Language change and the evolution of languages can be seen as an evolutionary process. Human brains form the environment that constrains language. Language acquisition provides the replication, recombination, and mutation for language. Thus, the whole process is a social transmission or evolution, but with specific constraints from the biology of the brain. Thus, it shares much in common with biological evolution, but also a lot in common with standard meme-theoretic social evolution.

What theories of the change and evolution of language has been studied? How complete and well understood is the "phylogenetic tree" of languages?


I am not targeting this question at the evolution of language (note the singular). I.e. I am not interested in how language faculties first appeared in ancestors of humans. I am interested in the change of languages since then.

I am also not that interested in the sort of micro-evolution of small changes in a specific language except in how it applies as a mechanism of mutation, recombination, or other form of novelty-provision within the longer-period of evolution of languages.

Specific references to the literature are preferred, but general summaries of trends that I can then use Google Scholar to follow are also fine.

One of the goals of this question is to look at both sides. On the one hand, understanding what models most resemble evolution of languages can give us insight into linguistics. On the other hand, insights from linguistics can provide us with a model with which to gain more insight about evolutionary processes (not all evolutionary process need be biological evolution).

  • 4
    So basically you're asking for a summary of the entire field of historical linguistics? Are you sure this isn't just a bit overly broad? – rintaun Sep 22 '11 at 11:19
  • 2
    @rintaun I am not trying to ask this. I am really asking what conceptual/theoretical model of evolution (in the broad sense) is most appropriate for the evolution of language. There are various models of evolution that are studied in mathematical biology with only a subset of them being classical Darwinian evolution (for instance memetic evolution or lamarckistic evolution are conceptually different and make different predictions). If you have any advice on how to make my question more clear I would really appreciate it. – Artem Kaznatcheev Sep 22 '11 at 11:24
  • 4
    I think you dismiss "changes in pronunciation" too quickly. Pronunciation changes are, more often than not, very regular and systematic, and actually formed the bedrock foundation for historical language reconstruction. – JoFrhwld Sep 22 '11 at 15:14
  • @JoFrhwld I removed the contested point. I was really just looking for a quick example and did not think too seriously about what I put in the parenthesis. Thank you for setting me straight. – Artem Kaznatcheev Sep 23 '11 at 3:52

I would take a look at some of Charles Yang's work, especially "Knowledge and Learning in Natural Language".

A lot of language change can be adequately modeled as genetic competition. For grammatical parameter X, there are competing variants a and A. These two variants have different fitness profiles, so one eventually replaces the other. This model works, fits the data well, and allows us to make interesting inferences and arguments.

As for a detailed model of inheritance, the world is very tricky. In biological inheritance, an organism has at most 2 parents. However, in linguistic inheritance, the "genotype" of your parents is often less important to your own genotype. Rather, there is some kind of aggregate inheritance from your community at large.

| improve this answer | |
  • 1
    the multiple inheritance is one of the things that makes the evolution of languages interesting from the point of view of studying evolution. It provides you with a model that can have features distinct from what you already have in a system like biological evolution. – Artem Kaznatcheev Sep 23 '11 at 3:45

Despite the superficial analogies, both at the large scale descriptive level (languages tend to bunch together in a tree-like fashion) and at the very theoretical level ('information' transfer over a 'noisy' channel), the mechanisms of language change are radically different from speciation.

  • Two languages, no matter how far apart, can be combined by a younger generation into a pidgin/creole that becomes a rich viable language incomprehensible to its 'parents'.
  • One language can become the source for vocabulary and the other for grammar.
  • Languages that started off from very different sources can become closer by proximity (via Sprachbunds, like the Balkan Languages).
  • There is nothing like DNA recombination in language.
  • There is no 'natural' selection
  • sex and death, two primary parts of the mechanism of biological evolution of languages, hardly correspond with the obvious language versions (overlapping speech language communities and extinction respectively).
  • there is no analogous competition for resources or mates in language
  • there are no 'niches' filled by lack of competition in language.

On the other hand one analogy that does supply change is the concept 'mutation', which for biological speciation comes from molecular transcription errors in DNA; in language it comes from aural distortions, production errors, and mental processing mistakes.

That all said, the major differences are that the mechanisms of change are very different (biology is about survival and reproduction, language is about either straightforward resolving noisy channels or sociological acceptance).

| improve this answer | |
  • thanks for the list of dis-analogies with biological evolution. But it seems harder to produce a similar list for dis-analogies with memetics... both are types of evolutionary processes. So I will take this answer as evidence for language evolution being closer to that of memes than genes, or would that be an incorrect interpretation on my part? – Artem Kaznatcheev Sep 22 '11 at 17:25
  • @Artem: Not knowing of any real technically describable evolution of 'memes' (other than what I can informally figure), I'd agree and say meme transfer and evolution is closer to that for languages than to genes (discounting that currently the most prominent technique for transfer of memes is human language itself). One might go so far as to -not- discount and say that memes really are just one aspect of human language, just not traditionally a matter for linguists (since memes seem to be (mostly) language independent). – Mitch Sep 22 '11 at 17:38
  • 6
    The no natural selection point is contestable. Grammatical variants "compete" for use by a speaker, and for interpretation by a hearer. If you have two competing grammars, there are three kinds of context they can arise in: X- either grammar could have produced the string, A- only grammar A could have produced the string, B- Only grammar B could have produced the string. If there are more B contexts than A contexts, then grammar B is more fit, and selected for. – JoFrhwld Sep 22 '11 at 17:43
  • @JoFrhwld: in your answer I see the possibility of some natural selection. Without seeing the data or definitions, I would informally expect that the analogy is too far of a stretch. I'll check it out. – Mitch Sep 23 '11 at 3:06
  • 2
    @Mitch, here's a paper that lays some of it out: ling.upenn.edu/~ycharles/papers/lvc.pdf. Even if it's an analogy that goes too far, it's an analogy that works. – JoFrhwld Sep 23 '11 at 3:54

Someone who's done a fair amount of work on studying the applicability of evolutionary metaphors (with greater or lesser degrees of abstraction) is Andrew Wedel (who has the additional benefit of PhDs in both biology and linguistics). In particular, he wrote "Exemplar models, evolution and language change" that explicitly addresses (i) pruning of lines of inheritance, (ii) blending inheritance, and (iii) natural selection as pathways of evolutionary language change.

He has several other papers along with this one that discuss evolutionary dynamics in language change and acquisition, most of which deal with fairly low-level sound change, and many of which involve simulation/modelling couched in a particular non-generative framework (exemplar/memory-based). Following the links on his pubs page and the references therein should net you much.

| improve this answer | |

William Croft argues for a model of language that is usage-based (rather than child-based) and is based on David Hull's Generalized Analysis of Selection evolutionary model.

see Croft's (2000) book "Explaining language change", also this (2009) article by Baxter, Blythe & Croft for more.

| improve this answer | |

Have a look at the following book:

Keller, Rudi. 1994. On language change: The invisible hand in language. London: Routledge.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.