Chapter 2: Order of possessor and possessum

Feature information for this chapter can be found in feature 2.

1. Feature description

This map shows the word order within attributive possessive constructions, i.e. the order of the possessor noun phrase with respect to the head noun (or possessum). It is based on WALS feature 86A (Dryer 2011d).

We restrict ourselves to the position of possessive noun phrases containing (full) nouns, rather than those involving only a pronominal word or a proper noun. This is because in some languages pronominal and/or proper noun possessors occur in different positions than nominal ones. Compare Pidgin Hindustani, where pronominal possessors follow the possessum but nominal possessors precede it (Siegel 2013):

wau. (pronominal possessor)
I fetched my hat.
kuta (nominal possessor)
Daya's dog

Whether or not the construction involves other words or affixes on the head or the possessor noun is irrelevant for this feature. For example, in a number of pidgins and creoles a third person possessive determiner is placed between the possessor and the possessum (cf. Chapter 38 “Marking of possessor noun phrases”), as illustrated by the possessor-initial construction of Berbice Dutch:

the old man's eye (Kouwenberg 2013a)

The term possession is used in this context in a broad sense, including of course ownership like the old lady’s dog but also

  1. kinship relations and body-part relations, as in German Maria-s Mutter [Mary-gen mother] ‘Mary’s mother’ and Maria-s Fuß [Mary-gen foot] ‘Mary’s foot’,
  2. the subjective and objective genitive, where the possessor would be the subject or object in a sentential paraphrase, e.g. the teacher’s efforts or the pupil’s detention.

2. The values

There are two possible orders of possessor and possessum:


Value 1. The possessor-possessum order is attested in e.g. African American English:

Sue's house (Green 2013)

Value 2. The possessum-possessor order is illustrated by an example from Cape Verdean Creole of São Vicente:

of= det
the house of a gentleman (Swolkien 2013)

Feature 1 is a multiple-choice feature, and 27 languages in our sample actually have both orders. Compare e.g.

Mr Jessie's good friend's son.
the woman's child(ren)

Note that the corresponding WALS feature 86A “Order of genitive and noun” (Dryer 2011) records only the dominant value and disregards the less frequent alternative (except when a language lacks a dominant order = value 3, as in e.g. Norwegian).

3. Distribution

Numerical. Overall, possessor-final is the somewhat more common order in the APiCS sample. This tendency is a little stronger when one considers only those languages that show just one order (“excl” in the value box above) and also holds – albeit to a lesser extent – when the APiCS values are reduced to match WALS: predominantly possessor-initial 33, predominantly possessor-final 36, no dominant order 8. This is somewhat against the trend among the world’s languages: predominantly possessor-initial 685, predominantly possessor-final 467, no dominant order 96 (Dryer 2011) but it has to be kept in mind that the APiCS sample is not typologically balanced.

By lexifier/substrate. Most African slaves in the Atlantic region originated from areas where Niger-Congo languages were spoken. These languages show a rather well-defined distribution regarding the order of possessor and possessum (cf. Dryer 2011): in the extreme west, the Atlantic subfamily is mostly possessor-final. Further east, and stretching to the western border of Nigeria, other Niger-Congo subfamilies are possessor-initial, while Nigerian languages und the large Bantu genus again show a possessor-final order. Substratal input in the Atlantic contact languages thus often provided a mix of both orders.

While English also has both orders, it is interesting that of the 19 English-lexicon contact languages in the Atlantic nine are exclusively possessor-initial but none is exclusively possessor-final. Also, in those languages that exhibit both orders, possessor-final is never the dominant one, with the sole exception of Early Sranan.

On the other hand, in nine of the ten Portuguese-lexified Atlantic creoles the possessor categorically follows the possessum. The explanation is probably that both Portuguese and the dominant substrates of the Gulf of Guinea and the Cape Verdean Creoles, Bantu and Atlantic languages, are also of this type. Papiamentu is the only Atlantic Ibero-Romance-lexified creole that has both orders. Slaves from the Togo-Benin area (cf. Maurer 2013d) or indeed the Dutch construction (e.g. Jan zijn huis [Jan his house] ‘Jan’s house’) may have contributed the minority possessor-initial order.

The French creoles of the Caribbean and the Americas are exclusively possessor-final, following the lexifier French, although there must have been possessor-initial substratal input.

The Dutch-lexified languages Afrikaans and Negerhollands show both orders, but Berbice Dutch is an exclusive possessor-initial language, which may be due to the strong substratal influence of Ijo, a southern Nigerian possessor-initial language (see Smith 1999: 254ff).

In all the African- and Arabic-lexified contact varieties in central and southern Africa the possessor can only follow the possessum, maybe because this is the dominant order in the main lexifiers and African substrates. On the other hand, Afrikaans allows both orders, just like its lexifier Dutch.

The picture is different when we turn to the Indian Ocean and South Asia: except for Reunion Creole, all French and Portuguese creoles allow the possessor-initial order to varying degrees. In the South Asian varieties Diu Indo-Portuguese, Korlai and Sri Lanka Portuguese, the possessor-initial order can be explained by the fact that Indian languages overwhelmingly prefer this order. This may also be the reason why Sri Lankan Malay does not follow Standard Malay’s possessor-final order (Prentice 1990: 928). The island creoles of Mauritius, Reunion and the Seychelles received an early and strong substratal possessor-final input through slaves from Madagascar and later from the Kenya region. This agreed with the lexifier order and seems to have been the reason why there is a strong preference in these varieties for possessor-final, even though the other order was also imported to Mauritius and Reunion, from West Africa (Mandinka) and India. Where possessor-initial also occurs (Mauritian Creole, Seychelles Creole) it is in the minority and a more recent development, cf. Mauritian Creole, where it “is not attested until about 1880” (Baker & Kriegel 2013a).

In the Pacific, Tok Pisin and the closely related Bislama are strictly possessor-final. For Tok Pisin this could be seen as surprising, since mainland New Guinea is predominantly possessor-initial (Dryer 2011; and Yimas-Arafundi Pidgin also has this order) and Tok Pisin’s lexifier English has both orders. As with Features 1 “Order of subject, object, and verb” and 3 “Order of adjective and noun”, the answer is probably found in Tok Pisin’s strong substratal input from the Bismarck Archipelago, where the majority of languages, including the important substrate Tolai, have a dominant possessor-final order (Dryer 2011). The languages of Vanuatu are also overwhelmingly possessor-final and Bislama unsurprisingly follows this. Note also that Tok Pisin’s and Bislama’s possessive constructions of the house belong father type may have been consolidated by the phrasal correspondence in English (the house belongs to father or the house that belongs to father). The possessor-initial order of Chinese Pidgin English is easier to explain, Cantonese preferring that order. Kriol and Gurindji Kriol in Australia allow both orders, maybe because the Aboriginal languages in the Northern Territories are also rather heterogeneous in this respect. In Hawai‘i Creole, there is a strong preference for possessor-initial, in spite of the fact that the substrates provided a mix of orders: Chinese and Japanese = possessor-initial, Hawaiian and Portuguese = possessor-final. The possessor can both precede and follow the possessum in Norf’k, probably because both English and the contributing language Tahitian (cf. Mühlhäusler 2013a) allow both orders.