James R. Hurford 2014. The Origins of Language. A slim guide. Oxford: Oxford University Press, 173 pp.

A wonderfully concise and beautifully written book from which I learned many interesting things. If one were to strive to find a critical comment it might be that Hurford has not so much to offer on the earliest steps in language evolution as in the later development of languages that are already sophisticated. But perhaps I need to read his other not-so-slim books.

The first chapter is a short, accurate and fascinating account of the prehistory and evolution of Homo sapiens and our antecedents. A brilliant summary that gives the reader great confidence in the rest of the book.

Hurford presents what seems (to me) a highly significant example of an experiment in self organising language development. I couldn’t locate the page reference again quickly but, if I recall correctly, researchers gave volunteers nonsense words and structures and asked them to try and develop ways of communicating meaning using these meaningless words. Then the researchers gave the system the first group developed to a new group of subjects with the same instruction. After only 7 iterations, a meaningful proto-language emerged.

A further inspirational example (and the best-documented such case) of the origin and evolution of a language from scratch is Nicaraguan Sign Language in which a group of deaf children in Nicaragua, isolated from other such children, spontaneously evolved their own independent sign language. In about 20 years it became a fully-fledged language, and the story was observed and studied by researchers. With this exciting case so well documented, it isn’t hard to imagine swift origin of many languages in our own evolution. (Well, obviously it must be so since it did happen.)

Nicaraguan Sign Language, and many other cases among languages, show that languages spoken by small isolated groups become complex; languages whose speakers mix with others and become larger populations speaking a shared language lead to simpler constructions as speakers arrive at simpler common meanings and structures.

Words, sounds and other language components are selected for. Only those that are able to be spoken and understood most easily and accurately will be useful, and will prevail.

Clauses, of at most 4 objects, are a useful unit of meaning in grammar. Corresponding with the number of objects we can grok without counting. This observation applies to apes too, so is likely an ancient constraing. (Contra the number 7 prposed by psychologist George Miller as the maximum number we can grasp without writing and store in short term memory.

Children learn languages to fit in. Adults do so to communicate, and thus learn pragmatically and minimally.

The Khoisan languages (click-based languages spoken in the Kalahari region) are unique in that they can be spoken independently of breathing. Handy for folk walking long distances. These peoples are a group that have been genetically isolated for about 100,000 years.

Computer modelling of the origins of vowels (as recognisably distinct sounds arranged discretely in “sound space”) often arrives at 5, as in English. Or sometimes more, sometimes fewer, but always close to 5 (p. 152).

Learned some great words too: Fricatives are sounds made through almost-closed lips (like “the”). Lenition is the softening of sounds (p. 160).

The bibliography is as slim and concise as the rest of the book but there is a whole lot of reading including some pdfs at Jim Hurford’s web page. Jim Hurford was one of the first researchers to use computers to simulate and model language.