Dominique Sportiche, Hilda Koopman, and Edward Stabler [1] make the following claim about Affixation in section 2.3.2 of their Introduction to Syntactic Analysis and Theory:

There are 5!=120 different orderings of the five morphemes in de-nation-al-ize-ation, but only one order forms a word... We claim: (1) A speaker of English, even one who is unfamiliar with this word, will only accept one of these sequences as a possible English word.

This isn't because only one of them is an English word:

We could try another one, even a non-existing word (e.g. denodalization, from node - nodal - nodalize, etc)... What are the regularities that a speaker needs to know in order to accept denationalization and reject the other 119 forms?

This seems to be untrue because the combination de-nation-ize-ation-al (denationizational) is equally valid - to quote two of the rules they give:

  • "-ize c-selects root N to form V, e.g. symbol-ize" (I would have thought that "nation" is just as much a root N as "symbol" and can therefore legitimately form "nationize")
  • "-al c-selects N to form A, e.g. natur-al" (Indeed, they themselves note that "from denationalization, we could form denationalizational")

Again, the words "nationize" and "denationizational" might not yet have been coined, but the authors stressed that their point about "denationalization" being the sole legitimate ordering of the five morphemes isn't dependent on the word actually existing. So it seems that their "empirical" discovery that only one of the 120 orderings is valid is flawed - or is it I who have misunderstood?

I am not sure whether Sportiche et al. view "-ize" as two separate suffixes (that just happen to have the same form), one attaching to adjectives and the other to nouns. But this way of defending their proposition would fail to support their claim that "a speaker of English, even one who is unfamiliar with this word, will only accept one of these sequences as a possible English word".

[1] Sportiche, D., Koopman, H., Stabler, E. (2014) An Introduction to Syntactic Analysis and Theory. Wiley Blackwell.

  • The title question is a simple empirical question, which calls for a psych experiment with a pile of speakers. I don't know of any test of this particular collocation of affixes, but I tried come alternative orders on other native speakers, who agree with me that *denationizalation, *denationationizal, *denationizational. I believe that the prediction about behavior is correct. You question really seems to be whether their account actually makes the prediction that they claim, or did they construct an account that post hoc explains that fact. – user6726 Oct 29 '20 at 20:40
  • I agree that nationize is possible even if icky, and it means something like "make into a nation". "Denationizational" also seems acceptabl-ish, and has is an adjective having to do with dissolving a nation. Denationalization means something totally different, about restoring property that was previously taken by the government. What role does semantics have in their theory? – user6726 Oct 29 '20 at 20:41
  • 1
    I will look into the role of semantics (it doesn't have much role in the sections I've read so far) but their assertion is that the other 119 orderings are impermissible (not that they would be permissible but with different meanings) - semantics seems irrelevant there. – rjpond Oct 29 '20 at 21:20
  • I agree, of course, that denationizational and denationalizational have or would have different meanings, but as far as I understand it, the question is whether my "denationizational" is validly formed (as a hypothetical word) - which as a native speaker, I think it is. I suspect those saying otherwise may unconsciously be over-influenced by the fact it isn't an actual word (yet?) (whereas denationalization is) - which isn't supposed to be a factor. – rjpond Oct 29 '20 at 21:20

I think you've just found an error here. I don't think they actually checked all of the possible permutations of the morphemes, so they missed that de-nation-ize-ation-al was also valid.


Honestly that claim is flawed to begin with. Certain morphemes are only suffixes or only prefixes, and some only are used on certain parts of speech. So the only knowledge an English speaker needs to have is the meaning of the morpheme (i.e., "-ize" means "to make [object] [word being affixed]").

  • Thanks for the reply, but I don't agree. Of course, I tend to agree that their claim is flawed in its specifics, but your answer gives their account too little credit. They fully acknowledge that certain affixes can only be used in certain ways. While I didn't go into the details of their theory on that, surely the fact that they argue for those limitations is implicit in the claim that only one of the 120 variations is valid. – rjpond Oct 29 '20 at 21:25

Very nice point. I think their point is the relative order of those morphemes. So if you take your newly coined de-nation-iz-ation-al, you could form a similar word by taking the authors' de-nation-al-ize-ation and adding one mor -al, getting de-nation-al-ize-ation-al.

To be clear, I think their point is that in English no-one would say de-nation-al-ation-ize, since the composition of the morphemes' semantics doesn't arrive at what de-nation-al-iz-ation means -- rather, it would be a verb, to begin with. I think their point is that morphological order reflects compositional semantic hierarchy.

But I admit this doesn't answer if the authors' argumentation is valid, I'll leave that to others.

  • 1
    Thanks. I think you are right about their main point (that the affixes cannot combine in just any order), but they did say that no one would accept any other order "as a possible English word" (not, for example, "as a noun"). I think I agree with their overall point, although obviously, the fact that I think they may be mistaken about one of the specifics also prompts me to wonder what else I might decide they're mistaken about as I think about it further. But that's probably healthy. – rjpond Oct 29 '20 at 23:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.