Reduplication is a pattern in which a linguistic form is (fully or partially) repeated directly before or after the base form in order to express a modification of its meaning. Most frequently, reduplication expresses intensity and iteration, as in (1)-(2).
Reduplication is notable in the context of pidgin and creole languages, as these have been found to exhibit a great variety of reduplicative phenomena, which are used frequently (e.g. Kouwenberg (ed.) 2003, Bartens 2004). What is not so clear is whether pidgins and creoles show an unusual amount of reduplication when seen in the context of the world’s languages in general, rather than in comparison with the European lexifier languages. Reduplication is widely found in the world’s languages (Moravcsik 1978, Hurch (ed.) 2005ff.), but it is poorly represented in the languages of Europe (Rubino 2005), which linguists often take as an intuitive baseline.
Almost all instances of reduplication found in the APiCS languages are cases of full reduplication, where the entire morpheme or even the entire word is repeated (as in example (1)). Much rarer in our data is partial reduplication, where only part of a morpheme is repeated, most often a syllable (e.g. Papiá Kristang kren-krensa ‘children’, Angolar fo-foga ‘asthma’, from foga ‘suffocate’). Partial reduplication is widely found in the world’s languages (Rubino 2005), but not in APiCS. It is not clear whether this is because the substrate languages tended to lack partial reduplication or because partial reduplication is more difficult to transfer from a substrate.
A special case of partial reduplication is duplifixation (Haspelmath 2002: 24), where a fixed segment sequence is added to the reduplicant, as illustrated by (3).
In the Korlai pattern, only the part of the word that follows the first syllable is reduplicated, while the first syllable is replaced by the fixed element bi-. Sri Lanka Portuguese has a similar pattern, with the fixed element ki-. Such duplifixation patterns are widespread in South Asian languages, where they are often called “echo words” or “echo compounds” (cf. Stolz 2008). The creole patterns have clearly been adopted from the South Asian substrates or adstrates.
Full reduplication is sometimes difficult to distinguish from (re-)iteration (Stolz 2006, Aboh et al. 2012), where entire words or even short phrases are repeated. Such examples are also found in our data (cf. 4-5), but the great majority of our examples are repetitions of simple roots. Only the latter are regarded as reduplication here.
Since the great majority of APiCS languages exhibit reduplication, we focused on the question what functions reduplicative patterns may have. We distinguish between iconic functions, the attenuating function, and the word-class-changing function.
Iconic functions subsume all those functions in which the reduplicated pattern expresses intensity, iteration, plurality, or distributivity. Almost all the examples seen so far are of this type, and we found that this is by far the most common function of reduplication. In fact, there is no APiCS language with reduplication that lacks the iconic function of reduplication. The attenuating function expresses the opposite of intensity: It refers to a low degree of intensity, e.g. Singlish cough-cough ‘cough a little’. Reduplication may also have the purely grammatical function of transposition (word-class-changing), e.g. Cameroon Pidgin English kwik-kwik ‘quickly’ (adverb) from kwik (adjective).
Since all languages with reduplication have the iconic function, there are five different types of languages:
No reduplication | 10 | |
Only iconic functions | 43 | |
Attenuating function | 7 | |
Word-class-changing function | 9 | |
Attenuating and word-class-changing function | 7 | |
Representation: | 76 |
As mentioned earlier, the subfunctions of the iconic function are intensity, iteration, plurality and distributivity. It is very common for gradable words (adjectives such as ‘small’, ‘good’, ‘dark’, or adverbs) to have the intensity sense (cf. 6) when reduplicated, while action words have the iteration or continuation sense (cf. 7). Example (6b) shows that “triplication” is possible as well.
Nouns have a plural sense in a few languages, e.g. Ambon Malay rusa rusa 'deer (pl)', Diu Indo-Portuguese muyɛ-muyɛr ‘women’, Kriol olgaman-olgaman ‘old women’. In Kriol, it is also possible for attributive adjectives to be reduplicated to show plurality of the noun phrase: lilwan-lilwan kokiroj [little-little-cockroach] ‘little cockroaches’.
The distributive sense occurs commonly with numerals (see Chapter 34), e.g. Tok Pisin Ol i kam tupela tupela [they pm come two two] ‘They came two by two’ (Smith & Siegel 2013). But it is also found in other kinds of situations, e.g. with temporal nouns referring to regular occurrence (día día [day day] ‘every day’ in Ternate Chabacano), with adjectives (teya teya 'torn in many places, ragged' in Norf'k, teya < tear), or with verbs, to indicate that the action distributes over the participants:
The attenuating function is mostly found with adjectives:
Nengee | fatu-fatu 'fattish' |
Saramaccan | wɛti-wɛti 'whitish' |
Jamaican | yala-yala 'yellowish' |
Seychelles Creole | vilenn-vilenn 'a little ugly' |
But verbal actions can also be attenuated (cf. also the Singlish example cough-cough mentioned earlier):
And Haitian Creole makes use of reduplication for diminutives, especially in child-directed speech:
The word-class-changing function of reduplication occurs much less systematically, and often seems to be restricted to a small group of lexical items. It never involves a change to a verb, creating only nouns and adjectives in our data. Here are a few examples (in addition to Cameroon Pidgin English kwik-kwik ‘quickly’ mentioned earlier):