures and prosodic in 12 6-year-olds. This investigation found that children
prominence in speech develops. To assess this question, the produced beat gestures which were linked to lexical words.
current study examines changes in gesture production in terms However, no clear/significant relation with pitch accents was
of referentiality and temporal alignment properties in narrative found, even though some children had already started to
speech using a longitudinal database consisting of 332 produce beats which were aligned with pitch accented syllables.
narratives performed by 83 children at two different time points The current study will complement existing research using a
in development (e.g., 5-6 years of age and two years later). longitudinal database, with more than 80 participants.
Results revealed that non-referential beat gesture production This paper aims to assess gesture development and gesture-
increases significantly during these periods of development. speech alignment in children’s narrative discourse. In line with
Furthermore, this study provides evidence that referential prior research ([4], [5], [6]), we hypothesize that by the age of
iconics and non-referential beat gestures behave similarly in 5, children will already produce beat gestures, and some of them
terms of temporal alignment, showing a high percentage of will have already started to develop the ability to align gestural
gesture integration with prosodic prominence (i.e., between prominence with prosodic prominence, as previous studies have
75% and 85%). shown (e.g., [6]). We think that by the age of 7, the number of
Index Terms: non-referential beat gestures, referential iconic beats used by children will increase significantly, essentially
gestures, temporal alignment, narrative speech taking up a larger proportion of the children’s total gestural
repertoire, reflecting a more adult-like performance.
1. Introduction
2. Methods
Previous research has provided evidence that the development
of gesture in children starts in the earliest stages of language 2.1. Participants
development. By the age of 5-6, children already start
producing non-referential beat gestures in narrative speech (i.e., Eighty-three children from the Girona area of Catalonia took
pragmatically meaningful gestures which highlight important part in [9]’s corpus of Catalan narrative speech development.
linguistic functions in discourse ([1], [2])), and use them to The children were recorded at two different time points: 5 to 6
years of age (Time 1; M = 5.9; SD = 0.55), and two years later
mark discursive relationships in speech ([3], [4], [5]). These
(Time 2; M = 7.98; SD = 0.60). All participants were typically
studies found that as children grow, the number of beat gestures
produced increases. However, further research is still needed to developing children without any prior history of language or
examine the relationship between beats and other types of communication disorders. Parents filled out a family language
gestures in narrative discourse. While some authors have found questionnaire [10] and consented to having their children be
that referential iconics tend to occur more often than non- video-recorded. All children were Catalan-dominant
referential gestures ([4], [5]), as far as we are concerned, no Catalan/Spanish bilinguals (88.39%; SD = 9.6).
prior study has directly compared the production of these
2.2. Materials and Procedure
gestures during development from a longitudinal point of view,
examining the same participants. The current investigation will At age 5-6, children participated in a narrative retelling task, in
contribute to this discussion and will assess the differences in which they were asked to watch two wordless cartoons and
gesture production in children’s narrative speech comparing the retell them to the experimenter. Two years later, the same
use of iconic and beat gestures at two points in development (at children did the same task with the same cartoons they had
the age of 5-6, and two years later). watched at Time 1. For further details on the procedure, see [1]
Gesture and speech have been found to be temporally and [11]. All in all, [9]’s database contains a total of 332 stories
integrated with prosodic prominence in adult’s discourse ([2],
(83 children × 2 stories × 2 time points), totaling approximately 4. Conclusions
166 minutes of speech.
The current study is in accordance with previous research ([4],
2.3. Gesture and Speech coding system [5]), as it shows that by the age of 5, children already produce
non-referential beats. Also, our results show that non-referential
To assess gesture-speech temporal alignment, all target video-
beats increase prominently with age, but outnumber the
recordings were coded in terms of gesture type and gesture
production of iconics, and thus suggest that the increase of beats
phasing and the sections containing gestures were transcribed
is an indication of gestural development in children’s narrative
orthographically. Each story was annotated following the
discourse, as prior studies pointed out (e.g., [4], [5]).
gesture labeling scheme proposed by [12]. Pitch-accented
syllables were also annotated to assess alignment between the Expanding on initial findings from the literature (e.g., [6]),
gesture stroke (i.e., the mandatory phase of the gesture) and we claim that by the age of 5 and 6, children already produce
speech. non-referentials which are aligned with the prosodic structure
of speech. This temporal integration of gesture and speech has
3. Results also been found in referential gestures. Therefore, we argue that
both referential and non-referential gestures align with prosodic
A GLMM analysis was run for each dependent variable: the prominence in children’s narratives.
overall number of gestures, and gesture alignment (coded as
either aligned or misaligned). In both analyses, Time, Gesture 5. Acknowledgements
Type and their interaction were set as fixed factors. Participant
was set as random factor. This research was benefited from funding awarded by the
Spanish Ministry and Competitiveness (FFI2015-66533-P), by
Results showed a main effect of Time (χ2(1) = 18.977, p < the Spanish Ministry of Science, Innovation and Universities,
.001), showing that children produced more gestures at Time 2
Agencia Estatal de Investigación, and Fondo Europeo de
(d = 0.76, p < .001) than at Time 1, regardless of Gesture Type;
Desarrollo Regional (PGC2018-097007-B-100) and by the
and also, a main effect of Gesture Type (χ2(1) = 11.688, p =
Generalitat de Catalunya (2014 SGR_925; 2017 SGR_971).
.001), with more beats (d = 0.41, p = .007) than iconics,
regardless of Time. Moreover, an interaction between Time and
Gesture Type was found to be significant (χ2(1) = 5.212, p = 6. References
.022). Post-hoc analyses revealed that while gesture production [1] I. Vilà-Giménez and P. Prieto, “Encouraging kids to beat:
increased from Time 1 to Time 2 for both gesture types (beat: d Children’s beat gesture production boosts their narrative
= 1.00, p < .001; iconic: d = 0.52, p = .020), there were performance,” Developmental Science, in press, 2020.
significantly more beat gestures than iconics at Time 2 (d = [2] S. Shattuck-Hufnagel and A. Ren, “The prosodic characteristics
0.65, p < .001) (Figure 1). of non-referential co-speech gestures in a sample of academic-
lecture-style speech,” Frontiers in Psychology, vol. 9, no. 1514,
pp. 1–13, 2018.
[3] J. M. Colletta, “Comparative analysis of children’s narratives at
different ages: A multimodal approach,” Gesture, vol. 9, no. 1, pp.
61–69, 2009.
[4] J. M. Colletta, C. Pellenq, and M. Guidetti, “Age-related changes
in co-speech gesture and narrative: evidence from French children
and adults,” Speech Communication, vol. 52, no. 6, pp. 565–576,
2010.
[5] J. M. Colletta et al., “Effects of age and language on co-speech
gesture production: an investigation of French, American, and
Italian children’s narratives,” Journal of Child Language, vol. 42,
no. 1, pp. 122–145, 2015.
[6] M. Mathew, I. Yuen, and K. Demuth, “Talking to the beat: Six-
year-old’s use of stroke-defined non-referential gestures,” First
Language, vol. 38, no. 2, pp. 11–128, 2018.
Figure 1: Number of non-referential beat gestures and [7] S. Shattuck-Hufnagel, A. Ren, M. Mathew, I. Yuen, and K.
referential iconic gestures in Time 1 and Time 2 Demuth, “Non-referential gestures in adult and child speech: Are
they prosodic?,” in Speech Prosody 2016, J. Barnes, A. Brugos,
S. Shattuck-Hufnagel, and N. Veilleux, Eds., 2016, pp. 836–839.
In terms of temporal alignment, results showed a main [8] N. Esteve-Gibert and P. Prieto, “Infants temporally coordinate
effect of Alignment (χ2(1) = 194.949, p < .001), indicating that gesture-speech combinations before they produce their first
both in Time 1 and in Time 2 there were more gestures aligned words,” Speech Communication, vol. 57, pp. 301–316, 2014.
(d = 1.36, p < .001) than misaligned. However, no interactions [9] I. Vilà-Giménez, J. Florit-Pons, G. Gurrado, P. L. Rohrer, and P.
were found to be significant (Time × Alignment, χ2(1) = 0.994, Prieto, “Audiovisual corpus of Catalan children’s narrative
p = .319; Gesture × Type Alignment, χ2(1) = 0.284, p = .594; speech development,” CHILDES corpus, in preparation.
Time × Gesture Type × Alignment, χ2(1) = 2.435, p = .119). [10] L. Bosch and N. Sebastián-Gallés, “Evidence of early language
discrimination abilities in infants from bilingual environments,”
These results show that there are no clear differences between
Infancy, vol. 2, pp. 29–49, 2001.
how children align gestures in the two time points, and that the [11] I. Vilà-Giménez, A. Igualada, and P. Prieto, “Observing
alignment patterns for non-referential beat gestures and storytellers who use rhythmic beat gestures improves children’s
referential iconic gestures are similar (76.67% for beats and narrative discourse performance,” Developmental Psychology,
84.62% for iconics in Time 1, and 79.42% and 76.96%, vol. 55, no. 2, pp. 250–262, 2019.
respectively, in Time 2). [12] P. L. Rohrer et al., “The MultiModal MultiDimensional (M3D)
labelling scheme for the annotation of audiovisual corpora,”
submitted at this conference.

You might also like