[Corpora-List] Google searches as linguistic evidence

David L. Hoover david.hoover at nyu.edu
Mon Dec 11 09:55:00 CET 2006

I had an undergraduate student who was confused when I commented on his
use of "an man", but that's the only time I've had that experience with
a native speaker.

The Google hits are virtually all simple errors. I just now did a search
for "an workshop", and, of the first 20, all but one are clearly just

In a couple of examples, "an" is actually "-an" and is part of a
preceding compound word, so it really isn't an example of "an workshop"
at all, just an artifact of Google's algorythm.

In some the error seems phonetic:
Presentation an[d] Workship =and
The annual SORPASŪ users meeting in North America and an Workshop =and
Workshop [my guess is an original an=and error, erroneously corrected]
Work with the client to develop an workshop prospectus [here "workshop"
is a verb, so it isn't an a/an error]
A shuttle-bus will run on April 5 between hotel an workshop =and workshop

Other fairly clear errors:
Workshop on Technology Transfer in Lilafured (June 2002) in cooperation
with the University of Miskolc, Hungary, and the TU Košice, Slovakia, an
workshop “International Workshop devoted to problems of technology
transfer of Intelligent Solutions into
= an “International Workshop...”

and evaluation and supervision of this growing industry were discussed
during an workshop on MSE Finance and Credit Guarantee, = an MSE Finance
and Credit Guarantee workshop?

An workshop in advanced techniques of stop-motion animation,
=an advanced workshop

In other cases the document with "an workshop" contains several examples
of "a workshop" as well, which casts doubt on its authenticity.

I have yet to see any significant evidence from Google that "an
workshop" is a valid example of intentional native speaker usage. But
this exercise does suggest that any evidence from Google hits needs to
be carefully examined when the results seem strange. Googling other
combinations of "an" and a consonant-initial word gives similar errors.
Many of them would also be found in print, but electronic text/editing
creates makes these kinds of errors easier.

"many researches" will freqeuntly be an error for "many researchers", no?


Geoffrey Sampson wrote:

>An amazing experience I had a few years ago was being asked in all

>seriousness by one of my part-time researchers whether "a bad egg" or

>"an bad egg" was correct. With another part of his time he worked for a

>company alongside another man who had to do some documentation and

>insisted that the correct form was "an bad egg". So far as I could make

>out, this other man (who, like my researcher, was as I understood it a

>native speaker) thought he had learned a rule that "a" v. "an" depends

>on whether the following noun begins with a vowel, and this explicit

>rule overrode in his mind what must surely have been a large weight of

>experience implying that it is not the following noun, but the

>immediately-following word, that matters. The third party was quite

>sure that only "an bad egg" would do in writing; my researcher was

>dubious, but felt he needed my professorial authority to contradict his

>colleague. This seemed to me very striking counter-evidence against the

>idea that native speakers "know" the rules of their language.

>Comparable misunderstandings of the a/an rule might perhaps explain

>sporadic cases of "an w..." written by people who would surely _say_ "a

>w..." when they were speaking spontaneously, without thinking about

>language issues.


>Geoffrey Sampson




> Prof. Geoffrey Sampson MA PhD MBCS CITP ILTM


> author of "The 'Language Instinct' Debate"


> Department of Informatics, University of Sussex

> Falmer, Brighton BN1 9QH, England


> www.grsampson.net +44 1273 678525






David L. Hoover, Professor of English & Webmaster
NYU English Department, 212-998-8832

"Nothing, not even moonshine, goes to the head quicker
than saving democracy with other people's money."
--Ellen Glasgow, _They Stooped to Folly_ (1929)

More information about the Corpora-archive mailing list