The World Wide Web provides access to an extremely rich collection of written texts and transcriptions of dialogue from both current and historical sources. But I often have trouble searching for specific things.

I encountered this recently when I was looking at the word "so". I was interested in two particular uses of "so". One of them was the ubiquitous use of "so" to start a sentence. The other was a one word sentence form of so: "So?"

But I couldn't come up with with any way to structure a search in Google to narrow it down that much. Does Google have more precise search options that would let me do so, or are there other search engines that allow me to construct searches more precisely? I'd be satisfied to structure a query saying "Only show me uses of 'so' at the beginning of a sentence", which is adequate for my purposes even though I'd still have to weed out "So." and "So!".

  • 1
    Google doesn't support searching with punctuation. Maybe scrape some webpages yourself, then use regex to search? – WavesWashSands May 3 '18 at 15:10
  • Google is not the only search engine. And scraping individual web pages would be insanely labor-intensive if we're talking about thousands or hundreds of thousands of pages. Surely for serious academic lexicographical research there must be better tools and techniques than what you describe. – user316117 May 3 '18 at 16:02

A web search engine isn't a good tool to answer your question. You need a large corpus of English text and a Corpus Query Processor.

With the cqp Corpus Query Processor, the query for So? looks like

[lemma="so"] [word="\?"]

And this query finds so at the beginning of a sentence (assuming that the previous sentence ends in a full stop, so the recall is not perfect)

[word="\."] [lemma="so"]

The queries can be refined to exclude unwanted hits, they are here for demonstration of the capabilities of cqp.

There are really great corpora for spoken English, both recent and historical. For example, there is the Old Bailey Corpus hosted at the CLARIN-D centre at Universität des Saarlandes for historical spoken English. More corpora can be found via the CLARIN Virtual language observatory, here is a bookmarked query for spoken English corpora

  • And you'd have to ignore almost all published texts, since written English is not where the action is with these two uses of so; they're almost strictly informal oral dialog, rarely written (and poorly transcribed when they are). – jlawler May 3 '18 at 18:51
  • There's also MICASE, which is strictly oral, but transcribed as well as recorded. It's American academic English from the turn of this century. – jlawler May 3 '18 at 23:05
  • @jlawler Make an answer out of your comment suggesting MICASE. It seems that the VLO query does not find it. – jk - Reinstate Monica May 4 '18 at 9:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.