Learning how to inflect verbs in different languages

This project considered how children learn to inflect verbs to convey different meanings such as person (1st vs 2nd: in English, "I" vs. "you"), number (singular vs. plural: in English, "is" vs. "are") and tense (present vs. past: in English, "is" vs. "was").

English has a very limited range of inflections, but many other languages have far more, and we know little about the ways in which children start to use them and develop the ability to use the full set of inflections across a wide range of contexts.

In this project, we looked at two highly inflected languages (Polish and Finnish). We studied the development of children’s use of these inflections and, in particular, the development of productivity, (i.e. children’s ability to use them across a wide range of verbs and even on verbs they have never heard before). We ran large-scale elicited production studies, where native learners of Finnish (N = 77; 35–63 months) and Polish (N = 81; 35–59 months), were shown animations of different characters (1st, 2nd, 3rd person; singular and plural) performing various actions, and asked to produce both the pronoun and the inflected present-tense form of the verb. Verbs were chosen across a range of token frequencies (e.g. very commonly used to very rarely used) and from verb classes varying in phonological neighbourhood density (PND) (e.g. words with lots of other similar sounding words to words with very few similar sounding words), with counts taken from child directed speech corpora and standard grammar dictionaries respectively.

Analysis with mixed-effects models revealed that, for both languages, despite overall low error rates, children made more errors with verb forms with lower token frequencies in the input  and with verbs belonging to classes with lower PNDs These findings suggest that successful models of children’s acquisition of verb morphology need to be sensitive to the statistical properties of children’s input, i.e., both token frequency (reflecting children’s retrieval of individually stored verb forms) and PND (children’s use of phonological analogy).

We then built and tested a connectionist computational model of the acquisition of person/number marking which not only acquires near adult-like mastery of the system (including generalisation to unseen items), but also yields all of the key phenomena observed in the elicited-production studies; specifically, effects of token frequency and phonological neighbourhood density of the target form, and a pattern whereby errors generally reflect the replacement of low frequency targets by higher-frequency forms of the same verb, or forms with the same person/number as the target, but with a suffix from an inappropriate conjugation class. The findings demonstrate that acquisition of even highly complex systems of inflectional morphology can be accounted for by a theoretical model that assumes rote storage and phonological analogy, as opposed to formal symbolic rules.

We also intended to compare methods of testing productivity in Chintang (a Tibeto-Burman language of East Nepal).  Data was obtained for this study but having examined it, it was unclear that there was a sufficient quantity to apply the productivity method developed by Pine (see Aguado-Orea, J., & Pine, J. M. (2015). Comparing Different Models of the Development of Verb Inflection in Early Child Spanish. PLOS ONE, 10(3) DOI: 10.1371/journal.pone.0119613).

Project Team: Elena Lieven (Lead), Julian Pine, Sabine Stoll, Grzegorz Krajewski, Jeffrey Elman, Joanna Kolak, Sonia Granlund and Felix Engelmann

Start Date: March 2015

Duration: 3 years

(Work Package 6)

Selected Outputs

Engelmann, F., Granlund, S., Kolak, J., Szreder, M., Ambridge, B., Pine, J., Theakston, A., and Lieven, E. (2019). How the input shapes the acquisition of verb morphology: Elicited production and computational modelling in two highly inflected languages. Cognitive Psychology, 110, 30-69. https://doi.org/10.1016/j.cogpsych.2019.02.001

Engelmann, F., Kolak, J., Vihman, V., Granlund, S., Ambridge, B., Pine, J., Theakston, A., Lieven, E. (2018). How the input shapes the acquisition of inflectional morphology: Computational modelling across three highly inflected languages. In Proceedings of the 24th AMLaP conference, Architectures and Mechanisms of Language Processing, Berlin, Germany.

Engelmann, F., Kolak, J., Granlund, S., Szreder, M., Ambridge, B., Pine, J., Theakston, A. & Lieven, E. (2017). The acquisition of Polish and Finnish verb inflection in a connectionist model. Paper presented at the 50th Annual Meeting of Societas Linguistica Europaea (SLE), Zurich, Switzerland.

Granlund, S., Kolak, J., Szreder, M., Engelmann, F., Ambridge, B., Pine, J., Theakston, A. & Lieven, E. (2017). Predicting errors in children’s production of verb morphology: evidence from person/number marking in Finnish and Polish. Paper presented at the 14th Congress of the International Association for the Study of Child Language (IASCL), Lyon, France.


Engelmann, Felix and Granlund, Sonia and Kolak, Joanna and Szreder, Marta and Ambridge, Ben and Pine, Julian and Theakston, Anna and Lieven, Elena (2021). International Centre for Language and Communicative Development: How the Input Shapes the Acquisition of Verb Morphology: Elicited Production and Computational Modelling in Two Highly Inflected Languages, 2014-2020. [Data Collection]. Colchester, Essex: UK Data Service. 10.5255/UKDA-SN-853914