Communicative cues in the absence of a human interaction partner enhance 12-month-old infants’ word learning

Tsuji, Sho; Jincho, Nobuyuki; Mazuka, Reiko; Cristià, Alejandrina

doi:10.1016/j.jecp.2019.104740

Cited by 19 publications

(24 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Together, these results indicate that infants identified the objects targeted by pointing as referents and linked them with the co-occurring words, albeit their looking response was short-lived. Such temporal profile of response conforms with the dynamics of infants' looking behavior reported in the looking-while-listening tasks (e.g., Schaffer& Plunkett, 1998;Tsuji et al, 2020). A complementary analysis of infants' first looks performed after the test question indicated that their referent selection and ensuing word mapping were also evident at the level of first gaze shifts executed in response to the test words (i.e., more saccades directed to the targets of pointing than to distractors in the trained-word condition, see Supplementary Materials SM2 First Looks).Interestingly, upon hearing the novel words, infants initially oriented towards the distractor objects (test bin 0-1 s: M = −.29, SD =.51).…”

supporting

confidence: 81%

“…However, no studies to date have provided evidence that the expectation of co-reference between words and actions contributes to referent selection for novel words (Hollich et al, 2000). Although 12-to 13-month-olds were shown to acquire word-object mappings coupled with communicative actions (Woodward et al, 1994;Tsuji et al, 2020), their performance could be explained without appealing to action interpretation or reference. Since even non-communicative object-directed actions orient infants' attention towards targeted items (Daum & Gredebäck, 2010;Daum et al, 2009), successful word mapping following gaze shifts or pointing might have been supported solely by the formation of associative links between stimuli that co-occur (i.e., the attended objects and the concurrently uttered labels).…”

Section: Introductionmentioning

confidence: 97%

“…Therefore, in production, they appreciate the referential nature and communicative significance of pointing. Finally, infants at this age rapidly form associations between words and objects trained in lab settings (Pomiechowska & Gliga, 2019;Tsuji et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Nonverbal action interpretation guides novel word disambiguation in 12-month-olds

Pomiechowska¹,

Csibra²

2020

Preprint

View full text Add to dashboard Cite

Whether young infants can exploit socio-pragmatic information to interpret new words is a matter of debate. Based on findings and theories from the action interpretation literature, we hypothesized that 12-month-olds should distinguish communicative object-directed actions expressing reference from instrumental object-directed actions indicative of one’s goals, and selectively use the former to identify referents of novel linguistic expressions. This hypothesis was tested across four eye-tracking experiments. Infants watched pairs of unfamiliar objects, one of which was first targeted by either a communicative action (e.g., pointing) or an instrumental action (e.g., grasping) and then labeled with a novel word. As predicted, infants fast-mapped the novel words onto the targeted objects after pointing (Experiments 1 and 4) but not after grasping (Experiment 2) unless the grasping action was preceded by an ostensive signal (Experiment 3). Moreover, whenever infants mapped a novel word onto the object indicated by a communicative action, they tended to map a different novel word onto the distractor object, displaying a mutual exclusivity effect. This reliance on nonverbal action interpretation in the disambiguation of novel words indicates that socio-pragmatic inferences about reference likely supplement associative and statistical learning mechanisms from the outset of word learning.

show abstract

supporting

confidence: 81%

Section: Introductionmentioning

confidence: 97%

See 1 more Smart Citation

Nonverbal action interpretation guides novel word disambiguation in 12-month-olds

Pomiechowska¹,

Csibra²

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Thus, it is still an open question whether the temporal contingency manipulation would be successful in the absence of a broader social context containing human agents. A more recent study controlled for these factors by displaying a virtual agent that was contingently reacting to 12-month-old infants' gaze via gaze-contingent eyetracking, and teaching them novel word-object associations (Tsuji, Jincho, Mazuka, & Cristia, 2020). The contingent reactions displayed by the on-screen avatar included mutual gaze and gaze following, but no broader social context such as a prolonged preceding interaction phase.…”

Section: Learning From Interactive Media In the Absence Of Humansmentioning

confidence: 99%

“…Thus, instead of seeing the toddler displayed on screen in real time, the experimenter saw the toddler's gaze position in real time, and was instructed to react accordingly. The third group of toddlers saw a virtual agent identical to the one used in Tsuji et al (2020). Script and reactions of the virtual agent were matched to those of the experimenter in the video chat group.…”

Section: The Present Studymentioning

confidence: 99%

Toddler word learning from contingent screens with and without human presence

Tsuji

Fiévet

Cristià

2021

Infant Behavior and Development

Self Cite

View full text Add to dashboard Cite

While previous studies have documented that toddlers learn less well from passive screens than from live interaction, the rise of interactive, digital screen media opens new perspectives, since some work has shown that toddlers can learn similarly well from a human present via video chat as from live exposure. The present study aimed to disentangle the role of human presence from other aspects of social interactions on learning advantages in contingent screen settings. We assessed 16-month-old toddlers' fast mapping of novel words from screen in three conditions: in-person , video chat, and virtual agent. All conditions built on the same controlled and scripted interaction. In the in-person condition, toddlers learned two novel word-object associations from an experimenter present in the same room and reacting contingently to infants' gaze direction. In the video chat condition,tthe toddler saw the experimenter in real time on screen, while the experimenter only had access to the toddler's real-time gaze position as captured by an eyetracker. This setup allowed contingent reactivity to the toddler's gaze while controlling for any cues beyond these instructions. The virtual agent condition was programmed to follow the infant's gaze, smile, and name the object with the same parameters as the experimenter in the other conditions. After the learning phase, all toddlers were tested on their word recognition in a looking-while-listening paradigm. Comparisons against chance revealed that toddlers showed above-chance word learning in the in-person group only. Toddlers in the virtual agent group showed significantly worse performance than those in the in-person group, while performance in the video chat group overlapped with the other two groups. These results confirm that in-person interaction leads to best learning outcomes even in the absence of rich social cues They also elucidate that contingency is not sufficient either, and that in order for toddlers to learn from interactive digital media, more cues to social agency are required.

show abstract

Grounding the Vector Space of an Octopus: Word Meaning from Raw Text

Søgaard

2023

Minds & Machines

View full text Add to dashboard Cite

Most, if not all, philosophers agree that computers cannot learn what words refers to from raw text alone. While many attacked Searle’s Chinese Room thought experiment, no one seemed to question this most basic assumption. For how can computers learn something that is not in the data? Emily Bender and Alexander Koller (2020) recently presented a related thought experiment—the so-called Octopus thought experiment, which replaces the rule-based interlocutor of Searle’s thought experiment with a neural language model. The Octopus thought experiment was awarded a best paper prize and was widely debated in the AI community. Again, however, even its fiercest opponents accepted the premise that what a word refers to cannot be induced in the absence of direct supervision. I will argue that what a word refers to is probably learnable from raw text alone. Here’s why: higher-order concept co-occurrence statistics are stable across languages and across modalities, because language use (universally) reflects the world we live in (which is relatively stable). Such statistics are sufficient to establish what words refer to. My conjecture is supported by a literature survey, a thought experiment, and an actual experiment.

show abstract

Communicative cues in the absence of a human interaction partner enhance 12-month-old infants’ word learning

Cited by 19 publications

References 40 publications

Nonverbal action interpretation guides novel word disambiguation in 12-month-olds

Nonverbal action interpretation guides novel word disambiguation in 12-month-olds

Toddler word learning from contingent screens with and without human presence

Grounding the Vector Space of an Octopus: Word Meaning from Raw Text

Contact Info

Product

Resources

About