Thursday, 5 September 2019

Learning Latin: why conjugations?

Verb conjugations were invented to torment first-year Latin students. OK, that wasn’t really the aim. But it is the effect. Learning four sets of endings is a pain. It’s also a waste of time.

Things would be hugely simplified for beginner Latin students if textbooks would explain thematic vowels properly.
What do you mean I need to learn four conjugations?
Latin has four basic types of verbs, or conjugations. Here’s a typical table for the verbs clamare shout, habere have, dicere say, and audire hear. The table shows the forms corresponding to I shout, you shout, she shouts, and so on. Each conjugation behaves slightly differently: I’ve added highlighting where the forms are distinctive to one conjugation.
1st conj. 2nd 3rd 4th
1 sg. (I) clamo habeo dico audio
2 sg. (you) clamas habes dicis audis
3 sg. (she, he) clamat habet dicit audit
1 pl. (we) clamamus habemus dicimus audimus
2 pl. (y’all) clamatis habetis dicitis auditis
3 pl. (they) clamant habent dicunt audiunt
The endings, at least, are nice and systematic. Looking down each column, you see -o, -s, -t, -mus, -tis, -nt (I shout, you shout, she shouts, etc.). The problem is the bit in between the stem and the ending, which changes depending on which conjugation the verb belongs to.

The outcome is that students have to memorise four separate sets of forms. There are 22 distinctive forms highlighted in the table, and it’s a pain to learn every single one of them.

As it happens, it’s also wasted effort. In reality, there are only two genuinely anomalous forms. Except for those two forms, you can have a full understanding of Latin verb conjugations -- in the present tense, at least -- in terms of just two things:
  1. Stem-sounds, that is, the sound that comes at the end of the verb stem: this will be either a consonant, or one of three possible vowels.
  2. Thematic vowels, that is, an unstressed vowel that appears in between a consonant stem and a consonant ending -- like in English hated.
Textbooks hint at these things -- some textbooks, anyway -- but they never explain them properly, they never get to the point.
A fairly typical introduction to Latin verb conjugations. (The yellow highlighting is my doing.)
As things stand, students have to memorise four sets of endings, with 22 distinctive forms. BOOOO-RING. Things are much easier if you learn (1) what the four conjugations really are, and (2) how thematic vowels work.

A better verb table

Here’s a revised table. This time, the highlighting isn’t there to mark differences between conjugations. Instead,
  • yellow marks the stem-sound, that is, the sound at the end of the stem;
  • green marks a contraction;
  • pink marks a thematic vowel.
1st conj.
1 sg. (I) clamo habeo dico audio
2 sg. (you) clamas habes dicis audis
3 sg. (she, he) clamat habet dicit audit
1 pl. (we) clamamus habemus dicimus audimus
2 pl. (y’all) clamatis habetis dicitis auditis
3 pl. (they) clamant habent dicunt audiunt
The real difference between the conjugations isn’t that they take different endings, it’s that they have different stem-sounds. ‘1st conjugation’ simply means a-stem, 2nd is e-stem, 3rd is consonant-stem, and 4th is i-stem.

This way, you can see that the verb system is actually extremely consistent. The actual endings -- the -o, -s, -t, etc. -- are still identical for every verb-type. 1st, 2nd, and 4th conjugation verbs have vowel stems, highlighted in yellow. And in most forms, the stem vowel is there in plain view. So, for example, e-stem verbs have
  • e + -o  >  -eo
  • e + -s  >  -es
  • e + -t  >  -et
  • e + -mus  >  -emus
  • e + -tis  >  -etis
  • e + -nt  >  -ent
3rd conjugation verbs have consonant stems. There, too, the consonant is in plain view, highlighted in yellow.

But take a look at the endings -- -o, -s, -t, -mus, -tis, -nt -- and notice that most endings are consonantal. Because of that, consonantal endings need a thematic vowel, to make the word pronounceable: to separate the dic- from the -s, -t, -mus, etc. This vowel is normally /i/, but when there’s a nasal sound it moves to the back of the mouth, /u/.
  • c + -o  >  -co
  • c + -s  >  -cis
  • c + -t  >  -cit
  • c + -mus  >  -cimus
  • c + -tis  >  -citis
  • c + -nt  >  -cunt
(When there’s an /r/ the thematic vowel becomes more open, producing /e/, as in dicere and diceris.)

The net result is that, in this revised version of the table, there are only two anomalies, instead of twenty-two:
  1. the contraction in clamo;
  2. the surplus thematic vowel in audiunt.
Compare that to the 22 distinctive features in the first version of the table. Easier, huh?

(Noun declensions developed in a similar way, but their history is more complex and you can’t simplify them as much. The declensions are, respectively, a-stem nouns (1st declension), o-stem (2nd), consonant-stem and i-stem (3rd), u-stem (4th), and e-stem (5th). In early Latin their endings were more similar to each other than they were by Caesar’s time.)

The mixed conjugation

The mixed conjugation is the reason we can’t have nice things. It is Latin’s fifth verb-type. And it is not determined by the sound at the end of a verb stem -- or at least, not straightforwardly. The mixed conjugation is the miscellaneous conjugation; the whatever-I-don’t-care conjugation; the Lord-High-Everything-Else conjugation.

Mixed conjugation verbs are hybrid, with elements of both consonant-stem and i-stem verbs, the 3rd and 4th conjugations. That’s exactly how they’re normally taught, and I can’t really tidy it up any further, because the mixed conjugation didn’t come about as systematically as the other ones. Mixed-conjugation verbs aren’t consistent in their linguistic origins, and there’s no getting around that.

Some mixed conjugation verbs are relatively friendly. Take cupio desire. That’s a very tidy case, because for some of its history, cupio can actually be treated as a 4th conjugation verb in disguise. If you look at its principal parts, except the infinitive --
cupiō, ______, cupīvi/cupiī, cupītum
-- a trained Latinist will immediately recognise all the hallmarks of a standard 4th conjugation verb. If you look up an etymological dictionary, you’ll see it comes from an i-stem. At heart, cupio is a 4th conjugation verb, with an irregular second principal part.

(Of course it’s not as simple as that ... it never is. cupio went through a phase shift in the 1st century BCE. In Lucretius, its imperfect subjunctive appears as cupīret, just as you’d expect for a 4th conjugation verb. Just a few years later, in Cicero, it appears as cuperet, treated as mixed conjugation.)

Some other verbs shift between consonant stem or i-stem depending on a variety of factors, like facio do, capio catch, iacio throw. They’ll be i-stem in the present indicative, but consonant-stem in the past participle -- stuff like that. Some are much more consonant-ish: rabio rave is solidly consonant-stem in its origins, but somehow acquired elements of an i-stem verb. Conversely, pario give birth was originally i-stem, but somehow lost the i altogether in the important participle parens parent. Often a mixed-conjugation verb is one that was originally i-stem in the present tense but consonant-stem in the past tense, like fugio flee, past tense fug-i I fled.

The perfect tense

Next time I teach Latin, I’m going to abandon conjugations altogether for the present tense. I’ll teach thematic vowels instead. Only the mixed conjugation needs special treatment.

But maybe Latinists will raise objections over the perfect tense: aha! the four conjugations do behave differently in the perfect! So you do need conjugations! Well, kind of. Here’s how the traditional conjugations look in the perfect tense:
  • 1st (a-stem): para-  >  parav-i
  • 2nd (e-stem): habe-  >  habu-i
  • 3rd (consonant-stem): inconsistent
  • 4th (i-stem): audi-  >  audi-i / audiv-i
Well, you could do it that way. Or you could learn what a glide is.

A glide is the opposite of a thematic vowel. Remember how a thematic vowel is a sound that pops up between two consonants to make the word pronounceable? Well, a glide is a sound that pops up between two vowels.

English has glides too. Try saying this sentence out loud: The law is just. The sound that comes up between law and is will vary depending on your dialect. For most people, it’ll be either /r/, /w/, or a glottal stop.

In Latin, glides are normally /w/ or /y/. (/r/ only in the genitive plural of 1st/2nd/5th declension nouns.) For example, the word filius son is going to be pronounced /feeleeyus/. With that in mind, let’s look at some perfect tense verbs.

Take the a-stem verb para- prepare. The perfect tense should, by rights, be para-i. But that’s hard to say ... so you add a glide /w/. The result is /parawee/ (as pronounced in the republican period, at least), spelled paravi I have prepared.

The same thing happens with i-stem verbs, but there you have a choice as to which glide you use. So audi- hear can become either /audeewee/ or /audeeyee/, spelled audivi and audii I have heard.

Consonant-stem verbs don’t need a glide, so you often end up with the ending attached directly to the stem: ascend-i I have climbed, vert-i I have turned, em-i I have bought. (More often the stem will be mutated, but let’s not go into that just now: you have to deal with irregularities like that with the other conjugations too.)

It’s the e-stem verbs that are the odd ones out. There, the stem-vowel stays short and the glide moves it back in the mouth, so instead of /habewee/ you get /habuwee/, written habui I have had.

Now, I can see a case for keeping conjugations around for the sake of the future tense. When I suggest abandoning conjugations, it’s really absolute beginners that I have in mind. There’s no point memorising four paradigms. Much better to learn a single paradigm, with just a couple of anomalies.


  1. Very good explanation for present tense.
    The perfect tense of traho is however traxi, not trahi.

    1. Yup, that was a pretty silly mistake on my part. Sorry. Fixed with a different verb now!

  2. I think the teaching of verbs should start from the infinitive (pr. act.). Drop the -re and the endings are perfectly regular in 1st, 2nd, and 4th – with two exceptions (clamo, audiunt). This does not take into account the vowel lengths, which are however always the same for the same persons.

    The 3rd declension, then, behaves badly.

    As to stem analysis, I don’t exactly know how close analysis young pupils (say 10–13 years of age) can take. Maybe they actually can take more of it than I think. I do know nowadays most start Latin a little later in life, maybe at 16 or even 19.

    1. Sorry, conjugation, not declension! Conjugations, if used, are then obviously needed only for present, imperfect, 1st future, imperative, and gerund(ive).

  3. I love this and plan on using this; however, just one comment in terms of your choice of Third Conjugation verb, as someone coming from an all-boys teaching environment. If showing this to high school students, I would probably choose to use rego, regere as the exemplar verb, so that when you break down the stem with with its attendant thematic "u" and personal ending "nt" ... young, dirty minds are not given the chance to give in to their baser instincts upon reaching the third, plural ending of dico in isolation.

  4. This post seems to confuse thematic vowels with epenthetic vowels.

    The "glide" (epenthetic consonant) explanation also sounds wrong since we don't see v surfacing in other such cases of hiatus.

  5. Very interesting! One correction:

    > This vowel is normally /i/, but when there’s a nasal sound it moves to the back of the mouth, /u/.

    If this were true, then it would apply to -mus just as to -nt, because /m/ is a nasal. I think there isn't any single phonetic feature that can quite this neatly explain why only -nt gets /u/.

    One observation is that /i/ puts the tongue in nearly the same position as /n/ and /t/; and that it's only when we get (or would get) all three of those together, for the 3pl of the 3rd and 4th conjugations, that we get an epenthetic vowel of /u/ instead of /i/. I'm not sure if that can be made into a plausible causal story (perhaps /-int/ would sound confusingly similar to /-it/?), but it does at least descriptively "explain" both the limitation to those two conjugations and to that one row of the table.