As you are probably aware, Spanish owes a significant portion of its vocabulary to Latin. An interesting difference however is that Spanish has only two genders for nouns - feminine and masculine. The coincidence of the Spanish gender with that of Latin is very high (in fact, so far I've not come across a Spanish noun which has the opposite gender in Latin).

Now, as a native speaker of Spanish, I can sometimes sense the gender (m or f) of a Latin word I encounter for the first time by drawing from Spanish. This is quite useful, as I can then try to get the declension, mode and understand its meaning, without having to go to a dictionary.

For instance, if I read insulae, my Spanish tells me this is feminine. If I read, fluvium, my Spanish tells me this is masculine. But if I read altaris, I would say this is masculine. And yet, it is neuter!

Thus, when it comes to neuter nouns in Latin, my method utterly fails, because there are no neuter nouns in Spanish! Thus, I am looking for some intuitive way, a rule of thumb, or way of sensing when a Latin noun might be neuter. Is there such a thing? Or is this ultimately arbitrary, and then purely memory?

More generally, is there a way to predict the gender of nouns? At least in Spanish, as far as I know, there is no ontological underpinning to the fact that a cuchara (spoon) is feminine and a tenedor (fork) is masculine. Thus, it might well be that the gender of many Latin nouns is so because it is so. Memorā!

  • 1
    Arbor is feminine, árbol is masculine. Just to let you know there are exceptions
    – Rafael
    Commented Aug 13, 2018 at 10:10
  • 1
    BTW your answer is probably in the line of your last paragraph: grammatical gender is mostly arbitrary, except for the most obviously gendered words, and even those can have exceptions
    – Rafael
    Commented Aug 13, 2018 at 10:19
  • Several fruits are neuter second declension in Latin but end up as feminine in romance. Perhaps the plural in -a got "regularized" into feminine singular, especially for small fruits that you never eat just one of (e.g. cherries). Commented Jun 12, 2023 at 13:30

2 Answers 2


I'm not sure this covers all relevant ideas, so any addition/clarification is appreciated.

Four ideas that can help you:

  1. Regarding logics as to certain types of nouns being predictably neuter by semantics alone, the main rule is... genders are arbitrary.

In view of that, the next best thing you can do is to pay attention to syntax and etymology:

  1. There is a number of rules of thumb for declension genders you may or may not be aware of, that can be helpful:

    • 1st decl. nouns are mostly feminine (with the notable exception of occupations like poeta, agricola, nauta, among other exceptions).
    • 2nd decl. nouns are mostly masculine or neuter (with exceptions like trees, such as ficus, fig tree).
    • 3rd decl. nouns may be either gender, but there is a number of lesser rules depending on endings. For example, most nouns ending in -men (e.g. flumen) are neuter [thanks to Colin Fine].
    • 4th decl. nouns are mainly masculine (if ending in -us) and neuter (those in -u).
    • 5th decl. nouns are nearly all feminine (with the notable exception of dies that occurs as both m. and f., and compounds like meridies).

    See this book for more detailed (yet not exhaustive) rules. You can navigate the pages with the < and > buttons in the top-left corner. The rules include etymological derivations driving gender, like the one for -men/-minis.

  2. Cases themselves give some clues too:

    • Neuter nouns do not change between nominative, vocative and accusative (e.g. templum). Neither in singular or plural. Seemingly there is absolutely no exception to this rule.
    • Most neuter nouns and adjectives form plurals in -a (including -ia, -ua; I think this is the case of all 2nd declension neuter nouns, and most if not all of those belonging to 3rd and 4th declension). Examples: principia (2nd), corpora (3rd), cornua (4th).
  3. Adjectives are also important clues, since some of them are more predictable and must agree in gender with nouns:

    • Rules for neuter nouns apply to adjectives too: Not changing in voc. and acc., and forming plurals in -a.
    • Many adjectives follow the paradigm of bonus/-a/-um (sometimes called 1st/2nd declension adjectives).
    • 3rd declension adjectives can either distinguish all three genders (e.g. acer, acris, acre), contrast m. and f. against n. (levis, leve), or make no distinction (atrox). In the first two cases, you still have the same clues (pl. in -a, acc.=nom.). Most of the neuter versions of 3rd decl. adjectives form sing. nom. and acc. in -e.
    • Comparatives in -ior are also 3rd declension and form the neuter in -ius. Example: prolixior (m. and f.), prolixius (n.)
  • Thanks! That is surely useful, but the spirit of my question (perhaps not clear in the wording) was rather different. In a sense, you reversed the directionality. By knowing declension rules, you can predict the noun gender. My logic is to derive declension rules from gender. Thus, I was hoping for the rules of thumb to come from elsewhere. My logic of why insula was feminine was because in Spanish it is. But I cannot do this with neuter nouns. Thus, is there an ontological reason behind the gender of nouns in Latin? I guess, as you pointed out in a comment, there is not.
    – luchonacho
    Commented Aug 14, 2018 at 8:29
  • 1
    @luchonacho I should have started from the 4th idea (which I think Alex B. conveys better): in his words, semantic assignment is somewhat marginal (only 8.9%). Semantics is what you are asking for (non technically, meaning, ontology). For example, things tend to be neuter. Besides that, syntax and etymology is the best you can have. See also the wp article I linked in the "it's not completely random" bullet.
    – Rafael
    Commented Aug 14, 2018 at 14:06
  • @luchonacho fixed. Old no. 4 is now no. 1.
    – Rafael
    Commented Aug 14, 2018 at 20:01
  • 2
    @Rafael Semantic assignment of gender is “somewhat marginal” in Latin because it only happens with animate nouns, and even then there are some, albeit rare, exceptions.
    – Alex B.
    Commented Aug 14, 2018 at 23:13
  • 2 Decl has an interesting exception: humus, humi, f.
    – Lenny
    Commented Aug 20, 2019 at 7:56

I was reading a new corpus study on gender assignment the other day - Hoffmann 2016 Gender in Latin and in language typology (in Latinitas Rationes, ed. P. Poccetti).

Based on two corpus studies, Hoffmann 2016 argues the following:

  • grammatical (morphological) gender assignment is predominant in Latin (71.4%), it includes gender assignment based on declension type (43.6%) and derivation (i.e. suffixes, 27.8%);
  • phonetic assignment also plays its role in Latin (12.2%);
  • semantic assignment is somewhat marginal (only 8.9%);
  • exceptions (4.9%) and Greek foreign words (2.6%) are rather marginal.

Corpus 1

  • 2,500 nouns total;

  • The first 500 nouns were counted, irrespective of repetitions, from each of the following works: Plautus, Amphitruo; Caesar, Bellum Gallicum; Cicero, De officiis; Sallust, Bellum Iugurthinum; Catullus, carm. 1-36.

Corpus 2

  • 3,545 nouns total;

  • It includes the lexicon of the following two authors: Cicero (philosophical works only); Virgil;

Roland Hoffmann determined the kind of gender assignment (listed above in the first bulleted list) of the nouns from the corpora using the following, commonly used, declension types:

  1. a-stems;
  2. o-stems, with the two syb-types,-us/-er and -um;
  3. cons. + -i declension;
  4. u-declension;
  5. e-declension.

For more information, see Tables 8-13.

  • So, point 1 means that gender is "chosen" primarily for language economy/efficiency? E.g. so that you do not have ten declension types but only five?
    – luchonacho
    Commented Aug 14, 2018 at 8:31
  • 3
    @luchonacho Of course. What is gender (in language) anyway? Gender is "a property that separates nouns into classes" and even though these classes are often linked to biological sex, that is not the only option. linguistics.oxfordre.com/view/10.1093/acrefore/…
    – Alex B.
    Commented Aug 14, 2018 at 13:28
  • 1
    Superb article! "...Nigerian Fula with more than 20 genders..." :0 Amazing! "Generally speaking, there are three types of assignment rule: semantic, phonological, and morphological." I guess it is the former the one I was thinking about, but according to your answer (the OUP article has little to mention about Latin), the latter is the one prevailing in Latin. If you have the time, maybe you could expand your answer this this broader picture. At least a reference to that article would be helpful.
    – luchonacho
    Commented Aug 14, 2018 at 13:43
  • What are those percentages exactly? Are they fractions of nouns listed in a dictionary or other word list, or fractions of nouns in a corpus, or something else? I imagine that weighting by frequency will change the percentages but I have no idea which way. It's not obvious whether res should have equal weight compared to a hapax legomenon in such comparisons.
    – Joonas Ilmavirta
    Commented Sep 6, 2021 at 23:59
  • 1
    @AlexB. Thanks! It certainly is.
    – Joonas Ilmavirta
    Commented Sep 7, 2021 at 5:39

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.