Corpora: Language Engineering

[Yorick Wilks]

Ive just caught up with this (what is..)language-engineering
discussion after a while, but Lou's contribution (below) has provoked me
to protest--this is reduction to two silly choices, neither of which
is anywhere bear reality. The joke-positivist definition (X is what X-ists do)
is sometimes fun but has been used so much and is only helpful where
definitions are really no use --AI may be a case, and AI probably
is what AIers do. But where there is a working/useful definition,
that's egregious. Similarly, the use he supports with the Aitchison
quote would (for me) require a quote from a sociolinguist (not a general survey
of linguistics) before I believed it--Lou seems to think that the
quote from Aitchison ABOUT this usage in some way attests it---the
lexicographic fallacy if ever Ive seen it!

More seriously, there surely isnt much doubt that, whatever
its precise origins, in Manchester or wherever, LE was meant to point
to the possibility of maturity in NLP where computer language applications
could possibly be engineered to a high standard of robustness and
reliability of the sort that sofware engineers talk about--SE is the
analogical origin of LE surely. Derek Partridge and I wrote
articles a decade ago with titles like 'AI and SE'arguing that
some areas of AI (like NLP) could not conform to SE norms because of
certain undecidable properties of language use. But that was not to deny,
nor is it now, that 'LE' has a  reasonable, aspirational, function.
Yorick Wilks