Title: Protolanguages that are Semi-holophrastic
1Protolanguages that are Semi-holophrastic
Mike Dowman mike_at_sacral.c.u-tokyo.ac.jpCo
mplex Systems and Artificial Life Lab, University
of Tokyo
- 1. What were Pre-syntactic Languages Like?
- Words for entities and actions (Bickerton)
- ? Strings of words with no syntax
- One word for whole complex meaning (Wray)
- 2. Modelling Protolanguage
- Meanings made from ten semantic primitives
- Complex meanings made from a set of three
primitives - Agents could use a limited number of words
- Agents remembered words
- and the meanings they were
- used to express
- 3. Iterated Learning
- Agent in each of 10 generations names 1000
randomly selected meanings - Each time they choose the best match to the
meaning, based on past uses of the words they know
4. Many Meanings, Few Words Ten words, ten
semantic primitives ? Most words denoted
single concepts
5. Many Words, Few Meanings 150 words, ten
semantic primitives ? All words were
holophrastic
6. A Semi-holophrastic Protolanguage 50 words,
ten semantic primitives Words varied
in their degree of holophrasticity
7. Coevolution
- 8. Conclusions
- Words dont have to denote single concepts or
holophrases - ? There are all sorts of possibilities in between
- Protolanguages could have contained mixtures of
different types of words - Protolanguages probably changed gradually over
the course of human history