• UABDivulga

How do we acquire language: does frequency in the input matter?

aprendre un llenguatge
Most of the time we adults take language for granted, unless one day we have to learn a new one. Then, things change radically. However, things seem to work differently for her (here a picture of a baby). She can’t tie a simple knot, draw a decent circle or eat without making a mess, but she can learn on average 10 new words per day. At 17 months she cannot produce even a two-word utterance, but she can produce multiword sentences with little deviation from their target language just one year later. For this reason, the question as to how do children acquire their native language has prompted a lively theoretical debate and a great deal of empirical work.

A study by researchers at the Center of Theoretical Linguistics de la Universitat Autònoma de Barcelona (CLT-UAB) analyses whether the acquisition of the first language (L1) is precipitous or is gradual in response to input frequency against the empirical domain of subject and (direct) object drop in Mandarin. Unlike Spanish or Catalan, Mandarin Chinese is considered as a discourse-oriented language, meaning that both grammatical subjects and objects may be omitted as long as their references can be retrieved from the preceding discourse context. Still, not all objects can be optionally dropped; in particular, the drop of objects in the so-called ba construction is ungrammatical in Mandarin.

Thus, at which age do the Mandarin-speaking children realize this peculiarity of their mother tongue? The answer is at one year and eight months. This answer has been confirmed from a corpus-based study of the naturalist speech of 47 Mandarin-speaking children aged one year and two months (1;2) to six years and five months (6;5) and their caregivers from the CHILDES database.

The results have been published in Journal of Child Language (accessible at https://doi.org/10.1017/S0305000919000114), which show that before age 1;8, all the children omitted subjects and objects in an adult-like manner, even if the presence of disambiguating evidence for this is very low. Besides, children’s ba constructions, which require an object obligatorily, included this object from the first occurrence although its frequency was scare in the input (only 3.1% in adult speech).

Taking into account the evidence, the authors suggest that the acquisition of L1, at least for certain structures, occurs early independently of the input (against the claims of Yang, 2002, 2004). Of course, this does not exclude the role of frequency in language acquisition, as has been highlighted by the authors in the article. Rather, it provides a reconsideration of the role of input in language acquisition, which indicates children are not simply accelerating (or delaying) the acquisition of certain structures based on what appears more (or less) frequently in their input.

Zhu Jingtao i Anna Gavarr
Centre de Lingüística Teòrica
Universitat Autònoma de Barcelona


Yang, C. (2002). Knowledge and learning in natural language. New York: Oxford University Press.
Yang, C. (2004). Universal grammar, statistics, or both? Trends in Cognitive Science, 8(10), 451-6.
Zhu, J.T., & Gavarró, A. (2019). Testing language acquisition models: Null and overt topics in Mandarin. Journal of Child Language, 46(4), 707-732. DOI: https://doi.org/10.1017/S0305000919000114.

View low-bandwidth version