There are many cases of prefixes changing their forms. For example

  • ex- can change to ef- in front of f, e.g. effusion.

  • ad- becomes a- in front of b, e.g. abate.

Are there some more general rules or a summary of such changes that can help me figure out the meanings of unfamiliar words, and recognize the prefixes in the words?

3 Answers 3


There were several stages in the history of English where prefixes took different forms, resulting in the variety we have now:

  1. At some stage in (early?) Latin, prefixes often assimilated to the following consonant, acquiring e.g. a second f as in effusion. This was mainly determined by rules of pronunciation; just as we say a man but find *a apple jarring in English, so the Romans found *exfusio jarring and effusio more harmonious. We also pronounce the v in have often like an f before t/p/c: we always pronounce have to as hef to or hefta. Many other rules based on pronounceability also emerged in this stage (see below).
  2. At some (later?) stage of Latin, the Romans also used unassimilated, unchanged forms at the same time, where the basic form of the prefix was used regardless of the sound that followed: you will often read e.g. both adlatus and allatus "brought to", sometimes even by the same author. Dictionaries have huge lists of entries like "adl-: see all-", because they don't want to list two forms of the same word twice.
  3. In premodern French, different rules applied, and some Latin words were changed in different ways. The prefix ad- often changed into a- or à-. This could mean losing the geminated ("doubled") consonant it used to have: so Vulgar Latin abbatere (from ad- + battuere, "beat") became abattre at some point in French. English borrowed some words from French before these forms had changed, other words after.

It is hard to predict which rules will apply to which word, so you have to take all three options into consideration. A few common variations of prefixes in Latin, all mainly in stage 1:

  • a/ab/abs/aps/as/au- "away from"; as a preposition a/ab/abs.

    There may not be a strict rule. Normally abs- or aps- before c, p, t; sometimes as- before p; sometimes au- before f; otherwise a- before consonants; ab- before vowels and h.

  • ad/a/ab/ac/af/ag/al/an/ap/ar/as/at- (French often à/a-) "towards, in addition"; preposition ad.

    Rule: ac- before q and c; ad- before h and vowels; a- before gn, often before sp, sc, st; rarely ag- before n (usually an-); otherwise assimilated before consonants.

  • ambi/amb/am/an- "both sides"; no preposition; noun ambo.

  • con/co/cog/col/com/cor- "together with", as a preposition or suffix cum.

    There is no very strict rule. Often co- before vowels, g, h, and n, but not only or necessarily; rarely cog- before n (usually con-); col- before l; cor- before r; com- before b, p, m, sometimes before vowels; otherwise/usually con-.

  • de- "down, away, not", preposition de.

  • dis/di/dif/dir- "apart, away, not"; no preposition.

    Rule: dis- before c, p, t, s [except before sc, sp, st]; di- before sc, sp, st; assimilated dif- before f; dir- before vowels and h; di- before other consonants.

  • e/ex/ef/ec- "out, away"; preposition e/ex.

    Rule: ex- before vowels and h, c, p, t; ec- or (assimilated) ef- before f; after ex-, a following -s- may be dropped (exsanguis/exanguis both exist); e- in all other cases.

  • in/im/il/ir- (French often en/em-) "in, to"; preposition in.

  • in/i/im/il/ir- "un-"; no preposition; related to non and other nasal negations.

    These two prefixes have the same form but a different origin. The same rule applies: il- before l; ir- before r; im- before b, p, m; i- before gn; before other letters in-.

  • ob/oc/of/op/o- "in front of, blocking"; preposition ob.

    Originally op- (operio); oc- before c; of- before f; sometimes shortened to o- (omitto); otherwise ob- or sometimes op-.

  • per- (French often par-) "through"; preposition per.

  • por/pol- "forward"; no preposition; adverb porro.

    Rule: pol- before l; otherwise por-.

  • prae- (French pré-) "front, forward"; preposition prae.

  • pro/prod- "before, forward, instead"; preposition pro.

    Rule: prod- often but not always before vowels, otherwise pro-.

  • re/red- (re-in- often contracted as ren- in French) "back, again, properly"; no preposition.

    Rule: red- usually before vowels and h.

  • se- "apart"; no preposition.

  • sub/suc/suf/sup/sur/sus- (French often sou/sous-) "below, up from below, supporting"; preposition sub.

    Originally sup; su- before sp; sus- [from subs-] sometimes before t, c, p; otherwise assimilated before c, f, p, r; otherwise sub- or sometimes sup-.

  • super- (French often sur-) "over, above"; preposition super.

  • trans/tran/tra- "over, beyond"; preposition trans.

    Rule: tra- often before i and consonants; tran- before s; otherwise trans-.

  • 1
    I have not looked into this, but it would be more logical if your point/stage 1 was early, and point/stage 2 later. At the earliest stages of Latin, these assimilations were not yet entirely carried out (as can be seen from the very early Dioscuri inscription that reads Castorei Podlouqueique qurois; later sound changes would yield Classical Latin Castorī Pollūquīque curīs), so it makes sense that writers at one point vacillated between the assimilated and the non-assimilated forms. Later on, the assimilations were of course phonemicised and no vacillation would be expected to persist. Oct 14, 2013 at 10:11
  • 1
    @JanusBahsJacquet: Hmm that is very well possible. You make it sound plausible, even. This calls for research! I'm going to search the Bibliotheca Teubneriana for the occurrence of adfer* versus affer*. You know what, in late Antiquity and during the Middle Ages, various archaic forms regained popularity, possibly from Vulgar Latin. And I know many/most authors vacillate all the time. So perhaps it's just...a mess. We'll see. Oct 14, 2013 at 10:19
  • In later ages, of course, new words were also coined. At some point, the sound laws that brought about the assimilations here stopped being productive, and newly coined words would not be subject to them. Oct 14, 2013 at 10:21

This is actually much simpler than one might expect, if one only paid attention to spelling.

The name of the phenomenon is Assimilation,
and it's a very natural process that occurs in every human language.

If a prefix ends in a consonant, and then is attached before some other consonant,
then the consonant at the end of the prefix is going to change in contact.

That's called assimilation; and in fact assimilation is an example of assimilation,
because assimilation comes from Latin ad 'toward' + simil 'same'; i.e, 'become similar'.

But the D at the end of ad changed to an S, making a double SS
(which was probly pronounced as a double /s/ in Latin; this happened in Latin).

In the examples above, you'll find that an underlying final consonant (say B in sub-)
will appear before vowels, but before /k/ (C in Latin) it becomes /k/ (suc-), before /p/ or /t/ it becomes /p/, before /r/ it becomes /r/, etc. There are individual rules for lots of prepositions and prefixes (which constitute the same class of morphemes, but define different uses for them).

As for a general rule, it's generally the following consonant that determines the fate of the consonant at the end of the prefix. That's because by the time you've gotten to the end of the prefix/position you know what it is and you don't need to hear the end, so you slur over it to get to the beginning of the next morpheme.

This happens all the time, I repeat. But with Latin prefixes, the assimilations were already done and fixed in place before they came to English, so they're not productive anymore and have to be learned by rote (or by learning Latin, as it used to be done).

  • 3
    Also worth mentioning is that these assimilations were not limited to just preverbs/prepositions: most were regular sound laws that occurred throughout the language. For example, the ks (written x) in ex- is lost before an initial n with compensatory lengthening to the vowel: exnūmerāre -> ēnūmerāre ‘enumerate’; and the exact same thing happened anywhere else you get the sequence -ksn-: louksna -> lūna ‘moon’. Oct 14, 2013 at 7:24
  • 1
    When I studied this long ago in college, I was confused by anticipatory versus regressive assimilation. The second always seemed backwards to me. The normal anticipatory assimilation happens all over the place, but I once wondered whether the way voiced stops occurring intervocalically turn into the respective fricative of that series in either European Portuguese or in Spanish of any flavor counts as a form of regressive assimilation, or if it isn’t really just a common form of lenition. I vote for lenition, as in Pg lua < luna, but then what are examples of regressive assim.?
    – tchrist
    Oct 14, 2013 at 11:41
  • @tchrist: what kinds of courses and books do you studied in college for this? I prefer books that are more technical.
    – Tim
    Oct 14, 2013 at 14:40
  • 2
    Phonology is plenty technical; trust me on this. Especially if you do your own phonetics. Oct 14, 2013 at 16:33
  • 1
    @tchrist: anticipatory and regressive assimilation are the exact same thing (the type of assimilation seen in the word "assimilation", in fact). The other type of assimilation is lag or progressive assimilation; it's the kind that makes the word "cats" be realized with /kæts/ even though "dogs" is realized with /dɒgz/.
    – herisson
    Apr 17, 2015 at 0:03

These prefixes come from Latin and there they fulfill the roles of prepositions. These prepositions have multiple forms, in many dictionaries they will list it like this: 'ex/e' and 'ab/a'. In your case the two words 'ex' and 'fundo' have been agglutinated already in Latin. When Romans formed composita, they often doubled liquids (e.g. l's and r's) and fricatives (f's or s's) when the second word started with one of those. I'm not sure whether it is a rule, but it's quite often the case (effeminate, illiterate, illogical etc.).

So basically, when they turned a preposition in a prefix, they took the first letter of the second word and added it to the short form of the pronoun.

All you need to do is look at the first letter. If it's 'e', 'a' or 'i', it's likely to be the short form of a latin pronoun. If it's a consonant it can still be a prefix like 'dis', 'mis', 'iso' etc. but many of the longer ones come from greek.

Abate means abbot in italian, but doesn't actually come from Latin, so there is neither 'ab' nor 'ab' within the word.

  • ‘Abbot’ (and abate) does come from Latin—which took it from Greek, which took it from Aramaic. Just looking at the first letter of a word isn’t very likely to tell you much. ‘Enhance’, for example, is from Latin (via French), but from the preverb in-; ‘endemic’, conversely, is from Greek; ‘abate’ is from Latin (ab-), but ‘abatic’ is from Greek. And I think you mean the short form of the preposition, not the pronoun. Oct 14, 2013 at 10:14
  • It may came via Latin, as many Greek words did too, but being originally Aramaic, it carries no added prefixes. You cannot say that abbot is a compositum of 'ab' + 'ate', because 'ate' is no stem per se. As for enhance: it's not my fault that it has undergone a drastic phonologic change. In these cases it helps to know the origin, or alternatively a romance language equivalent; e.g. in this case, the Italian verb 'innalzare', which has the same root, the same prefix and verb stem, and even a doubled consonant. As i said, there is no rule, but he asked for something that would help
    – Matthaeus
    Oct 14, 2013 at 13:04

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.