The second half of chapter 5 asks what shapes individual languages settle into as generations pass. The key term is “cultural attraction.”
Grammaticalization: words become grammar (5-5)
Ostensive communication lets speakers use existing words with new meanings, because hearers infer the rest. Accumulated over time, content words gradually take on grammatical functions. This is grammaticalization; “be going to” (movement to future), seen at the start of chapter 5, is the classic case. The author reinforces the view that grammar was not there from the start but grows out of accumulated conventions.
Transmission is rebuilding, not copying (5-6)
The book contrasts cultural transmission with biological replication (pp. 239–241).
- In biological evolution, genes are copied as they are.
- In cultural transmission, things are made anew at each step. A student taking lecture notes does not copy the lecturer’s spelling; the student rewrites the correct spelling.
At each rebuild the input is nudged toward what suits the person’s purposes and mind. The nudging force is a factor of attraction; where things end up is an attractor. Given enough time, cultural traits move toward forms that fit the environment of human bodies, minds, and behaviour, much as biological adaptation does.
The book gives three examples.
- Basic colour terms: the physiology of the eye decides which colours are easy to divide, and word systems drift toward that.
- Tone: a correlation has been reported between the distribution of certain genes and of tonal languages. The author is cautious: correlation alone does not show causation, and the pattern can also be explained by shared history (pp. 239–240).
- Segmentation: in a new experiment, young children showed manner and path of motion with separate gestures, while older children and adults combined them. Since language is passed on through small children, a tendency to break information into linear pieces shapes language that way (p. 242).
The author expects this tendency to bear on the hierarchical structure of language as well (p. 243).
Languages that decay (5-7)
The second pillar is that how a language is used shapes its form. The author describes a transmission-chain experiment (pp. 244–247).
Setup. Participants learn an artificial language of 27 words. Each word refers to a combination of shape (square, triangle, circle), colour (red, blue, black), and motion (straight, spinning, bouncing). The first language is random. One person learns it, and that person’s output becomes the next person’s training data. This runs for ten generations.
Result one. Left alone, the language degrades. Within a few generations many meanings attach to one word (for example, every square is “bubime” regardless of colour or motion). The language simplifies toward what is easy to learn and recall. One participant afterwards wondered whether the aliens were colour-blind.
Result two. If the experimenters remove homonyms (one word form with two or more unrelated meanings) from the training data, things change. Over generations the language becomes structured: separate syllables attach to each component of shape, colour, and motion. For example, all black figures share a syllable regardless of shape or motion. Thus morphemes, the smallest meaning-bearing units, arise inside words.
The author’s reading is that avoiding homonymy (keeping things distinguishable) and ease of learning (simple composition) act together as two attractors. At their balance point, the language moves to a structure with the fewest elements that still tell things apart, one that also fits the human perceptual categories of colour, shape, and motion (p. 246).
Against “ambiguity is a flaw”
Chomsky argued that language is ambiguous and poorly designed for communication. The author says this is exactly backwards (pp. 248–250).
- Speakers need not spell out what context can supply, so short ambiguous expressions are efficient.
- By relevance theory (article 279), the hearer’s processing effort should be small.
- Prediction: the easier a word is to use (short, frequent), the more ambiguous it will be. The book reports this in English, German, and French: the most usable words have the most meanings (p. 250).
So ambiguity is an expected property of an ostensive communication system, not a defect, because context fills the gaps.
Where this section stands
The author presents this section as a research programme for why individual languages have their current forms, not a finished answer (pp. 251–252). It is also an alternative to the view that Universal Grammar fixes the form of languages, which leads to the next article (282).
Note
The experiments described here are the book’s summary; I have not checked the original papers’ conditions or figures. On tone and genes, note that the book itself does not assert causation.