Evolang previews: Holistic or synthetic protolanguage: evidence from iterated learning of whistled signals

Guest post by Tessa Verhoef

Evolang is busy this year – 4 parallel sessions and over 50 posters.  We’ll be positing a series of previews to help you decide what to go and see.  If you’d like to post a preview of your work, get in touch and we’ll give you a guest slot.

Tessa Verhoef, Bart de Boer and Simon Kirby Holistic or synthetic protolanguage: evidence from iterated learning of whistled signals.
Lecture room 3, Fri. 16th, 14.25

In this talk we will present results of an iterated learning experiment about the emergence of structure in sets of whistle sounds produced with a slide whistle. We will link these results to the debate on the nature of human protolanguage.

In the protolanguage debate, it is being discussed whether human protolanguage was holistic or synthetic. We present data that corroborates arguments in favour of holistic protolanguage and forms a counterexample to some arguments that were proposed against it. Tallerman (2007) argues that the rapid change of language would prevent analysis of a holistic language into components. In addition, she argues that the task of segmenting signals into components is impossible because even the tiniest distinctions between signals should be considered potentially significant.

The slide whistle used in the experiment


The experiment involves iterated learning and recall (based on Kirby et al., 2008) of a set of (initially) holistic whistles produced with a slide whistle. Chains were created in which a participant was exposed to the recall output of the previous participant in the same chain (or to the initial set).


Cultural transmission causes the systems of whistles to become regularized and more learnable. They become more constrained and predictable, which allows learners to focus on relevant signal variations (Verhoef, Kirby, & Padden, 2011). A significant cumulative decrease of recall error shows that the whistle sets become more learnable and a significant cumulative decrease of entropy indicates an increase of structure. This figure shows and example fragment of an emerged structured set with whistles plotted as pitch tracks:


The results show that iterated learning and recall causes continuous, holistic signals to transform into a structured set of combinatorial signals. Increased predictability helps participants to focus on relevant distinctions, allowing segmentation. Therefore the arguments that the changeability of holistic language and the lack of distinctive features would prevent emergence of structure is not supported by empirical evidence from modern human behavior. Although our findings refute arguments against holistic protolanguage, they also show that an extended holistic phase would perhaps not be likely. It would not be stable for many consecutive generations because, as we see in our experiments, combinatorial structure would emerge rapidly. We do not exclude the possibility that aspects of both holistic and synthetic protolanguage existed in one system.


Kirby, S., Cornish, H., & Smith, K. (2008). Cumulative cultural evolution in the laboratory: An experimental approach to the origins of structure in human language Proceedings of the National Academy of Sciences, 105 (31), 10681-10686 DOI: 10.1073/pnas.0707835105

Tallerman, M. (2007). Did our ancestors speak a holistic protolanguage? Lingua, 117 (3), 579-604 DOI: 10.1016/j.lingua.2005.05.004

Verhoef, T., Kirby, S., & Padden, C. (2011). Cultural emergence of combinatorial structure in an Artificial whistled language. In Proceedings of CogSci 33 (pp. 483–488).

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.