Synforms are lexical items similar in their phonological, graphic and/or
morphological features. Laufer (1988) showed that they could create problems not only for learners of English (including advanced learners) but also for native speakers. She divided them into ten categories (which you can find in the article linked to below by Kokic, 2008) dependent on the reason for their similarity - for example, derivation from a similar root, differences in spelling due to a single vowel or consonant etc. Some examples of synforms taken from Laufer's categories would be historic/historical, staff/stuff, price/prize, manual/menial, loose/lose
Both Laufer and Kocic, and other researchers have looked at which of these features cause more difficulty than others. However, their research seems to have been purely quantative - did the respondents make more errors in one of the ten categories than another - and based on multiple choice tests, where the items were presented to the respondents, both the correct choice and its synform amongst the distractors.
What I've not been able to find is any research that has been done on inaccuracies in the spontaneous use of these words, or any investigations of the reasons for these inaccuracies, including whether they are genuine errors or just mistakes. The sort of research I would envisage might use written texts containing the inaccuracies and retrospective think-aloud procedure to get the writers to explain what they were thinking when they chose the erroneous word and why they used it.
I spend a lot of my time marking Cambridge Delta assignments written by both native speaker and near-native speaker teachers of English, and find that synform inaccuracy abounds. I don't claim to have done any systematic research at all (if anyone is looking for a topic for a MA thesis there might be something to consider here) but here are a few synform problems that I've found come up again and again and some hypotheses as to the reasons why they might have been made.
The first category is banal - I hypothesise that the error is due to a simple typing mistake and that if the text were in handwriting wouldn't be there at all. It's typified by the use or leaner instead of learner. I don't think there's any question about the writer not understanding the difference, nor any chance that they don't know how learner should be spelt. It's a simple typo which only creates a synform by pure chance.
The next category is a bit more interesting - hands up anyone who has never typed their when they meant to type there or they're, or it's when they meant to type its. (Incidentally, this type of synform involving or created by punctuation differences was not included in Laufer's categories but should, I think, be there.) Why do we make this mistake? I'd suggest it's because, as we are actually physically typing (or writing) the word, our brains have moved on to the next part of what we want to say, and without full attention on the word we are actually typing, we retrieve its synform due to the fact that phonologically it's identical. (Guess what I had to go back and correct two words before this bracket started?) But again, at least in my case and - I would suggest - those of the highly educated trainees I work with - it's not that we don't know the difference. It's a mistake, not an error.
Phonological identity of words - ie the fact that they are homophones - also plays a part in the confusion between my next two item bare vs bear and, although it may not initially seem like it, upmost vs utmost.
The email I once received where the writer asked me to "bare with me" was probably caused by exactly the same process as the their/there problem I talked about before, but what about utmost and upmost?
Although in their citation forms they are pronounced differently - /ʌpməʊst/ vs /ʌtməʊst/ - in connected speech assimilation causes the alveolar /t/ of utmost to change to a bilabial /p/ in order to facilitate the pronunciation of the following bilabial /m/ - so both words end up sounding like /ʌpməʊst/. This means that anyone who has heard and used utmost in speech, but not "noticed" the difference in its written form, is liable to confuse the two when writing. They would likely be using upmost rather than utmost in speech as well, but because of the assimilation, it would go unnoticed.
So now we are in the realm of errors made because of genuinely not knowing the difference between the two forms. The same problem occurs with the pair affect (verb) and effect (noun) - but this time it is more complicated. Again, a phonological feature can make them homophones in one of their meanings ("influence") - the stress on the second syllable in both leads to the first vowel being pronounced as /ə/ so that the pronunciation is identical - /əˈfekt/ This could explain the confusion created for writers using them in this sense. However, both words can also change word class and meaning - noun affect (psychological variables) and verb to effect (ie carry out). I've never noticed synform problems with affect /ˈæfekt/ and would hypothesise that the pronunciation difference allows the brain to store it as a "separate" item from the others. However, the existence of the verb to effect is probably another reason for the synform error when it's used to mean to affect.
Going back to utmost and upmost, there is also the possibility of confusion occurring because, in a sentence like It is of the utmost importance, the substitution of upmost seems logical in terms of meaning. That, I would suggest, is also the reason for a trainee recently using failproof rather than foolproof in an essay. Both forms exist, and are so close in both meaning and pronunciation that they can easily be confused.
And finally - my favourite. The one where I think the brain does a backflip and attempts (erroneously) to "construct" new knowledge based on its interpretation of previously acquired knowledge. It's exemplified by gemination vs germination. Gemination is a technical term that few people have met before taking the Delta or a similar course. It therefore has to be understood and assimilated by the brain which, in the effort of doing so, associates it with the most familiar "known" word in its store - and learns and retrieves the term as germination. The same occurs with another phonological term, juncture, which is frequently retrieved by the writer's brain as junction.
These are only hypotheses, and I'd love to see some serious research done into the reasons for the synform mistakes of both native speakers and learners. What are the processes that cause them? Are they the same for the two groups or do they differ? If anyone knows of anything I've missed that looks at the problem in this way - ie qualitatively rather than quantatively, please let me know. However, I would suggest that phonological similarities, similarity to a more familiar word, and retrieval processes in the brain all play an important part in determining if a problem arises. And I'd also suggest that the problem also has to be considered in terms of whether it is a real error or just a mistake.
References
Laufer, B., (1988), 'The concept of 'synforms' (similar lexical forms) in L2 vocabulary acquisition'. Language and Education, 2, 113–132.
Kokcic, A. The problem of Synforms (Similar Lexical forms) Linguistics and Literature Vol. 6, No
1, 2008, pp. 51 - 59
