Linguistic typology started out in the early nineteenth century with work by the brothers Friedrich and August Schlegel and Wilhelm von Humboldt. At that time, it was the history and evolution of languages which enjoyed primary interest. Thus, instead of developing synchronic linguistic description, scholars launched more or less speculative hypotheses about the development of languages. Humboldt (1836) sketches an evolutive typology which is often dubbed ‘morphological typology’. This is not accurate in view of the fact that Humboldt (o.c., 653f) clearly stated that the types were ones of ‘sentence formation’. These are the isolating, flexional, agglutinative and incorporating type. The incorporative type will be left aside in what follows.

The model of the isolating type at that time was Old Chinese. The idea was that a language of the isolating type only has roots, no different word forms. There was no grammatical morphology; the sentence was just a combination of roots. Using a modern example here (), it was thought that morphemes like yòng are nothing but lexical roots. Even morphemes like wǒ seemed to fit in the picture since they do not show any inflection, either.

The flexional type is contrasted with the agglutinative type as follows (1836: 490f; Lehmann 1979): While the agglutinative type – standardly represented by Turkish, but in fact widespread in the world – provides a separate expression unit for the lexeme and for each of the grammatical features born by it, the flexional type – standardly represented by an ancient Indo-European language, but also found in other linguistic families – merges all of these semantic components in one expression unit. More precisely, it signals grammatical categories on the stem by modifying it (this being the original intuition behind the term ‘inflection’). Thus, Turkish () has a morpheme for each of the notion ‘year’, the genitive and the plural. Each grammatical feature is coded when needed and otherwise absent. Thus ‘year’ without any specification of number or case (e.g. in subject position) is simply yıl.

By contrast, Latin () has an ending for the genitive singular, another ending for the genitive plural and no word form corresponding to the bare lexeme, since neither the root (ann-) nor the stem (anno-) can be used as a word form in the construction of a sentence. In 20th century morphology, this insight has been reframed as the distinction between an ‘item-in-arrangement’ and a ‘word-in-paradigm’ morphology: The Turkish word form can be profitably described as a sequential arrangement of morphs each of which codes a value of a parameter. The Latin word form is more efficiently described as a (possibly unanalyzed) member of a paradigm.

Humboldt turned this relatively simple holistic typology into a theory of the evolution of language: A language starts out at the level of the isolating type. Grammaticalization (the term was not yet available, though the idea was) then deranks some of the words to the status of grammatical affixes, which produces the agglutinative type. Finally, the bond between the affixes and the root becomes tighter, distinct morphemes merge into one, and the entire set of morphological categories ends up being conveyed by a modification of the stem. This produces the flexional type.

Towards the end of the century, Georg von der Gabelentz (1891) threw overboard the prejudiced idea of the primitivity of the isolating type and proposed the idea of an evolutionary spiral. He added a further phase to the Humboldtian evolution path, viz. the loss of all inflectional morphology in a language of the flexional type, adducing English as an example. As a result, the development ends where it started. However, since 19th century scientists believed in cognitive and communicative development of mankind, the development did not simply return to its point of departure, but followed the upwards movement of a three-dimensional spiral:

Evolutive typology

As already indicated, the conception is simplistic. Even in 1818, August Schlegel had already introduced the distinction between synthetic and analytic morphology, which complicates the picture. The actual development of languages cannot be described as the passage from one type to another. Nevertheless, with additional delimitation, some aspects of the model are still useful:

• There are attested historical cases of the passage from isolation to agglutination. The development of aspect marking in Chinese over the past millennia is an example. While their exact status between enclitic free forms and agglutinative suffixes may remain to be settled, they do illustrate the first phase of the evolution.
• Any serious proposal of the origin of human language assumes that at the beginning, there were only lexical roots. Over the millennia, grammaticalization then produces the first grammatical formatives on this basis.
• As already said, some aspects of the history of English illustrate the passage from a flexional to an isolating language. For instance, the entire inflection of adjectives was lost, and from the rich inflection of nouns, only the plural suffix remained.

For the rest, more fine-grained typologies are available today.