I'm computationally working on an agglutinative language, a morphological analyser. The thing is that roots can form verbs (not all) but some roots like lüg 'white' is classified as Adjective if it is used as an independent word lüg ruka 'white house', and classified as verb when followed by suffixes lügüy 'it became white', literally 'it whitened'.

So, my hypothesis is that really these kind of roots belong to a "conceptual category" where the root holds the concept it displays when used, I mean it is a verb when followed by suffixes and adjective when not, but originally it has not a determined category.

The question is, in linguistics is this possible? or am I just delutional? if this is possible, is there any linguist who support this idea? who?

  • Parts of speech are morphological and syntactic categories, not semantic ones. Many languages form different parts of speech from a single root: Semitic languages are particularly well known for it, but English allows not just the same root but the same form of the root to act as verb, noun, adjective. I'm not quite sure what you mean by "conceptual category", but I'm pretty sure it's possible.
    – Colin Fine
    Oct 2, 2020 at 14:54
  • hypothesis: there is a general category, let's calle it "conceptual category" where there are the words not used yet, unclasified. Anyway these words hold a concept, for instance the form "work" holds the concept of laboring, when it is used it concretizes that concept becoming an action (verb) or the name of something (noun), for instance 'I work hard' and 'at work' in the first phrase work is verb, in the second is noun, but only because it is used one way or another, but basically the "original" word only holds a conception of something. Oct 2, 2020 at 17:26
  • On the contrary, it could be that words are just homonyms "work" verb has the same form of "work" noun. Oct 2, 2020 at 17:27
  • Arguably you made a good point with your first example. A root that describes actions would be inherently verbal, and a root that pertains to substantives inherently nominal (piece of work, Werk). However, analytical languages more than others make it crystal clear that the difference is difficult to split. The problem is probably that semantic drift and all the word formation processes are often inscrutable from a syncronic perspective, and if that is the case, then diachronic analysis won't fare much better because isolating a unique proto-language down the line you are back at square one.
    – vectory
    Oct 3, 2020 at 0:48
  • "... where there are the words not used yet, ..." metaphors derived from programming don't work particularly well in cognitive sciences, i.e. associative memory layout, registers, discrete values are all bizare abstractions. In a relational algebra e.g. you could say for the base case that every word has as a concept its own form (primary key). Thus you could even have a null morpheme, theoretically. You can then further conceive of its context (secondary keys), to distinguish homonymy. The only morpheme that has only itself as context is the null morpheme, logically. ...
    – vectory
    Oct 3, 2020 at 1:01

3 Answers 3


Yes, some linguists consider this possible. Here are some such concepts/authors:

"roots": Pesetsky, David. 1995.Zero syntax: Experiencers and cascades (CurrentStudies in Linguistics 27). Cambridge: The MIT Press.

"listemes": Borer, Hagit. 2005. Structuring Sense Volume I. Online:http://dx.doi.org/10.1093/acprof:oso/9780199263905.001.0001.

"bare V": Landau, Idan. 2006. Chain resolution in Hebrew V(P)-fronting.Syntax9.32–66. Online:https://doi.org/10.1111/j.1467-9612.2006.00084.x

Categorization is also called "labelling", or "projection":

  • Chomsky, N. (2013). Problems of projection. Lingua, 130, 33–49. doi.org/10.1016/j.lingua.2012.12.003

  • Chomsky, N. (2015). Problems of projection. Structures, Strategies and Beyond, 1–16. doi:10.1075/la.223.01cho

From the 2013 one: (http://www.eact.ir/wp-content/uploads/2013/12/chomsky-2013.pdf p. 37) "(...) separation of projection (labeling) from the principles of construction of expressions"

I haven't heard of a "conceptual category" as you suggest; in these approaches, it's rather "lack of category".


Your hypothesis is true, partially. Tamil employs agglutinative grammar. Suffixes may be used to mark noun class, number, case, verb tense and other grammatical categories.

Wikipedia has a great example of agglutination in Tamil.

The only place where I would differ from your hypothesis is that all words by themselves would belong to some 'category' as you are calling it, at least in Tamil.


This is a common issue in Austronesian linguistics where the notion of precategorial (=functionally unspecified) roots is often employed to explain the fact that roots don't have a POS category until they're employed in an utterance, and then the same root can be used in many different POS categories. This may match the situation you describe, where the root can be thought of as holding primarily semantic information, but also may be specified for a range of potential POS categories.

Some links:

Nêlêmwa (Austronesian, New Caledonia)

Late Archaic Chinese

Lexical precategoriality in Polynesian languages

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.