for most of the cognitive and linguistic literature on speech errors, it's something like 'an utterance that doesn't correspond to the utterance the speaker intended to produce.' people that use this definition are generally studying instances like "cattleships and bruisers" being produced when "battleships and cruisers" was intended. these are known as errors of performance. what this definition (sometimes explicitly) excludes are errors of competence. these are substitutions like "spicket" for "spigot." for most linguists, if you think that word is "spicket," intend to say it that way, and execute it properly, you haven't made an error.
or have you? clearly, for grammarians, using the word "spicket" where better-educated speakers would use "spigot" is an error. it's a violation of, as seb says, "the way things are" in the language at this point in time.
what's the disconnect here? grammarians are talking about the Language as it is recorded in dictionaries and linguists are talking about the intention a speaker has to make a particular utterance without regard to its correctness in the Language as a whole. this seems to be Saussure's parole and langue distinction in action. langue was his term for the language as a whole unified entity, whereas parole was the individual utterance spoken at a particular moment in an absolutely unique context.
the two domains are so distinct that they even have completely different ways of defining what an error is. yet it strikes me as part of the mystery of language that langue is built completely from parole. and, in turn, parole is built on langue; we each learn how to form utterances by taking in a ton of language, riddled with error though they may be.
what interests me further is this: the two constructions stand in very different relations to the speaking subject. he's sort of surrounded by langue while in control of parole. but both definitions of error have to do with the speakers' intention. what does this mean for educational policy? what does it mean for humanity? stay tuned?
I give you:
Common Errors in English
It is a compilation of "errors" in the English language. It is quite impressive, covering everything from the use of "spicket" to mean "spigot" to the alleged redundancy of "cheese quesadilla."
This raises a question for me, though: isn't it a contradiction for something to be a common linguistic error? Paul Brians, the creator of the site (and the book it's based off of) is not ignorant of the point, and presents his work in a particular, practical context:
What is an error in English?
The concept of language errors is a fuzzy one. I’ll leave to linguists the technical definitions. Here we’re concerned only with deviations from the standard use of English as judged by sophisticated users such as professional writers, editors, teachers, and literate executives and personnel officers. The aim of this site is to help you avoid low grades, lost employment opportunities, lost business, and titters of amusement at the way you write or speak.
But isn’t one person’s mistake another’s standard usage?
Often enough, but if your standard usage causes other people to consider you stupid or ignorant, you may want to consider changing it. You have the right to express yourself in any manner you please, but if you wish to communicate effectively, you should use nonstandard English only when you intend to, rather than fall into it because you don’t know any better.
What do we think about this? Paul Brians seems innocent enough. But how should we deal with the problem of language standardization in general? I once heard a great definition of a speech error, which was simply a speech act that doesn't have the effects that it was intended to have. If that's the case, the Brians' catalogue would be appropriately named--he is not universally prescribing a usage as "true" to the language, but rather providing a kind of instrumental prescription: "To not be laughed at, don't say 'spicket.'"
Perhaps this awareness of the scope of the prescription is all that's needed in linguistic education to make it less insidious. If language is taught not as "correct" or "incorrect" but rather as "(in the present context) effective" and "(in the present context) ineffective)", politicization of speech 'errors' (with its racist and classist effects) might not make as much sense to anybody.
On the other hand, linguistic competence seems to be such a basic faculty, I have doubts that conscious realization of the contingency of linguistic 'rules' will do much to stop the automatic perception of those that make 'errors' as being ignorant of some True Way Things Are.
But on a third hand--is incompetence withing the scope of the present linguistic conditions any better than a kind of ignorance of some (perceived as) universalized facts about language? Maybe speech errors ought to be looked down on, if only as a way of providing an incentive for people to get up to speed.