Fig 8.2 Architecture of early single-symbol stage
At this point, then, the syste mhas an architecture like Fig. 8.2. The sy mbols of the syste mare long-ter m me mory
associationsbetweenmeanings and auditory/motorcodes;wemightcallthem“paleo-lexicalitems.”Thearrows inthis
diagra mstand for interfaces of the sort described in Chapter 5.
8.4 Open class of symbols
To go beyond single symbols toward modern language, we need two major innovations. Thefirst is to permit an
unlimitedlylarge class ofsymbolsin thesystem (a large lexicon);thesecond is theconcatenation of symbols intolarger
utterances (the beginning of syntax). These two are logically independent: one could have a communicative system
involvingonly oneor theother. This is reflected in Fig. 8.1 in thetwoparalleltracks descending from“use of symbols.
”^123
Let's thinkfirst about the open vocabulary, the repertoire of meaningful linguistic units stored in long-term memory.
By contrast with primate call repertoires(the closest appropriate comparison), which number roughly in the dozens at
most, thevocabulary ofan average speaker is estimatedtorun intothetensofthousands. Beginning around theage of
two, children learn these in droves, and we keep picking up new words all our lives (Carey 1978). As stressed by
Donald (1998), such a large vocabulary places significant demands on long-term memory and rapid retrieval.
The language-trained apes, by contrast, acquire at most several hundred symbols, mostly through extensive training,
but in some instances (e.g. Savage-Rumbaugh's bonobo Kanzi) appearing to“just pick the mup.”At present it is
unknown what accounts for the hundredfold difference in vocabulary size. It might be a consequence of the larger
human brain, or alternatively of some special human“tuning”that makes vocabulary learning vast and effortless.
Some indirect evidence suggests the latter. Reports of children subjected to early hemispherectomy (Vargha-Khadem
et al. 1991) donotobserve massive vocabulary
AN EVOLUTIONARY PERSPECTIVE 241
(^123) Intuitively, it makes sense that development of an open vocabulary probably went on concurrently with the advent of combinatoriality. I keep them separate so as to make
clear the logical independence of the two developments.