Linguists on the Internet

The current issue of the Economist (1/20/2005) has an article on linguists on the web and linguablogging. Its main focus is on linguists' use of Google and the web in general as a data resource -- and the pitfalls such research presents. The article also mentions Philip Resnik's Linguists' Search Engine and recent discussions on Language Log.

I personally have not yet used the web as a ready-made corpus, except for checking on the status of particular linguistic forms. For instance, tense/aspect/marker combinations were among the most basilectal features of Hawai'i Creole English, and the past progressive been/wen stay VERB was the most frequently attested TMA combination in historical texts (e.g. You wen stay wait by the tree how long? "How long were you waiting by the tree?"). Derek Bickerton's research in the 1970s found them still in existence, but they are not attested in Viveka Vilupillai's recent book on tense and aspect in contemporary HCE. Do they still exist or are they found only among certain speakers or in certain registers? With Google, I was able to find some examples on the web:

(1) Lyk da ada dae... he wen stay making troubLe... so den she wen make hym go in tha breakout room... en den Lyk he was pLaying wif everything in there. | link

(2) HO! BOY My eye wen lite up. Da back wen stay swell jus like wan KP! | link

Past searches also provided other examples, which no longer exist on the web (the ephemeral nature of the web is an interesting issue). However with the internet, we need to be careful about the quality of the evidence. In general, how representative is literary HCE of the spoken vernacular? Are some features distorted or exaggerated in terms of frequency? What kinds of sampling errors do we expect with this kind of corpus? And most important — do we know whether these writers are actually native speakers of HCE at all? These are important questions and they pertain to the study of historical pidgin/creole texts as well.

UPDATE: Language Log has some comments on the Economist article.