John Fields in his 2003 psycholinguistics textbooks says:
Note that the there -> their example is not
a spelling error - the writer is fully aware of the difference between the two forms;
but, under the pressure of writing, one form (often the more frequent one) is substituted
for the other.
And he is correct about the relative corpus frequency of those words (google ngram below):
If we apply the same theory to possibly -> possible, the difference in relative frequency is even greater:
This idea of higher-frequency word "takeover" (my term) is backed up by some more empirical research, but contradicted by others:
With regard to lexical properties, some studies of spontaneous speech errors (del Viso et al., 1991; Kelly, 1986) as well as errors arising in cases of acquired spoken language deficits (Blanken, 1990, 1998; Gagnon et al., 1997; Martin, Dell, Saffran & Schwartz, 1994) have reported that error responses are biased to be higher in frequency than their targets. However, a number of other studies have found no such effect (spontaneous errors: Harley & MacAndrew, 2001; Vitevitch, 1997; experimentally induced errors: Dell, 1990; aphasic errors in spoken production: Best, 1996; written production: Romani, Olson, Ward, & Ercolani, 2002).
Another interesting issue is that written errors seem to cross syntactic category boundaries more often than speech errors:
In addition, a large number of studies have reported very high rates of syntactic category preservation in spoken errors produced both spontaneously (Abd-El-Jawad & Abu-Salim, 1987; Arnaud, 1999; Berg, 1992; del Viso et al., 1991; Fay & Cutler, 1977; Fromkin, 1971; Garrett 1975, 1980; Harley, 1990; Harley & MacAndrew, 2001; Leuninger & Keller, 1994; Nooteboom, 1969; Rossi & Defare, 1995; Silverberg, 1998; Stemberger, 1985) as well as subsequent to neurological impairment (Best, 1996; Berg, 1992; Blanken, 1990; Dell et al., 1997; Gagnon et al. 1997; Martin et al., 1994; but see Blanken, 1998). Some studies of written errors have reported that the tendency to preserve grammatical category is weaker in written than in spoken production—and is perhaps non-significant (spontaneous: Hotopf, 1980; aphasic errors: Romani et al. 2002).
Perhaps this partly explains why people don't seem to make this possibly -> possible substitution in speech. (One is an adjective, the other an adverb.)
And as far terminology goes, this is sometimes called an "atomic typo", at least among writers:
Another kind of typo—informally called an "atomic typo"—is a typo that happens to result in a correctly spelled word that is different from the intended one, and since it is spelled correctly, the spellchecker cannot find the mistake. Examples include "unclear" instead of "nuclear", "you" instead of "your", "Sudan" instead of "sedan" (leading to a diplomatic incident in 2005 between Sudan and the United States regarding a nuclear test code-named Sedan), "Untied States" instead of "United States", and "the" instead of "they". The ["atomic typo"] term was used at least as early as 1995 by Robert Terry.
And regarding speech errors (which a lot of answers here point to). Depending whether you are willing to call typos spelling errors or not (Field won't), this is or isn't a (written) malapropism. There's not complete consensus whether accidental pronunciations (or typos) that produce word substitions should be called that:
Definitions differ somewhat in terms of the cause of the error. Some scholars include only errors that result from a temporary failure to produce the word which the speaker intended. Such errors are sometimes called "Fay-Cutler malapropism", after David Fay and Anne Cutler, who described the occurrence of such errors in ordinary speech. Most definitions, however, include any actual word that is wrongly or accidentally used in place of a similar sounding, correct word. This broader definition is sometimes called "classical malapropism", or simply "malapropism".
And I'm not entirely convinced Wikipedia has this issue right
Under the name of malapropisms, Fay and Cutler (1977) have examined a variety of speech error. However, malapropisms in the classical sense are not speech errors: they are what the speaker intended to say and would be willing to repeat. This paper analyzes 158 classical malapropisms in English. Though some probably result from imperfect learning, most represent errors of storage in the mental lexicon. Statistical assessment of the examples indicates that they most resemble the ‘tip of the tongue” phenomenon, rather than slips of the ear or malapropisms in the Fay/Cutler sense.
Anyway, what seems to be a reasonable conclusion here is that "atomic typos" include the written equivalent of Fay-Cutler malapropisms.
The UPenn Language Log discusses a recent example of the latter, which is a bit long to reproduce here with the accompanying discussion in full. But the point there is that "committed of a crime":
is an "error" because if you showed the writer the sentence and invited him to review it, he would (I surmise) say "oops" and correct "committed" to "convicted". In contrast, a (classical) malapropism, as deployed by one of the Mrs. Malaprops of the world, is a case where the speaker or writer has entered the wrong word into their mental lexical, and would not be aware that they're making a mistake even if they had the opportunity to review what they said or wrote.
The error in this case is "phonological" because committed and convicted are both three-syllable words with medial stress, whose initial and final syllables are almost the same. And the error is "semantic" because someone is convicted of having committed an offense, so that there's a close relationship in the psychological space of meanings.
Why do I think that this case is a Fay-Cutler malapropism, i.e. a (written form of a) speech error?
For one thing, if we look at a somewhat edited collection, namely the COCA corpus of 425 million words of recent American English, the pattern "committed of a crime" gets no hits, while "convicted of a crime" gets 99, and "committed a crime" gets 236. [...]
the substitution has the general lineaments of a Fay-Cutler-type speech error: same number of syllables, same stress pattern, same lexical category, same first and last syllables.
So we can also call possibly -> possible "a typo with the general lineaments of a Fay-Cutler-type speech error". Ha ha. Linguists need more roundabout terminology. Actually not all the criteria are met: in particular, it's not the same lexical category in the case of possibly -> possible, but rather adverb -> adjective. I don't know if Fay & Clutler would have allowed for this.
This how they actually defined their notion of malapropism
First, the erroneous intrusion is a real word-not the intended word, of course, but not
a meaningless string of phonemes either. Second, the target and error seem to be
unrelated in meaning. Finally, there is a close relation between the pronunciation of the
target and the pronunciation of the error.
So possibly -> possible doesn't quite fit that definition because the words are semantically related. It goes to show how difficult it is to come up with a good categorization of errors.
And since (rereading your question) you think the word "typo" does not apply to this; linguists use "slip of the keyboard". These include two sub-categories (cf. Field, p. 70)
- motor errors, where there is a failure in the signal that the brain sends to the hand
or in the contact between hand and keyboard
- lexical errors, where the wrong written form has been selected
You seem to think that "typo" only describes the first sub-category. But the Wikipedia definition of typo excludes spelling errors (that are a result of ignorance). And since your [lexical] error is not a result of ignorance, it's not excluded, so it is a typo.
What you probably mean to say is that it is a "nonspecified substitution". That means that neither key adjacency nor left/right mirroring were the cause. These two are common causes of "motor" (in the sense of Field) substitutions in typing.