• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Derivation: inputs and input restrictions

Affixes are specified for the kinds of bases they take. Some affixes take more kinds of bases. An example is the diminutive suffix: although most of its bases are nominal, it may also take adjectives and verbs as inputs, and even intransitive prepositions (used as adverbs). Sometimes diminutive bases can even be non-lexical (Van Marle 1981).

Table 1
Input category Base Derived word
N vrouw woman vrouwtje little woman, sweetheart
A lief dear liefje sweetheart
V dut nap dutje nap
Num tien ten tientje ten-euro note
P/Adv uit out uitje outing
NP twaalf uur twelve o'clock twaalfuurtje wrapped lunch
PP onder ons among each other onderonsje private chat
Det dit en dat this and that ditjes en datjes odds and ends
For many affixes the nature of possible inputs is restricted by phonological, morphological, lexical, syntactic, semantic, or pragmatic conditions.

[+]Phonological restrictions

Affixes may impose requirements on the phonological shape of their bases. For instance, the suffix -aar can only be attached to stems ending in a coronal consonant. In other cases, restrictions are of a prosodic nature. For example, the distribution of the two productive plural suffixes -en and -s can be accounted for by the generalization that -en occurs after a stressed syllable, -s after an unstressed syllable. Alternatively, the condition could be formulated as a restriction on the output rather than on the input: plural nouns should end in a trochee (see here for examples and discussion).

[+]Morphological restrictions

An example of a morphological constraint is that suffixed nouns cannot function as inputs for denominal -ig suffixation (Booij and Van Santen 1998: 56):

Table 2
Base Derived adjective
nuf prim nuffig prim, prissy
Sinterklaas St. Nicolas sinterklazig St. Nicolas-like
vijand enemy vijandig hostile
heldin hero-SUFF heroine *heldinnig
scholier school-SUFF pupil *scholierig
violist violin-SUFF violinist *violistig
Another example of a morphological input restriction is that the diminutive suffix, which is otherwise very productive, cannot be attached to nouns ending in the suffix -e, as in blinde blind person > *blindetje, kampioene female champion > *kampioenetje. Since monomorphemic nouns ending in schwa do allow for diminutive formation (e.g. tantetje aunty), we know that the restriction is morphological rather than phonological. In general, it appears that the suffix -e functions as a closing suffix, barring further derivations.

A more general tendency concerning the morphological complexity of base words is that derivational processes are not recursive, that is, they do not apply to their own outputs. For instance, we do not find adjectives like *groenigig greenishish. An exception is the occasional doubling of a prefix for intensification, as in hyperhypergroot hyper-hyper big. One might reason that the restriction is semantic rather than formal, since it is not clear what kind of meaningful use a word like groenigig greenishish would have. However, the diminutive suffix in Dutch can also be used to express endearment, yet, double diminutives such as *meisjetje meis-je-tje girl-DIM-DIM are impossible. This implies that we do need a formal prohibition on recursivity, at least for suffixes.


As observed by (Krott 1999), there is a general tendency for complex words to be underrepresented as bases of (more) complex words. On the basis of facts concerning Dutch word formation, they conclude that short and highly frequent words are mostly used in word formation. However, this generally reduces the chance of complex words to occur as constituents of complex words, since they are longer and less frequent than simplex words. The availability of external, psycholinguistic explanations makes it difficult to determine whether we really need specific morphological restrictions on base words in the grammar of Dutch. The issue of constraints on affix sequences in Dutch is also discussed in (Booij 2002).

[+]Semantic restrictions

The negative prefix on- usually fails to attach to various sets of adjectives, for instance colour adjectives or adjectives with a lexicalized antonym, such as rijk rich and dom stupid, and adjectives in -loos : *onademloos not breathless, *onzinloos not meaningless. The latter is a subset of the adjectives with a negative meaning, therefore we do not have to assume a specific morphological restriction for on-‘s failure to attach to adjectives ending in -loos. Note, however, that this does not follow from a general semantic incompatibility of the negative meaning of the prefix on- with negative expressions: a phrase like niet onhandig not clumsy, with a combination of two negative elements, is unproblematic. The double negative construction (litotes) even serves to make a normally unacceptable on--adjective usable: onknap unhandsome is normally not used, but does occur with niet, as in Mijn buurman is niet onknap My neighbour is not unhandsome (see more on -onhere). Moreover, each adjective can be turned into its contradictory counterpart by means of the negative prefix niet-, and if it is non-native by means of the prefix non-.

[+]Syntactic restrictions

The classic example of a syntactic restriction on derivation is that the deverbal suffix -baar usually attaches to transitive verbs (there are a few exceptions, such as werkbaar workable, dansbaar danceable, studeerbaar lit. study-able), at least in its productive use (e.g. in a novel form such as skypebaar skypeable). This restriction is obviously directly related to the passive meaning contribution of this suffix, ‘being able to be V-ed’, a meaning that can only be applied to a base verb that is transitive.

[+]Pragmatic restrictions

Pragmatic restrictions pertain to improbable rather than impossible words. Generally, a complex word will not be coined if there is no use for it. For instance, the diminutive of gevaar danger, gevaartje, is not a concept that we will often need, and hence such a word will not be readily coined, though it is a possible diminutive.


(Van Santen 1992) also advocates the distinction between the notions 'possible word' and 'probable word'. The formal and semantic restrictions discussed so far are seem to be absolute restrictions. If they are indeed absolute, they form part of the definition of the notion 'possible complex word'. However, as argued by (Mackenzie 1985), restrictions may also be violable. This means that in that case they define the prototypical cases of a word formation type only. For instance, Mackenzie observed that deverbal ge-nominalization applies preferably to simplex verbs (as in gehuil crying), but that base verbs with a particle or a prefix are not absolutely impossible (for example opgebel making phone-calls < opbellen to phone, geverhuis  moving < verhuizen move). This also means that the words of the different subsets will differ as to the probability of their use.

Read more on:

The meaning of affixes

Non-native morphology

  • Booij, Geert2002Prosodic restrictions on affixation in DutchBooij, Geert & Marle, Jaap van (eds.)Yearbook of Morphology 2001DordrechtKluwer183-202
  • Booij, Geert & Santen, Ariane van1998Morfologie. De woordstructuur van het NederlandsAmsterdamAmsterdam University Press
  • Krott, A., Schreuder, R., and Baayen, R. H1999Complex Words in Complex WordsLinguistics37905-26
  • Mackenzie, Lachlan1985GenominaliseerValentie in Functionele Grammatica. Interdisciplinair Tijdschrift voor Taal- en Tekstwetenschap5177-198
  • Marle, Jaap van1981Over de dynamiek van morfologische categorieënForum der Letteren2251-63
  • Santen, Ariane van1992Produktiviteit in taal en taalgebruikLeidenUniversity of Leiden
printreport errorcite