Fawcett's Computational Solution To His Unnecessary Theoretical Problem

Fawcett (2010: 258-9):

The solution to this problem proposed here — and already implemented in the computer model — is to build the relevant choices into the system networks. And the realisation of the choices is variation in the depth of exponence.

In the system network for QUANTITY, for example, the feature [cardinal] leads to a system in which one of the options is [sub-hundred]. When this is chosen the grammar will generate a simple cardinal number such as nine or ninety-nine. The feature [sub-hundred], together with several other features, then enters a system in which the choice is between 'adjusting' the quantity and 'not adjusting' it. It is only if both of the features [sub-hundred] and [cardinal unadjusted] are chosen that the grammar will generate an item that directly expounds the qd. If they are not, the realisation rules associated with the other features in the relevant systems trigger a re-entry to the system network to generate a new syntactic unit. So if 'adjustment' is required (e.g., to generate (c) in Figure 20) the generation of the cardinal is postponed till later. The reentry to the network then generates a quantity group, which opens up the range of meanings that follow from choosing to 'adjust' the cardinal number.

If a feature other than [sub-hundred] is chosen (and there is no 'adjustment') the grammar is re-entered to generate a nominal group, as in (b). And of course the two types of meaning may combine, as in over two hundred books.

Blogger Comments:

[1] Significantly, Fawcett does not provide any of these system networks.

[2] The reader is invited to consider (i) how many options would feature in this system, given an infinity of cardinal numbers, and (ii) how an infinity of numbers could be generated from this type of system.

[3] Clearly, this is not a model of language, but a method of generating texts by computer.