It may just be that I'm demonstrating my gross ignorance, but I can't seem to find a 'why' for stem-changing verbs in Spanish. I understand that there is some sort of perceived weakness in the vowel that is emphasized by the addition of stress to the syllable, and that the vowel is somehow strengthened by the shift. I just don't understand the governing principles of that shift. Why do some vowels shift and not others? What is it that says that the 'e' or 'o' in this verb will change, but not the 'e' or 'o' in that verb?

Every search I've done just comes back with a lot of people saying 'that's just how it is, memorize it', and that's not really an answer. Any help out there?

  • The change has to do with the Yod sound of some verbs.. The yod affects the preceiding vowel... – user9437 Mar 27 '15 at 1:56

The diphthongization of front and back mid vowels that's referred to here is an historical process moving from Classical Latin to Vulgar Latin to Castilian Spanish, over about a millennium.

This is the way it works:

  • Classical Latin (ca 0 CE) had long and short vowels.
  • Vulgar Latin (ca 0-1500 CE) lost the Classical vowel length distinction.
  • Vulgar Latin innovated a new open/closed distinction for mid vowels e and o.
  • Mid vowels that were short in Classical Latin became open in Vulgar Latin.
  • Castilian Spanish (ca 1200 CE) changed Vulgar Latin open e [ɛ] to ie, and open o [ɔ] to ue.

Here's what it says in the Mexican standard high school textbook on Greek and Latin etymology of Spanish (Mateos M, A, Etimologīas Grecolatinas del Espanol, Editorial Esfinge, Mexico DF):

(51) La e breve y el diptongo ae del latín clásico se convirtieron en la e abierta del latín vulgar y se transformaron generalment in ie, al pasar al castellano.

(54) La o breve del latín clásico se convirtieron en la o abierta del latín vulgar y se diptongó in ue (pasando por uo), por regla general.

In essence, if the mid vowel was short in Classical Latin, it's likely to be a diphthong in Spanish now

  • Ahh, the missing link: LATIN. With a heavy dose of historical phonology. Thank you. I was hoping to find something that would make it easier to explain to my young HS minds, but I doubt they will benefit from this knowledge. I, on the other hand, am deeply grateful. – jrob Apr 18 '13 at 19:24
  • It doesn't seem to bother Mexican HS minds; Mexico requires everyone to pass a standardized exam on this book to get into college. – jlawler Apr 18 '13 at 20:23
  • You might also find this useful; it's the handout from a talk I gave on historical linguistics at a science fiction conference. There's nothing in it -- or in linguistics in general -- that's beyond high school minds. That's where there rest of the first world teaches and learns it, just like trigonometry and chemistry. – jlawler Apr 18 '13 at 23:08
  • 1
    this answer has major problems. Vowel length didn't change within the conjugation of verbs the way stem-changing verbs change their vowel. This is because the explanation misses that the breaking only occurred in stressed syllables. The second and third elements of the bullet point list should probably also be combined. As it is it implies the distinction between high-mid and low-mid vowels was innovated de novo after the loss of vowel length, rather than developing out of the exact same process that led to the loss of contrastive vowel length – Tristan Apr 23 at 13:12
  • 1
    @Tristan I think you're right, but the thing to do was to write a new answer rather than completely rewriting an accepted answer with so many upvotes. – TKR Apr 23 at 18:24

The 'governing principle' is stress. In Spanish, [ɛ] and [ɔ] become diphthongs in stressed positions, explaining why niego, niegas, niega, niegan have diphthongs, while negamos does not (the stressed penultimate is [a] there).

See the Wiki article on vowel breaking for other examples of diphthongisation in stressed syllables.

  • The stress aspect of it I understand, but if it is simply the presence of the 'e' or the 'o', then why don't 'beber' or 'comer' or 'colocar' or any of dozens (hundreds?) of verbs with those same sounds have that same stem-change? There is something more to it than just stress, some sort of interaction with other sounds in the word that I can't see or have some sort of mental block towards... – jrob Apr 18 '13 at 15:21
  • your answer was not lacking, it was my lack of understanding of what was meant by IPA vowels you mentioned. With the other answer, I now see what you were trying to indicate. Thank you for adding to my understanding. – jrob Apr 18 '13 at 19:27
  • 1
    @jrob If it's like Portuguese, the answer is because of vowel harmony. In words with medium height root vowels (o and e in the root) and low theme vowel (-ar), the root vowel will try to match the height of the theme vowel. The same is true in verbs derived from long theme vowels in Latin Therefore you have ciego from cegar, tiene from tener (which is tenēre with a long ē), but como from comer (which is from comedere with a short e). siento from sentir from sentīre, BUT beber is from bibere which contains a short theme vowel, thus no vowel harmony. – Ryan Ward Jun 13 '14 at 15:45

The e -> ie and o -> ue stem-changing verbs are the product of the interaction between two different factors

The first is the "breaking" of the Early Western Romance low-mid vowels /ɛ/ & /ɔ/ (which developed from the Latin short mid vowels /ĕ/ & /ŏ/) to /jɛ/ & /wɛ/ in stressed syllables followed by a merger of any remaining low-mid vowels (including the /ɛ/ in these new diphthongs) into the Early Western Romance high-mid vowels /e/ & /o/ (which developed from a merger of the Latin long mid vowels /ē/ & /ō/ with the short high vowels /ĭ/ & /ŭ/)

This was a regular sound change, and so affected the entirety of the lexicon across all parts of speech so can also be seen in nouns like (note that Romance nouns generally derive from the accusative):

  • miel < mél vs meloso < melṓsum
  • mes < mnsem
  • fuego < fócum vs hogar < focā́rium (note that the on-glide in the diphthong also led to the preservation of the initial f)
  • flor < flrem

The second is the position of the stress in Latin. Stress was on the penult if the syllable was "heavy" (containing a long vowel or a consonant coda), or else on the antepenult. This leads to a pattern where in the 1st, 2nd, & 4th conjugations the thematic vowel of the ending is stressed in the infinitive and 1st & 2nd person plural, but in all other forms the stem vowel is stressed

  • 1st conjugation (these become -ar verbs in Spanish) e.g. amā́re:
    • áámās ámat amā́mus amā́tis ámant
  • 2nd conjugation (these become -er verbs in Spanish) e.g. vidḗre:
    • vídeō vídēs vídet vidḗmus vidḗtis vídent
  • 3rd conjugation (these mostly become -er verbs in Spanish, but some, especially those in -iō become -ir verbs) e.g. dū́cere:
    • dū́cō dū́cis dū́cit dū́cimus dū́citis dū́cunt
  • 4th conjugation (these become -ir verb in Spanish) e.g. audī́re:
    • áudiō áudīs áudit audī́mus audī́tis áudiunt

Put these two factors together, and we see that Latin verbs in the 1st, 2nd, or 4th conjugations with either an /ĕ/ or /ŏ/ as their stem vowel will break in all present tense forms except the 1st & 2nd person plural

  • 1st conjugation:
    • segar < secā́re
      • siego siegas siegat segamos segáis siegan < sécō sécās sécat secā́mus secā́tis sécant
    • volar < volā́re
      • vuelo vuelas vuela volamos voláis vuelant < vólō vólās vólat volā́mus volā́tis vólant
  • 2nd conjugation:
    • tener < tenḗre
      • tengo tienes tiene tenemos tenéis tienen < téneō ténēs ténet tenḗmus tenḗtis ténent
    • doler < dolḗre
      • duelo dueles duele dolemos doléis duelen < dóleō dólēs dólet dolḗmus dolḗtis dólent
  • 4th conjugation:
    • sentir < sentī́re
      • siento sientes siente sentimos sentís sienten < séntiō séntīs séntit sentī́mus sentī́tis séntiunt
    • morir < morī́re
      • muero mueres muere morimos morís mueren < móriō mórīs mórit morī́mus morī́tis móriunt

The e -> i & o > u stem-changes are more complicated

At a stage before the breaking of stressed low-mid vowels, a following /j/ (which can develop from the loss of intervocalic voiced stops) caused the mid vowels to raise one step i.e. /ɛ/ -> /e/ -> /i/ & /ɔ/ -> /o/ -> /u/. As /ɛ/ & /ɔ/ merge into /e/ & /o/ in unstressed syllables the only affect this has on low-mid vowels is causing them not to break in stressed syllables

Instead, this effect is mostly noticed on the high-mid vowels where it results in an e -> i or o -> u shift. In many instances, especially with nouns and adjectives, this shift applies across the entire paradigm, and the /j/ may no longer be visible, but in -ir verbs it can occur in some forms, but not others

In many of these instances, the /j/ is usually still present, as in the 3rd person preterites, and the imperfect subjunctive. In some other instances, especially those that were originally 3rd conjugation verbs in -iō (or became such) the raised alternant caused by the /j/ in the 1st person singular and 3rd person plural was generalised across the present (except for the 1st & 2nd person plurals) by analogy to the e -> ie & o -> ue stem-change

Depending on its origin, and the specific sounds making up the stem of its Latin ancestor, a given verb might have no alternations, one set of alternations on its own, or both sets of alternations together. Over their history, many verbs have also changed conjugation, especially between 3rd & 4th conjugations, via the 3rd conjugation in -iō and, depending on when this occurred, this may or may not lead to unexpected raising stem-changes. Additionally, some verbs have lost their stem-changes over time due to analogy from the infinitive. In general as with most forms of analogy, this affects rarer verbs more than more common ones

  • Excellent answer, though some of the part about the e>i and o>u change is a bit unclear to me. Since low-mid, unstressed vowels are raised to high vowels (*sɛntió > sintió), surely the non-breaking in stressed syllables isn’t the only affect of the umlauting? Or are you excluding those since they later merge with the high-mid vowels anyway? Some more examples in the second half would probably make your meaning clearer. (If you can explain why focāri(um) becomes hogar but iocārī/-e gets raised to the now entirely unique position of stem-changing jugar, that would be an extra plus!) – Janus Bahs Jacquet Apr 26 at 17:38
  • @JanusBahsJacquet the jod caused vowels to raise one-step in the chain /ɛ/ -> /e/ -> /i/ so sɛntió > sentió, not *sintió. I still don't entirely understand the cause of jugar's raising but I'll see if I can find something – Tristan Apr 27 at 9:00
  • But the actual form is sintió, not *sentió, and the same with murió, etc. If the raising took place before breaking (and thus also before the later merger of high- and low-mid vowels), then you’d expect *sentió and *morió, etc., but that’s not what we find. Were there two separate rounds of raising? – Janus Bahs Jacquet Apr 27 at 9:11
  • Although… given that there’s also raising in unstressed forms of the present subjunctive where there was never a /j/ (sintamos, muráis), perhaps there’s something additional going on here as well? – Janus Bahs Jacquet Apr 27 at 9:15
  • sentir had a long vowel in its perfect stem which likely accounts for what appears to be double raising. I'm not unsure what's going on with morir though as morior, and later morio don't seem to have had such an alternation. The fact morir & jugar both derive from deponent verbs makes me wonder if that's part of the reason, but I'm not sure – Tristan Apr 27 at 9:55

One last thing. Vulgar latin, like Italian and Catalan, only had the distinction of ɛ-e and ɔ-o in stressed syllables. They were reduced (ie. merged) to e and o in unstressed syll.

So, the alternation 'mover' - 'muevo' comes from Vulgar latin /mo'vere/ - /'mɔ.vo/. The stem, /mɔv/, had been reduced to /mov/ in unstressed positions, so it wasn't diphthongized.


It's basically for the sound. Without stem-changes, certain words would sound awkward would be harder to say in a sentence, and the sentence would not flow smoothly. I don't know about stress on vowels, though. Comer and Beber aren't conjugated is because they sound fine.

  • 2
    This is complete nonsense. Duermo would absolutely not sound awkward or be harder to say if it were dormo, because that’s how it would be then, and the sentence would flow perfectly smoothly. If comer had had cuemo as its present form, you’d be saying the same thing about that verb too – the only reason some verbs sound ‘awkward’ and others don’t is because that’s how it happens to actually be. So your reasoning boils down to ‘because it is’, which is what the asker explicitly did not want. – Janus Bahs Jacquet Apr 23 at 0:14

Stem changes allow for better and more fluent pronunciation of certain words in Spanish. Some vowels like O and E are stressed in some words but unstressed in others

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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