Fawcett (2010: 283-4):
Appendix B summarises the main facts of what filling relationships are possible. In other words, statements about where a unit can and cannot occur have to be made for each unit and for each element that it may fill. This approach to the relations between units seems to correspond more closely to the patternings that we find in naturally-occurring texts than the picture that emerges when one tries to apply the 'rank scale' hypothesis that is embodied in "Categories" and IFG. In Appendix B, the differences in the probability that a unit will fill one element or another are indicated for conversation, and we must expect that some adjustments will be needed for other registers, such as academic writing.
However, the ultimate source of these probabilities is in the generative grammar. Here the probabilities are shown as percentages on features in systems, so that the theoretical-generative version of the model is capable of great refinement. Indeed, the probabilities can be changed in the light of specific contextual or systemic contexts. (See Fawcett, Tucker & Lin 1993 for a fuller picture.)
Blogger Comments:
[1] This is misleading. On the one hand, this is a bare assertion, unsupported by evidence. On the other hand, it is nonsensical, because Fawcett has the relation between data and theory backwards. The "patternings that we find in naturally-occurring texts" depend on which theory is used to construe the data as patternings.
[2] To be clear, the ultimate source of these probabilities is the data.
No comments:
Post a Comment