Animal grammar learning
This week’s tutorial picks up on a topics we discussed last week, on the grammar learning capacities of non-humans. In this tutorial you’ll read and discuss Truswell (2017), which is a short critique of a well-known monkey artificial grammar learning experiment, plus an interesting analysis of the comprehension abilities of a language-trained ape. The author, Prof. Rob Truswell, is based here in Edinburgh, you might know him.
The first part of the paper provides a discussion and critique of Fitch & Hauser (2004). The description is quite brief (although the 2004 paper is quite brief, so if you are interested just read it!), so to elaborate a little bit: Fitch & Hauser exposed cotton-top tamarins (small new world monkeys) and humans to one of two sets of sequences generated by simple artificial grammars. Both grammars involved sequences of syllables spoken by a female and male speaker - A syllables were spoken by the female and were ba, di, yo, tu, la, mi, no, or wu; B syllables were spoken by the male and were pa, li, mo, nu, ka, bi, do, or gu. In the (AB)n condition sequences consisted of repetitions of AB pairs - e.g. you might hear “di mo yo pa” and “di pa wu bi tu nu” where syllables in italics were spoken by the female. In the AnBn condition sequences consisted of a sequence of A syllables followed by a sequence of B syllables, e.g. you might hear “di yo gu do” and “no tu wu ka mo gu”. Participants were then tested on whether they could differentiate between novel (i.e. unheard) (AB)n and AnBn sequences. Humans trained on either grammar could do this, and tamarins trained on the (AB)n grammar succeeded; however, the tamarins trained on the AnBn sequences did not differentiate between sequences generated by the grammar they were trained on and the other sequence type, i.e. they appear not to have learned the AnBn grammar. Fitch & Hauser conclude “the current findings suggest that tamarins suffer from a specific and fundamental computational limitation on their ability to spontaneously recognize or remember hierarchically organized acoustic structures.”
One rather sad footnote to this paper: Marc Hauser was found to have committed scientific misconduct in other scientific papers. I am not aware of any suggestion that this particular paper is unreliable, and the first author (Tecumseh Fitch) is an extremely careful and rigorous scientist whose work I trust implicitly. But the data in this case comes from the Hauser lab. That’s not actually very relevant in evaluating Truswell’s critique of this paper, but worth knowing in general.
The second part of the paper is an evaluation of Savage-Rumbaugh et al.’s (1993) data on Kanzi’s interpretation of spoken English commands, with a focus on what they tell us about Kanzi’s knowledge of English grammar, specifically his sensitivity to linear order and hierarchical structure. Kanzi is a male bonobo (a species closely related to common chimpanzees) who was raised by humans in a research lab in quite unusual circumstances. You can see a video of Kanzi performing a similar task (or maybe this is some of the actual data?) on youtube, worth watching all the way through once you have read the paper, applying Truswell’s eye to Kanzi’s actions!
Questions for discussion:
Fitch, W.T. and Hauser, M. 2004: Computational constraints on syntactic processing in a nonhuman primate. Science, 303, 377–80. https://doi.org/10.1126/science.1089401
Savage-Rumbaugh, E.S., Murphy, J., Sevcik, R., Brakke, K., Williams, S., Rumbaugh, D. and Bates, E. 1993. Language comprehension in ape and child. Monographs of the Society for Research in Child Development, 58, 1–252. https://www.jstor.org/stable/1166068
Truswell, R. (2017), Dendrophobia in Bonobo Comprehension of Spoken English. Mind and Language, 32, 395-415. https://doi.org/10.1111/mila.12150
All aspects of this work are licensed under a Creative Commons Attribution 4.0 International License.