This is not a naif question asked by a layman just out of curiosity. I am presently editing a book by a colleague which is devoted to the notion of grammatical feature (with a special focus on morphology). The author's approach is quite unorthodox, yet very convincing. The fact is, however, that there is no really "traditional" or at least "commonly accepted" definition of this notion. At least, not to my knowledge. Such a recent publication as Features: Perspectives on a Key Notion in Linguistics, while providing many interesting details and discussions on specific features in specific languages, does not give a general and fundamental definition of this notion.

(To make a comparison: the notion of phoneme may have different descriptions, but there can hardly be anyone who does not share the widespread definition of the phoneme as, say, "meaning distinguishing segment").

Therefore, I try polling the colleagues' opinions on what the grammatical features (also called grammatical categories) are. Consider this list of examples of grammatical features:

  • gender
  • number
  • case
  • person
  • time

In case you are asked to define what do all the items listed above have in common, how would you answer?

(Possibly, avoid referring to syntax, I am trying to remain within the domain of morphology).

  • Ah, one of the mostly studied questions in Soviet and Russian linguistics. I strongly recommend reading chapters 1 and 2 in Plungian 2011 (Введение в грамматическую семантику. Грамматические значения и грамматические системы языков мира), esp. pp. 21-24, and his bibliographic commentary at the end of each chapter too.
    – Alex B.
    Jun 24, 2021 at 19:37
  • How do you intend to avoid referring to syntax? You can't describe the full range of facts of morphology without crucial references to syntax: case, agreement, noun incorporation, valency, languages with special verb inflections for subordinate clauses (ex. Greenlandic), 'dummy' pronouns, gerunds vs. nominalizations, 'quirky subjects' like in Icelandic, reflexive and reciprocal forms, clitics, polysynthesis, and so on...You can't describe morphological features in isolation from syntax without sacrificing a whole lot of power, both descriptively and explantorily.
    – Khove
    Jun 24, 2021 at 22:44
  • @AlexB. I know that it’s a very popular topic in Russian linguistics and I know Plungian’s work. I could have specified that I am interested in extra-Russian approaches but that would have sounded a bit racist. Jun 25, 2021 at 2:13
  • I see. I’m sure you read most relevant research by Osten Dahl as well? He uses the term gram.
    – Alex B.
    Jun 25, 2021 at 3:38
  • 1
    @AlexB. thank you for the reference to Osten Dahl's work, I did not know it. And yes, I also have the impression that the study of grammatical categories is something typically European, especially Russian. Jun 25, 2021 at 15:10

1 Answer 1


By way of context, I am a phonologist, and I do not share your definition of the phoneme as a "meaning distinguishing segment", moreover I claim that socially speaking that the meaning definition is a view of phoneme not generally shared by phonologists, though it seems to shared by non-phonologists. I'm not pointing this out to harass you, I say this because it is an example of a view of theoretical concepts in linguistics that I think is mistaken. Specifically: the confusion between the definition of a concept, and empirical discoveries about the thing that the concept refers to. The "meaning-distinguishing" characterization is a far-removed theorem from something about what "phonemes" and other things are. If sounds X and Y don't happen to enter into a minimal pair in some language, that doesn't mean that they aren't distinct phonemes. The meaning thing is a potential consequence of something else, it is not the defining essence of "feature".

In that vein, then, there may be a "definition" of feature, but then there are empirical theories of the essential properties of features. Your question is not, as I see it, about the definition of features, it is a question about the nature of features in morphology. There is no reason to say that "feature" is a different concept in syntax, phonology or morphology, but these three kinds of feature may have somewhat different properties. "Feature" in grammar is one thing, and the relation "precedes" is a totally different thing – those concepts have totally different definitions.

The desideratum of avoiding syntax is highly problematic since most theories of morphological features are syntax-heavy (case, for example, can't sensibly be understood without reference to syntax).

In asking what the features of morphology are, the question apparently presupposes that there is a fixed list of available features, but that itself is a controversial assumption. For example, Ethiopian Semitic languages make a distinction at least between type A, B and C verbs, which plays a role in morphology. Bantu languages make distinctions in the tense system not just between completive and noncompletive, and past, present and future, they also distinguish a number of degrees of past and future, other distinctions such as aspectual focusing, "persistive", various expressions of speaker "attitude", not to mention clause type (main clause vs. subordinate, affirmative vs. negative). There are dozens of such verb forms as reflected in verb morphology.

There is a view of "feature" that simply says that there are "features", which are attributes, and the type that a particular feature has is language-specific. Given that perspective, there is no "list of features". Given a different perspective, one with a baked-in list of parameters (e.g. Optimality Theory), there probably has to be a fixed list.

  • Thank you for your answer. Honestly, I did not intend to discuss phonology here. I can even partly agree with your criticism of "meaning-distinguishing characterization" of phonemes. Let's invert the implication. Not every phoneme generates minimal pairs (simply because not any sequence of phonemes is a word in the lexicon); still, but every minimal pair is distinguished by phonemes. Jun 25, 2021 at 15:16
  • Concerning the essential part of your answer. I am afraid you have misunderstood my question (or I could have not expressed myself clearly). I am not asking you for the list of all the possible features (my own list being just an example). What I am looking for is some kind of abstract definition that characterizes them all. Something like this: «a feature is a set of values automatically selected by the syntactical environment and expressed by bound morphemes». Jun 25, 2021 at 15:24
  • Is your claim that there is a special definition of "feature" for morphology that sets it apart from "feature" in syntax or phonology? How are "gender; number; case; person; time" relevant to your question? Is it that you reject the distinction between "definition of X" and "theory of X"?
    – user6726
    Jun 25, 2021 at 15:41
  • All "features" exist in "feature systems", which are artifacts of analysis, and not facts about language or any particular language. There are so many analytic modes and so many definitions (and lists) of "feature" that you simply can't define it -- like theology, you have to accept it without proof. Also like theology, you can only accept one, and they're contradictory. This sounds like yet another one. Consult your confessor.
    – jlawler
    Jul 3, 2021 at 20:49

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.