Instead of giving the past tense form bended, the verb bend fuses together bend and -ed and removes voice, producing bent.

Lent and sent are produced in similar fashion.

What's the word for the fusion here?

  • For examples, look here dictionary.cambridge.org/grammar/british-grammar/…
    – GEdgar
    Commented Sep 21, 2023 at 20:59
  • 2
    The process is not quite as you've described, because it is not only words ending in /d/: consider leant, spelt, learnt and dreamt. And actually caught, brought, and thought are also examples, thought other processes have intervened in those.
    – Colin Fine
    Commented Sep 21, 2023 at 21:10
  • 2
    It should be noted that this isn't a process applying to all words; the past tense of mend is mended, not ment.
    – alphabet
    Commented Sep 21, 2023 at 22:31
  • 1
    No, ed is added to bend+ed. And in bent, the d becomes t.
    – Lambie
    Commented Sep 22, 2023 at 17:53
  • 2
    @Lambie - Adding +ed to bend and then changing a d to a t does not give bent. Some kind of "fusion" or "assimilation" is involved. What exactly this is and what it should be called, if it has a name, is the topic of the question.
    – ool
    Commented Sep 23, 2023 at 20:13

2 Answers 2


Short answer (tl;dr):

According to the Cambridge Grammar of the English Language (Huddleston & Pullum 2002; pp. 1601—1602) the two processes involved are simply:

  1. devoicing of the suffix (d→t)
  2. consonant reduction (dd→d, tt→t)

NB, the term coalescence emphatically doesn't apply here. Coalescence is when two different sounds are replaced by a third new one. An example might be when the the sequence of two phonemes /d/ (as in did) and /j/ (as in you) are replaced by the affricate /d͡ʒ/, for example in the sequence Did you see that?:

  • /dɪd͡ʒu si: ðæʔ/ "Di joo see that?"

For Wikipedia description of coalescence see here. Not that I recommend Wikipedia in such matters!

Full answer

The past participle and past tense forms of regular English verbs are identical, so I'll simplify things, here, by just talking about past tense forms.

The formation of English past tense forms:

We create the past tense forms of regular English verbs using the following rule:

  1. Add the voiced aleveolar plosive, /d/, to the base of the verb.

This is both simple and elegant. However, the phonological rules of English—the rules of how the sounds of the language fit together—must still be observed, and for this reason we need a couple of sub-rules. Firstly, it is a cross linguistic rule that once the voicing has been turned of in a syllable, it cannot go back on again. In other words any following consonants must be voiceless. We therefore need the following caveat to rule (1):

  1. If the base ends in a voiceless consonant, change the suffix to an unvoiced alveolar plosive, /t/, accordingly.

Another phonological rule in English, this time a language-specific one, is that geminate consonants (double consonants) are not allowed within syllables. This means that there is a problem if the base of the verb already ends in /d/ or /t/, because this would give us an illicit geminate consonant at the end of the word. For example, the verb mend would have the illicit form */mendd/, and the verb insist would have the form */ɪnsɪstt/. Thus the language requires one last sub-rule:

  1. If the base already ends in /d/ or /t/: add the KIT vowel, /ɪ/, to the base before applying rule (1) above.

Note that rule (3) above handily creates a second syllable to add the regular /d/ to.

Some cantankerous verbs:

The Cambridge Grammar of the English Language (Hudleston & Pullum, 2002 pp. 1601—1602) enumerates seven classes of semi-regular verbs. I say semi-regular, because they all involve a past-tense suffix on the verb, but also involve one or more additional phonological processes. Relevant to our discussion here are three particular classes:

Class 1A: devoicing of the suffix. With eight verbs ending in /l/ or /n/ there is both a regular form with /d/ and one in which the suffix is /t/, despite the voicing of the final consonant of the base. The variation is normally reflected in the spelling, with t instead of ed (and ll reduced to l).


  • smell /smel/ /smeld/, /smelt/ (smell smelled, smelt)
  • burn /bɜ:ʳn/ /bɜ:ʳnd/, /bɜ:ʳnt/ (burn: burned, burnt)


  • Also: dwell, earn, learn, spell, spill, spoil

Earn is an exception orthographically: it has only the regular form earned.

[CGEL, p. 1600]

Notice that the last sound in the base in all of these cases is alveolar, in other words it is made at the same place in the mouth as the consonants /d/ and /t/. We might speculate that the devoicing occurs in part to make the ending more clearly audible. Anyhow, the terminology used in CGEL for this is simply devoicing of the suffix.

The next group seems to derive from an alternative strategy to avoid geminate endings on past tense verbs. In other words it is an alternative to rule (3) for regular verbs:

Class 1c: consonant reduction. With some verbs the preterite and past participle form is identical with the base:


  • hit /hit/ /hit/ (hit hit)
  • spread /spred/ /spred/ (spread spread)


  • Also: bet R, bid,, burst, bust R, cast, cost, cut, fit R, hurt, let, put, quit R, rid R, set, shed, shut, slit, split, thrust, wedR, wet R

As indicated above, the notation 'R' means that the verb has regular forms as variants: He had wet/wetted it. BrE has only regular forms for fit; otherwise, there is generally a preference for the irregular forms, especially in BrE. The regular forms of the colloquial bust are found primarily in AmE or when it has the sense "arrest". Irregular rid is required in adjectival passives like At last we were rid of them. The taboo verb shit (again with regular alternants) also belongs here, but it has an additional preterite variant with vowel change, shat.

There are phonologically similar verbs such as knit, shred, sweat that have only the regular forms. Similarly, although cost, "be priced" is in class 1c, cost, "estimate the cost of" is always regular. Bid, "make an offer" has different forms from the bid, of bid farewell, bid someone do something, which belongs in Class 3E. In writing read belongs here too, but in speech it is like bleed and is therefore listed in Class 1F.

The identity of the preterite and past participle with the base might seem to suggest that there is no suffix and that these verbs are like the base plural nouns of §4.1.3, but it may be argued that it is significant that all of these verbs end in an alveolar plosive /t/ or /d/. An alternative analysis, therefore, is that these verbs are irregular by virtue of having the suffixes /t/ and /d/ rather than regular /id/ - just like verbs with bases ending in consonants other than /t/ and /d/. Addition of /t/ and /d/ to the bases here would give */hitt/ and */spredd/, but English phonology does not permit doubled consonants at the end of a word, so these forms would be reduced to /hit/ and /spred/. This is plausible in itself, and it also receives support from the discussion of Class 1F below.

[CGEL, p. 1601]

And here we see that the term used in CGEL is simply consonant reduction, which is in effect similar to the Original Poster's description of it as fusing.

Of course, we can get both of these processes working in tandem:

Class 1E: consonant reduction with devoicing of the suffix. Six verbs combine these two operations:

  • bend /bend/ /bent/ (bend bent)
  • build /bild/ /bilt/ (build built)


  • Also: lend, rend, send, spend

The verbs in this class all end in an alveolar sonorant plus voiced plosive (/n/ or /l/ followed by /d/) in the base, but in a sonorant plus voiceless /t/ in the preterite and past participle. This can be accounted for by the combination of consonant reduction (*/bendd/ and */bildd/ giving */bend/ and */bild/) and devoicing of the suffix (giving /bent/ and /bilt/). The archaic gird (with girt /g3:'t/ as its preterite and past participle) is similar but would need to be explained in terms of another sonorant, /r/; this would be correct for rhotic accents but, as we have noted, the /r/ has been lost in modern BrE and is reflected by vowel quality alone.

[CGEL, pp. 1601—1602]

The Original Poster's question

The two terms, then, for these two processes seem to be, in the case of CGEL:

  1. devoicing of the suffix
  2. consonant reduction

There may be different terms used elsewhere. It's probably worth mentioning that devoicing is somewhat misleading here because those nominally voiced (the technical term is lenis) consonants that appear at the ends of words are usually devoiced unless followed by another voiced (i.e. lenis) consonant. This is a phonetic (not phonological) process, caused by the vocal cords ceasing to vibrate in anticipation of voicelessness. This is not what is happening in the case of the irregular past forms, where we see not a devoiced /d/, but an unvoiced, or fortis /t/. So devoicing, then, is not an ideal description.

Be that as it may, there are other terms which we definitely should avoid in such discussions. One of them is coalescence. In phonetics/phonology, the term coalescence refers to a situation in which two different sounds are replaced by a third sound, in turn different from the first two. An example might be when a sequence of /d/ and /j/ are replaced by the single affricate /d͡ʒ/. This is very, very different from a process which takes two sounds which are the same and reduces them to a single instance of the same sound!

It seems then, that if there are specific catchy well-known fossilized terms for these processes, they aren't used by CGEL. Instead, we have the simple descriptive terms shown in (1, 2) above.

  • The screenshots of text are not accessible. Can you transcribe the IPA at least? Regular OCR software would be able to help with the rest.
    – Laurel
    Commented Sep 22, 2023 at 14:41
  • @Laurel Hmm, they're visible for me. How are these ones? Is it because of the resizing? They're huge when not resized. Commented Sep 22, 2023 at 14:45
  • I can see the images. They're not accessible (as in web accessibility) because "enter image description here" is not a useful way to describe what's there to anyone who can't see the images (screen reader users, search engines, people behind firewalls). They really should be plain text instead.
    – Laurel
    Commented Sep 22, 2023 at 14:48
  • @Laurel That's not possible in this case. I don't have a plain text copy. (And the plain text IPA wouldn't be read anyway, right?) Commented Sep 22, 2023 at 14:49
  • @Laurel Let's do this is main site chat. Commented Sep 22, 2023 at 14:50

The two key terms you want are devoicing (going from /d/ to /t/) and coalescence (removing the multiple /d*d/).

The historical explanations would get into the class of Old English verbs, dialects of Middle English, and how each past tense form was standardized in Early Modern English. The question Send, sent; end, ent? has an answer that exemplifies that approach.

You however ask for a word specific to the change in phoneme: what do you call a change from something like /ˈbɛndɪd/ to /bɛnt/?

First I would caution that I'm not sure the change was synchronous, that is, whether people really were going from [bended] to [bent] or whether both usages were extant and [bent] prevailed over time (the Oxford English Dictionary describes both forms occurring at the same time). So describing a change in phoneme need not entail that the pronunciation historically changed in that way.

Going from /d/ to /t/ is called devoicing, since /d/ is a voiced alveolar plosive and /t/ is a voiceless alveolar plosive. Since this occurs at the final consonant, we could call it final consonant devoicing (see, e.g., Brian Hopkins, Lancaster Glossary of Child Development). As well, in going from /dɪd/ to /d/, the final consonants would have coalesced or contracted to remove the repetition of /d/ and the vowel in between. Thus, this is how Wikipedia ("English irregular verbs") describes the sound change you mean:

Verbs with coalescence of consonants and devoicing of the ending: bend, build, lend, rend, send, spend

  • How about "wend"? Commented Sep 22, 2023 at 22:11
  • Wend tended to wended over time rather than went, perhaps to differentiate from the past tense of go. Verbs don't have to follow the pattern: see tend*/*tended and trend*/*trended. Commented Sep 22, 2023 at 23:46
  • 2
    @TaliesinMerlin I believe the suppletive past tense of go, the word went, actually is the old past tense of wend! That's where it was suppleted from. :) Commented Sep 23, 2023 at 16:02
  • @TaliesinMerlin : I don't think there's such a thing as a difference between the past tense of "wend" and the present-day past tense of "go". Commented Sep 24, 2023 at 3:08
  • @MicharlHardy Wended and went are different words. Two syllables, one syllable. There is a difference. Commented Sep 24, 2023 at 11:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.