Many languages have processes of derivation which provide a stem with a derivational suffix. We here forego prefixal derivation, as in unnecessary. Some derivational suffixes which have a documented history originate in the way symbolized in the following diagram.

In an endocentric compound, the category of the determinatum – C2 in the schema – determines the category of the compound.¹ In the following English examples, C2 is N (noun). When the construction becomes a derived stem, the erstwhile determinatum, now derivational suffix, still determines the category of the result.

Consider the suffix -hood as in brotherhood or even a linguistic neologism such as nounhood. It originates in an Old English noun hād ‘condition, status, rank’. At that stage, it forms compounds like cild-hād ‘child's condition’, which becomes childhood. The noun hād had a very general meaning, so it was widely used to form abstract compound nouns. In this way, a pattern [X-hād]_N ‘X-hood’ arises, which develops into a derivational pattern using -hood as a derivational operator in the formation of abstract nouns.

The case of the English derivational suffix -dom is largely analogous. In Old English, this is the noun dōm ‘state, rank, domain’. It is the determinatum of an endocentric compound with stems of any category functioning as determinans, in words like kingdom or freedom. As the pattern becomes productive, dom becomes a derivational suffix in Middle English.

One of the few derivational processes of Mandarin Chinese is the formation of diminutives of the form [N -ér] ‘little N, N-let’. The suffix, now written 儿, derives from the noun 兒 ér ‘child’. Among other contexts, it appears in diminutives like pén-r (basin-DIM) ‘small basin’ and xiǎogǒu-r (dog-DIM) ‘puppy’ (Wiktionary s.v. 兒).

Such examples show, in the first place, that a compounding pattern may develop into a derivational pattern and, by the same token, a noun which is productively used as the determinatum of a left-branching compounding pattern may develop into a derivational suffix.

The question is how this process is related to grammaticalization. It shares with grammaticalization the reduction of both semantic and phonological weight; but as said before, just any word can undergo these processes. In the course of reduction, the erstwhile determinatum does not merge phonologically with its determinans – later on, its derivational base –, as would typically happen in lexicalization. This is a feature which the development from determinatum of a compound to derivational suffix shares with grammaticalization.

Another feature shared by these two processes is the condition that it is the reduced item which becomes an operator determining the category of the complex. This condition holds in the above cases of transition of determinatum of compound to derivational suffix. It also holds for many inflectional suffixes. The case suffix on a noun determines the syntactic function that the noun can fulfill, e.g. subject or adverbial. Likewise the conjugation suffix determines the syntactic function that the verb can fulfill: if it is a suffix of finite conjugation, the form may be the clause predicate; but if it is a suffix forming participles, the word form may function as an attribute. In this sense, both certain derivational and certain inflectional modifications function as grammatical operators.

A visualization of a process leading from compounding to derivation within the significative system similar to the visualization of lexicalization and grammaticalization would, at any rate, point in the direction of grammaticalization rather than in the direction of lexicalization. Some derivation is hardly distinguishable from inflection. Especially the formation of non-finite verb forms such as participles is treated variously as a form of conjugation or of deverbal derivation; and languages may, in fact, differ in this respect. Such considerations suggest that processes which transfer a compounding pattern into a derivational pattern are processes of grammaticalization.

Finally, if the change of the determinatum of a compound into a derivational suffix is a process of grammaticalization, it should be possible that the process goes on in the same direction, so that a derivational affix is changed into an inflectional suffix. There are at least some examples to show this. English -ing as reviewed in the section on the progressive form goes back to an Old English derivational suffix -unge which forms abstract nouns on verbal bases. In its modern form -ing, it is a conjugation suffix forming a gerund or present participle from every verb. In this case, one is lucky to have the historical evidence. Otherwise we would be faced with the synchronic situation in Modern English, where the derivational suffix is exceedingly productive, forming abstract nouns on nominal bases, too, as in rafting. As noted before, not every actual case of grammaticalization is discoverable on the basis of purely synchronic evidence.

The net result of this discussion is the following: The transition from a lexical stem functioning as the determinatum of an endocentric compound to a derivational suffix is a case of grammaticalization, but not a typical one. It is not typical because its course lies, so to speak, beside the path indicated by the grammaticalization arrow referred to above, as the source is not a syntactic construction, but a lexeme, and the result is still a lexeme, not an inflected form.

¹ There is an unfortunate, but hardly avoidable terminological clash here. The base of an endocentric compound determines the category of the complex. However, in this compounding pattern, the base is traditionally called the determinatum, in a semantic sense, because the other component, the determinans, specifies its meaning.