If you hear someone say ‘I goed’ and you are a fluent speaker of English, you probably assume it’s a child in the midst of learning their native tongue, albeit with a few over-generalizations of word-formation rules, or an adult who is learning English as a second language and hasn’t paid attention yet to the irregular verb paradigms. 🙂 It could have been otherwise, though.
Old English gan (to go) used other forms for its past tense/past participle: eode/eodon. The origin of eode is not certain, but it appears to also have been originally another word separate from OE gan (oede is perhaps related to Gothic iddja (went)). During the 15th century (early Modern English), went — the past participle of a different verb, wenden (to turn, depart),  replaces eode as the past tense of go. (However, in some dialects in Northern England and Scotland, eode was instead replaced by gaed, a construction based on the verb go.) At some point the verb wend(en) adopted wended as its new past tense form (perhaps to avoid semantic confusion with went’s new role?). Today it continues with the meaning of heading in a direction, often slowly, deliberately, e.g., ‘we wended our way through the perfume counters at Bloomingdales’.
This case illustrates a type of linguistic change or syncretism, the merging of historically distinct morphs (word forms) into allomorphic variants of a single word, go. The distribution of these forms in linguistic contexts changed as a consequence of the merge; today go and went are in complementary distribution, went occurs where the past tense is intended and go/goes elsewhere. Prior to the merger, go and went were contrastive in meaning (they were different verbs altogether) so could appear in the same linguistic contexts.
A parallel from phonology can further illustrate this difference in distribution patterns: in English the phoneme /p/ has two allophones, one illustrated in the pronunciation of ‘p’ in the word pot and the other in the pronunciation of ‘p’ in the word spot. In the first example, ‘p’ has a small puff of air, associated with it, in the second example it does not. Convince yourself of this fact by lighting a match and saying both words in front of the flame; the flame will not dance when you say the word spot. The point here is that word-initial (or word final) position ‘p’ in English is the ‘puffy’ sort, whereas ‘p’ following an ‘s’ (or between vowels) is not puffy. The distribution of these two variants of a single phoneme is complementary, parallel to the case above for the two modern-day morphemic variants, go and went: the choice of which form to use is dictated by a linguistic rule.
In terms of contrastive distribution, the phoneme /p/ occurs in many contexts where other non-related English phonemes also occur, e.g., in word initial position ‘hat’ contrasts with ‘pat’, thus /h/ and /p/ are not the same phoneme, because replacing one with the other changes the meaning of the word. (If you choose to say spot with a puffy ‘p’ it will sound odd, but won’t change the meaning of the word.) Â Similarly, before went became an inflectional variant of go, it contrasted with go in the sense that replacing go with went would alter the meaning of the sentence — an alteration of meaning beyond just a difference in present/past tense, a difference more akin to the contrast between ‘I drove the car’ and ‘I started the car’.
I mentioned earlier the merging of went with go may have given rise to wended as past tense of modern wend. Two other words from Old English, bendan, sendan retain the earlier parallels to older went, i.e. today we have bend/bent, send/sent.  Interestingly, forms which came via Old French have final -ed, e.g., defend/defended/*defent,  attend/attended/*attent.  Then there is Old English blondan (to blend), is it too weird to say ‘I blent it in the food processor’?  To my ear it sounds rather loutish and uneducated, but not ungrammatical (there is a difference!). Even more acceptable is modern lend/lended/lent. The Old English word was laenan, which added a ‘d’ to the stem during the time of Middle English, probably by analogy to bend and send.
One can wonder why, having adopted one unrelated word, eode, to serve as past tense for the verb go, that form became replaced by yet a different unrelated word, went. As noted above, the change did not go through in all dialect regions at the same time. I find myself wondering whether the verb left could be next in line.  On the face of it, there would seem to be reasons this won’t happen. Unlike go, the verb leave is transitive and takes objects of  all kinds (‘he left his umbrella’, ‘they left their children at the daycare’, ‘we left Paris two days later’), and one of its fundamental meanings is nearly the opposite of go, ‘no traces were left’ means almost the same thing as ‘no traces remained’, remain is opposite of go. Yet, there are some parallel constructions close in meaning: he left/he went, I think she left for the grocery store/I think she went to the grocery store.
Today the verb go has an immense usage and range of meaning, even standing in for ‘say’ in colloquial speech: Â ‘he goes “what do you mean?” and she goes “you know darn well what I mean!”‘. Â Will the whole go paradigm hang together as it now stands, or will it begin to split along certain semantic fissures?
Word Travels 