The above typology seems to also be called "Humboldt-Schleicherian".

While reading this answer in the question "Is there really a difference between agglutinative and non-agglutinative languages when spoken?", I was seized by a desire to know what is being used for morphological typology for languages.

I'm hoping the answer isn't that they are all unique snowflakes, since I rather enjoyed using words like "isolating" as a shorthand for "languages like English, Chinese, Khmer".

  • So you mean alternates for each of those items, additional such items in that framework, alternate orthogonal scales or systematizations.
    – Mitch
    Nov 28, 2011 at 16:11
  • 1
    Taxonomies are a sort of model that descriptively describe how things are and explains certain observations. So I guess Chinese and English appear similar because they are both isolating. But if that is the wrong reason & taxonomy, maybe is it because they have a similar set of Chomskian parameters? I don't know. I was kind of surprised to see that isolating/agglutinating/etc was passe. Nov 28, 2011 at 16:52
  • Don't forget templatic.
    – Nate Glenn
    Mar 29, 2012 at 17:39

3 Answers 3


Besides the terminological confusion (see James C.'s answer), there are other reasons why that particular approach has been abandoned:

  1. Even Humboldt himself realized that those language types are idealizations, i.e. a pure agglutinative or isolating language doesn't exist. That is why Sapir modified that typology into something less crude, for example, the number of morphemes per word (analytic, synthetic, polysynthetic) or the degree of morphemic alteration (isolating, agglutinative, fusional, and symbolic). One of the undesirable consequences was that you had to admit that the same language could simultaneously belong to the different types, e.g. Cambodian being fusional-isolating (Shibatani & Bynon 1999). Greenberg went even further and proposed to calculate various indices. However, as Croft 2003 notices, you could calculate those indices for different word classes and you might end up with an observation that in the same language the nominal system is agglutinative whereas the verbal system is inflectional. Or, as Booij 2007 observes, "Germanic languages are fusional in their inflectional systems, but agglutinative in their system of derivational word-formation" (p. 43).

Comrie (see jlovegreen's answer) proposed two indices, the index of synthesis and the index of fusion.

  1. Kroeber 1954 posed a very good question, "What do we do with a morphological classification of the world's languages when we have it?" Bauer 2003 argued that "the value of typology qua typology is [..] very much in doubt."

A partial solution proposed by Comrie (1989:42--52) is to break the traditional scale up into two orthogonal scales, an index of fusion and an index of synthesis. The index of fusion relates to the extent to which morphemes are segmentable within a word, and the index of synthesis relates to the number of morphemes on average in a word. Polysynthetic languages would have a low index of fusion and a high index of synthesis. Agglutinating languages would have a low index of fusion, and an intermediate index of synthesis. Comrie notes labels associated with morphological typology have been around for a long time, "...and it is to be hoped that general linguistics textbooks will not continue indefinitely to give the impression that this is the only, or most insightful, way of classifying languages typologically." (1989:52)

I might add that another useful pair of distinctions in doing morphological comparison is that between segmental and non-segmental morphology, and between concatenative and non-concatenative morphology.

However, the interesting discussion that came up in the context of the related question concerns a more general issue which has provoked much discussion lately. The issues is whether descriptive labels used by linguists, e.g. "word", "subject", "dative case", "[+voice]", "noun" can be recognized as being instantiated in individual languages. Some representative works are Dryer (1997), Croft (2000), and especially Haspelmath (2010). These authors suggest that the answer to this question is no (though their position is considered controversial). These labels do not mean the same thing in different languages, but instead refer to comparative concepts that tend to be useful in comparing languages. In other words, inconsistencies are bound to result when we try to apply the same set of tests for, say, wordhood, to different languages. This finding doesn't mean that the concept of word is not useful in any given language, it just means that the notion will mean different things in different languages, and may be more or less useful in describing grammatical patterns in different languages.

Comrie, B. (1989) Language universals and linguistic typology. 2nd ed.

Croft, W. (2000) Parts of speech as typological universals and as language particular categories. Approaches to the typology of word classes, ed. Petra Maria Vogel and Bernard Comrie, 65-102. Berlin: Mouton de Gruyter.

Dryer, Matthew S. 1997 "Are Grammatical Relations Universal?" In Essays on Language Function and Language Type: Dedicated to T. Givon, edited by Joan Bybee, John Haiman, and Sandra Thompson, pp. 115 - 143. Amsterdam: John Benjamins.

Haspelmath, M. (2010) Comparative concepts and descriptive categories in cross-linguistic studies. Language 86(3). 663-687

  • Index of fusion and index of synthesis were first suggested by Sapir, Language (1921), to whom Comrie is indebted. Mar 31, 2018 at 11:19
  • @ArtemijKeidan aren't we all?
    – user483
    Apr 2, 2018 at 1:39

Those terms are still used as typological shorthands in linguistics, but none seems to be well defined in the literature, or else the various definitions and usages are so disjunct that there is no widespread agreement on their denotations. Basically, a linguist can still say ‘isolating’ and get their basic point across – a language with a morphosyntax similar to Mandarin, Vietnamese, or English – but after that initial introduction more soundly defined phenomena should be used instead.

To take one term as an exemplar, the term ‘polysynthetic’ lately seems to mean something like “has a lot of morphemes per word and exhibits some kind of noun incorporation”. But there are competing definitions that completely disagree with each other, for example Mark Baker’s extremely limiting definition in his The polysynthesis parameter (1996) that nobody else seems to be willing to accept. Randomly choosing a contemporary article, I find that Russell (1999) works under a different definition of polysynthesis given that he applies the term to languages that Baker has excluded from his definition. In fact, Russell doesn’t really bother to define his idea of polysynthesis, only using it in the title of his paper. That’s basically the status quo – use the term to get the basic idea across and then abandon it as soon as is convenient because it can’t be clearly defined.

(BTW, the name should be “Humboldt”, from philosopher and linguist Wilhelm von Humboldt who was the older brother of the more famous naturalist Alexander von Humboldt.)

  • Baker, Mark C. 1996. The polysynthesis parameter. (Oxford studies in comparative syntax). Oxford: OUP.
  • Russell, Kevin. 1999. The “word” in two polysynthetic languages. In T. Alan Hall & Ursula Kleinhenz (eds.), Studies on the phonological word, pp. 203–221. (Current issues in linguistic theory 174). Amsterdam: John Benjamins.
  • 1
    I should note that the definition of ‘fusional’ as meaning “has lots of portmanteau morphemes” is particularly flawed, since many ‘polysynthetic’ languages have lots of portmanteau morphemes and they are clearly not what is meant by ‘fusional’ ≡ “like Latin and Greek morphology”.
    – James C.
    Nov 28, 2011 at 20:47

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.