Distributed Morphology is right: There is no “lexicon”

The terms lexicon and lexical have a range of different meanings in linguistics, and I have been wondering whether they could perhaps be replaced by something better. Thus, in the spirit of my 2006 paper against markedness, I now distinguish four senses of the traditional term “lexical (item)” (in a paper that will appear in the journal Lexique):

(i) lexical entity as word-form (or simply word)
(ii) lexical item as lexeme (an abstract element based on a root)
(iii) lexical item as inventorial item (an item of the inventorium)
(iv) lexical item as mental item (an item of the “mental lexicon” or mentalicon)

Now interestingly, there is some convergence between my paper and the “Distributed Morphology (DM)” approach, which has gained some prominence over the last three decades (it seems that there is even a handbook in the making).

DM rejects the “lexicalist” views of authors such as Selkirk (1982) and Wunderlich (1996), and in fact there is no single lexicon. In their (2003) overview article, Harley & Noyer distinguish three different senses of “lexical(ized)” and spell out the the corresponding concepts of DM (see the excerpt below). They say that there is no sense in which anything “happens” in the lexicon (i.e. there are no “lexical rules”), and this view has apparently been adopted much more widely in recent years (e.g. Bruening 2018). If “word” and “phrase” were easy to tell apart, one might want to have two distinct sets of rules, but since there is no clear way of identifying words across languages (Haspelmath 2011), this makes good sense, I think.

Harley & Noyer’s first sense of “lexical(ized)” that they distinguish is roughly the same as my sense (iii) (inventorial). An inventorial item is a conventional pairing of a shape and a content (basically the same as Goldberg’s 1995 “construction”), and similar to what Di Sciullo & Williams (1987) called listeme (though it may be that latter authors have in mind items that are listed, not only items that must be listed, i.e. what I call mental items, sense (iv) above). Harley & Noyer are quite right to point out that many inventorial items are multi-word expressions (as was also emphasized by Jackendoff 1995; 1997), so they are “word-like” only in that they would have to be recorded in a dictionary (Wörter-buch in German). DM has a notion of “Encyclopedia”, which may be similar to the inventorium, but I won’t comment on this here.

Harley & Noyer’s second sense of “lexical(ized)”, namely “not constructed in the syntax”, does not have a direct counterpart in my paper, and I am not sure what to do with it, because I do not presuppose that “syntax” is a separate component of grammar. But the notion of “postsyntactic morphological operations” is similar to the concepts that inflectional morphologists have traditionally worked with. The tradition of distinguishing between “words” and “phrases/clauses” is the basis for the traditional distinction between morphology and syntax, and even though it is unclear whether languages really deal with these domains differently, I decided in 2023 that I wanted to have a definition of “word” in the sense of “word-form” (i.e. a text word, not a dictionary word) that can be applied to all languages (see my 2023 paper on defining the word).

Harley & Noyer’s third sense of “lexical(ized)”concerns phonological domains and regularities, and here I have nothing to contribute – but I agree that it does seem to be the case that the old distinction between “lexical” and “postlexical” rules is not tenable, and the notion of the “phonological word” is not robust either (and in any event, it depends on the notion of the (morpho)syntactic word, as is emphasized by Svenonius 2018).

Harley & Noyer do not make reference to the idea of a “Parallel Architecture” (Jackendoff 1997; 2023), but their negative view of the “lexicon” is quite in line with Jackendoff’s perspective, which I find much more suitable for understanding language structures than other generative-grammar views. There is no “lexicon component” that is opposed to one or more “grammar components”, but inventorial items (morphs, complex expressions, constructions) come in degrees of generality and specificity.

It is good that the old componentialist view (made famous by Chomsky’s Aspects) seems to be fading away, but the term “lexicon” will persist – and even more so, the adjective “lexical”, which is deeply entrenched in the linguistic literature. I do not expect it to be replaced (by new terms such as “lexemic”, “inventorial”, or “mentalic”) anytime soon, but it is probably good to have these four alternatives – at least I feel that it’s useful to have these terms because now each time I read “lexical”, I ask myself which of the four senses is intended (or maybe yet another sense?)

A final remark on “lexicalization”: In Harley & Noyer’s paper, there is no distinction between “lexical” and “lexicalized”, but linguists normally use the latter only if some process has happened, generally a diachronic process. But what is “lexicalization” as a diachronic process? This term is widely used, but has rarely been discussed in a clear way, so my forthcoming paper has a section where I point out multiple ways in which it could be understood. I end up saying that the most usual sense of “lexicalization” is what one should call inventorization, i.e. the passing of a previously regular and predictable combination into the inventorium (the set of conventional expressions and constructions). More can be said about inventorization (and its relation to the notion of “institutionalization”), but this will have to wait for another occasion.


Bruening, Benjamin. 2018. The lexicalist hypothesis: Both wrong and superfluous. Language 94(1). 1–42. (doi:10.1353/lan.2018.0000)

Di Sciullo, Anna-Maria & Williams, Edwin. 1987. On the definition of word. Cambridge, MA: MIT Press.

Goldberg, Adele E. 1995. Constructions: A construction grammar approach to argument structure. Chicago: The University of Chicago Press.

Harley, Heidi & Noyer, Rolf. 2003. Distributed Morphology. In Cheng, Lisa & Sybesma, Rint (eds.), The second Glot International state-of-the-article book, 463–496. Berlin: De Gruyter Mouton.

Haspelmath, Martin. 2006. Against markedness (and what to replace it with). Journal of Linguistics 42(1). 25–70. (doi:10.1017/S0022226705003683)

Haspelmath, Martin. 2011. The indeterminacy of word segmentation and the nature of morphology and syntax. Folia Linguistica 45(1). 31–80. (doi:10.1515/flin-2017-1005)

Haspelmath, Martin. 2023. Defining the word. WORD 69(3). 283–297. (doi:10.1080/00437956.2023.2237272)

Haspelmath, Martin. 2024. Four kinds of lexical items: Words, lexemes, inventorial items, and mental items. Lexique (to appear). (https://ling.auf.net/lingbuzz/007880)

Jackendoff, Ray S. 1995. The boundaries of the lexicon. In Everaert, Martin & van der Linden, Erik-Jan & Schenk, André & Schreuder, Rob (eds.), Idioms: Structural and psychological perspectives, 133–165. Hoboken: Taylor and Francis. (Accessed August 15, 2022.)

Jackendoff, Ray S. 1997. The architecture of the language faculty (Linguistic Inquiry Monographs 28). Cambridge, MA: MIT Press.

Jackendoff, Ray S. 2023. The Parallel Architecture in language and elsewhere. Topics in Cognitive Science n/a(n/a). (doi:10.1111/tops.12698)

Selkirk, Elisabeth O. 1982. The syntax of words. Cambridge, MA: The MIT Press.

Svenonius, Peter. 2018. Delimiting the syntactic word. Manuscript. (https://blogg.uit.no/psv000/wp-content/uploads/sites/55/2018/04/syntacticword.pdf)

Wunderlich, Dieter. 1996. Minimalist morphology: The role of paradigms. In Booij, Geert & van Marle, Jaap (eds.), Yearbook of Morphology 1995, 93–114. Dordrecht: Springer. (doi:10.1007/978-94-017-3716-6_6)





Cite this blog post
Martin Haspelmath (2024, May 12). Distributed Morphology is right: There is no “lexicon”. Diversity Linguistics Comment. Retrieved May 31, 2024, from https://doi.org/10.58079/11ntb

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.