The psycholinguistic study of human language acquisition can make important contributions to computational linguistics, and vice versa. The cognitive mechanisms proposed in psycholinguistic theories can serve as constraints on computational models that would enable them to learn language as efficiently as humans do. Experimental designs from psycholinguistics can inspire computational linguists to evaluate their computational models based on the similarity of the models to human performance. In the other direction, computational approaches to language acquisition could be beneficial for the psycholinguistic community by providing the means to quantitatively evaluate possible theories of language learning.
L2HM is an interdisciplinary 2-day workshop aimed at bringing together researchers who work on language acquisition from both psycholinguistic and computational perspectives. We are especially interested in:
1. Computational studies of language acquisition related questions, using realistic, large scale data (natural and/or artificial).
2. Psycholinguistic work on language acquisition, either theoretical or experimental.
We invite submissions for poster presentations on experimental and/or modeling approaches to human and machine language learning by April 10th 2018 (23.59 CET).
Submissions are now closed.
February 9th: abstract submission opens
April 10th: abstract submission deadline
May 10th: notification of abstract acceptance
June 29th: poster uploaded to L2HM's OSF (see https://osf.io/view/L2HM/ )
Abstracts must be submitted as plain text.
Abstract text can be a maximum of 500 words (including references if used)
The body of the abstract should be fully anonymous and indicate the current state of the work (i.e. complete or in progress—both will be considered).
Figures and tables may be appended to the submission in .PNG, .JPG, .TIFF, or .PDF format (up to four files of max. 4 MB each).
Each abstract will be rated by two reviewers for fit with the goals of L2HM, completeness, methodological and analytical soundness, originality/innovation/novelty, and clarity. Authors will be notified by May 10th, 2018.
By its nature, this workshop will be interdisciplinary and methodologically broad. We provide here some example topics for abstract submissions. This list is far from exhaustive, so if you are unsure whether your topic is relevant for the workshop, please don’t hesitate to contact us at l2hm2018@sciencesconf.org..
One- and “few”-shot learning
Scalability and natural datasets for modelling human development
Incorporating cognitive constraints into word learning models
How might infants and artificial agents use extra-linguistic cues to learn new words and understand sentences?
What are the differences and similarities in the learning of nouns, verbs and function words?
Semantic and syntactic bootstrapping: from developmental evidence to modelling
Need help? Email us at l2hm2018@sciencesconf.org.