Grammatical gender often seems arbitrary from a semantic point of view. When I was taking French many years ago, we were told that one must simply memorize the gender for each noun. Are there any languages with similar markings on the verb--obligatory morphological markers for the verb that typically seem arbitrary from a semantic point of view?

  • Arbitrary in what sense, and to whom? Would you consider the obligatory irregular past forms of some English verbs but not others arbitrary (i.e. "is it he speaked, he spoke, or he spake?")?
    – Cerberus
    Commented May 13, 2012 at 7:12
  • 1
    Semantically arbitrary, as the question states. Obviously, selection of 'speaked,' 'spoke,' and 'spake' would not be pragmatically arbitrary or syntactically arbitrary, but they would all have the same meaning. Commented May 13, 2012 at 7:19
  • Some verbs have a regular past (kiss, kissed), other have an irregular past (do, did). Does this distinction count as "semantically arbitrary" for you? Learners go though the same process of memorizing long lists of words.
    – Cerberus
    Commented May 13, 2012 at 7:51
  • 1
    The relevance of the regular/irregular distinction hadn't occurred to me until your comment. But I suppose that the distinction between the conjugation of regular and various types of irregular verbs might be analogous to grammatical gender. Yes it seems semantically arbitrary to me, since "did" and "*do-ed" would mean the same thing. Commented May 13, 2012 at 17:45
  • The fact that you don't know why a noun is inflected a certain way should not be sufficient ground to call it arbitrary. It would be arbitrary only if you could arbitrarily change it. The nominal gender allows a determiner in a relative clause to unambiguously select one of multiple possible referents from the main clause. This makes it semantic. If you don't think of this and that as based on gender, nor who, which, that, but Ger. dies-, jen- + -er/-e/-es or der, die, das and e.g. Fr. quel- + -_/-le (not even audible), or lui, leur, then it is the question that is arbitrary.
    – vectory
    Commented Jul 13, 2019 at 12:46

5 Answers 5


Romance languages have (generally speaking) three classes of verbs, inherited from Latin. In French, these correspond to the infinitive endings -re, -er, and -ir. These are like gender in that they don't correspond to any semantic property of the verb. Unlike gender (as expressed in Romance), it's difficult to point to a single morpheme as the "gender marker." However, the different verb classes have differences that pervade the paradigm of tense and agreement marking. In the Romance context, this is sometimes called the "theme vowel," hearkening back to a stage of the language when the morphology was more segmentable.

Many other languages have a notion of "verb class" that is even less morphologically transparent. Old English for example had seven classes of strong verbs; each of these had a different ablaut pattern. (There were also three different weak verb classes.)

  • 1
    Latin is considered to have 4 conjugations. Just to avoid any misunderstandings.
    – dainichi
    Commented May 15, 2012 at 4:49
  • 4
    Yes, but the gender of the noun is ruling over the adjective (enforcing forms based on the gender of noun), and nothing similar happens with the classes of the verbs in above examples. Commented Jun 4, 2012 at 11:47
  • 2
    To me that is the counterpart of the nominal declinations (5 in Latin). But words from a certain declination can be of almost any gender anyway (femina vs. agricola). I do not see what has this to do with the question. Commented Jul 13, 2019 at 20:10
  • @VladimirF The first declension is almost always feminine, second declension almost always masculine, 3rd 1/2 and 1/2? 4th rare, fifth all feminine except 'dies'. So gender maybe arbitrary as in not deterministic by category, but it is somewhat predictable. But ignore gender for the moment. It i-is- arbitrary what (semantic) noun goes in which (syntactic) declension, and that is similar to how it is arbitrary what verb goes in what conjugation. For the verbs there happens to be no convenient canonical verbs with meanings like nouns do (with male things in the masculine group).
    – Mitch
    Commented Jul 15, 2019 at 14:53

French and other Romance languages provide an example, actually – verbs are arbitrarily associated with theme vowels.

For example, in Spanish (I am not familiar with French), the infinitive form of to have is tener, and that of to take is tomar. The bolded vowel, traditionally called "theme", affects most of the inflected forms of the verb: tienes (you have) vs. tomas (you take), tiene vs toma (third person singular), tenéis vs. tomáis (second person plural), etc.

Crucially, these are inherent properties of the verbs, not simply a part of their phonological form: in the Spanish subjunctive, verbs with an e or i theme take a in the theme position and verbs with an a theme take e in the theme position. For example, tomas becomes tomes, and tienes becomes tengas. (Obviously, there are some other things going on here morphophonologically, but the important part is the theme vowel.)

More generally, grammatical gender, theme vowels, and the Germanic verb classes mentioned in Aaron's answer, are kinds of inflectional class (in the modern sense, not the more limited traditional sense), i.e. arbitrary groups of roots/stems identifiable by a morphological reflex.


I'll also throw in that some languages also have grammatical gender in their verbs. Polish, for example (I assume other Slavic languages do as well, but I'm only familiar with Polish).


  • chciałbym = I would like (male speaking)
  • chciałabym = I would like (female speaking)
  • byłem = I was (male speaking)
  • byłam = I was (female speaking)
  • 5
    These are (gender-marked) participles fused with finite verbs. Hebrew marks the gender on finite verbs themselves (though normally only in the second or third person singular). But I don't think this is actually at all relevant to the question.
    – Colin Fine
    Commented May 16, 2012 at 23:25
  • 1
    Yes, but it is a gender of the noun that is ruling over the verb. Commented Jun 4, 2012 at 11:46
  • Well, I do think it's relevant to the question because the subject's gender is shown in the verb. The subject isn't always a person; it could be something that we English speakers don't generally consider to have gender. In this way, the verb requires morphological markers that "seem arbitrary from a semantic point of view." Commented Jun 18, 2012 at 17:49
  • Wonder how this compares historically to the romance thematic vowels mentioned in other answers.
    – vectory
    Commented Jul 13, 2019 at 12:18
  • @vectory Slavic verb classes are divided according to the present stem suffixes. However, they are not just the wovels (-e, -ne, -je, -i, -a in Czech, similarly in Proto Slavic en.wiktionary.org/wiki/Appendix:Proto-Slavic_verbs#Verb_classes Russian, for example, has -aje(t) where Czech has just -a). Commented Jul 13, 2019 at 20:16

English might only have one pattern of regular verb conjugation (depending on how you define "regular", the strong Germanic verbs have several sub-regularities), but many languages have several classes of regular verbs which have separate rules for conjugation. Usually these are independent from semantics.

Other than the Romance ones that others have mentioned, examples are

  1. Danish has two regular ways to form past tense, -ede and -te.
  2. Swedish has 3 regular conjugations, -er verbs, -ar verbs and -r verbs.
  3. Modern Japanese has two regular conjugations, godankatsuyo (5-row conjugation) and ichidankatsuyo (1-row conjugation).
  • 1
    This does not seem to have anything to do with nominal gender though, or the answer at least does not show it.
    – vectory
    Commented Jul 13, 2019 at 12:20

The perfect example of verbal gender/verb classes would be a language that assigns every verb to one of a small number of classes and has morphology elsewhere in the sentence that agrees with the verb, e.g. declining adverbs to agree with the verb. I don't know of any slam dunk examples.

French itself, though, has an example of something similar. The selection of avoir (to have) or être (to be) for forming compound tenses is partly lexical.

A closed class of intransitive verbs use être (e.g. aller (to go), devenir (to become), naître (to be born)) and every reflexive verb uses être.

Je             suis     allé(e).
1sg.NOM be.1sg.PRES   go.PP.M/F
I went / I have gone.

J'                 ai     dormi
1sg.NOM have.1sg.PRES  sleep.PP
I slept / I have slept

Selection of être is highly correlated with unaccusativity, but in French the choice of auxiliary cannot be predicted perfectly from semantic criteria alone.

This paper goes into some detail about the verbs that form compound tenses with to have and to be in French and Italian.

  • Yes, in Italian and German too, but it's not very arbitrary, if that's a criterion here. Commented Jul 13, 2019 at 10:27
  • @AdamBittlingmayer it is reasonably arbitrary, if L2 speakers getting it wrong frequently is significant enough. Active "Have kept" is pretty logical, "was got" follows maybe reasonably, but "have lost" is an outright oxymoron.
    – vectory
    Commented Jul 13, 2019 at 12:06
  • However, in a strict sense of analogy, I'd have to say that I do not see how this could have been derived by analogy from nominal gender, unless e.g. the morpheme -t related to "it" at least in some cases.
    – vectory
    Commented Jul 13, 2019 at 12:14

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.