Professional Documents
Culture Documents
≥
The Expression of Cognitive Categories
ECC 2
Editors
Wolfgang Klein
Stephen Levinson
Mouton de Gruyter
Berlin · New York
The Expression
of Possession
edited by
William B. McGregor
Mouton de Gruyter
Berlin · New York
Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.
앪
앝 Printed on acid-free paper which falls within the guidelines
of the ANSI to ensure permanence and durability.
ISBN 978-3-11-018437-2 hb
ISBN 978-3-11-018438-9 pb
쑔 Copyright 2009 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin.
All rights reserved, including those of translation into foreign languages. No part of this
book may be reproduced in any form or by any means, electronic or mechanical, including
photocopy, recording, or any information storage and retrieval system, without permission
in writing from the publisher.
Cover design: Frank Benno Junghanns, Berlin.
Printed in Germany.
Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
William B. McGregor
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
Index of subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
Index of languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
Index of persons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
Introduction
William B. McGregor
the ‘closest’ and most inherent relations (e.g. body parts, kin), the latter
with less close, less inherent relations (e.g. owned material objects). (See
further Chappell and McGregor 1995.)
To assist the the reader track their way through the volume and to iden-
tify the main issues, we now provide detailed summaries of each of the pa-
pers. It is hoped that these summaries will whet the reader’s apetite for the
detailed discussions and arguments in the individual papers, which of ne-
cessity can be at best hinted at here.
But before we begin, it is necessary to establish some basic notions and
terminology. Throughout the book the term possessum (abbreviated PM) is
used in reference to that which is possessed; it is also sometimes used of
the linguistic expression that denotes this item. Correspondingly, the term
possessor (abbreviated PR) is used in reference to the person, animal, or
whatever, that possesses the PM.
Three primary and general types of possessive construction are usually
distinguished, attributive, predicative, and external. As it is usually used,
the term attributive possession refers to constructions in which the PM and
PR expressions form an NP, as in my dog, the king of France’s bald pate,
and Cliff’s ankle. These constructions are also termed adnominal possession.
By contrast, predicative possession is used of constructions in which the
possessive relation is expressed in a predicate, often by a possessive verb, as
in I have a dog and The king of France has a bald pate. External possession
constructions (EPCs) are constructions in which the possessive relation is
not specified either by the lexical verb or within the NP – the PM and PR
expressions, that is, do not belong to an NP – but rather at the level of a
clausal construction, as in The dog bit Cliff on the ankle (see further Payne
and Barshi 1999). Sometimes the term internal possession construction
(IPC) is used instead of adnominal possession, especially when a contrast is
being drawn with EPCs.
In the first paper, English possessives as reference-point constructions
and their function in the discourse, Peter Willemse, Kristin Davidse and
Liesbet Heyvaert enquire into the information status of the PM referent in
English adnominal possessive constructions with prenominal possessors
(PRs), as in Greta Garbo’s knickers. Based on corpus data, they argue that
the standard analyses of possessives as mere definite NPs (as per e.g. Quirk
et al. 1985; Lyons 1999; Rosenbach 2002), or as NPs presupposing the
identifiability of their referent (Du Bois 1980; Martin 1992) are problematic.
At the same time, Taylor’s (1996) contrary prediction – based on the theory
of possessive NPs as reference-point constructions (as per e.g. Langacker
1995; Taylor 1996) – that PM referents will be overwhelmingly discourse-
Introduction 3
ternal possession in German, although English and other languages are also
mentioned. In contrast to many acquisition studies, this one focusses on the
encoding of possessive relations, rather than merely uses examples of pos-
sessive constructions as instances of morpho-syntactic phenomena to be
acquired. It is based on a corpus of children acquiring German monolin-
gually, which includes data elicited utterances by stimuli designed to elicit
possessive relations; child-directed speech is also included.
It is shown that the target constructions emerge step by step, and that
similarly the range of possessive functions encoded in the constructions in-
creases over time. In the earliest stages, adnominal possession constructions
appear not to exist. In the first stages, possessive relations need not be spo-
ken of at all; when they subsequently are, it may begin with single word ut-
terances that just identify the PR, even if the child is in the two-word stage.
Prepositional possessive constructions involving von ‘of’ emerge later, the
–s genitive construction even later. One of the interesting issues in the ac-
quisition of the genitive –s concerns the acquisition of language specific
constraints: in German, its restriction to unmodified PR nouns. The child
acquiring German appears not to generalize the genitive to nominals with
modifiers; instead, they adopt a strategy such as the omission of a required
modifier (even where otherwise they would use the modifier), or omission
of the marker itself. The developmental stages may overlap: constructions
involving target morphemes obligatory in the adult language may alternate
with constructions lacking them in the child’s speech. Moreover, there are
often lexical restrictions, whereby the morphemes are initially restricted to
particular lexical nouns, and only later generalize.
Less well studied is the acquisition of predicative possession construc-
tions and EPCs, especially the latter, and comparatively little data is avail-
able. This paper thus presents significant new data on the acquisition of
these constructions. Eisenbeiß, Matsuo, and Sonnenstuhl show that HAVE-
constructions emerge prior to BELONG-constructions, and show fewer de-
viations from the adult norm, consistent with the notion that HAVE-
constructions are less marked. The German child language data shows
EPCs only rarely, and quite late. In many circumstances in which adult
speakers prefer a dative EPC, children often employ IPCs. Where they do
produce something different, it often differs from the target dative EPC. In-
terestingly, Japanese children show no tendency to extend the Japanese da-
tive construction to EPCs. How and when the double subject and double
object EPCs of Japanese (e.g. Tsunoda 1995) are acquired is not known.
More generally, research on the acquisition of predicative possession and
EPCs in a wider sample of languages is called for.
6 William B. McGregor
volve verbal elements and retain some clausal characteristics. Thus in some
languages the possessive construction involves a ligature deriving from a
‘have’ verb, while in others the possessive construction involves attach-
ment of a marker cross-referencing the PR to the PM nominal, this marker
being identical or almost identical with the corresponding verbal subject or
object marker. Languages differ, however, in the extent to which the con-
struction is grammaticalised; in some languages the possessive marker is
clearly verbal, in others it is a ‘ligature’. The word order in possessive con-
structions is PR–PM rather than PM–PR as elsewhere in Indonesia and
throughout the Austronesian world; this is also a reflection of the verbal/
clausal nature of the construction. Showing as it does that the distinction
between predicative and attributive possession need not always be system-
atically maintained, van Staden’s paper nicely complements Jan Rijkhoff’s
paper, which, it will be recalled, argues that adnominal attributive posses-
sion in Dutch and English do not constitute a single (emic) construction
type.
In Possessive expressions in the Southwestern Amazon, Hein van der
Voort discusses adnominal possession constructions in eight unrelated lan-
guages of the linguistically diverse southwest Amazon region: three lan-
guage isolates and five languages belonging to different genetic families.
van der Voort identifies two general types of adnominal possessive con-
struction (each of which occur in a range of subtypes): Type I in which a
general possessive marker is added to the PR; and Type II in which the PM
hosts PR agreement or person marking morphemes. In Kwaza and Aikanã
(both isolates), both types of marking are found, though Type I predomi-
nates, and Type II is reduced and used only with third person singular PRs.
Baure is also unusual in that it has a full set of PR marking morphemes, as
well as a general possessive marker that is attached to optionally possessed
nouns (but not to obligatorily possessed or unpossessable nouns), and not to
the PR as in Type I.
The main focus of van der Voort’s paper is on Kwaza. Kwaza is interest-
ing not just because it shows both types of possessive construction, but also
because of the way Type I possession is marked, by the form -dyh. The
analysis of this form is somewhat uncertain. The balance of evidence seems
to indicate that it is synchronically unanalysable. However, it is possible
that it is analysable, at least diachronically, into a possessive marker -dy-
(homophonous with a range of morphemes with related meanings, includ-
ing causative/benefactive) and a nominalising derivational morpheme -h,
which is also used as a neutral classifier (see van der Voort 2004 for dis-
cussion of the classifiers). Interestingly, in possessive constructions -h can
Introduction 9
References
Bally, Charles
1926 / The expression of concepts of the personal domain and indivisibility
1995 in Indo-European languages. In The Grammar of Inalienability: A
Typological Perspective on Body Part Terms and the Part-Whole
Relation, Hilary Chappell and William B. McGregor (eds.), 31–61.
Berlin /New York: Mouton de Gruyter.
Chappell, Hilary and William B. McGregor
1989 Alienability, inalienability and nominal classification. Berkeley Lin-
guistics Society Proceedings 15: 24 –36.
Chappell, Hilary and William B. McGregor
1995 Prolegomena to a theory of inalienability. In The Grammar of Inal-
ienability: A Typological Perspective on Body Part Terms and the
Part-Whole Relation, Hilary Chappell and William B. McGregor
(eds.), 3–30. Berlin /New York: Mouton de Gruyter.
Goldberg, Adele E.
1995 Constructions: A Construction Grammar Approach to Argument
Structure. Chicago and London: University of Chicago Press.
Du Bois, John W.
1980 Beyond definiteness: The trace of identity in discourse. In The Pear
Stories: Cognitive, Cultural and Linguistic Aspects of Narrative Pro-
duction, Wallace L. Chafe (ed.), 203–274. Norwood: Ablex.
Heine, Bernd
1997 Possession: Cognitive Sources, Forces and Grammaticalization.
Cambridge: Cambridge University Press.
Haude, Katharina
2006 A Grammar of Movima. PhD thesis, Radboud Universiteit Nijmegen.
(available on http://webdoc.ubn.ru.nl/mono/h/haude_k/gramofmo.pdf)
Herslund, Michael and Irène Baron
2001 Introduction: Dimensions of possession. In Dimensions of Posses-
sion, Irène Baron and Michael Herslund (eds.), 1–25. Amsterdam:
John Benjamins.
Langacker, Ronald W.
1995 Possession and possessive constructions. In Language and the Cog-
nitive Construal of the World, John R. Taylor and R. E. MacLaury
(eds.), 51–79. Berlin: Mouton de Gruyter.
Lyons, Christopher
1999 Definiteness. Cambridge: Cambridge University Press.
Martin, James R.
1992 English Text: System and Structure. Amsterdam: John Benjamins.
12 William B. McGregor
McGregor, William B.
2001 Non-verbal predicative possession in Nyulnyulan languages. In Forty
Years on: Ken Hale and Australian Languages, Jane Simpson, David
Nash, Mary Laughren, Peter Austin and Barry Alpher (eds.), 337–
352. Canberra: Pacific Linguistics.
Payne, Doris L. and Immanuel Barshi (eds.)
1999 External Possession. Typological Studies in Language, 39. Amster-
dam: John Benjamins.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik
1985 A Comprehensive Grammar of the English Language. London: Long-
man.
Rijkhoff, Jan
2002 The Noun Phrase. Oxford: Oxford University Press.
Rosenbach, Annette
2002 Genitive Variation in English. Berlin /New York: Mouton de Gruyter.
Seiler, Hansjakob
1983 Possession as an Operational Dimension of Language. Tübingen:
Narr.
Taylor, John R.
1989 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford:
Clarendon Press.
Taylor, John R.
1996 Possessives in English. Oxford: Oxford University Press.
Tsunoda, Tasaku
1995 The possession cline in Japanese and other languages. In The Gram-
mar of Inalienability: A Typological Perspective on Body Part Terms
and the Part-Whole Relation, Hilary Chappell and William B.
McGregor, (eds.), 565–630. Berlin /New York: Mouton de Gruyter.
Voort, Hein van der
2004 A Grammar of Kwaza. Mouton Grammar Library 29. Berlin /New
York: Mouton de Gruyter.
English possessives as reference-point constructions
and their function in the discourse
1. Introduction
In this article we will be concerned with the discourse status of the posses-
sum (PM) referent of prenominal 1 possessive NPs such as his car, John’s
car.2 With regard to this issue we find two opposed, indeed paradoxical,
claims in the literature: (a) possessive NPs are definite NPs (e.g. Quirk et
al. 1985; Lyons 1999; Rosenbach 2002) or they presume the identifiability
of the PM referent (e.g. Du Bois 1980; Martin 1992), and (b) possessive
NPs introduce overwhelmingly new, previously unknown, PM referents into
the discourse by linking them to typically given possessor (PR) referents
(Taylor 1996). In this article we will argue, basing ourselves on real usage
data,3 that the analysis of possessive NPs as either mere definite NPs, or as
NPs which typically introduce new referents in the discourse, cannot be
maintained. We will propose that (a) possessive NPs have an identification
mechanism different from that found in NPs with definite articles or de-
monstratives, and (b) the question of the discourse status of PM referents of
possessive NPs cannot be reduced to a binary distinction between new or
given in the discourse. On the basis of a qualitative and quantitative analy-
sis of a corpus of possessive NPs in extensive discourse contexts, we will
argue that many PM referents have a discourse status in between fully
given and fully new. For this range of discourse statuses we will propose a
continuum-like classification.
1
As per Taylor (1996: 2), we use the term prenominal possessives to refer to pos-
sessive NPs in which a genitive or a possessive determiner precedes the head
noun, as opposed to ‘postnominal’ possessives such as a friend of John’s.
2
This excludes NPs with ‘classifying’ or other non-determining genitives like the
lion’s share, a mother’s boy.
3
The data we used were extracted from the COBUILD Bank of English corpus
(examples are marked ‘CB’) and the Lancaster-Oslo-Bergen corpus (examples
are marked ‘LOB’). The data from the COBUILD corpus are reproduced with
the kind permission of HarperCollins Publishers Ltd.
14 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
(3) … when he turns around his hat flies off. (Du Bois 1980: 243)
4
It is not fully clear whether Du Bois uses “presupposed identifiable” in the tech-
nical sense of ‘already available in the discourse context as a singled-out entity’
(Langacker 2002: 33).
16 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
as: the cat she her dinner (1992: 143), the boy his frog (1992:
144). His phoric chains, in other words, track the identifiability status of the
PR, not of the PM referent.
We can now sum up the main elements in the tradition that views pos-
sessive NPs as definite and their referents as identifiable, and formulate our
main criticisms with regard to them. Firstly, linguists adhering to this
analysis of possessive NPs often invoke systematic alternation of posses-
sive NPs with definite NPs as a grammatical argument for according
definiteness to the former. However, there is textual evidence that this al-
ternation is not as systematic as claimed. Possessive NPs in predicative
copular sentences, for instance, do alternate with indefinite NPs, e.g. You’re
my friend – You’re a friend of mine (see also Declerck 1986: 32). Likewise,
NPs with indefinite genitives such as a friend’s friend in (4) can, against
Rosenbach’s (2002: 14) claim, alternate with indefinite NPs, e.g. a friend of
a friend in (5).
Our data contained other examples in which possessive NPs alternate with
indefinite NPs, such as the following:
(7) Goldfinger, the third Bond movie, was released in December of that
year, and with it was founded an industry that would turn Fleming’s
cardboard booby into a product that 30 years later rivals Mickey
Mouse in terms of global penetration. (CB)
5
It should be noted that indefinite grounding genitives are by no means anoma-
lous; see Willemse (2005: 183–200).
20 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
set up not only with reference to the literature, but also by shunting between
the description and the data to ensure optimal coverage of the patterns
emerging from the latter.
The remainder of this article is structured as follows. In section 3, we will
look more closely at functional equivalents of (in)definiteness discussed in
the literature. We will focus in particular on the important distinction be-
tween recoverability (relevant to the reference point analysis of possessive
NPs) and that of first – subsequent mention (relevant to the discourse status
of the PM referent). In section 4, we will present the qualitative and quanti-
tative results of the data analysis. In the first place, this involves setting out
our descriptive classification of the different discourse statuses, ranging
from fully given (coreferential), over a number of statuses in between given
and new, to fully new. Secondly, we also report on the quantitative results
of Willemse’s (2005) corpus study, i.e. on the relative frequency with which
these different statuses occur in the data-base. Finally, we will also situate
our descriptive analysis in a broader discourse perspective: possessive NPs
turn out to serve a variety of specific discourse functions such as reiteration
of the PR as a central discourse participant or reclassification of the PM
referent, which cautions against viewing them only as referential identifica-
tion mechanisms along a strict reference point logic. In section 5, we will
summarize our main findings and point to their implications for the analysis
of possessive NPs as reference point constructions and as so-called ‘defi-
nite’ NPs.
signals that reference is made to the totality of objects or mass in the context
satisfying the description. ‘Inclusive’ reference is treated as secondary to
identifiability by Lyons and as an implicature, rather than a meaning com-
ponent, of definiteness by Hawkins (1978) and Declerck (1986).
Martin (1992) uses the notion of recoverability, rather than identifiabil-
ity, which more clearly suggests that other processes than actual knowledge
of the referent may be involved. According to Martin (1992: 98) “every time
a participant is mentioned, English codes the identity of that participant as
recoverable from the context or not”. When the grammatical resources used
in a specific NP signal that the identity of its referent is in some way recov-
erable, the NP is said to be phoric. Phoric NPs are, consequently, different
types of definite NPs (proper names, pronouns, NPs grounded by the defi-
nite article or by a demonstrative) which all embody “directives indicating
that information is to be retrieved from elsewhere” (Halliday and Hasan
1976: 31). Like Lyons, Martin also stresses that ‘this information’ may per-
tain to the referent itself or to another referent with which it is indirectly
associated. It is in accordance with this logic that Martin (1992: 133), de-
spite recognizing that possessive NPs refer to two referents, codes them
only once for phoricity, as the PM referent is recoverable through the PR
referent. Whilst this is reminiscent of the reference point analysis, it has to
be pointed out that Martin’s recoverability analysis is restricted to function-
ally definite possessive NPs, and does not apply to examples such as (4)
and (6) above. By contrast, the reference point analysis is broader – at least
as a theoretical concept, rather than in Taylor’s (1996) interpretation of it in
relation to possessive NPs. It foregrounds the conceptual relation between
reference point and target and can also be applied to functionally indefinite
possessive NPs.
As recoverability of a referent via association or bridging is not dis-
cussed much in general reference works, we will consider the specialized
literature on it more closely. When a bridging relation holds between two
elements, it may be motivated by one of several types of conceptual rela-
tions, i.e., the ‘source’ of bridging may differ. Generally speaking, the basis
for bridging is some sort of strong associational relationship between enti-
ties, one which is strong enough to allow the identification of one entity on
the basis of an earlier mention of the other, associated entity. This translates
to several more specific relation types, the most important one of which is
the part-whole relation, e.g. (8):
(8) Peter has bought a new car. There is much more room in the boot than
there was in his old car.
22 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
In example (8) the part ‘boot’ can be marked as recoverable on the basis of
its relation with the whole that it forms part of, ‘car’. The processing of
bridging reference thus requires some inferencing on the part of the ad-
dressee, who needs to retrieve the conceptual relation which forms the basis
for the identification of the referent from background knowledge (cf. Ariel
1990: 185), while the role of the immediate discourse context is to provide
an element which activates this inference (a ‘trigger’, Hawkins 1978: 123).
Besides part-whole relations, other relations of strong and habitual asso-
ciation between two entities may be strong enough to motivate bridging, as
in (9).
(9) He was very interested in buying that old house up the road, but the
owner wouldn’t sell.
(10) It was dark and stormy the night the millionaire was murdered. The
killer left no clues for the police to trace. (Brown and Yule 1983: 258)
cream cheese out of a carton and introduces the referent ‘spoon’ with a
definite NP.
(11) and if you just sort of rinse the spoon off afterwards, you don’t really
have to wash dishes […]. (Chafe 1996: 39)
which they are inferable such as the handbrake to the Mitsubishi Starwagon
in (12).
(12) Police prosecutor Snr-Sgt Geoff Jackson told the court Pizzino was
one of eight passengers in the Mitsubishi Starwagon which crashed
on Robina Parkway at the Gold Coast about 5.10 am on Wednesday.
Snr-Sgt Jackson said Pizzino activated the vehicle’s handbrake,
causing the driver to lose control. (CB)
(12’) […] Snr-Sgt Jackson said Pizzino activated the handbrake, causing
the driver to lose control.
Moreover, PM referents may also be ‘more new’ to the discourse than in-
ferable first mentions: they may have some link to a conceptual ‘anchor’ in
the discourse that makes their occurrence not entirely unpredictable, but is
not strong enough to intrinsically convey uniqueness/inclusiveness on them
so as to make them ‘recoverable’. This can be illustrated with possessive
NPs referring to clothes and other alienable possessions, which have a cer-
tain conceptual link to their ‘owners’. If the possessions referred to are con-
textually unique or include all the instances in the given context, then such
NPs are functionally definite. In (13) the detective is wearing only one hat,
and (14) is concerned with all of Paxman’s ties.
(14) Most of Jeremy Paxman’s ties don’t go with his shirts. (CB)
However, if this is not the case, then they are functionally indefinite, as in
example (6) above, Christie’s to sell Garbo’s knickers, and example (15), in
which Ruth has been stealing some, but not all (*the), clothes, jewellery
and accessories of Elizabeth’s. These examples show that the conceptual
relation of alienable possession is not in itself strong enough to make a ref-
erent inferable.
26 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
(16) Point two, Milner was the only person who entered Stevens’s apart-
ment the night of the murder – he was positively identified by the
hallman. Point three, we found Milner’s prints [sic] four places in
the apartment including the library. (CB)
4. Data analysis
We pointed out in section 2 that one of the most important remaining ques-
tions in the literature on possessive NPs pertains to the discourse status of
their PM referent. In section 3, we linked up the notion of discourse status
with the discussion of initial-subsequent mentions, but we differentiated
this contrast into degrees of discourse newness and givenness. Thus, we are
now in a position to tackle the discourse status of PM referents in real us-
age. We compiled a set of 400 instances of possessive NPs, composed as
listed in Table 1.
English possessives as reference-point constructions 27
The central question of our analysis was whether the PM entity is a subse-
quent mention of a discourse referent or a first mention. Seeing the impor-
tance of distinguishing a range of discourse statutes, we propose five cate-
gories which form a continuum from discourse-given (coreferential posses-
sive NP) to discourse-new (possessive NP introducing a new referent).
6
From the genitive sample, a few tokens had to be removed because the PM was a
proper name, e.g. The electro sound, which grew out of Bambaataa’s pioneering
28 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
‘Planet Rock’, kept the street end of rap cooking (CB). As noted in the previous
section, proper names do not construe the contrast between initial – subsequent
mention. Another small set of tokens of the dataset was left out of the classifica-
tion, because the possessive NP in these cases functioned as part of a fixed ex-
pression (often a prepositional phrase), e.g.
(i) If Laurie noticed she chose not to comment, but she was curious in her way.
(CB)
(ii) In a stern reply, he reminded him that ‘Our troops are elated and confident;
those on the enemy’s side cannot but be depressed.’ (CB)
These expressions are not processed compositionally and, as a result, they do not
have a distinct PM referent with a discourse status of its own. For each category,
the number of data left out of the classification was the following: genitive (15);
its (2); their (4); my (4); her (4). The percentages in the tables are calculated on
the basis of the number of tokens included in the classification.
English possessives as reference-point constructions 29
Coreferential possessives account for about 10% of the total data. Still, in a
comprehensive account of the discourse status of possessive NPs, this
makes them a non-negligible category. Let us consider some examples:
‘verdict’ and ‘jury’ is made clear in the preceding context (the jury found
against him in a majority verdict), as well as in the following example:
(19) […] Her salvation was also a cause for celebration among the 60
volunteers and 12 staff who run Britain’s only national charity set up
to find missing people. This Christmas, the runaway was one of
14,000 people on the charity’s computer database, which is housed
in a spartan, donated office above a supermarket in East Sheen,
southwest London. […] A lot of adolescent girls aged around 14 and
15 do not get on with their parents. […] Some fall prey to prostitution
and others, among the most urgent on the charity’s database, become
caught up in paedophile rings. (CB)
In such cases, the reference chain which the possessive NP forms part of
consists of several possessive NPs, in which not only the PR (in this case
‘charity’) is linked up with all its previous mentions, but the PM (in this case
‘(computer) database’) as well (this use of possessive NPs is mentioned by
Barker (2000: 214)).
The coreferential use of possessive NPs has important implications for
the reference-point model in its application to possessive constructions, and
underlines the importance of studying their discourse context, instead of
treating them as isolated syntagms and looking at their internal identifica-
tion mechanism only. In actual usage, the reference-point mechanism can
be adapted to specific discourse needs and used for particular ‘rhetorical’
purposes. Thus, the use of a possessive NP to refer to a given referent ap-
pears to be often motivated by the desire to ensure non-ambiguity of the
referent in contexts where there is potential confusion. For example, in (17),
English possessives as reference-point constructions 31
7
Brown and Yule (1983: 202) give the following example (involving pronominal
reference):
(i) Kill an active, plump chicken. Prepare it for the oven, cut it into four pieces
and roast it with thyme for 1 hour.
In such cases, they argue, although the identity of the referent ‘chicken’ does not
change, its description does: “A reader who simply went back up the endophoric
32 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
(21) With the Bond books, as his friend Ernest L. Cuneo wrote, Fleming’s
‘objective was the making of money’ and he succeeded. But it wasn’t
until after his death in August 1964 that Bondmania erupted and the
money really began to flow. Film was responsible. Goldfinger, the
third Bond movie, was released in December of that year, and with it
was founded an industry that would turn Fleming’s cardboard booby
into a product that 30 years later rivals Mickey Mouse in terms of
global penetration. (CB)
(22) One bae insider said last week that there could be no formal discus-
sions until Daimler and the Dutch government had resolved the
problems of Fokker, the ailing short-haul aircraft maker that will
collapse unless it receives an emergency cash injection of almost £
1.4 billion. […] But Fokker’s crisis is only one contributor to the
problems of Daimler-Benz Aerospace […] (CB)
chain and substituted the expression an active plump chicken for it in the last
clause would, in a significant sense, have failed to understand the text” (Brown
and Yule 1983: 202).
English possessives as reference-point constructions 33
Examples (22) and (23) illustrate a distinction between two kinds of text
reference in relation to possessive NPs. In one type, illustrated by (23), the
possessive NP contains a metatextual noun, that is, the possessive NP
summarizes and categorizes something which has been described in the
preceding discourse as a semiotic phenomenon, i.e. a symbolically proc-
essed phenomenon. The second type, illustrated by (22), summarizes and
categorizes something which has been described in the preceding discourse
as a non-semiotic phenomenon, i.e. as a phenomenon (event, state, activity,
etc.) in reality. Following Takahashi (1997: 63), we will call this the sum-
mative type. Table 4 shows how many of the possessive NPs in our data-
base have text reference, either metatextual or summative.
While the number of tokens for this category is relatively small in our cor-
pus, the cases which occur shed an interesting light on possessive NPs, as
they reveal their potential to categorize complex referents built up over
longer stretches of text. Again, this demonstrates the rhetorical versatility
of possessive NPs in interaction with the surrounding discourse.
34 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
In (23), offer in the possessive NP the government’s offer refers back to the
special advantages and conditions offered by the British government to
banks and building societies willing to grant loans to students, which have
been described in the preceding discourse. In (24), sentiments in the posses-
sive NP Mr Battle’s sentiments refers back to the opinions and ideas of
John Battle which have been represented explicitly through indirect speech
in the preceding discourse. Note that the possessive NP not only refers back
to what has been stated in the text, but also categorizes (parts of) the pre-
ceding text as a text referent. Thus, a possessive NP realizing text reference
not only refers back to a preceding stretch of discourse, but also actively
construes a text referent in the sense that it categorizes and ‘labels’ the text
it refers to. This is also the case in (25), where more inferencing is required
from the addressee than in (23). The decision referred to by the possessive
English possessives as reference-point constructions 35
(26) A little over a month ago the Federal Government announced a $ 17.8
million grant to Indonesia to help combat a HIV/AIDS epidemic
which may infect 2.5 million people by 2000. […] I spent a few
weeks in Thailand in 1989 and am convinced bureaucrats misled the
population into thinking the virus did not infect Asians, largely to
protect their rich international sex-tour industry. If Thailand was
lulling the population and the rather stupid sex tourists into a false
sense of security, why not other Asian countries? So now we have to
cough up $ 17 million to Indonesia alone trying to hold back the
scourge. The irony is that a month after our generous grant was an-
nounced, Indonesian President Suharto’s son, Mr Tommy, bought a
majority stake in Italy’s glamour sports car maker, Lamborghini. […]
What he paid is anyone’s guess, a trifle more than our $ 17 million, I
would think. Sounds mighty like he is in a position to match our gen-
erosity in his country’s fight against AIDS, does it not? (CB)
(27) Among the regions, London heads the list as the centre with the lion’s
share of venture-capital deals; 35 of companies involved in raising
money last year were based in the capital. Scotland showed a high
level of activity with 13 of the deals, closely followed by northwest
England. The northeast, despite its high levels of inward investment,
is not, on these statistics, generating much fresh entrepreneurial ac-
tivity; it accounted for 2 of the deals. The capital’s dominance is
confirmed by the regional breakdown for flotations. (CB)
Table 5 lists the number of cases in which the PM referent is inferable from
a referent or another element in the preceding discourse.
(28) a. As she sprang to her feet and ran to Alistair, the book fell to the
floor, face downwards, its pages doubling up in disorder. (CB)
b. […] the book fell to the floor, face downwards, the pages doubling
up in disorder.
If we want to recognize different degrees of discourse-newness of posses-
sive NPs, PM referents inferable from other entity referents, activities or
larger scenarios evoked by the context, can be viewed as constituting the
first, lowest degree of discourse newness. The possibility of designating
such inferable referents with NPs containing a definite article is evidence of
this relatively low degree of newness, and it can also be used as a formal
indication of which referents to include in this category. This can be illus-
trated with the following example:
8
We are not claiming that there is free variation between these two possible reali-
zations of the inferable referent. Different semantic and pragmatic restrictions
apply to NPs with prenominal possessives and definite determiners, and many
instances of bridging have no possessive alternative and vice versa.
38 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
The PM referent of the possessive NP, ‘moat’, even though strictly speaking
new, has a relation of strong association with the referent ‘castle’ which is
mentioned in the preceding context (a castle is very often surrounded by a
moat). On the basis of this conceptual relation the PM referent ‘moat’ is
immediately recoverable in the sense that it can be inferred from the referent
‘castle’. In fact, in this example, the ‘moat’ could also have been referred to
with a definite NP:
(29’) […] an exhibition of the astronomical work of the 3rd and 4th Earls
and scientific galleries in the moat.
Distance between antecedent and associated entity is in this case clearly not
the problem, as the referent ‘(comic) strip’ is mentioned right before the
associated referent ‘originator’. However, the referent ‘comic strip’ is syn-
tactically embedded in an of-phrase in the preceding clause, which reduces
its accessibility. Consequently, although an NP with definite article is pos-
sible (The originator simply fell for the blonde and blue-eyed charms…),
the speaker opts for a possessive realization of the referent.
However, the data show that a clear motivation for the use of a posses-
sive NP to refer to a referent which could have been realized by a bridging
NP is certainly not always present. Consider, for instance, the following
example:
(31) But even this hardly prepared one for the spectacle that the house
itself presented on closer view. It stood, as it were, knee-deep in
weeds – like some forlorn prehistoric creature in an edible pasture.
Its grey surfaces were flaked and cracked; its woodwork was de-
nuded of paint; many of the lower windows showed tattered curtains
pulled awry, and some of the upper ones lacked entire panes of glass.
[…] If challenged to date it, Appleby would have said 1718; if chal-
lenged to the name of the builder, he would have said James Gibbs.
But now it spoke either of madness – which, indeed, was what was
attributed to its owner – or of penury. (LOB)
The inanimate referent ‘house’ is clearly the topic of this passage. Moreover,
all of the underlined NPs can be analyzed as realizing bridging reference,
since they name parts (surfaces, woodwork, windows) or associated human
referents (builder, owner). However, two of them are anaphorically bridged,
whereas the others are realized by means of a possessive NP. The antecedent
is clear for all of these referents, and no obvious other differences between
e.g. lower windows (referred to by means of bridging) and woodwork (re-
ferred to with a possessive NP) or between builder (bridging) and owner
(possessive) can account for their different realization. It seems, then, to be
a matter of personal choice on the part of the speaker for one construal or
the other in many cases.
40 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
(32) a. She is on sick leave because she has broken her leg .
b. *She is on sick leave because she has broken the leg .
(33) a. During cross-examination the woman said her eyes were not open
all the time, but she was sure she did not fall asleep. (CB)
b. *During cross-examination the woman said the eyes were not open
all the time, but she was sure she did not fall asleep.
(34) The burglar hit John on the head with a baseball bat.
(36) When hounds had run a stag to a standstill, the stag would turn to
defend itself. And few dogs fancied first toss on the antlers, so would
stand baying for the huntsmen to come in for the kill. (CB)
(37) Gut the mackerel, slit down the belly and open them out kipper-style,
or skewer fillets, season and grill. (CB)
possessive NPs, as illustrated by examples (4), (6) and (15) above. We will
refer to this class as ‘anchored’ PM referents: they are conceptually an-
chored to other elements in the discourse – to which, indeed, possessive
NPs ‘anchor’ them via their internal reference point mechanism – and, as a
consequence, cannot be classified as entirely new. At the same time, they
differ from the inferable PM referents. As we saw in section 4.3, these have
a stronger conceptual relation to their antecedents in the discourse, which
inherently guides the person processing the text to unique instances or to
inclusive sets of instances, and excludes functionally indefinite NPs. Spe-
cifically, two main kinds of anchoring relations can be distinguished: on the
one hand, kinship and other interhuman relations and, on the other hand,
alienable possessions. Table 6 gives an overview of the number of cases in
which the PM maintains a relation of ‘anchoring’ with the discourse context.
The first main type of anchoring relation is formed by kinship and similar
interhuman relations. They are evoked by certain nouns with an inherently
relational meaning (see Barker 1991 [1995], 2000; Fraurud 1996). Taylor
(1989: 675) gives a list of nouns which in addition to kinship nouns “in-
voke, in their semantic structure, various other kinds of interpersonal rela-
tionships”, including friend, fiancée, colleague, guest, fellow student, com-
petitor, confidant, etc. Taylor (1989) discusses such relational nouns in
terms of how they ‘steer’ the interpretation of the relationship between PR
and PM by evoking an unprofiled relationship in their semantic structure,
an element of which is elaborated by the PR nominal.
Possessive NPs whose PM is a relative, friend or associate can be func-
tionally indefinite, as illustrated by (4), a friend’s friend, described by let-
ter, should turn out an unrecognizable stranger, above, (38) and (39).
English possessives as reference-point constructions 43
(38) Created in 1968 and based on a true story, Ashton depicts Elgar’s
friends visiting him after he had sent his score of his Variations to the
famous Viennese composer Ritcher, to try and interest him in the
work. (CB)
(39) Dwight wrote Victor Serge that subscriptions were pouring in and
“all kinds of people were being stimulated to write for it […]”. […]
Dwight’s associates and contributors wrote long and constructive
criticisms. (CB)
(40) In the months leading up to the killing, Avent’s wife became pregnant
and he lost his job as a supermarket shelf stacker because he stole
$300. (CB)
(42) <ZF1> H h <ZF0> how do you get to the Fab Club? <M01> Well
FX c erm <ZF1> the d <ZF0> the daughter comes and picks us up
or we get taxis.9 (CB)
9
The codes in this example are part of the transcription of spoken language used
in the COBUILD corpus. <F01> and <M01> indicate the two participants in the
dialogue; <ZF1> and <ZF0> indicate a repetition of (parts of) words, and ‘FX’
and ‘MX’ replace, for privacy reasons, female and male proper names mentioned
in the dialogue.
44 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
(43) When a man loses his job, it deals a sickening blow to his ego and
sense of identity. Feelings of failure and rejection strain family rela-
tionships to breaking point and it’s often the wife who’s left to pick up
the pieces. (CB)
(44) No one checked up on us. At 14, I was cutting school in the lunch
hour and getting drunk in the afternoons usually at a friend’s house.
(CB)
(45) She smiles with some regret when the fans rip Brett’s shirts to shreds
during The Drowners. He’s lost so many beautiful clothes; […] (CB)
By contrast, (46) and (47) are functionally definite, as they refer to the one
car driven by Rachelle in (46), and to all contextually given instances of
garments of Christ in (47).
(46) […] “I saw Rachelle’s car slow in the final and figured she must have
struck problems,” Cowin said. (CB)
The quantitative results for the category of new PM referents in our data are
given in Table 7.
English possessives as reference-point constructions 45
These results show that although it is true that possessive NPs newly intro-
duce referents into the discourse in a number of cases (28% of our corpus),
new referents certainly do not represent the majority of PM referents.
This last category forms the ‘new’ end of the continuum of discourse
statuses that we propose. It contains cases in which the PM is a new refer-
ent in the discourse, not only in the sense of not having been mentioned as
such in the preceding discourse, but also in the sense of not being inferable
or predictable from elements in the preceding discourse. In many of these
cases, the PM is realized by a nominalization or deverbal nouns, as in (48)
or (49) below.
In these examples, the PR nominal designates the subject of the action de-
scribed by the nominalization. While the PR referent typically maintains an
anaphoric relation with the preceding discourse, the action designated by
the ‘PM noun’ is not inferable from the preceding text. In some of these
examples, there is some sort of general semantic relation between the PM
referent and the preceding discourse context, but this relation is not strong
enough to confer (even partial) givenness on the referent. This is for in-
stance the case in example (49), in which the PM referent fits in with the
scenario that is being evoked in the context, without being a predictable
part of this scenario. That no ‘givenness’ of the PM referent is present in
46 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
(49) Queensland cricket captain Stuart Law yesterday warmed up for this
week’s Sheffield Shield battle against New South Wales with a mighty
double century for Valley in a XXXX Brisbane club cricket game yes-
terday. […] In a punishing display, Law smacked 32 boundaries in-
cluding two sixes, and faced 266 balls. He came to the wicket when
the score was eight after opener Michael Ephraims departed, and im-
mediately went on the attack. […] There was little respite for Univer-
sity’s bowling attack which included Sheffield Shield bowlers Michael
Kasprowicz and Peter Jackson. (CB)
5. Conclusion
In this paper we set out to investigate the underresearched issue of the dis-
course status of the PM referent of possessive NPs in English. We investi-
gated the ways in which this referent can be an initial or non-initial men-
tion, paying attention to how it interacts with previously given information
in extensive discourse contexts. We demonstrated that a binary distinction
between given and new does not suffice to capture its discourse status, and
proposed instead a continuum of discourse statuses, ranging from fully dis-
course-given to fully discourse-new over a number of in-between statuses
in which the PM is only partially given. Our analysis has the following ma-
jor implications.
Firstly, our analysis reviews and refines the description of English pos-
sessive NPs in a number of ways. We have shown that possessive NPs can-
not be analyzed as mere definite NPs which presuppose the identifiability
English possessives as reference-point constructions 47
Acknowledgements
References
Abbott, Barbara
2004 Definiteness and indefiniteness. In The Handbook of Pragmatics,
Laurence R. Horn and Gregory Ward (eds.), 122–149. Oxford: Black-
well.
Ariel, Mira
1990 Accessing Noun-Phrase Antecedents. London: Routledge.
Barker, Chris
1991 Possessive descriptions. PhD thesis, University of California Santa
Cruz [published 1995, Stanford: CSLI].
Biber, Douglas, Stig Johansson, Geoffrey Leech and Randolph Quirk
1999 Longman Grammar of Spoken and Written English. London: Long-
man.
Blanche-Benveniste, Claire and André Chervel
1966 Recherches sur le syntagme substantif. Cahiers de Lexicologie IX (2):
3–33.
Brown, Gillian and George Yule
1983 Discourse Analysis. Cambridge: Cambridge University Press.
Chafe, Wallace L.
1972 Discourse structure and human knowledge. In Language Compre-
hension and the Acquisition of Knowledge, Roy O. Freedle and John
B. Carroll (eds.), 41– 69. Washington: Winston.
Chafe, Wallace L.
1976 Givenness, contrastiveness, definiteness, subjects and topics. In Sub-
ject and Topic, Charles N. Li (ed.), 25–55. New York: Academic Press.
English possessives as reference-point constructions 49
Chafe, Wallace L.
1994 Discourse, Consciousness, and Time. Chicago: The University of
Chicago Press.
Chafe, Wallace L.
1996 Inferring identifiability and accessibility. In Reference and Referent
Accessibility, Thorstein Fretheim and Jeannette K. Gundel (eds.),
37–46. Amsterdam: John Benjamins.
Chappell, Hillary and William B. McGregor
1996 The Grammar of Inalienability: A Typological Perspective on Body
Part Terms and the Part-Whole Relation. Berlin /New York: Mouton
de Gruyter.
Declerck, Renaat
1986 Two notes on the theory of definiteness. Journal of Linguistics 22:
25–39.
Du Bois, John W.
1980 Beyond definiteness: the trace of identity in discourse. In The Pear
Stories: Cognitive, Cultural and Linguistic Aspects of Narrative
Production, Wallace L. Chafe (ed.), 203–274. Norwood: Ablex.
Emmott, Catherine
1997 Narrative Comprehension: A Discourse Perspective. Oxford: Oxford
University Press.
Fraurud, Kari
1990 Definiteness and the processing of noun phrases in natural discourse.
Journal of Semantics 7: 395–433.
Fraurud, Kari
1996 Cognitive ontology and NP form. In Reference and Referent Acces-
sibility, Thorstein Fretheim and Jeannette K. Gundel (eds.), 65–87.
Amsterdam: John Benjamins.
Gundel, Jeannette K., Nancy Hedberg and Ron Zacharski
1993 Cognitive status and the form of referring expressions in discourse.
Language 69: 274–307.
Gundel, Jeannette K., Nancy Hedberg and Ron Zacharski
2001 Cognitive status and definite descriptions in English: why accommo-
dation is unnecessary. Journal of English Language and Linguistics
5: 273–295.
Halliday, Michael A.K. and Ruqaiya Hasan
1976 Cohesion in English. London: Longman.
Hawkins, John A.
1978 Definiteness and Indefiniteness: A Study in Reference and Gram-
maticality Prediction. London: Croom Helm.
Huddleston, Rodney
1988 English Grammar: An Outline. Cambridge: Cambridge University
Press.
50 Peter Willemse, Kristin Davidse and Liesbet Heyvaert
Langacker, Ronald W.
1991 Foundations of Cognitive Grammar, Volume 2: Descriptive Applica-
tion. Stanford: Stanford University Press.
Langacker, Ronald W.
1993 Reference-point constructions. Cognitive Linguistics 4: 1–38.
Langacker, Ronald W.
1995 Possession and possessive constructions. In Language and the Cog-
nitive Construal of the World, John R. Taylor and R. E. MacLaury
(eds.), 51–79. Berlin /New York: Mouton de Gruyter.
Langacker, Ronald W.
2002 Remarks on the English grounding systems. In Grounding: The Epis-
temic Footing of Deixis and Reference, Frank Brisard (ed.), 29–38.
Berlin /New York: Mouton de Gruyter.
Lyons, Christopher
1999 Definiteness. Cambridge: Cambridge University Press.
Martin, James R.
1992 English Text: System and Structure. Amsterdam: John Benjamins.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik
1985 A Comprehensive Grammar of the English Language. London: Long-
man.
Rosenbach, Annette
2002 Genitive Variation in English. Berlin /New York: Mouton de Gruyter.
Sanford, Anthony J. and Simon C. Garrod
1981 Understanding Written Language: Explorations of Comprehension
beyond the Sentence. Chichester: John Wiley and Sons.
Sanford, Anthony J. and Simon C. Garrod
1998 The role of scenario mapping in text comprehension. Discourse
Processes 26 (2 /3): 159–190.
Takahashi, Hidemitsu
1997 Indirect anaphors: definiteness and inference. Leuvense Bijdragen-
Leuven Contributions in Linguistics and Philology 86: 53–80.
Taylor, John R.
1989 Possessive genitives in English. Linguistics 27: 663–686.
Taylor, John R.
1991 Possessive genitives in English: a discourse perspective. South Afri-
can Journal of Linguistics 9 (3): 59–63.
Taylor, John R.
1996 Possessives in English. Oxford: Oxford University Press.
Willemse, Peter
2005 Nominal reference-point constructions: possessive and esphoric NPs
in English. PhD thesis, University of Leuven.
On the co-variation between form and function
of adnominal possessive modifiers in Dutch and
English
Jan Rijkhoff
1. Introduction
This contribution is concerned with Dutch and (to lesser extent) English
possessive modifiers introduced by the preposition of (Dutch van), as in a
woman OF INFLUENCE or (Dutch) de auto VAN MIJN BROER (the car OF MY
BROTHER) ‘my brother’s car’. The main goal of this paper is to demonstrate
that the remarkable variation in the grammatical properties of this posses-
sive construction directly correlates with the kind of modifier function it
has in the noun phrase.
It is first shown that lexical possessive modifiers with van /‘of’ (“ad-
nominal possessives” for short) are used to express most of the modifier
functions recognized in a semantic, five-layered model of the noun phrase
(section 2). I will then argue that the values for certain grammatical pa-
rameters (here subsumed under the labels MODIFICATION, PREDICATION,
REFERENCE) are determined by the kind of modifier function the adnominal
possessive has in the noun phrase (section 3 and 4); a tentative explanation
is given in section 5.
The more general point this paper wants to make is that functional modi-
fier categories like CLASSIFYING MODIFIER, QUALIFYING MODIFIER, and
LOCALIZING MODIFIER can be characterized in grammatical terms, which
makes it possible to capture grammatical differences between members of
the same form class (such as prepositional phrases, as shown in Table 3)
and grammatical similarities between members of different form classes
(e.g. adjectives and prepositional phrases, see Tables 1a–b) within and
across languages. This paper is restricted to possessive modifiers of nouns
that denote concrete objects.
52 Jan Rijkhoff
1
See Rijkhoff (2002) for details; the most recent accounts are Rijkhoff (2008a, b, d).
On the co-variation between form and function of adnominal possessives 53
Scope increase
ability, possibility or desirability of
the actual occurrence of event ei.
4. Discourse-Ref. modifiers (4, 4) 4. Discourse-Ref. modifiers (4, 4)
specify the existential status of thing xi or event ei in the World of Discourse.
Figure 1. Layers of modification in the noun phrase and in the clause (Discourse-
Ref. = Discourse-Referential; Greek characters symbolize grammatical
[ω/ , /] and lexical [τ /, σ /] modifiers in the NP and the clause).
54 Jan Rijkhoff
(1) Could you pass me the book that is lying next to you?
2
Synchronic and diachronic relationships between modifiers of the clause and the
NP are discussed in Rijkhoff and Seibt (2005) and Rijkhoff (2008a, d).
56 Jan Rijkhoff
Samoan
(2) Sa fau=siae e Tagaloaalagi fale e tolu …
PAST build=ES ERG Tagaloaalagi house GENR three …
‘Tagaloaalagi built three houses …’
(Mosel and Hovdhaugen 1992: 318)
Galela
(3) awi dohu i lalamo
his foot it be.big.PRT
‘his big foot’ (van Baarda 1908: 35)
denoting a concrete physical object (such as car, dog or tree). Here are
some examples (all non-asterisked examples were found on the Internet
using Google or Ask):3
3
Rosenbach (2006) is a recent study of genitives in English.
4
The same text also contains an example without a definite article: Julius Stinde
is niet alleen een man van geest, maar ook een man van het ware midden. Hij
weet maat te houden. ‘Julius Stinde is not just a man of mind/spirit, but also a
man of the true middle. He knows how to contain himself’.
58 Jan Rijkhoff
‘three year old children’). As a matter of fact, I am not aware of any lan-
guage that uses adnominal possessives to express the cardinality of the ref-
erent of the NP (Rijkhoff 2002: chapter 5).5 What we find instead are in-
stances of so-called dependency reversal (Malchukov 2000): the numeral
word is formally expressed as the head of the construction with the noun
denoting the counted entity displaying features of a possessive dependent.
We will return to this gap in the functions of the adnominal possessive in
section 3.2 below.
Dutch does not use lexical possessives as DISCOURSE-REFERENTIAL
SATELLITES (T4) either, but notice that in some languages a (bound) posses-
sive pronoun has been claimed to mark definiteness. If this is indeed the
case, the (erstwhile) possessive pronoun would serve as a grammatical DIS-
COURSE-REFERENTIAL GRAMMATICAL MODIFIER (cf. Siewierska et al. 1998:
811–812, notes 14 and 33). For example, Comrie (1988: 465), referring to
Tauli (1966: 148), writes that some Uralic languages “use the third person
singular possessive suffix as a general marker of definiteness” (see also
Englebretson 2003 on Indonesian –nya and Guillemin 2007 on Mauritian
Creole so). Similarly, in his monograph on the Turkic languages Menges
(1968: 113) states that “the possessive significance of the suffix of the 3rd
person can completely recede when it defines or determines a noun”, in
which case the suffix conveys “a determinative idea as expressed by the
article in Indo-European or Semitic languages”.
To what extent the possessive affix in these languages has actually gram-
maticalized into a bound marker of definiteness is still a matter of debate,
as it seems that the change in meaning has not (yet) clearly resulted in a
category switch (from suffixed possessive pronoun to suffixed definite arti-
cle; Fraurud 2001). Furthermore, in Turkish the possessive suffix does not
seem to occur on nouns.
5
Cardinality is indicated by an adnominal possessive in e.g. een group van 20
(mensen) ‘a group of 20 (persons)’ but this seems only possible when the phrase
is headed by a collective noun like ‘group’ (i.e. not a noun denoting an individ-
ual).
60 Jan Rijkhoff
Turkish
(22) Bun-u iste-mi-yor-um, baka-sın-ı ver
this-ACC want-NEG-PRES-1S other-POS.3SG-ACC give
‘I don’t want this (one), give me the other one’
(Gerjan van Schaaik, pers. comm.)
Furthermore I have only taken into account instances in which the adnomi-
nal possessive modifier is marked by a special form (such as the -s suffix or
the preposition van ‘of’) or is a member of the special set of possessive
pronouns), thus ignoring possessives that are unmarked (as in the Nasioi ex-
ample below) or not formally distinguished from other modifier categories.
Nasioi
(23) Máteasi bauran
Mateasi daughter
‘Mateasi’s daughter’ (Rausch 1912: 120)
6
On gradience, see e.g. Aarts (2004a/b, 2006), Rijkhoff (2008c), Sorace and
Keller (2005).
62 Jan Rijkhoff
7
See Hengeveld (1989: 142) on the diachronic development of clausal operators
and Song (2005) for a recent discussion of scope increase and scope decrease in
grammaticalization.
On the co-variation between form and function of adnominal possessives 63
fying satellites (section 3.2.) in that they do not specify some more or less
objective property of an entity (round table) or the speaker’s subjective atti-
tude towards the entity (beautiful picture), but rather a particular subclass of
the category denoted by the head noun. The semantic range of classifying
satellites is rather broad and may include such categories as material, pur-
pose and function, status and rank, origin, and mode of operation (Halliday
2004: 320). Essentially they relate to any feature that may serve to classify
entities into a system of smaller sets. In English classifying satellites may
take the form of an adjective (25a), but also of a possessive construction
(25b–c):
8
Actually a boy’s shirt is potentially ambiguous, meaning either ‘the shirt of an
unidentified boy’ (non-classifying) or ‘a particular kind of shirt’ (classifying), see
on this difference e.g. Taylor (1996: 665) and Willemse (2005); see also section
3.4.
9
Cf. Quirk et al. (1985: 1339); on non-predicative adjectives, see also e.g. Farsi
(1968) and Levi (1973).
10
One does find e.g. quick divorce lawyer, where quick modifies divorce, but in
such cases we seem to be dealing with a fixed expression (for example, divorce
cannot be modified by slow or short).
64 Jan Rijkhoff
woman’s in a woman’s hat does not refer to any particular woman but
merely serves to specify for the hearer that the speaker is talking about the
kind of hat that is worn by women. Furthermore, as in the case of classifying
adjectives, classifying possessives such as woman’s in (29a) and (30a) can-
not be modified (29b) or separated (30b) from the head noun by other modi-
fiers, at least not without changing its function:11
CLASSIFYING POSSESSIVE
(31) a. a woman’s hat b. *that hat is a woman’s12
CLASSIFYING POSSESSIVE
(32) a. a man of prayer b. * this man is of prayer
(33) a. De priester is dus een man God-s, een man van gebed, een man
the priest is thus a man God-GEN, a man of prayer a man
van de Kerk, een herder voor de mensen.
of the Church, a shepherd for the people
‘so the priest is a man of God, a man of prayer, a man of the
Church, a shepherd for the people’
b. *deze man is van gebed 13
this man is of prayer
11
Gunkel and Zifonun (2008) is a detailed study of one particular subcategory of
classifying satellites (‘relational adjectives’) in English, German and French.
12
Notice that it is possible to have a possessive predicate in e.g. That hat is Mary’s
(That hat belongs to Mary).
13
The same goes for een man Gods ‘a man of God’ and a man van de Kerk ‘a man
of the Church’.
On the co-variation between form and function of adnominal possessives 65
(34) a. Een mooi citaat is een diamant aan de vinger van een man
a nice quotation is a diamond on the finger of a man
van geest en een kei in de hand van een dwaas
of mind/spirit and a rock in the hand of a fool
‘a nice quotation is a diamond on the finger of a wise man and a
rock in the hand of a fool’
b. *die man is van geest
that man is of mind/spirit
Classifying possessives can also take the form of a case marked modifier,
as in these examples from Swedish and Lithuanian:
Swedish
(36) en folk-et-s teater
a:C people-DEF.C-GEN theatre
‘a theatre for the people’ (Koptjevskaja-Tamm 2003a: 539–540)
Lithuanian
(37) duon-os peilis
bread-GEN knife
‘a bread knife’ (Koptjevskaja-Tamm 2002: 155)
Roviana
(38) a. mamalaengi barikaleqe
voice woman
‘a woman’s voice/a female voice’ (classifying/non-referential)
b. mamalaengi-na [barikaleqe hoi]
voice-3SG.POSS [woman that]
‘that woman’s voice’ (referential/non-classifying)
(Corston-Oliver 2002)
66 Jan Rijkhoff
It turns out that prepositional phrases with van/‘of’ behave differently with
respect to the following three grammatical tests, depending on the kind of
modifier function they perform in the NP:
The data presented above indicate that classifying possessives have negative
scores on all counts:
CLASSIFYING — — —
Een meisje *een meisje *het meisje is *het meisje
van plezier van veel plezier van plezier van zijn plezier
‘a girl of pleasure’ ‘a girl of much ‘the girl is of ‘the girl of his
pleasure’ pleasure’ pleasure’
14
Apart from the close connection between compounding and classification, there
is also a relation between inalienability and classification (as in a mountain top
vs. the top of a mountain; for discussion see e.g. Chappell and McGregor 1989,
1996).
On the co-variation between form and function of adnominal possessives 67
CLASSIFYING — —
een medisch *een erg medisch *het onderzoek is
onderzoek onderzoek medisch (not applicable)
‘a medical ‘a very medical ‘the investigation
investigation’ investigation’ is medical’
Hausa
(39) mutum mài alheri / arzaki / hankali
person having kindness / prosperity / intelligence
‘a kind/prosperous/intelligent person’ 15 (Schachter 1985: 15)
Here are some adnominal possessives with van ‘of’ exemplifying some of
the major adjectival categories distinguished above.
DIMENSION
(40) kabeljauwen van enorme grootte
cod:fish(PL) of enormous size
‘cod fish of enormous size’
(41) een oudere man van geringe lengte
an older man of short length
‘an elderly man of short stature’
15
mài (also: màasú) is glossed as ‘owner, possessor of’ in Newman (1987: 721).
68 Jan Rijkhoff
PHYSICAL PROPERTY
(44) Dymfna was … een christelijke moeder van grote schoonheid
Dymfna was … a Christian mother of great beauty
‘Dymfna was … a Christian mother of great beauty’
COLOR16
(45) een Ferrari F40 van rode kleur
a Ferrari F40 of red color
‘a red Ferrari F40’
HUMAN PROPENSITY
(46) Hij was een man van grote charme17
he was a man of great charm
‘He was a very charming man’
AGE
(47) mensen van middelbare leeftijd
people of medium age
‘middle-aged persons’
16
In English the color term can occur by itself, i.e. without the abstract noun kleur
‘color’ (as in a Jovian moon of incredible redness, from the book Cosmos by
Carl Sagan), but in Dutch this seem to be restricted to NPs headed by nouns
denoting non-spatial entities: het effect van rood ‘the effect of red’, het karakter
van rood ‘the character of red’, de richting van de snelheid ‘the direction of the
speed’, de effecten van snelheid ‘the effects of velocity’.
17
There is a wide variety of qualities that can be expressed this way (here all with
the adjective grote ‘great’): een man van grote spontaniteit en een ontwapenende
openhartigheid ‘a man of great spontaneity and disarming sincerity’, een man
van grote gaven ‘a man of great gifts’, een man van grote wijsheid, geduld en
begrip ‘a man of great wisdom, patience and understanding’, een man van
grote kracht ‘a man of great power’, een man van grote, bombastische theo-
rieën ‘a man of big, bombastic theories’.
On the co-variation between form and function of adnominal possessives 69
VALUE
(49) een parel van grote waarde
a pearl of great value
‘a very valuable pearl’
(50) een huis van 4,5 ton of meer
a house of 4.5 TON or more (a ‘ton’ = 100.000 euros)
‘a house that is worth 450.000 euros or more’
SPEED
(51) The Boots of Great Speed (title of a story by R. P. Barnett)
Thus it appears that the core adjectival categories are also fairly well covered
by adnominal possessives, but the Internet search with Google and Ask also
seems to point to certain gaps, at least in Dutch and English. For example,
it turned out to be extremely difficult to find examples of adnominal pos-
sessives expressing notions concerning certain physical properties (e.g. of
softness, of smoothness), speed (e.g. of slowness/of slow speed), or taste
(e.g. of sweetness/of sweet taste). Such properties tend rather to be ex-
pressed with the entity denoting noun in the dependent (possessive) con-
struction as in the seductive softness of the Sheridan Ultrasoft towel, the
redness of berries, the slowness of the Concorde or the sweetness of honey
(see also below on Dependency Reversal) or in combination with a noun
denoting a higher order entity (de perceptie van bittere smaak ‘the percep-
tion of bitter taste’, de bron van bittere smaak ‘the source/cause of bitter
taste’ – see also note 16). Although one could construe more or less accept-
able Dutch NPs like een appel van bittere smaak ‘an apple of bitter taste’,
such adnominal possessives do not seem to occur in the actual (written)
language.18 The precise reason for the scarcity of certain adnominal posses-
18
One can say een man van goede smaak ‘a man of good taste’, but here smaak
‘taste’ is used metaphorically, i.e. it does not mean that the man tastes good but
rather that he HAS good taste. The fact that adnominal possessives are not used
to express the notion of speed is probably at least partly due to the fact that
SPEED is not an inherent property but rather a potential property of certain mov-
able objects, like cars and trains.
70 Jan Rijkhoff
sives remains obscure, but there is little doubt that the observed gaps are at
least partly due to the fact that the current study is limited to properties of
concrete objects, since we saw above that one can easily find examples out-
side this restricted ontological domain: tekenen van roodheid ‘signs of red-
ness’, de weg van zachtheid ‘the road of softness’, oorzaken van zuurheid
‘causes of sourness’ (cf. also English the benefits of softness, a touch of
softness, areas of softness, the concept of softness, a feeling of softness; a
touch of green, shades of green; the Well of Sweetness, acts of sweetness,
scoops of sweetness, the perils of sweetness; the taste of sweetness; the dis-
covery of slowness, in praise of slowness, a philosophy of Slowness, etc.).
19
This sentence is acceptable in Dutch if there is a qualifier such as totaal ‘totally’
or helemaal ‘completely’, but this results in an idiomatic expression with the
meaning ‘to be strange, confused’ or even ‘unconscious’.
On the co-variation between form and function of adnominal possessives 71
20
In some cases (e.g. with possessives of age or measure, i.e. height, length,
weight) the predicative variant is only possible without the possessive marker
in Dutch: een muur van twee meter hoog (lit. ‘a wall of two meter high’) vs. die
muur is twee meter hoog (lit. ‘that wall is two meter high’); see also note 22.
21
I will say more on the classifying flavor of some of these examples in section 4.1.
72 Jan Rijkhoff
Apart from possessives like van grote kwaliteit ‘of high quality’, this group
seems to be confined to possessives specifying some measure (of age, size,
length, width, height, depth, volume, weight etc.).22
Note, however, that the predicative variant is not possible when the meas-
ure is given a specific value. In such cases, a non-possessive predicate is
used (headed by an adjective rather than a noun), as in (62c).
22
In Dutch, there is also the variant with an adjectival form (instead of an abstract
noun), as in een boot van zeventig meter langA (a boat of seventy meter long) ‘a
boat of seventy meters’; see also example (62c).
On the co-variation between form and function of adnominal possessives 73
(66) Ik vind het dus veel geld voor een lens van zulke kwaliteit
I find it thus much money for a lens of such quality
‘So I think it is a lot of money for a lens of such quality’
(67) een van de laatste Attische grafmonumenten van deze hoge kwaliteit
one of the last Attic burial monuments of this high quality
‘one of the last Attic burial monuments of this high quality’
ADNOMINAL
POSSESSIVES MODIFICATION PREDICATION REFERENCE
QUALIFYING B+
(modifier compulsory)
beelden van beelden van de beelden zijn beelden van
grote kwaliteit grote kwaliteit van grote kwaliteit deze kwaliteit
‘statues of great ‘statues of high ‘the statues are of ‘statues of this
quality’ quality’ high quality’ quality’
(REFERENCE TO
A PROPERTY)
QUALIFYING B
een kroon van goud een kroon van de kroon is van —
‘a crown of gold’ (zuiver) goud (zuiver) goud
‘a crown of the crown is of
(pure) gold’ (pure) gold
‘the crown is made
of (pure) gold’
QUALIFYING A+
(modifier een man van — —
compulsory) vele g ezichten
een man van ‘a man of many
vele gezichten faces’
‘a man of many
faces’
QUALIFYING A
een man van gezag een man van — —
‘a man of authority’ (groot) gezag
‘a man of (great)
authority’
CLASSIFYING
een man van de — — —
wereld
‘a man of the world’
On the co-variation between form and function of adnominal possessives 75
Swedish
(69) a. en satan-s kärring
a:C Satan-GEN crone
‘a damned crone’
b. ett herra-n-s oväsen
a:N lord-DEF.C-GEN noise
‘a hell of a noise’
In some languages, such as English, we see that the same property (e.g. ‘(to
be) rich, richness’) can be expressed in the form of an adjective or in the
form of a possessive construction (Quirk et al. 1987: 1278, 1286):
Now compare: 23
23
“A Man of Riches Built His Team From a Trust Fund” is the title of an article by
Richard Justice about Jack Kent Cooke (The Washington Post, April 7, 1997).
76 Jan Rijkhoff
preposition of appears between the head noun (man) and the abstract noun
(riches). More specifically, whereas rich seems to refer to an actual prop-
erty, of riches seems to specify a property of a man for whom being rich is
a more characteristic or permanent feature, which may come with a set of
other features (the money was acquired in a proper manner, the man has
class, etc.). One could even say that the NP containing the possessive of
riches has a distinct classificatory flavor in that it refers to a certain kind of
man (as opposed to rich in a rich man). This is confirmed by the way NPs
that contain a qualifying adnominal possessive are used in actual speech
(see section 4.1. below).
Both in English and in Dutch we also find instances of adnominal pos-
sessives (sometimes referred to as METAPHORICAL CONSTRUCTIONS in
Dutch grammar) in which the ‘qualifying’ lexeme seems to serve as the
head of the construction:24
Dutch
(72) a. een boom van een vent
a tree of a man
‘a very strong man’
b. een schat van een kind
a treasure of a child
‘a very sweet child’
At least in Dutch these are rather idiomatic expressions, with limited possi-
bilities for morphosyntactic variation (*een hoge boom van een aardige vent
lit. ‘a tall tree of a kind man’ or *een grote schat van een Nederlands kind
lit. ‘a big treasure of a Dutch child’ are ungrammatical). Semantically, such
metaphorical constructions typically have a rather strong emotional value
and indicate a property in excess.
A similar construction is attested in a number of Austronesian languages
of the Western Oceanic group. In these languages an NP such as the big
house is ostensibly expressed as ‘the house’s big(ness)’. Ross (1996) ex-
24
Compare on this construction Paardekooper (1956), Quirk et al. (1987: 1279,
1284–1285), Everaert (1992), Foolen (2003), Plank (2003); also e.g. Ross (1998
a,b), Koptjevskaja-Tamm (2003a,b) and Lichtenberg (2005) and in particular
Keizer (2007: chapter 5). See Malchukov (2000) on DEPENDENCY REVERSAL.
On the co-variation between form and function of adnominal possessives 77
Possessive modifiers are also found in NPs that contain quantifying expres-
sions, but such NPs typically seem to involve constructions in which the
quantifying expression displays at least some nominal properties (such as
plural marking) and takes the quantified entity as its argument.
Russian
(75) pjat’ main
five:NOM car:GEN.PL
‘five cars’
It is not clear why there should be this gap in the modifier functions of a
possessive, but it may be relevant to point out that cardinal numerals have
been found to display certain properties that sets them apart form other
noun modifiers.
Firstly, whereas other grammatical modifier categories of the noun have
a strong tendency to (further) grammaticalize and increase their scope
(‘from inner to outer layer’), numerals tend to be rather resistant to such
diachronic processes. For example, collective markers (classifying opera-
tors/ωω 0) are a common source of plural markers (quantity/ω 2) and in many
languages demonstratives (ω 3) have grammaticalized into definite articles
(4). But as far as I know, only the lowest numerals are involved in gram-
maticalization processes. For example, in many languages the numeral ‘one’
78 Jan Rijkhoff
is the diachronic source of the indefinite article (Givón 1981) and it has
been established that at least in some languages the numeral ‘two’ is the
diachronic source of the dual marker (Dixon 1980: 323).
Secondly, whereas in many languages certain words from the spatial
domain may come to be used for temporal distinctions (see e.g. Haspelmath
1996), this is not so for numerals (Rijkhoff 2008d). For example, von Garnier
(1909) showed that there is clear diachronic relation between markers of
(spatial) collectivity and (temporal) perfectivity in languages of the Indo-
European family. Thus in Dutch and other Germanic languages the prefix
ge- still has a collectivizing meaning in nouns like gebroeders ‘(collection
of) brothers’ and gebergte ‘(collection of) mountains’ (modern German
Gebrüder and Gebirge respectively). At some point in the history of Ger-
manic this prefix came to be used with certain imperfective verbs to express
the notion of completeness (i.e. perfectivity) and ultimately it became asso-
ciated with the past participle form of the verb (Kirk 1923: 65), as in:
Dutch
(76) Heb jij dat ge-daan?
have you that done
‘Did you do that?’
25
Gil (1993: 281) claims that there is asymmetry between nominal and verbal
quantification: “…, while nominal quantifiers can show up on the verb, …, ver-
bal quantifiers can never show up on a noun.”
26
As to the origin of number markers, Frajzyngier (1997) argued that both nomi-
nal and verbal plural markers developed from the same source: a set of deictics,
determiners and anaphors (see on this subject also Lehmann 1982).
On the co-variation between form and function of adnominal possessives 79
27
Notice that ordinal numerals are often expressed in terms of spatial concepts
(Rijkhoff 2004: 167).
28
Evidence for the nonverbal representation of number in human cognition is also
offered in recent studies of infants (Jordan and Brannon 2006; Berger et al.
2006).
80 Jan Rijkhoff
One could say that an adpositional modifier such as op de fiets ‘on the bi-
cycle’ makes the referent of the matrix NP de man ‘the man’ locatable and
hence identifiable for the addressee in the world of discourse by establishing
a semantic relationship between referent of the matrix NP (the man) and the
referent of the adnominal possessive phrase (the bicycle). Notice that the
notion LOCATION is also closely related to the notion EXISTENCE: if an en-
tity has a location in the world of discourse it exists it that world of dis-
course, and vice versa (cf. Lyons 1977: 718–724; Bugenhagen 1986: 127).
The reason for treating (definite) possessives as localizing satellites has
to do with the fact that they, too, typically serve to ‘ground’ the (otherwise
unidentifiable) referent of the matrix NP: they license the definiteness of
the matrix NP by providing a referential anchor for the referent of that ma-
trix NP. In other words, without this modifier the addressee could not lo-
cate or identify the referent of the matrix in the shared world of discourse.
This can even be illustrated with these more or less isolated sentences. The
first example concerns a 71-year old man who in his younger days used to
participate in bicycle races):
(79) hij verzorgt zijn lichaam heel goed. Geen wonder dat men
he takes.care.of his body very well. No wonder that people
hem vroeg of dat de fiets van zijn vader was!
him asked if that the bicycle of his father was
‘He takes very good care of his body. No wonder that they asked him
if that was his father’s bicycle!’
Since de fiets ‘the bicycle’ is linked up with referent of the localizing satel-
lite (i.e. an identifiable entity: the father of the 71-year old man), the ad-
dressee has no problem identifying the referent of the bicycle in the shared
world of discourse. By contrast, without the adnominal possessive van zijn
vader ‘of his father’ the sentence would be pragmatically marked, as there
is no way in which the addressee could infer the existence or location of de
fiets ‘the bicycle’ in conversational space. The same goes for the next sen-
On the co-variation between form and function of adnominal possessives 81
tences, which also contain the phrase de fiets van zijn vader ‘the bicycle of
his father’.
(81) Beiden konden lopend terug, terwijl Tinus drie dagen later ook
both could walking back, while Tinus three days later also
de fiets van zijn vader diende af te staan.
the bicycle of his father must give.up
‘Both were allowed to walk back, but three days later Tinus also had
to give up his father’s bicycle’
Restrictive relative clauses are also typically used to provide the addressee
with a referential anchor for the identification of the referent of the matrix
NP. In fact there are languages in which relative clauses are almost exclu-
sively attested in definite NPs. Consider in this context, for example, these
remarks taken from Lehmann’s monograph on relative clauses (see also
Moravcsik 1969: 167, 170; Bach 1974: 192, 272; Givón 1990: 645ff.).
Mit einem Relativsatz kann man leicht einen bestimmten Gegenstand durch
Spezifikation der Situation, an der er teilhat, identifizieren. So erklärt es
sich, daß die typische Relativkonstruktion von einem Definitum begleitet ist,
wiewohl das natürlich prinzipiell nicht notwendig ist [With a relative clause
one can easily identify a certain object by specifying the situation in which
it is involved. This explains that the typical relative construction co-occurs
with a determiner, although in principle this is not necessary, of course – JR].
(Lehmann 1984: 402).
Das Adjektiv dient mehr der Begriffsbildung, der Relativsatz mehr der Ge-
genstandsidentifikation [The adjective primarily adds to the meaning, the
relative clause typically serves to identify an object – JR]
(Lehmann 1984: 405).
82 Jan Rijkhoff
In (82a) the man has not been mentioned before and one is inclined to ask
what man the speaker is referring to. The response to the b-sentence is
probably different, for example “I did not know you had a car” or “I had no
idea your car was stolen”. The definiteness marking of man in (82b) does
not depend on whether or not the addressee knows the referent of the NP
the man, but on the hearer’s ability to identify the speaker’s car referred to
in the relative clause. Due to presuppositions contained in this clause (‘the
speaker has a car and it was stolen a couple of days ago’), the addressee has
no problem identifying the referent of the NP the man, whose existence and
location in the world of discourse can be inferred on the basis of its semantic
relation to the identifiable entity in that world of discourse referred to in the
adnominal modifier (‘my car’ i.e. the speaker’s car; in fact, ultimately just
the referent of ‘my’, i.e. the speaker).
Relationships between locative, possessive, as well as existential con-
structions have been observed and discussed in many studies. Lyons (1967)
was among the first to point out that these constructions are related, both
synchronically and diachronically. Clark (1970, 1978) investigated the na-
ture of this relation in more detail and demonstrated on the basis of a sam-
ple of 65 languages that these constructions are systematically connected in
terms of word order and patterns of verb use. She interprets possessors as
locations, saying that, cognitively, possessed items can be argued to be lo-
cated ‘at’ the possessor (Clark 1970: 3):
psychologically it would appear quite plausible to argue that if an object is
in some place, and the ‘place’ is actually an animate being, then the object is
possessed by the ‘place’. In other words, it is the [+Animate] feature added
to the locative phrase that transforms it into a Possessor-nominal.
Lyons and Clark were only concerned with sentential constructions, but
obviously the localist account of possession (see section 2.3) also holds for
adnominal possessive modifiers. An example of a locative element that has
developed into a marker of possessorship can be found in the following ex-
ample from Ewe (Niger-Kordofanian), where “the relational noun φé ‘place’
was used as a vehicle to denote possession and developed into a general
On the co-variation between form and function of adnominal possessives 83
marker of nominal possession” (Claudi and Heine 1986; Heine and Kuteva
2002):29
Ewe
(83) fofó nye φé x
father my place house
‘my father’s house’ (‘the house at my father’s place’)
(Claudi and Heine 1986: 316)
29
The locative sense of Dutch van ‘of ’ is still present in expressions specifying a
source, as in Hij komt van buiten ‘He comes from outside’ or De jongens stalen
een fiets van het meisje ‘The boys stole a bike from the girl’ (where van het
meisje ‘from the girl’ is an oblique argument of the verb stelen (van) ‘to steal
(from)’.
84 Jan Rijkhoff
Notice that the figures in (84) are slightly exaggerated because I did not
quite catch the odd doublet and all the (rare) instances in which the van-
phrase happened to follow the NP de fiets ‘the bicycle’ as a prepositional
phrase at the level of the clause (i.e. not as a modifier) specifying a source,
as in this example from a published police report (see also note 29):
In this sentence, the buying event involves three parties (and the price of the
bicycle, of course: 12 Dutch guilders): the buyer (Agent), the item bought
(Patient), and the seller (Source), who in (85) is being referred to in the
prepositional phrase van de man ‘from the man’.
By contrast, the Internet search produced only 31 examples (31/ 23988 =
0.13% of the total number of instances) in which the matrix NP or the lo-
calizing possessive modifier is INDEFINITE.
30
A detailed study on definite and indefinite genitives in English (such as the/a
girl’s name) is Willemse (2005); see also Willemse et al. (this volume).
On the co-variation between form and function of adnominal possessives 85
DEFINITE + INDEFINITE
(de fiets van een man / vrouw/ kind/ jongen /meisje)
From police reports:
(89) Op maandag werd de fiets van een vrouw gestolen.
on Monday became the bicycle of a woman stolen
‘Last Monday the bicycle of a woman was stolen’ (first sentence of
the report)
(90) Bijna twee jaar nadat de fiets van een vrouw uit
almost two year after the bicycle of a woman from
Veenendaal werd gestolen, heeft zij haar fiets terug.
Veenendaal was stolen, has she her bicycle back
‘Nearly two years after the bicycle of a woman from Veendendaal
was stolen, she has got her bicycle back’
INDEFINITE +INDEFINITE
(een fiets van een man / vrouw/kind/ jongen/meisje).
From police reports:
(93) Daders pakten tevens een fiets van een jongen af
perpetrators took also a bicycle of a boy away
‘(the) perpetrators also stole a bicycle of a boy’
(94) Een politieman mocht echter een fiets van een meisje
a policeman was.allowed however a bicycle of a girl
gebruiken
use
‘a policeman was, however, allowed to use a bicycle of a girl’
On the co-variation between form and function of adnominal possessives 87
INDEFINITE +INDEFINITE
(een fiets van een man /vrouw/ kind/ jongen/meisje).
From a joke
(95) snel pakte de man een fiets van een vrouw
quickly grabbed the man a bicycle of a woman
‘Quickly the man grabbed a bicycle of a woman’
INDEFINITE + DEFINITE
(een fiets van de man /de vrouw/ het kind/de jongen /het meisje).
From a police report:
(96) De twee, een 13-jarige jongen en een 34-jarige man, hadden
the two a 13-year.old boy and a 34-year.old man had
een tuinverlichting en een fiets van de vrouw vernield en
a garden_lights and a bicycle of the woman destroyed and
haar verbaal bedreigd. …
her verbally threatened
‘the two, a 13-year old boy and a 34-year old man, had destroyed gar-
den lights and a bicycle of the woman and verbally threatened her’
Emphasis falls naturally on the two indefinite NPs een tuinverlichting and
een fiets (van de vrouw), which indicates that this particular construction
(INDEFINITE + DEFINITE) is used to highlight the listed items. This is also
true for one of the other three instances of the combination INDEFINITE +
DEFINITE. This example occurs in a book report from a teenage boy, after
he mentioned other items the man was not supposed to have in his posses-
sion:
88 Jan Rijkhoff
INDEFINITE+DEFINITE
(een fiets van de man / de vrouw/ het kind/de jongen /het meisje).
(97) Ze vonden ook nog een fiets van de man
they found also still a bicycle of the man
‘Additionally they found a bicycle of the man’
The remaining 12 cases can be divided in two main groups. In one group,
the adnominal possessive contains a modifier that specifies a property that
is relevant for the further development of the story:
(99) … de fiets van een man met een vies gele broek.
… the bicycle of a man with a dirty yellow trousers
Hij kent die broek …
He knows those trousers
‘… the bicycle of a man with dirty yellow trousers. He has seen those
trousers before.’
The second group has adnominal possessives that mention highly non-
specific entities (like the cases attested in the police reports). The NPs in
question typically occur in texts from chat sites or web-logs:
This is summarized in Table 3, which also shows that the layered model
(Figures 1 and 2) seems to iconically reflect the degree of grammatical in-
dependence of NP satellites (‘iconicity of distance’): the further away the
modifier is from the head noun in the model (i.e. the wider the scope of the
modifier) the less restricted it is with respect to MODIFICATION, PREDICA-
TION and REFERENCE.
90 Jan Rijkhoff
ADNOMINAL
POSSESSIVES MODIFICATION PREDICATION REFERENCE
LOCALIZING
de fiets van mijn de fiets van mijn die fiets is van mijn de fiets van
vader (oude) vader (oude) vader Peters vader
‘the bike of my ‘the bike of my the bike is of my ‘the bike of
father’ (old) father’ (old) father Peter’s father’
‘the bike belongs to (REFERENCE TO
my (old) father’ AN ENTITY)
QUALIFYING B+
beelden van beelden van de beelden zijn van beelden van
grote kwaliteit grote kwaliteit verschillende kwaliteit deze kwaliteit
‘statues of great ‘statues of high ‘the statues are of ‘statues of this
quality’ quality’ varying quality’ quality’
(REFERENCE TO A
PROPERTY OF AN
ENTITY; SECTION
4.2)
QUALIFYING B
een kroon van een kroon van de kroon is van —
goud (zuiver) goud (zuiver) goud
‘a crown of gold’ ‘a crown of (pure) the crown is of (pure)
gold’ gold
‘the crown is made of
(pure) gold’
QUALIFYING A+
een man van vele een man van vele — —
gezichten gezichten
‘a man of many ‘a man of many faces’
faces’
QUALIFYING A
een man van een man van — —
gezag (groot) gezag
‘a man of ‘a man of (great)
authority’ authority’
(SEE SECTION 4.1)
CLASSIFYING
een man van de — — —
wereld
‘a man of the world’
On the co-variation between form and function of adnominal possessives 91
Subject
(102) There is photographic evidence that native Australians of small stature
[…] lived in northern Queensland during the 19th and early 20th …
(103) Man of many pursuits insists arms trade isn’t among them.
Object
(104) The past couple of weeks we’ve been honored to have had riders of
such incredible stature racing their bikes once again in T-town.31
31
The phrase of stature is particularly frequent in biblical texts. For more exam-
ples, see: http://concordance.biblebrowser.com/s/stature.htm.
92 Jan Rijkhoff
It turns out that type A qualifying possessives are almost exclusively at-
tested in indefinite NPs that occur in predicate position as well as in cap-
tions, headings and book titles. From a grammatical perspective, titles are
rather similar to non-verbal clauses, except for the fact that non-verbal
clauses must contain a copula (at least in Dutch and English), which is
typically absent in book titles or article headings (instead we tend to find a
colon). Below are examples of the string ‘… man of many …’ in the contexts
mentioned above (all found on the Internet using the Ask search engine).
Fridtjof Nansen: Man of many facets. … His gentler qualities … came per-
haps from his quieter, more ascetic father, a lawyer of repute and a man of
unswerving integrity; Benjamin Franklin: A Man of Many Talents; David A.
Poulsen: A Man Of Many Hats; Alec Guinness – A Man of Many Parts;
Bryan Adams is a man of many faces; Nansen: man of many talents; A man
of many strong opinions … Carl Everett is a man of conviction; Maseda is
a man of many talents; Russell B. Farr: a man of many colors; Hip-Hop
Artist Madlib, Man of Many Names; Thomas Merton: Man of Many Jour-
neys; Man Of Many Voices – The Official Johnny Vallis Web Site; A Man of
Many Firsts – Otellini Finds New Ways to Reach Consumers; A man of
many parts; Ebbers is a man of many contradictions; Bruno Kirby: A Man of
Many Scenes; Ed Case is a Man of Integrity; Gideon Omnibus, a collection
of fictional stories about a Scotland Yard senior superintendent of enor-
mous stature and presence, …; Mhanda is a man of stature. … This is a
man of stature; And he slew an Egyptian, a man of great stature, five cubits
high.
The string ‘… woman of many …’ did not yield as many results, but here are
some examples:
These examples involve NPs with a type A qualifying possessive, but notice
that in each case it is the whole NP has a strong classifying flavor, not just
the possessive modifier. This ties in nicely with the fact that indefinite NPs
in predicate position are typically used to express class inclusion (Shelley is
On the co-variation between form and function of adnominal possessives 93
a communist – Dik 1997 Part 1: 205; cf. Hengeveld 1992: 89, Keizer 1992:
ch. 6).32
(106) Wie een paar keer koffie van deze kwaliteit gedronken heeft, …
Who a couple.of time coffee of this quality drunk has, …
‘Someone who has tasted coffee of this quality a couple of times, …’
32
By contrast, when a DEFINITE NP occurs in predicate position, it is typically
employed for identification purposes (Suzy is the singer). This is also true for
predicate NPs with a qualifying adnominal possessive of Type A, as in:
Dutch
(i) Huub is [de man van de computers]IDENTIFYING PREDICATE NP in ons kantoor
Huub is the man of the computers in our office
‘In our office, Huub is the man of the computers’
94 Jan Rijkhoff
SIZE, etc.) applies to an entity. One could say that type B+ qualifying pos-
sessives, which refer to PROPERTIES, form a bridge between the more typical
qualifying possessives, which cannot be used referentially, and localizing
possessives, which refer to ENTITIES and which makes it possible for the
addressee to identify the referent of the matrix NP.
These data indicate that the boundaries between the various modifier
categories (CLASSIFYING, QUALIFYING, LOCALIZING) are gradual rather
than distinct. Qualifying possessives of type A occur between other types
of qualifying possessives and classifying possessives, because the NPs in
which they occur often signal class inclusion, as in Ed Case is [a Man of
Integrity], i.e. Ed Case belongs to the class of men who are honest and
strong about what they believe to be right; section 4.1). Qualifying posses-
sives of type B+ are situated between other types of qualifying possessives
and localizing possessives because they are the only qualifying possessives
that can be used referentially (but recall that they refer to properties rather
than entities).
5. A tentative explanation
How can we explain the distribution of values for the three parameters in
Table 3? The short answer is: the different values reflect the functional re-
quirements of the various modifier functions in which the adnominal pos-
sessive is used. To put it differently, the possessive construction has a wide
range of semantic and syntactic possibilities that are exploited in different
degrees (depending on the modifier function it serves in the NP) by the lan-
guage user. From a formal perspective the possessive construction is simply
a phrasal structure with (potentially at least) the full array of grammatical
features associated with the three major lexical word class Adjective
(MODIFICATION), Verb (PREDICATION), and Noun (REFERENCE):
The values for the three parameters in Table 3 seem to suggest that there are
no grammatical restrictions when the adnominal possessive with van ‘of’
serves as a LOCALIZING SATELLITE (section 3.4), providing the addressee
with an entity (a referential anchor) that makes the referent of the matrix NP
identifiable in the shared world of discourse. There is, however, one prag-
matic constraint: since the referential anchor must be an identifiable entity,
the localizing adnominal possessive phrase must be definite (a marginal num-
ber of apparent counter examples were discussed in the previous section).
When the possessive phrase is used as a QUALIFYING SATELLITE (section
3.2), it specifies a more or less characteristic property of the referent (color,
value, size, etc.), a function typically associated with adjectives (if a lan-
guage has them). Therefore there is no need to fully exploit the referring
potential of the possessive phrase. In fact, there is only one type of qualify-
ing possessives (type B+) which has restricted referential potential in that it
can be used to refer to the extent to which some property (size, age etc.)
applies to an entity (as in munten van DEZE kwaliteit ‘coins of this quality’).
Since qualities are gradable properties (a rather/very /incredibly small ta-
ble), it is possible to modify the head of the qualifying adnominal posses-
sive (i.e. modification), as in een parel van g rote waarde ‘a pearl of great
value’. Qualities are also properties that can be predicated of an entity,
therefore possessives used as qualifying satellites may also appear as predi-
cates (just like adnominal possessives used as localizing satellites), as in een
paspoort is van g rote waarde ‘a passport is of great value’. The fact that
type A qualifying possessives cannot occur as predicates (which puts them
in between classifying possessives and qualifying possessives of type B)
nicely illustrates the fuzzy boundary between qualifying and classifying
satellites (see note 8 and section 4).
CLASSIFYING POSSESSIVES further specify what kind of entity is being
denoted by the head noun (section 3.1), e.g. steam train, electric train, ex-
press train all specify a particular kind of train. It seems, however, that one
can only linguistically predicate a property of an entity that is conceptually
complete. Thus, the reason why classifying satellites cannot serve as a sen-
tence predicate is precisely that the subject entity does not constitute a con-
ceptually complete entity yet to predicate a property of, as the classifying
satellite actually helps to define that entity.33 In other words, one cannot say
33
Alternatively one could say (Bill McGregor, pers. comm.): the reason why clas-
sifying possessives do not occur as predicates is that a classifying modifier
highlights a (proto)typical feature of a set of entities (a TYPE), and such features
cannot be predicated of any particular instance (a TOKEN).
96 Jan Rijkhoff
that *a hero is urban or that *a man is of the cloth, because there first has
to be ‘an urban hero’ or ‘a man of the cloth’ before one can predicate some
property of these entities (as in, for example, the urban hero is a myth).34
It is basically for the same reason that classifying possessives cannot be
modified themselves. Since the classifying modifier and the head noun con-
stitute a tight unit, not just conceptually (as we just saw) but also syntacti-
cally, one can only modify the combination of classifying modifier-plus-
noun. In other words, the strong conceptual and syntactic bond between
classifying modifier and head makes it impossible to single out the classify-
ing possessive for modification or predicative purposes.
6. Conclusion
We have seen that adnominal possessives with van/‘of’ can occur in three
of the four descriptive modifier functions distinguished in the layered model
of the NP presented in Figures 1 and 2: they can be used as CLASSIFYING,
QUALIFYING, and LOCALIZING NP satellites. They are, however, not attested
as quantifying NP satellites, which may have to do with the special role of
numerical concepts in human cognition (section 3.3). Furthermore there is a
positive correlation between the kind of modifier function of the adnominal
possessive in the NP and its grammatical properties. CLASSIFYING AD-
NOMINAL POSSESSIVES (section 3.1), which only have the noun in their
scope, are severely restricted with respect to the three grammatical parame-
ters Modification, Predication, and Reference. By contrast, LOCALIZING
ADNOMINAL POSSESSIVES, which have the widest scope, are not restricted
with respect to these parameters, and the various types of QUALIFYING AD-
NOMINAL POSSESSIVES occupy intermediate positions between the two ex-
tremes (section 3.2):
34
Notice that of the cloth does not refer to a particular cloth, i.e. the presence of
the definite article in of the cloth does not mean that the classifying possessive
is used referentially here.
On the co-variation between form and function of adnominal possessives 97
LOCALIZING MODIFIER + + +
(reference to an
entity)
QUALIFYING MODIFIER B+ + + +
(reference to a
property)
QUALIFYING MODIFIER B + + —
QUALIFYING MODIFIER A/A+ + — —
CLASSIFYING MODIFIER — — —
35
See McCawley (1985: 675) and Dik (1986) for a critical discussion of the ‘one
form – one function’ approach in linguistics (Newmeyer 1983).
98 Jan Rijkhoff
Acknowledgements
References
Aarts, Bas
2004a Modelling linguistic gradience. Studies in Language 28 (1): 1–49.
Aarts, Bas
2004b Conceptions of gradience in the history of linguistics. Language Sci-
ences 26 (4): 343–389.
Aarts, Bas
2006 Conceptions of categorization in the history of linguistics. Language
Sciences 28 (4): 361–385.
Aertsen, Henk, Mike Hannay and Rod Lyall (eds.)
2004 Words in their Places: A Festschrift for J. Lachlan Mackenzie. Ams-
terdam: Free University, Faculty of Arts.
Bach, Emmon
1974 Syntactic Theory. New York: Holt, Rinehart & Winston.
Barton, Irène, Michael Herslund and Finn Sørensen (eds.)
2001 Dimensions of Possession. Amsterdam: Benjamins.
Bauer, Laurie
1998 When is a sequence of two nouns a compound? English Language
and Linguistics 2: 65–86.
Bennis, Hans and Jan W. de Vries (eds.)
1992 De Binnenbouw van het Nederlands: Een Bundel Artikelen voor Piet
Paardekooper. Dordrecht: Foris.
Berger, Andrea, Gabriel Tzur and Michael I. Posner
2006 Infant brains detect arithmatic errors. Proceedings of the National
Academy of Sciences 103 (33): 12649–12653.
On the co-variation between form and function of adnominal possessives 99
Bugenhagen, Robert D.
1986 Possession in Mangap-Mbula: its syntax and semantics. Oceanic
Linguistics 25: 124 –166.
Butler, Christopher S.
2003 Structure and Function: A Guide to Three Major Structural-Function-
al Theories. Part 1: Approaches to the Simplex Clause; Part 2: From
Clause to Discourse and Beyond. Amsterdam/Philadelphia: Benjamins.
Chappell, Hillary and William B. McGregor
1989 Alienability, inalienability and nominal classification. In Hall et al.
(eds.), 24–36.
Chappell, Hilary and William B. McGregor (eds.)
1996 The Grammar of Inalienability: A Typological Perspective on Body
Part Terms and the Part-Whole Relation. Berlin /New York: Mouton
de Gruyter.
Clark, Eve V.
1970 Locationals: a study of the relation between ‘existential’, ‘locative’,
and ‘possessive’ constructions. Working Papers in Language Univer-
sals 3 (Stanford University), L1– L36 + xiii.
Clark, Eve V.
1978 Locationals: existential, locative, and possessive constructions. In
Greenberg et al. (eds.), 85 –126.
Claudi, Ulrike and Bernd Heine
1986 On the metaphorical basis of grammar. Studies in Language 10 (2):
297–335.
Cole, Peter (ed.).
1981 Radical Pragmatics. New York: Academic Press.
Comrie, Bernard
1981 Language Universals and Linguistic Typology. Oxford: Blackwell.
Comrie, Bernard
1988 General features of the Uralic languages. In Sinor (ed.), 451–477.
Comrie, Bernard (ed.)
1987 The World’s Major Languages. London: Croom Helm.
Connolly, John H. and Simon C. Dik (eds.)
1989 Functional Grammar and the Computer (Functional Grammar Series
10). Dordrecht: Foris.
Corston-Oliver, Simon
2002 Roviana. In Lynch et al. (eds.), 467–497.
Corum, Claudia, T. Cedric Smith-Stark and Ann Weiser (eds.)
1973 You Take the High Node and I’ll Take the Low Node. Papers from
the 9th Regional Meeting of the Chicago Linguistic Society, April
13–15, 1973. Chicago: Chicago Linguistic Society.
Dik, Simon C.
1986 On the notion ‘functional explanation’. Belgian Journal of Linguistics
1: 11–52.
100 Jan Rijkhoff
Dik, Simon C.
1989 The Theory of Functional Grammar (Functional Grammar Series 9).
Dordrecht: Foris.
Dik, Simon C.
1997 The Theory of Functional Grammar (2nd revised edition, edited by
Kees Hengeveld). Part 1: The Structure of the Clause. Part 2: Com-
plex and Derived Constructions. Berlin/New York: Mouton de Gruyter.
Dixon, Robert M.W.
1980 The Languages of Australia. Cambridge: Cambridge University Press.
Dixon, Robert M.W.
1982 Where have all the Adjectives Gone? And other Essays in Semantics
and Syntax. Berlin /New York: Mouton de Gruyter.
Englebretson, Robert
2003 Searching for Structure: The Problem of Complementation in Collo-
quial Indonesian Conversation [Studies in Discourse and Grammar
13]. Amsterdam /Philadelphia: Benjamins.
Everaert, Martin
1992 Nogmaals: Een schat van een kind. In Bennis and de Vries (eds.),
45–54.
Everett, Daniel L.
2005 Cultural constraints on grammar and cognition in Pirahã: another
look at the design features of human language. Current Anthropology
46 (4): 621–634 (plus comments and a reply from the author, 635–
646).
Foolen, Ad
2004 Expressive binominal NPs in Germanic and Romance languages. In
Radden and Panther (eds.), 75–100.
Farsi, A. A.
1968 Classification of adjectives. Language Learning 18: 45–60.
Frajzyngier, Zygmunt
1997 Grammaticalization of number: from demonstratives to nominal and
verbal plural. Linguistic Typology 1 (2): 193–242.
Fraurud, Kari
2001 Possessives with extensive use: a source of definite articles? In Bar-
ton et al. (eds.), 243–267.
García Velasco, Daniel and Jan Rijkhoff (eds.)
2008 The Noun Phrase in Functional Discourse Grammar [Trends in Lin-
guistics. Studies and Monographs 195]. Berlin /New York: Mouton
de Gruyter.
Giegerich, Heinz J.
2005 Associative adjectives in English and the lexicon-syntax interface.
Journal of Linguistics 41 (3): 571–591.
Gil, David
1993 Nominal and verbal quantification. Sprachtypologie und Universa-
lienforschung (STUF) 46 (4): 275–317.
On the co-variation between form and function of adnominal possessives 101
Gildea, Spike
1993 The development of tense markers from demonstrative pronouns in
Panare (Cariban). Studies in Language 17 (1): 53–73.
Givón, Talmy
1981 On the development of the numeral ‘one’ as an indefinite marker.
Folia Linguistica Historica 2 (1): 35–55.
Givón, Talmy
1990 Syntax: A Functional-Typological Introduction. Volume II. Amster-
dam: Benjamins.
Greenberg, Joseph H., Charles A. Ferguson and Edith A. Moravcsik (eds.)
1978 Universals of Human Language. Volume 4: Syntax. Stanford: Stanford
University Press.
Guillemin, Diana
2007 A look at so in Mauritian Creole: From possessive pronoun to em-
phatic determiner. In Huber and Velupillai (eds.), 279–296. Amster-
dam /Philadelphia: Benjamins.
Gunkel, Lutz and Gisela Zifonun
2008 Constraints on relational-adjective noun constructions: a comparative
view on English, German and French. Zeitschrift für Anglistik und
Amerikanistik 56: 283–302.
Hall, Kiria, Michael Meacham and Richard Shapiro (eds.).
1989 Proceedings of the Fifteenth Annual Meeting of the Berkeley Lin-
guistics Society, February 18–20, 1989. General Session and Para-
session on Theoretical Issues in Language Reconstruction. Berkeley:
Berkeley Linguistics Society.
Halliday, Michael Alexander Kirkwood
2004 An Introduction to Functional Grammar (third edition, revised by
Christian M.I. M. Matthiessen). London: Arnold.
Hammond, Michael and Michael Noonan (eds.).
1988 Theoretical Morphology: Approaches in Modern Linguistics. New
York: Academic Press.
Haspelmath, Martin
1996 From Space to Time: Temporal Adverbials in the World’s Languages
(LINCOM Studies in Theoretical Linguistics 03). München: Lincom
Europa.
Haspelmath, Martin
1999 Explaining article-possessor complementarity: economic motivation
in noun phrase syntax. Language 75 (2): 227–243.
Hawkins, John A.
1991 On (in)definite articles: implicatures and (un)grammaticality predic-
tion. Journal of Linguistics 27: 405–442.
Heine, Bernd
1997 Possession: Cognitive Sources, Forces, and Grammaticalization
(Cambridge Studies in Linguistics 83). Cambridge: Cambridge Uni-
versity Press.
102 Jan Rijkhoff
Koptjevskaja-Tamm, Maria
2003a A woman of sin, a man of duty, and a hell of a mess: non-determiner
genitives in Swedish. In Plank (ed.), 515–558.
Koptjevskaja-Tamm, Maria
2003b Possessive noun phrases in the languages of Europe. In Plank (ed.),
621–722.
Lehmann, Christian
1984 Der Relativsatz: Typologie seiner Strukturen, Theorie seiner Funk-
tionen, Kompendium seiner Grammatik. Tübingen: Narr.
Levi, Judith N.
1973 Where do all those other adjectives come from? Papers from the
Ninth Regional Meeting of the Chicago Linguistic Society, April 13–
15. In Corum et al. (eds.), 332–345.
Lichtenberg, Frantisek
2005 On the notion “adjective” in Toqabaqita. Oceanic Linguistics 44 (1):
113–144.
Lynch, John, Malcolm Ross and Terry Crowley (eds.).
2002 The Oceanic Languages. Richmond: Curzon.
Lyons, John
1967 A note on possessive, existential and locative sentences. Foundations
of Language 3: 390–396.
Lyons, John
1977 Semantics (2 volumes). Cambridge: Cambridge University Press.
Malchukov, Andrej L.
2000 Dependency Reversal in Noun-Attribute Constructions: Towards a
Typology (LINCOM Studies in Language Typology 03). München:
Lincom Europa.
McCawley, James
1985 Review of Newmeyer 1983. Language 61: 668–679.
Menges, Karl Heinrich
1968 The Turkic Languages and Peoples: An Introduction to Turkic Studies.
Wiesbaden: Harrassowitz.
Mithun, Marianne
1988 Lexical categories and the evolution of number marking. In
Hammond and Noonan (eds.), 211–234.
Moravcsik, Edith A.
1969 Determination. Working Papers on Language Universals 1: 64 –130.
Mosel, Ulrike and Even Hovdhaugen
1992 Samoan Reference Grammar. Oslo: Universitetsforlaget AS.
Newman, Paul
1987 Hausa and the Chadic languages. In Comrie (ed.), 705–723.
Newman, Paul
1990 Nominal and Verbal Plurality in Chadic (Publications in African
Languages and Linguistics 12). Dordrecht: Foris.
104 Jan Rijkhoff
Rijkhoff, Jan
2003 When can a language have nouns and verbs? Acta Linguistica Haf-
niensia 35: 7–38.
Rijkhoff, Jan
2004 Iconic and non-iconic word order patterns: on symmetry in the NP
and counter examples to Universal 20’. In Aertsen et al. (eds.), 169–
180.
Rijkhoff, Jan
2008a Layers, levels and contexts in functional discourse grammar. In Gar-
cía Velasco and Rijkhoff (eds.), 63–115.
Rijkhoff, Jan
2008b Layering and iconicity in the noun phrase: descriptive and interper-
sonal modifiers. Linguistics 46 (4): 789–829.
Rijkhoff, Jan
2008c On flexible and rigid nouns. Studies in Language 32 (3): 727–752
(Special issue, U. Ansaldo, J. Don and R. Pfau (eds.), Parts of Speech:
Descriptive Tools, Theoretical Constructs).
Rijkhoff, Jan
2008d Synchronic and diachronic evidence for parallels between noun
phrases and sentences. In Josephson and Söhrman (eds.), 13–42.
Rijkhoff, Jan and Johanna Seibt
2005 Mood, definiteness and specificity: a linguistic and a philosophical ac-
count of their similarities and differences. Tidsskrift for Sprogforskning
3 (2): 85–132. http://ojs.statsbiblioteket.dk/index.php/tfs/issue/archive.
Rosenbach, Anette
2006 Descriptive genitives in English. English Language and Linguistics
10 (1): 77–118.
Ross, Malcolm
1996 Adjectives with possessor nouns. LINGUIST List: Vol-7-1678, Sub-
ject 7.1678.
Ross, Malcolm
1998a Proto-Oceanic adjectival categories and their morphosyntax. Oce-
anic Linguistics 37: 85–119.
Ross, Malcolm
1998b Possessive-like attribute constructions in the Oceanic languages of
northwest Melanesia. Oceanic Linguistics 37: 234 –276.
Schachter, Paul
1985 Parts-of-speech systems. In Shopen (ed.), 3–61.
Seiler, Hansjakob
1983 Possession as an Operational Dimension of Language. Tübingen:
Narr.
Shopen, Timothy (ed.).
1985 Language Typology and Syntactic Description. Volume I: Clause
Structure. Cambridge: Cambridge University Press.
106 Jan Rijkhoff
Doris L. Payne
1. Introduction
1
I capitalize semantic role labels (Locative, Theme, etc.), but I use lower case
(location, possession, etc.) when representing the view that there is not necessar-
ily a theoretically relevant semantic role involved. When in quotes, “possession”
indicates the view that “Possessor” is a spurious semantic role and no different
from Locative or the predication of location. I will use the abbreviations PR
(possessor) and PM (possessum) only when I am not particularly concerned one
way or the other with the theoretical status of such notions as semantic roles.
108 Doris L. Payne
2
William McGregor (pers. comm.) notes that in Nyulnyul and related Australian
languages, a negative existential and a negative possessive construction share
highly similar syntax that must be related historically. Significantly, there is no
formally related negative locative construction to provide any link between the
two – i.e. the possessive and existential constructions are directly connected.
Is possession mere location? Contrary evidence from Maa 109
The view that “possession” is just location with a (likely animate) Locative
is well attested in the literature (though held to varying degrees of strength).
I will refer to this as the “possession-is-location” view. For just a few ex-
amples, Baron and Herslund (2001) argue that the fundamental meaning of
have is that one object is simply located with respect to another, though
additional part-whole or other meanings may be superimposed onto this
fundamental meaning. DeLancey (2002: 8) suggests that constructions which
predicate possession and existence/location have the “same underlying
structure, and differ only in the relative salience, inherent or contextually-
determined, of the two arguments”. Additional claims in this direction are
found in Gruber (1965), Lyons (1967, 1968), Clark (1978), Jackendoff (1983),
inter alia. These sorts of claims are generally grounded in the view that the
semantic role(s) of both putative “Possessor” and Locative derive from
one-and-the-same conceptual or cognitive model of [x BE.AT y], where x is
a Theme and y is Locative. In general, Theme and Locative derive only
from either BE.AT or GO.TO types of abstract predicate notions, which
may be understood in either literal or metaphorical ways. Agent derives
from a CAUSE (or DO) predicate notion. All more lexically-specific predi-
cates can be subsumed to one of these, and hence all core semantic roles in
any clause whatsoever can be subsumed to Agent, Theme or Locative.3
DeLancey (2002) argues that Theme and Locative are grounded in – if
not exactly equivalent to – the cognitive distinction between Figure and
Ground (cf. Wertheimer 1923). In cognitive studies, Figure designates an
entity which perceptually stands out against a Ground, it is the focus of at-
tention, and it is seen in detail. A Ground is that part of the perceptual field
against which the Figure stands out, it is in the periphery of attention, lacks
3
In this paper I cannot attempt to address the vast literature on theories or systems
of semantic roles. DeLancey (2002) is concerned with developing a constrained
theory of semantic roles that can account for surface case marking of core ar-
guments of predicates, and specifically not with oblique cases. He does not deny
the existence of semantic categories like Beneficiary or Instrument for obliques,
but denies they are ever warranted as roles for core arguments. My own approach
is to posit whatever is empirically driven, and believe the cross-linguistic evi-
dence supports a somewhat larger inventory than DeLancey posits (cf. Comrie
and van den Berg (2006) on case marking evidence for core Experiencier distinct
from core or any type of oblique Locative).
110 Doris L. Payne
Existential
(1) Ay anma yul conob mach scuy yuninal yin abxubal.
exist people in town not teach their.children in Jacaltec
‘There are people in town who do not teach their children Jacaltec.’
Location
(2) Ay-c’oj ha mam.
exist-DIR your father
‘Is your father here?’
Possession
(3) Ay no’ hin txitam.
exist CL my pig
‘I have a pig.’
In some very important historical and cognitive work, Heine (1997, 2001)
argues that there are eight conceptual sources of predicative possession
constructions, only one of which he calls locational. An ACTION conceptual
4
Abbreviations are as follows: A – most agent-like argument of canonical transi-
tive verb, ACC – accusative, ANTIP – antipassive, ASSOC – associative, CL –
noun class, CN – discourse connective, COMP – complementizer, COP – copula,
DAT – dative, DEF – definite, DIR – directional, DSCN – discontinuous discourse
connective, EP – epenthetic, F – feminine, IMPF – imperfective, INF – infinitive,
LOC – locative, MID – middle, NEG – negative, NMLZ – nominalizer, NOM –
nominative, NPF – non-perfect/non-perfective, OBJ – object, OBL – oblique, P –
most patient-like argument of canonical transitive verb, PASS – (impersonal)
passive, PF – perfect/perfective aspect, PL – plural, PM – possessum, POSS –
possessive, PR – possessor, PRF – perfect, PRT – particle, REL – relative, S –
single argument of canonical intransitive verb, SBJ – subject, SBJV – subjunc-
tive, SG – singular, VENT – ventive. Maa examples taken from texts are tagged
with a text name and line number.
112 Doris L. Payne
The only one of Heine’s schemas that at first glance might seem problematic
to the possession-is-location framework is ACTION because she would be
AGENT in She grabbed/took the book (i.e. these could well stand as an an-
swer to What did Mary do? She grabbed the book.) However, the fully-
specified event chain underlying Mary grabbed the book would be [Mary
CAUSE [ [book GO.TO Mary ] & [book BE.AT Mary] ] ]. Furthermore,
since the preceding schemas are presented as historical sources of posses-
sion constructions, the possession-is-location adherent simply notes that as
CAUSE or “doing” semantics is eventually bleached out of a verb like grab
and it comes to have an inherently stative meaning, the descendent verb is
no longer a CAUSE type of predicate (nor even a GO.TO predicate), but
simply a BE.AT predicate.
Is possession mere location? Contrary evidence from Maa 113
In some contrast to the preceding view is, I believe, the view of possession
articulated by Langacker (1993) and Velazquez-Castillo (1996), as well as
scholars who treat Posessor and Locative as distinct semantic roles (e.g.
Kemmer 2002; Tham 2005; Van Valin and LaPolla 1997: 125–127).5 Lang-
acker and Velazquez-Castillo point out how possession involves the cogni-
tive notion of singling out one entity, the “target” (or PM), for individual
conscious awareness in terms of a more salient entity which is the “refer-
ence point” (or PR). According to Langacker (1993: 6), a reference point
construction includes not only the “target” and the “reference point”, but
the relevant construction is applied within a particular contextually-deter-
mined “dominion” (or set of entities) which the reference point enables ac-
cess into.
Reference points are determined by ease of identifiability on the part of
the conceptualizer (e.g. speaker). Ease of identifiability can be context de-
pendent, but in some way the reference-point entity has greater perceptibil-
ity – whether because it is inherently more salient, or because it is an entity
for which the conceptualizer has greater empathy, familiarity, acquaintance,
etc. Strikingly, such features correspond to cognitive Figures, as opposed to
their Grounds; i.e., a Figure is conceptually the most salient entity. Thus,
conceptually the salient PR is naturally a primary Figure, which stands out
against its background. (If anything corresponds to Ground in Langacker’s
scenario, it would seem to be the “dominion”.) Thus, if Theme and Locative
are grounded in or equivalent to the conceptual notions of Figure and
Ground, overall this should lead to the designation of PRs as being tanta-
mount to Themes – contrary to what is assumed in the possession-is-loca-
tion view.
Now, of course, there can be hierarchical layers of what is Figure and
what is Ground – one Figure could be a Ground for yet another Figure as
the mind’s eye ranges across a conceptual scene – and this is doubtless part
of what is conceptually operative in possessive constructions. But very im-
portantly, this potential nesting of a Figure against Figure-as-relative-
Ground does not destroy the insight that PRs are highly perceptible and sa-
5
Van Valin and LaPolla’s (1997) multiple semantic roles are differentiated from
their small set of macro-roles, and their Locative and Possessor are claimed to
pattern together in various ways along with Experiencier, Attributant, etc.
Kemmer (2002) presents typological evidence for distinct conceptualization of
Possessor from Locative.
114 Doris L. Payne
lient. Crucially, they may be highly perceptible and salient relative to the
PM. Thus, predicating possession is different from just predicating loca-
tion. If possession is “location” at all, it is localizing something relative to a
highly perceptible, typically individuated, Figure-like entity. We might say
that the PR is a “big” Figure, relative to a “small” Figure that is the PM.
Note that these are relational (hence semantic or cognitive role-like) notions,
regardless of whether they are instantiated in a verbal or clausal predicate
or within an NP structure. That is, Figure and Ground mutually co-define
each other, and cannot be reduced to the features of solitary referents in-and-
of themselves.6 I will refer to this as the “possession-is-more-than-location”
view.
All this leads to the question: Is there empirical lexical and/or morpho-
syntactic evidence that languages treat predicative possession as a signifi-
cant, even basic, concept relevant to the structure of human language? That
is, is there evidence supporting a distinct Possessor role (and possibly also
Possessum role) related to certain predicates, different from Locative and
Theme – at least in certain languages? Or can lexical and constructional
behaviour be fully accounted for just in terms of the three core roles posited
by the possession-is-location theory?
My answer is “I don’t think that possession-is-location accords with all
the linguistic data.” Indeed, this answer should not be surprising given the
prima facie ramifications that understandings of possession have in ordi-
nary social relations.7 The most obvious empirical evidence is that many
languages do have distinct syntactic structures for predicating location ver-
sus possession. Indeed, even in many languages where a single predicate
may be used for possessive as well as locative and/or existential meanings,
there are additional required grammatical features, besides just the predi-
cate morpheme, which distinguish the meanings. This can be seen in the
6
One reason a Possessor may seem different from a Locative is because a Posses-
sor may often be topical. Proponents of the possession-is-location view would
presumably say that participant topicality has no direct bearing on the existence
of any particular semantic role in a clause, as topicality is a discourse or other
cognitive property not based in the relational nature of the predicate.
7
Indeed, referring to Lyons (1967: 392), DeLancey (2002: 9) suggests that physi-
cal possession is the more primitive concept in a developmental sense. Neither
DeLancey nor Lyons pursues empirical acquisition research; but if DeLancey’s
suggestion is true, it would seem reasonable to hypothesize that Possessor ought
to be the more basic role, and “location” just a derivative or metaphorical exten-
sion of that.
Is possession mere location? Contrary evidence from Maa 115
Jakaltek examples (1)–(3) above: though they share the same copular root
ay, they require distinct constructions – and something about human cogni-
tion is certainly driving the development of distinct constructions taken as a
whole.
In her focus on what is typologically common across constructions that
predicate possession, existence, and location, Clark’s (1978) exposition
does not particularly highlight how common differences are among these
constructions, even if they might happen to share a particular predicate, a
locative phrase at some point in history, or some other morphosyntactic
element. However, differences are evident from a closer examination of the
grammars of at least some languages on which she originally based her
conclusions. For just one example, Amharic is one of the 30 languages in
her sample. In Amharic, the same root may be used in existential (5) and
predicative possessive (6) clauses, but this hardly means that the construc-
tions are the same: the existential is intransitive, while the possessive is
transitive, as shown by argument marking on the verb (examples are from
Michael Ahland, pers. com.; see also Ahland 2009).
Existential
(5) a. (Be-t’erapeza laj) ms’ haf-ott all-u.
LOC-table top book-PL exist.PRF-3PL.SBJ
‘There are books/some specific books (on the table).’
Existential/Locative (vague)
b. Ms’ haf-ott be-t’erapeza laj all-u.
book-PL LOC-table top exist.PRF-3PL.SBJ
‘The/Some specific books are on the table.’
Possession
(6) L d
-itwa ms’ haf-ott all-u-at.
child-DEF.F book-PL exist.PRF-3PL.SBJ-3F.OBJ
‘The girl has books/some specific books.’
Location
(7) Ms’haf-ott b-t’rapeza laj n-attw.
book-PL LOC-table top COP.IMPF-3PL
‘The/some specific books are on the table.’
If one were to argue that the “Possessor” in (6) is somehow a Locative, akin
to the Locative in (7), it certainly is not the case that the constructions as a
whole are the same: first, (6) is transitive while (7) is intransitive; second,
the order of Themes and putative Locatives in (6) and (7) are reversed; and
third, the predicates are not the same.8
Clark (1978) goes on to argue that a single root (nw) is shared both for
predicating location (7) – in (7) this is shortened to n – and for what she
calls “Possession 2”, i.e. a construction that involves a genitive NP in a
copular construction akin to The book is John’s. But when we compare the
Locative in (7) with Clark’s Possession-2 structure, the constructions are
still not the same because the Possession-2 structure necessarily requires a
genitive NP, i.e. something like John’s or John’s thing – the possessive
sense does not emerge simply from the copular construction alone.9
In sum, I do not dismiss the clear linguistic evidence that human beings
can and very often do see conceptual connections between predicating loca-
tion of an object, possession of an object, and existence of an object – else
why would a Jakaltek or an Amharic root used at one point in time to
predicate one of these notions be adapted to predicate another of these no-
tions? I do, however, believe it is too reductionist to say that they are sim-
ply the same.
My concern in this paper is to explore usage patterns of lexical and con-
structional forms for expressing location and possession, so as to bring fur-
ther empirical evidence to bear on our understanding of how distinct these
conceptions may be for human cognizers. I conclude that the Maa evidence
points towards location being closely linked conceptually to existence; and
of possession being somewhat closely connected to existence. But the data
examined here present no evidence that possession and location are directly
linked conceptually. In Maa, they do not share any lexical or constructional
8
Ahland (2009) presents arguments that the PR in the Amharic structure in (6)
developed out of a formal locative phrase at some point in history. This does not
destroy the clear fact that speakers perceived the meaning of possession to be
sufficiently distinct from location that they did develop distinct constructions.
9
Clark (1978) does not give a full Amharic clause illustrating what she calls Posses-
sion-2, as she is concerned with reporting results from a large language sample.
Is possession mere location? Contrary evidence from Maa 117
4. Key definitions
10
Though there is important conceptual overlap, Kuna (2003) observes that
predicative possession is more precise or constrained semantically, while NP
possession is polysemous.
11
In a corpus study of Guaraní, Velazquez-Castillo (1996: 70–77) compared two
predicative structures used for expressing possession. She found that the primary
usage of a non-verbal predicate construction was to express intimate relation-
ships like ‘I have arms’; she also found that no instances of the construction
were used to designate spatial relations. A second verbal predicate construction
with reko ‘have’ was used to express that a PR is consciously in charge of or in
control of the PM. She argues that the latter structure can sometimes also place
emphasis on an unusual location when used with a body part, but this is not a
consistent meaning to the structure; rather, conscious control is the consistent
meaning. See McGregor (2001a, b) for similar distinctions in Nyulnyul. The
Maa ata Possessive construction described further below apparently covers both
types of meanings.
118 Doris L. Payne
12
Other scholars have come up with other lists, with differing numbers, of pos-
sessive subsenses (e.g. Langacker 1995; Heine 1997). My point here is just to
sketch some of the typical semantics as a basis for later evaluating the hypothe-
sis that Maa ata codes possession. The English predicate own is semantically
much narrower than have, being mostly restricted to subsenses (8) and (9). I have
excluded aspectual and modal auxiliary uses of have. Though the modal and
aspectual senses are historically connected to the possessive sense, the former
are not central to the core possession meaning.
Is possession mere location? Contrary evidence from Maa 119
(12) Predicate a part-whole relationship. (In their context of use, all the fol-
lowing are likely more concerned with predicating characteristics of
the possessors, than on the part-whole relationship.)
a. I have a herniated disk.
b. … and then I met a man who had no feet.
c. The better half thinks I have awful feet.
d. I have eyes with two different colors.
e. Do I have a big nose?
What holds all the preceding together is the predication of a control rela-
tionship (in (8)–(9)) or particularly intimate relationship (in (10)–(13)) be-
tween two entities. In English predicative possession with have, these two
entities are generally expressed as Subject (PR) and Direct Object (PM).
Possessor. Given the preceding exposition, I define Possessor (for predi-
cative possession) as the argument of a predicate or clause (a) presented as
a salient reference point relative to which the speaker identifies another par-
ticipant, and (b) which predicates it as having an intimate and/or controlling
relationship to another participant.
Agent is the argument of a predicate or clause which the speaker pre-
sents as doing or causing something; it corresponds to x in “What did x
do?” (Chafe 1970; Van Valin and LaPolla 1997).
120 Doris L. Payne
13
See Payne, Hamaya and Jacobs (1994) for further discussion of the Inverse-
Direct nature of the person prefixes, and Rasmussen (2002) for tonal behaviour
of these prefixes.
Is possession mere location? Contrary evidence from Maa 121
Transitive Intransitive
INVERSE DIRECT
P A 1SG 2SG 3 S
1SG áá- a- a-
2SG k
-
-
-
3 áà- k
- - -
Maa uses tone to mark two case distinctions on nouns and modifiers of
nouns in a marked-nominative typological pattern (König 2006). There are
multiple tone classes for nouns and adjectives, but each lexeme has two
tonal case forms. Postverbal subjects, objects of the oblique preposition t,
and vocatives occur in what Tucker and Mpaayei (1955) called the “Nomi-
native” tone. Other post-verbal NPs, all preverbal NPs regardless of gram-
matical relation, objects of the oblique Associative preposition , nouns and
adjectives in their citation form, and NP-internal PRs occur in their “Accu-
sative” tone pattern.
Though phrases can occur before the verb for specific pragmatic reasons,
the most neutral order is for clauses to begin with a verb. Thus, a basic
transitive or ditransitive clause will have the following structure (numeric
indices simply refer to relative order): [V (NP1) (NP2) (NP3)]. Because NPs
are marked for case in postverbal position, subject and object may vary in
order. In this structure, at most one NP can be Nominative, and this could
be either either NP1 or NP2/3. Invariably, NP1 is more discourse-topical than
NP2/3 (Payne, Hamaya and Jacobs 1994). Text data show that order of nom-
inals does not correspond to identifiability (definiteness) per se. Compare:
Tests for transitivity of a stem include its ability to take an Inverse prefix,
and Antipassive and Middle suffixes (which then derive an intransitive
stem). Thus, the following show that the root du is lexically transitive.
(16) Áà-dû.
3>1SG-cut
‘He/she/ they will cut me.’
Basic intransitive clauses have the structure [V (NP)]. Evidence that a stem
is intransitive includes its inability to take an Inverse prefix, a Middle, or an
Antipassive suffix. Compare:
(20) *Áà-d.
3>1SG-be.red
Another construction that will figure in our discussion is the Impersonal (or
“Impersonal Passive”). As Greenberg (1959) demonstrated, the Maa Imper-
sonal developed historically out of a third person plural suffix -
. This ap-
parently satisfied the subject requirement of the verb, and hence no free
Nominative NP could (nor can) co-occur in the sentence. Also, an Agent
cannot be expressed in the clause in an oblique phrase. Certainly in the
Maasai variety of Modern Maa, the Impersonal can occur with both transi-
tive and intransitive stems.
Is possession mere location? Contrary evidence from Maa 123
(24) -d-
.
3-be.red-PASS
‘People [in general, nonspecific] are red’ / ‘Being red happens’.
We will return below to what is the appropriate semantic role of ‘boy’ in (27).
124 Doris L. Payne
With this background, we now turn to the senses of tii and ata, which can-
not really be considered apart from the constructions that the roots occur in.
Thus, I will review senses of each of these roots according to construction.
We will see below that tii can occur in both the simple transitive and in-
transitive syntactic constructions. However, it cannot take the Antipassive
(28a), Middle (28b), or Inverse (29) affixes. These properties argue that it is
not a lexically transitive root.
In the [V NP1 NP2] construction, tii primarily has a locative sense, possibly
an existential sense, and in some contexts is possibly in the process of
grammaticalization as an imperfective aspect marker.
Tii most commonly predicates location of objects at literal physical
places. In (30), the river is Ground (or Locative) and the boy is Figure (or
Theme). This is a prototypical example of the predication of location. Cer-
tainly neither the river nor the boy have discretionary control over each
other. Additional examples follow, demonstrating a range of locative situa-
tions including metaphorical places. Note how relative order of Nominative
and Accusative phrases can vary.
The following is similar. Here the protagonist has traveled for some distance
to see some cattle, eventually locating the herd. The last line of this excerpt
states that a striped ox (which had consumed multiple people) is located /
exists within this herd:
126 Doris L. Payne
(35) ‘…until he has gone to see bulls far away. There he is, until he has
gone where they are,’
e-tí í knâ kíshú l-k
t
3-be.at these.FPL.ACC cattle.ACC MSG-cow.NOM
ó-írìm-ò …
MSG.REL.NOM-make.spots-MID.NPF
‘in this herd of cows there is a striped ox …’ (enamuke1.0012)
Another shade of meaning for tii in the [V NP1 NP2] construction is the
predication of metaphorical location. This is infrequent in the corpus data –
and it is in this point only, it seems, where the use of tii comes closest to
anything that could be thought of as possession. For example, in (36) one
might opine that the Maasai character “has” the good feature of generosity.
Note that the free translation below, using an English locative (not an English
possessive) was provided by a highly fluent bilingual speaker, perhaps sug-
gesting that he conceptualized it as closer to a locative than to English has.
(37) n-k
smá è-tí í ó l-mórùàk.
FSG-education.ACC 3-be.at ASSOC MPL-elders.ACC
‘They are learning with the elder.’
(lit: ‘They are at education with the elders.’) (bulunoto.070b)
in one written story, the writer recounted a famous event of when a British
District Commissioner (DC) seized cattle from a warrior and fought over a
particular bull that belonged to the warrior:
(38) a. -shm
l dìsí à-
b
l-móí
3-go.PF that.M.NOM DC.NOM SG.INF-seize MPL-oxen.ACC
l-nyná t èn-gólòn à-lòt-ú
M-3SG.ACC.PR.of.PL.PM OBL FSG-strength.NOM SG.INF-go-VENT
‘that DC went to take his oxen by force’
b. à-m
r t l-mnàndà t Mórìjò
SG.INF-sell OBL MSG-market.place.NOM OBL Morijo.NOM
Lóítà.
Loita.NOM
‘to come to sell them in the market place in Morijo Loita.’
c. N-é-m-é-tí í àpá
l mrràní
nâ
CN-NEG-3-be.at long.ago that.M.NOM warrior.NOM that.F.ACC
kátá
time.ACC
‘And that warrior was not (there) [in the market] at that time.’
(DC 6–8)
With the root tii, it is rare to express the Locative in an oblique t phrase,
rather than in a simple Accusative NP; indeed, this has been rejected in
elicited material. However, it is attested and three such instances occurred
in the 130-item sample from the corpus (section 8). In at least the following
example, the oblique t with ‘our home’ is conceivably motivated by the
fact that a Comitative oblique occurred before the Locative phrase, making
it more difficult for a comprehender to know how to interpret an otherwise
bare NP so structurally distant from the verb tii.
128 Doris L. Payne
14
Contra Gutierrez (2006) and others, existential constructions do not always “re-
quire locative complementation” in any overt way – clearly they do not in Maa.
Is possession mere location? Contrary evidence from Maa 129
A short time later, the same consultant listened to the following, and said
that this was a good way to begin a story, involving repetition of nétìí. With
tone sandhi and vowel coalesence, the first phrase sounds like [nétìápà].
In the corpus study reported further below, if the sequence nétìí àpá ap-
peared to have strictly a formulaic story-opening function as in (45), it was
130 Doris L. Payne
counted as such for the study. Otherwise, as in (43) and (44) above, it was
counted as an existential. Clearly, however, these two functions and con-
structions shade into each other historically.
7. Senses of Maa a ta
Unlike tii, ata is completely grammatical with Inverse prefixes and certainly
can take 1st and 2nd person objects. But like tii it does not take Antipassive
or Middle suffixes (46). The fact that it can take Inverse prefixes suggests it
is lexically (more) transitive. One hypothesis as to the reason why the Anti-
passive and Middle suffixes are not acceptable with ata may be because the
Aktionsart is stative, but this needs further investigation.
In contrast to tii, ata does predicate possession in the [V NP1 NP2] construc-
tion. It is attested with a range of possessive subsenses, as described in Sec-
tion 4 above. In the following examples, the woman is an individuated hu-
man who can exercise control over the Accusative NPs, to the extent of
selling or killing them.
The construction can also be used for predicating the experience of an event
or situation, relative to a participant.
(53) K--átà
l-k
shr l-shr.
DSCN-3-have MPL-Kishuru.NOM MSG-senior.elder.initiation.ACC
‘The Ilkishuru age-set has a senior-elder ceremony.’ (i.e., to make
them senior elders)
The closest example I have encountered of where the predicate ata could be
argued to express location of an object is the following elicited clause,
where the barrel could not (in most worlds) be said to have control over the
water; though ‘having water’ could certainly be a way of predicating a par-
ticular characteristic of a barrel. However, out of any particular discourse
context it is hard to say whether the speaker’s communicative intent would
be predication of a characteristic or of a location.
Finally, marginal uses of ata may include indicating something like value
or need, i.e. the possible beginning of modal sense. The data are so sparse
on this point that I have nothing further to say about it here.
As with tii, the Accusative argument of ata can be expressed with a
definite null when it is clearly known from the context. In the following,
the understood Accusative is ‘the story about the strap to heaven’, which
was mentioned two clauses previously. The person posing the question
comes from a different region of the Maa-speaking territory than the ad-
dressee, and is asking whether people in the il-Chamus region also ‘have’
this story:
For an example like (57), both arguments would be known from context. I
consider all such definite-null examples to also be instances of a transitive
construction.
(57) K-áà-àtà.
DSCN-3>1SG-have
‘He owns me.’
Is possession mere location? Contrary evidence from Maa 133
(59) N--àtá-
aké n
ny òl-òtùnó.
CN-3-exist-PASS still emphatic MSG-leader.of.age.set.ACC
‘There is still the leader of the age-set.’ [called Olotuno, the-one-
who-planted]. (aibartisho.004b)
The corpus study reveals a construction which is unusual for Maa, and to
my knowledge unique to the verb root ata. This is an intransitive clause
with a single Accusative argument, but without the Impersonal suffix -
8. Corpus results
The table incorporated into Figure 1 reports raw numbers based on a sample
of 130 tokens of ata and 130 tokens of tii, extracted from a larger text corpus
of approximately 20,000 clauses (containing mostly spoken but also some
written material, and a variety of genres). The original methodological intent
was to simply consider the absolute first 130 instances of each of these
roots; but due to some false starts or other lack of clarity among the first
130 instances, some of those examples were discarded and subsequent in-
stances were considered until a data set of 130 instances of each was com-
piled. I will henceforth refer to these 260 clauses as “the corpus”.
In the corpus, 47% of tii instances expressed existence, 47% expressed
location, and 6% expressed other senses including stative aspect; see Figure
1. There is no evidence whatsoever from this corpus that tii predicates pos-
session.
In the corpus, 65% of ata instances expressed existence, 34% expressed
possession, and 1% arguably expressed modal senses; see Figure 1. There is
no evidence from the corpus study that ata is used for predicating location.15
There is also essentially no evidence from elicitation that ata predicates loca-
15
Figure 1. Corpus distribution of ata and tii by sense (N = 130 for each root)
Figure 2 shows the constructions used for predicating location. The vast ma-
jority are the [V NP1 NP2] transitive construction with tii. The few instances
of the [V NPNOM] construction have definite null locations.
Here the -
suffix might be interpreted as referring to a plural possessive
subject, but the sense is almost existential. In sum, my conclusion is that
there is really only one predicative construction for indicating possession:
the ata-Transitive clause construction.
Where the two roots tii and ata do overlap is for predicating existence.
Figure 4 shows that the intransitive [tii NPNOM] (with no definite null locative
referent), and the intransitive Impersonal [ata -
NPACC] are the constructions
of choice for predicating existence. Though the intransitive [ata NPACC] con-
struction without the Impersonal suffix -
is robust, it is clearly less frequent.
Strikingly, for predicating existence the use of a surface transitive-like
construction with overt expression of a location (even as a placeholder of
sorts) is almost non-existent in the Maa corpus. This is notable, given the
clear attestation in so many other languages of the use of a locative phrase
in existential constructions – even if it is a semantic dummy, as in There is
an idea that democracy is the preferred political system, where there does
not refer to any location whatsoever. In Maa the near absence of any such
phrase is possibly because an abstract notion of Locative is lexicalized into
the root tii; so this language may ultimately not be particularly exceptional.
9. Conclusions
The corpus study, however, argues that ata and tii are in complementary
distribution for predicating possession versus location – unlike English have
or the French locative copular construction. The corpus data are supported
by the general study of senses-by-construction which also included elicited
material. If “Possessor” were always just a metaphorical extension of Loca-
tive, the non-overlap of these two roots for predicating location and posses-
sion would be surprising. In particular, if all “possession” is cognitively the
predication of (e.g., animate) Locative (Clark 1978: 89), then why would ata
have zero attestation for spatial locational uses? And if all “possession” is
cognitively the predication of location, then why would tii have zero at-
testation for possessive uses? The wheels of grammatical and lexical
change are by all accounts slow, so finding doubling of a form across two
functions (or senses) is normal and expected in the process of change. The
lack of doubling – either for the roots or for constructions – suggests the
lack of cognitive “sameness” or identity about possession and location.
The apparent non-overlap between these two roots suggests that Posses-
sor is indeed a salient relation that Maa speakers distinguish from Locative.
Even where there are conceptual linkages that motivate sharing certain
lexical roots or pieces of the morphosyntax (in constructional terms, “in-
heritance” of subconstructions; cf. Heine 1997), distinct constructions do
emerge via the accretion and stablization of extra morphemes, and become
grammaticalized. The concept of possession (and hence Possessor) is es-
sential in explaining the specialization of both dedicated linguistic con-
structions and senses of forms.
Finally, what merits closer cross-linguistic and historical investigation is
whether human beings in general do not find closer conceptual links be-
tween possession and existence, and between location and existence, than
between possession and location. Clark (1978: 113) herself noted that loca-
tion and existence more often share the same verb root, than does either
existence with possession, or location with possession. With reference to
Maa, it could be that further usage-based research might show overlap in
sense between tii and ata. However, since the [tii NPNOM] existential con-
struction is intransitive, I would not expect it to directly develop into a pos-
sessive construction since possession necessarily involves predicating a
relationship between two entities. Similarly, given that both the [ata-
Acknowledgements
References
Ahland, Michael
2009 From topic to subject: grammatical change in the Amharic possessive
construction. Studies in Language 33 (3): 685–717.
Baron, Irene and Michael Herslund
2001 Semantics of the verb HAVE. In Dimensions of Possession, Irene
Baron, Michael Herslund and Finn Sørensen (eds.), 85–98. Amster-
dam: John Benjamins.
Clark, Eve V.
1978 Locationals: existential, locative and possessive constructions. In
Universals of Human Language, Vol. 4: Syntax, Joseph H. Greenberg
(ed.), 85–126. Stanford: Stanford University Press.
Chafe, Wallace
1970 Meaning and the Structure of Language. Chicago: University of Chi-
cago Press.
Comrie, Bernard and Helma van den Berg
2006 Experiencer constructions in Daghestanian languages. In Semantic
Role Universals and Argument Linking: Theoretical, Typological, and
Psycholinguistic Perspectives, Ina Bornkessel, Matthias Schlesewsky,
Bernard Comrie and Angela Friederici (eds.), 127–154. Berlin /New
York: Mouton de Gruyter.
Craig, Colette Grinevald
1977 Jacaltec: The Structure of Jacaltec. Austin: University of Texas Press.
DeLancey, Scott
2002 The universal basis of case. Logos and Language 1 (2): 1–15.
Freeze, Ray
2001 Existential constructions. In Language Typology and Language Uni-
versals: An International Handbook, Martin Haspelmath (ed.), 941–
953. Berlin /New York: Mouton de Gruyter.
140 Doris L. Payne
Goldberg, Adele
1995 Argument Structure Constructions. Chicago: University of Chicago
Press.
Greenberg, Joseph
1959 The origin of the Masai passive. Africa 29: 171–176.
Gruber, Jeffrey
1965 Studies in lexical relations. Ph.D. dissertation. MIT. [Reprinted 1976,
as part of Lexical Structures in Syntax and Semantics. Amsterdam:
North Holland.]
Gutierrez Morales, Salome
2006 Morphosyntactic expressions of possession and existence in Sinhala.
In Proceedings from the Workshop on Sinhala Linguistics (Santa
Barbara Papers in Linguistics 17), Robert Englebretson and Carol
Genetti (eds.), 20 –28.
Heine, Bernd
1997 Possession: Cognitive Sources, Forces, and Grammaticalization.
Cambridge Studies in Linguistics. Cambridge: Cambridge University
Press.
Heine, Bernd
2001 Ways of explaining possession. In Dimensions of Possession, Irene
Baron, Michael Herslund and Finn Sørensen (eds.), 311–328. Amster-
dam: John Benjamins.
Jackendoff, Ray
1983 Semantics and Cognition. Cambridge, MA: MIT Press.
Kemmer, Suzanne
2002 Human cognition and the elaboration of events: some universal con-
ceptual categories. In The New Psychology of Language: Cognitive
and Functional Approaches to Language Structures, Michael
Tomasello (ed.), 89 –118. Mahwah, NJ: Lawrence Earlbaum.
König, Christa
2006 Marked nominative in Africa. Studies in Language 30: 655 –732.
Kuna, Branko
2003 Izmedu atributne i predikatne posvojnosti. (Between attributive and
predicative possession.) Rasprave Instituta za Hrvatski Jezik i Jezi-
koslovlje 29: 157–171.
Lamoureaux, Siri van dorn
2004 Applicative constructions in Maasai. MA thesis, University of Oregon.
Langacker, Ronald
1993 Reference-point constructions. Cognitive Linguistics 4: 1–38.
Langacker, Ronald
1995 Possession and possessive constructions. In Language and the Cog-
nitive Construal of the World, John R. Taylor and E. MacLaury (eds.),
51–79. Berlin /New York: Mouton de Gruyter.
Is possession mere location? Contrary evidence from Maa 141
Lyons, John
1967 A note on possessive, existential, and locative sentences. Foundations
of Language 3: 390 –396.
Lyons, John
1968 Introduction to Theoretical Linguistics. Cambridge: Cambridge Uni-
versity Press.
McGregor, William B.
2001a Non-verbal predicative possession in Nyulnyulan languages. In
Forty Years On: Ken Hale and Australian Languages, Jane Simpson,
David Nash, Mary Laughren, Peter Austin and Barry Alpher (eds.),
337–352. Canberra: Pacific Linguistics.
McGregor, William B.
2001b The verb HAVE in Nyulnyulan languages. In Dimensions of Posses-
sion, ed. by Irène Baron and Michael Herslund, 67–84. Amsterdam:
John Benjamins.
Mol, Fr. Frans
1996 Maasai Language and Culture Dictionary. Limuru, Kenya: Kolbe
Press / Maasai Centre Lemek.
Payne, Doris L.
1997 The Maasai external possessor construction. In Essays on Language
Function and Language Type, Joan Bybee, John Haiman and Sandra
Thompson (eds.), 395–422. Amsterdam: John Benjamins.
Payne, Doris L. and Immanuel Barshi (eds.)
1999 External Possession. Amsterdam: John Benjamins.
Payne, Doris L., Mitsuyo Hamaya and Peter Jacobs
1994 Active, passive, and inverse in Maasai. In Voice in Discourse, T. Givón
(ed.), 283–315. Amsterdam: John Benjamins.
Rasmussen, Kent
2002 Tone in il-Keekonyokie Maa. MA Thesis, University of Oregon.
Sørensen, Finn
2001 Possession spaces in Danish. In Dimensions of Possession, Irene
Baron, Michael Herslund and Finn Sørensen (ed.), 57– 66. Amster-
dam: John Benjamins.
Tham, Shiao Wei
2005 Representing possessive predication: semantic dimensions and
pragmatic bases. Ph.D. dissertation, Stanford University.
Tucker, Archibald N. and John T. Mpaayei
1955 Maasai Grammar with Vocabulary. London: Longmans, Green & Co.
Van Valin, R. and R. LaPolla
1997 Syntax: Structure, Meaning and Function. Cambridge: Cambridge
University Press.
Velazquez-Castillo, Maura
1996 The Grammar of Possession: Inalienability, Incorporation and Pos-
sessor Ascension in Guarani. Amsterdam: John Benjamins.
142 Doris L. Payne
Wertheimer, Max
1923 Untersuchungen zur Lehre von der Gestalt II. Psycologische For-
schung 4: 301–350. [Translation published in W. Ellis. 1938. A Source
Book of Gestalt Psychology, 71–88. London: Routledge and Kegan
Paul.]
Learning to encode possession
1. Introduction
Learning and talking about their own possessions and the possessions of
their peers and caretakers plays a central role in children’s daily life. It is un-
surprising then that relationships between possessors and their possessions
are amongst the first relationships that children encode when they start to
string words together (see e.g. Brown 1973); and it is no wonder that many
psycholinguists have made use of this rich data source to address questions
about the mechanisms that drive children’s linguistic development.
However, most of the available studies of the acquisition of possessive
constructions that we will discuss have investigated only one or two posses-
sion-encoding constructions in an individual language. Moreover, the focus
has typically not been on the encoding of the possessive relation itself, but
on other aspects of the respective possessive construction. For instance,
possessive -s markers in German and English (e.g. Susi-s Huhn ‘Sue’s
chicken’) were analysed in studies that investigated whether the syntactic
categories of the target language were already present in early child gram-
mars (e.g. Eisenbeiß 2000; Marinis 2002, 2003; Radford 1990). In these
studies, possessive markers were simply treated as morpho-syntactic reali-
sations of syntactic categories; and semantic aspects were largely ignored.
Similarly, possessive constructions with two-place verbs like have and be-
long were investigated in studies of the acquisition of syntax-semantic
mappings, but these constructions were just treated as one type of two-
argument construction and not compared to other constructions encoding
possession (see e.g. Bowerman 1985; Pinker 1984). To our knowledge, no
study has yet provided a comprehensive cross-linguistic overview that fo-
cuses on the different ways in which possessive relationships are encoded
linguistically.
In order to fill this gap, we will provide a cross-linguistic overview of
studies of children’s acquisition of the constructions that their target lan-
guage employs to encode possession. In addition, we will present new data
from German child language and child-directed speech, and discuss the im-
144 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
on linguistic aspects, i.e., we will investigate how children acquire the pos-
sessive constructions of their target languages – and what this can tell us
about the mechanisms that drive children’s linguistic development.
Current research on the mechanisms underlying children’s language ac-
quisition is characterised by an opposition between generative approaches
(see Eisenbeiß 2009 for overview) and functionalist or usage-based ap-
proaches (Bates and MacWhinney 1987; Tomasello 2003, 2006; Goldberg
2006). Moreover, acquisition researchers have provided competing ac-
counts for the time course of linguistic development and for the orders in
which children acquire the properties of their target language. In the fol-
lowing, we will first provide an overview of these debates before we pre-
sent data on the acquisition of possession constructions that can help us
evaluate competing models.
they overcome these “errors”, and how they learn to use morpho-syntactic
forms appropriately.
One could argue that children “unlearn” errors on the basis of negative
evidence, i.e. information about the ungrammaticality of their utterances.
However, many studies have shown that explicit corrections are not system-
atically available to all children at all developmental stages (Marcus 1993).
Moreover, even for explicit corrections such as You can’t say that, it is not
obvious whether the correction refers to the phonological or morpho-
syntactic structure, the use of lexical elements, or the appropriateness of the
utterance in the social context. In addition, children do not always take up
corrections – and even if they do seem to take them up, they might later go
back to their non-target like structures; see Marcus (1993) for an overview
and the following example from Simone Miller (2;4,1 see Miller 1976; Eisen-
beiß 2003: 45):2
1
Age information is provided in the following format: Year; Month.
2
We have used the following abbreviations and glosses: [#] – pause; [ /] – inter-
ruption; ACC – accusative; DAT – dative; D-elements – case /gender /number-
marked articles, possessive pronouns, demonstratives, quantifiers; EPC – external
possession construction (I hit him on the head); FEM – feminine; GEN – genitive;
IPC – internal possession construction (I hit his head); MASC – masculine;
NEUT – neuter; NOM – nominative; PART – particle (note that as it is difficult
to provide exact translations for German focus and other particles, particles will
simply be glossed as PART, without further information); PM – Possesssum; PR –
Possessor; SG – singular; and TAG – tag question.
Learning to encode possession 147
(2) a. Look there are so many mice in this picture: There is one mouse
under the table and one mouse under the chair and two mice un-
der the bed and three mice in the corner.
b. Child: There are the two mouses again!
Adult: Yes, we have seen these two mice before.
150 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
markers are not uncommon in natural languages – what is more, they occur
in English, which has adnominal possessive constructions that are otherwise
very similar to the corresponding German constructions.
Thus, when we discuss the acquisition of possessive constructions in the
following sections, we will address the logical problem and explain how
children manage to avoid or overcome deviations from the target language
without recourse to reliable negative evidence. In addition, we also have to
capture the time-course of children’s linguistic development, i.e. explain
why children acquire particular properties of their target language at a par-
ticular time and in a particular order.
Rather, any deviations from the target must be attributed to other factors;
and Full-Competence proponents have suggested a number of such factors,
e.g. the late maturation of general cognitive constraints, problems with the
morphological or phonological realization of unstressed morphemes and
underdeveloped pragmatic knowledge (see Guasti 2002; Eisenbeiß 2009 for
overviews).
In contrast to such Full-Competence approaches, Structure-Building ap-
proaches do not assume adult-like representations for the early two-word
stage. Hence, they must account for children’s early grammatical represen-
tations and explain how children acquire adult-like representations. Most
current Structure-Building approaches adopt versions of the Lexical Learn-
ing Hypothesis (see Pinker 1984; Eisenbeiß 2007, 2009 for overviews; see
Radford 1990 for an earlier maturational approach). According to this hy-
pothesis, children have adult-like categorisation abilities, but they still need
to determine the grammatical features and properties of the input elements
they encounter. Recent versions of Lexical-Learning approaches assume
that this process is incremental: grammatical distinctions are acquired one
by one, lexeme by lexeme, and with initial restrictions of inflections to in-
dividual lexemes that frequently appear with these markers in the input (see
e.g. Eisenbeiß 2003, 2007). For instance, possessive markers should initially
be restricted to individual words – and only later generalised to all words or
phrases that can carry this marker in the target language (Eisenbeiß 2000).
Usage-based approaches make similar predictions as Lexical-Learning
approaches, though on the basis of slightly different assumptions. Accord-
ing to them, adults grammars are based on schemas or constructions, i.e.
interrelated form/meaning pairs that are characterized by various degrees of
abstractness, ranging from idioms with concrete lexical items (e.g. kick the
bucket ‘die’), to abstract templates characterized by grammatical roles such
as subject-predicate (e.g. Goldberg 1995, 2006). Children are assumed to
acquire such templates step by step; beginning with limited generalisations
that are centred on individual words or phrases and then gradually extending
these generalizations by analogy. For instance, Tomasello (2003, 2006) ar-
gues that grammatical morphemes such as case markers and agreement
markers are initially associated with individual (high frequency) verbs.
Only when a critical mass of such “verb-islands” is learned do children ac-
quire more general constructions, e.g. the transitive construction.
Thus, taken together, both generative and usage-based approaches pre-
dict that children acquire the general properties of the possessive construc-
tions that they hear in their input already in the two-word stage. Moreover,
structure-building approaches as well as usage-based approaches specifi-
Learning to encode possession 153
3
For instance, Radford (1996) argues that morpho-syntactic realizations of the
functional category COMP (complementizers, wh-elements, etc.) are acquired
later than realizations of the functional category INFL (tense and agreement in-
flections, etc.). Moreover, some generative psycholinguists argue that the ability
to produce adult-like passive sentences only develops around the fourth birthday,
due to neural maturation of the underlying mechanisms (e.g. Borer and Wexler
1987). See Eisenbeiß (2009) and Tomasello (2003) for critical discussion.
154 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
their body part. However, in the EPC, both relations have to be mapped
onto a single argument hierarchy as both PR and PM are realised as verb
arguments. By contrast, in IPCs, the Agent-Patient relation is realised on
the sentential level while the possessive relationship is encoded within the
Patient noun phrase. We will argue that this is conceptually simpler than the
integration of agentive and possessive relations in the EPC – which should
be reflected in acquisition orders and possibly also in deviations from the
target language.
STAGE I
Children frequently omit possessive markers and other morphological
markers. They also omit D-elements, i.e. function words in the noun
Learning to encode possession 155
4. Adnominal possession
PR phrase has to follow the PR in English as in (3b, d), and it tends to ap-
pear in the same position in German as in (3b, d). However, in German, at
least some speakers find it acceptable to position the prepositional PR
phrase to the left of the PR for emphasis as in (3e) or to extract the preposi-
tional PR phrase in a question – see (3f). Note, however, that structures
where the prepositional phrase does not follow the PR are highly marked –
if not unacceptable.
In addition to the adnominal possession constructions in (3a) to (3f),
German has two constructions that do not have an equivalent in English.
The first one is a genitive construction, where the PR noun phrase typically
follows the PM as in (3g), but may also precede it as in (3h). This genitive
construction is more characteristic of formal and written German, has so far
not been observed in the speech of pre-school children (Clahsen et al. 1994;
Eisenbeiß 2000; Mills 1985), and does not occur in the corpora we analysed.
Note that the genitive construction is different from the possessive -s con-
struction though this is not immediately obvious for masculine nouns that
take -s as their genitive ending (see (3g, h)). The difference between the -s
possessive and the “real” genitive can be seen when one looks at masculine
nouns with a different genitive ending or femine nouns, which do not carry
any overt marker in the genitive (see (3i)).
The second German possessive construction without an English equiva-
lent appears in some spoken variants of German and involves a PM that is
preceded by a dative-marked PR and a resumptive possessive pronoun, as
in (3j). Due to its regional character, this construction is rarely discussed in
the acquisition literature (see Penner and Weissenborn 1994 for some initial
observations), and we only found three instances of this type in the data of
the German boy Carsten that we analysed. Therefore, we will not discuss
adnominal genitive and dative constructions in a lot of detail.
Given the range of constructions mentioned above, both German and Eng-
lish children have to learn (i) when to use pronominal or a non-pronominal
constructions and (ii) when to use a construction with a non-prepositional
PR rather than a construction with a PR-PP. Moreover, German children
have to acquire adnominal constructions with genitive PRs or a combination
of dative PRs and possessive pronouns – though we won’t be able to inves-
tigate this aspect of the acquisition process due to a lack of relevant data.
With respect to the choice of pronominal vs. non-pronominal construc-
tions, German and English are similar: possessive pronouns are preferred
when the PRs can be identified on the basis of contextual or discourse in-
formation. As this is particularly easy for speakers and hearers, adults tend
to use 1st and 2nd person possessive pronouns rather than names when they
talk to other adults (e.g. mein-e/dein-e Henne ‘my/your chicken’). However,
in children’s early language and in language directed at young children, we
can often observe the use of names and kinship terms instead of 1st and 2nd
person possessive or personal pronouns (e.g. Ruff 2000). For instance, a
mother might tell her daughter Jane: Das ist Jane-s Auto; und das ist
158 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
Mama-s Auto ‘this is Jane’s car; and this is mommy’s car’. Thus, we will
have to investigate when and for which types of PRs children use posses-
sive pronouns.
Whether a prepositional or a non-prepositional construction is chosen,
depends on a range of factors. The first factor is the syntactic status of the
PM: if it is an unmodified noun such as the proper name Paul, it can either
appear with a prepositional PR as in (3d) or with a non-prepositional PR as
in (3c). However, possessive pronouns as well as -s-marked PRs induce a
definite reading and cannot co-occur with determiners or wh-elements
(*ein / dieser/welcher Pauls/ mein Freund ‘a/ this / which Paul’s / my friend).
Thus, the use of a wh-question or an intended indefinite reading may require
a prepositional construction. We will now look at factors that play a role if
the choice of construction is not already determined by the type of PR.
For pronominal PRs, English or German speakers have a choice between
a possessive pronoun construction as in (3a) and a prepositional construction
with a pronoun as in (3b); the possessive pronoun construction is typically
preferred as PR phrases tend to proceed the PR, especially when they are
short, animate and topical and the PR is not used contrastively, see e.g. my
friend vs. ?the friend of mine.
For non-pronominal PRs, English and German differ with respect to the
factors that determine their choice of construction. In English, -s can attach
to non-pronominal PR phrases of any syntactic complexity (e.g. Jack’s/the
old farmer’s chicken farm). Similarly, prepositional constructions can in-
volve simple PR nouns or complex PR noun phrases (e.g. a teacher of Jane/
my little daughter). Thus, for non-pronominal PR phrases, English speakers
have a choice between -s and of. This choice is determined by the animacy,
topicality and syntactic weight of the PR and by the type of possessive rela-
tionship: -s is preferred for prototypical and inalienable possessive relations
and when the PR is animate, topical and short (e.g. Sue’s eyes) – whereas of
is preferred when the PR is inanimate, not topical and syntactically modi-
fied and the relationship between PR and possesum is not a close and proto-
typical possessive relation (e.g. the fumes of a shabby old car; see e.g.
Rosenbach 2002, 2005, 2008; Jäger and Rosenbach 2006; Denison, Scott,
and Börjars 2008 and references cited there for the discussion of these fac-
tors and their interaction). Note, however, that while these factors can con-
spire to make one construction highly preferable over the other, none of
these factors on its own can determine the choice of construction.
By contrast, German exhibits a constraint for the use of possessive -s
that cannot be violated and hence can uniquely determine the choice of
non-pronominal possessive construction: -s can only be combined with PR
Learning to encode possession 159
nominals that lack an article or any other modifiers. Typically, these nouns
are proper names like Susi, but some kinship terms can also be used with-
out a determiner and can thus be combined with -s (e.g. Mamas/Papas Auto
‘mommy’s / daddy’s car’). This has led some linguists to distinguish posses-
sive -s markers from genitive markers by calling them “proper name posses-
sive markers” (see e.g. Harbert 2007: 161ff.). We will argue that the con-
straint on the use of -s is a syntactic constraint and not a restriction of -s to
a particular semantic class of nouns. In many regional variants of colloquial
German, kinship terms or proper names appear with determiners (e.g. die/
eine Mama ‘the/a daddy’ or die Emma ‘the Emma’). However, when such
phrases are used as PRs, -s cannot be used (e.g. *der Mamas/ Emmas Auto
‘the mommy’s/Emma’s car’). Rather, a prepositional construction is chosen
(e.g. das Auto von der Mama/ Emma ‘the car of the mommy/ Emma). Thus,
it is not the type of noun per se that determines whether -s can appear, but
the lack vs. presence of determiners or other modifiers.
Given the syntactic constraints for -s, German speakers can only choose
between -s and the prepositional von-construction when the PR is an un-
modified proper name such as Emma or name-like kinship term such as
Mama ‘mommy’. As the referents of these nouns are all animate and the
length of the PR phrase is limited to one word, animacy or syntactic weight
cannot determine the choice between an -s construction like Annas Auto
‘Anna’s car’ and a prepositional construction like das Auto von Anna ‘the
car of Anna’. However, the type of possessive relationship might play a
role, with -s being preferred for closer and more prototypical possessive
relationships. To our knowledge, this has not yet been investigated thor-
oughly.
To summarise, German and English both have possessive constructions
with pronominal and non-pronomial PRs; and the choice of pronominal vs.
non-pronominal PRs is determined by similar pragmatic factors. Moreover,
in both languages, possessive pronouns are preferred to prepositional
phrases with pronominal PRs – unless the PR requires a modifier. However,
in German and English non-pronominal possession constructions, additional
factors play a role: the choice between -s and prepositional constructions
for non-pronominal PRs is determined by semantic and discourse factors in
English, whereas the use of the German possessive marker is restricted to
particular syntactic environments – i.e. unmodified nominals. Thus, a choice
between -s and the prepositional construction is only available for proper
names and a few kinship terms that can appear without a determiner.
Not all languages show such a competition between a prepositional con-
struction and a construction with a possessive marker. For instance, in Japa-
160 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
nese, all PR phrases, whether they are personal pronouns, unmodified nouns
or more complex noun phrases, are marked by the postposition no and pre-
cede the PM (e.g. watashi/Toshiko no kuruma ‘I/Toshiko’s car). In Hebrew,
the PR is marked by the preposition shel and follows the possesum (e.g. pe
shel buba ‘(the) mouth of (the) doll’). The PR < PR order can also be ob-
served in constructions with possessive pronouns (e.g. hasefer sheli ‘the
book (of) my/mine’; see e.g. Armon-Lotem, Crain, and Varlokosta 2005).
Even in a language with only one possessive marker for non-pronominal
constructions, children may have to acquire different word order patterns:
For instance, in Standard Modern Greek, the genitive-marked PR is com-
patible with determiners for the PR and can either precede the PM as in
(4a) or it can follow it as in (4b); see Marinis (2002, 2003):
In the following, we will first focus on the developmental problem and in-
vestigate when children start to produce the adnominal possessive construc-
tions of their target language and whether their earliest uses of these con-
structions are restricted to particular lexical items – as predicted by
structure-building and usage-base approaches. Then, we will study whether
children show early sensitivity to the language-specific constraints that
govern the choice of construction. Against this background, we will then
try to provide an account for the order in which adnominal possessive con-
structions are acquired and used to encode different types of possessive re-
lations.
Learning to encode possession 161
Pronominal PR Non-Pronominal PR
Stage Child possessive von + PR+PR PR+PR von + NP Total
pronoun pronoun w/o -s w. -s
Ann 6 – 1 – – 7
Han – – – – – 0
I Leo 0 – 17 – – 17
Mat 3 – 3 – – 6
Total 9 (30%) 0 (0%) 21 (70%) 0 (0%) 0 (0%) 30
Ann 3 – 1 – – 4
Han – – – – –
II Leo 1 – 6 1 – 8
Mat – – – – – –
Total 4 (33%) 0 (0%) 7 (58%) 1 (8%) 0 (0%) 12
And 63 – 2 2 – 67
Ann 15 – 3 – – 18
Han – – – 1 – 1
III
Leo 17 – 2 20 – 39
Mat 7 – – – – 7
Total 102 (77%) 0 (0%) 7 (5%) 23 (17%) 0 (0%) 132
Ann 25 – – – – 25
Car 105 1 – – 4 110
Han 4 – – 1 1 6
IV Leo 26 – – 7 – 33
Mat 31 – – 5 1 37
Sve 75 1 – 17 5 98
Total 266 (86%) 2 (<1%) 0 (0%) 30 (10) 11 (4%) 309
Total 381 (79%) 2 (<1%) 35 (7%) 54 (11%) 11 (2%) 483
This tabulation suggests that there might be an early stage without adnomi-
nal possessive constructions: Hannah does not produce any of these con-
structions in stages I and II and only one proper name possessive construc-
tion in stage II. In stage I, she mostly labels or points out objects and
actions or asks for things, but does not talk about possessions. From stage
II on, possessive relations become a talking point for her and the possessive
162 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
4
Recording information is provided in the following format: (Name/number of
recording /age).
Learning to encode possession 163
The observed shift towards pronominal PRs cannot simply be due to chil-
dren initially preferring their own name for self-reference: of the seven
children, only Annelie, Hannah, and Leonie ever employ their own name as
PRs; and Annelie and Leonie use both the possessive pronoun and their
own name as soon as they talk about themselves as PRs. Hannah produces
her name only once in an adnominal possessive construction (stage II) be-
fore she starts using both the possessive pronoun and her own name (stage
III). But recall that she already used the possessive pronoun in one-word
utterances in stage II. Thus, those children who use their own name for self-
reference in adnominal possessive constructions do not seem to do so be-
cause they have not yet acquired pronouns. Rather, they exhibit an alterna-
tion between pronouns and their names in these constructions.
As we do not have video data available to study the details of the situ-
ational context, we cannot investigate all factors that determine when a pos-
sessive pronoun is chosen. However, an alternation between pronominal
and non-pronominal references to the speaker as PR has also been observed
by Ruff (2000) who argues that possessive pronouns are initially used in
demands and conflict situations, whereas non-pronominal PRs appear in
descriptive utterances. Moreover, Ruff reports that German children use pro-
nouns more frequently when the speaker is the PR than when the addressee
is the PR.
Taken together, we observe an incremental development of possessive
constructions, with a late acquisition of prepositional constructions and a
164 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
An early stage with omissions of possessive markers has also been ob-
served for other German children and in studies of the corresponding pos-
sessive markers in English, Greek, Hebrew and Japanese (Clahsen, Eisen-
beiß and Vainikka 1994; Penner and Weissenborn 1996; Brown 1973;
Radford 1990; Radford and Galasso 1998; Marinis 2002, 2003; Berman
1985; Armon-Lotem 1998; Clancy 1985). Proponents of full-competence
approaches have tried to show that possessive markers appear early even if
they are not always realised. For instance, Bohnacker (1997) argued that
possessive markers appeared in the data of a young Swedish child (Embla;
age: 1;8–2;1). However, the entire corpus only contained 14 possessive
markers; and none of these occur at the beginning of the recording period
(Eisenbeiß 2003). Still, it is not clear whether all children in all languages go
through a stage in which possessive markers are omitted. Note, however,
that structure-building and usage-based approaches do not in principle rule
out the possibility that possessive markers appear early – they simply allow
for the gradual and incremental acquisition of such markers and the possi-
bility of an early, marker-less stage. In fact, for languages where the posses-
sive marker is frequent, obligatory, salient (e.g. syllabic) and not homony-
mous with any other marker, proponents of structure-building and usage-
based approaches would expect this marker to be acquired quite early.
Learning to encode possession 165
One could attribute the early omissions of possessive markers and the initial
lexical restrictions to non-syntactic factors, e.g. to the fact that possessive
markers are typically unstressed and that some combinations of PRs and
possessive markers might be more difficult to hear or pronounce. However,
Eisenbeiß (2000, 2003) has argued that phonological factors alone cannot
account for the distribution of possessive markers in the corpora we ana-
lysed. For instance, the word Papa ‘daddy’, which Andreas uses with -s, and
the word Mama ‘mommy’, which appears without -s, have the same syllable
structure and the same final vowel.
Moreover, a distributional analysis of the possessive pronouns in our
corpora offers further support for the assumption of initial lexical restric-
tions. In stages I and II, only Annelie and Mathias produce more than one
possessive pronoun in an adnominal possessive construction; and the 9 in-
stances of possessive pronouns in Annelie’s early data only involve 5 high-
frequency nouns (puppe ‘doll’, mama ‘mommy’, schuhe ‘shoes’, zimmer
‘room’, and bilder ‘pictures’). Similarly, in stages I and II, Mathias only
produces 3 types of adnominal possessive constructions with the possessive
pronouns. Finally, all possessive pronouns in stages I and II are forms of
mein ‘my’; forms of dein ‘your’, unser ‘our’ or sein ‘his’ as in (9) do not
occur until stages III and IV.
At the same time, agreement with the PR is not target-like: except for one
use of the correct plural form meine in stage II (meine bilder ‘my pictures’),
Annelie only uses the uninflected form mein, which leads to agreement er-
rors for feminine and plural nouns that require the form meine (e.g. *mein
mama ‘my mommy’). Mathias produces the uninflected form mein once, in
Learning to encode possession 167
While the previous section showed quite a few studies of the time course of
the acquisition process in the domain of adnominal possessive constructions,
there are fewer studies of the constraints that govern the use and choice of
these constructions, which is somewhat surprising given the rich theoretical
literature on the topic. We have already discussed the choice between pro-
nominal and non-pronominal constructions above (see Ruff 2000 for more
details).
With respect to non-pronominal possessives, we observed earlier that
the possessive marker -s is restricted to unmodified PR nouns in German –
typically proper names like Susi, and some kinship terms (e.g. Mamas/Papas
Auto ‘mommy’s/daddy’s car’). Thus, for modified PR nouns (e.g. das
kleine Huhn ‘the (little) chicken’), German speakers have to use a preposi-
tional or genitive construction (e.g. der Kopf von dem kleinen Huhn or der
Kopf desGEN kleinenGEN HuhnsGEN ‘the head of the little chicken’). This
raises questions with respect to German child language: as we observed
earlier, prepositional and genitive constructions do not appear in our data in
stages I– III and we only found a few prepositional constructions in some
files from late stage IV. At the same time, noun-modifying determiners
emerge in stage III and appear in nearly all obligatory contexts in stage IV.
Thus, what do young German children do when the PR is a common noun
that requires a determiner or some other modifier – and is thus incompatible
with -s? In principle, they could refrain from using any non-pronominal ad-
nominal possessive constructions with PRs that require a modifier until
they have acquired the prepositional or genitive constructions of their target
language, but – as we will see – that does not seem to be what happens.
Thus, children could do one of three things. The first option is to use an -s
possessive but to omit the required noun modifier. This would not violate
the German constraint for -s, though it would mean treating a common noun
like a proper noun. As reported in Eisenbeiß (2000), two of the children we
investigated exhibited this option: Svenja produced one and Leonie pro-
168 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
While this is clearly a very small set of examples, more can be found in the
literature: see (11) for -s overgeneralizations reported by Mills (1985: 185).
What is striking about this set of examples is that it contains a plural noun,
which can definitely not be interpreted as a proper name. Thus, children do
not seem to violate the constraint on the affixation of -s to modified nouns.
They DO violate the requirement to use determiners with count nouns.
Note, however, that some German speakers who we showed the examples
to commented that adnominal possessive constructions such as affes ba-
nane are acceptable if one analyses the respective PR noun as a proper
name, which would not require a determiner in German.
The second option for children faced with the dilemma of either violating
the -s or the determiner constraint would be to use a modified PR and omit
-s, thus producing a construction that exhibits a modifier, but does not show
the required morphological marking. We do not have a lot of evidence for
this option, only one example, which we did not include in our counts as
the interruption in it makes it difficult to interpret:
What is striking, however, is that we did not find any evidence for the third
option: overgeneralisation of -s to a modified PR noun, i.e. a violation of
the constraint on -s that would avoid treating the PR noun like a proper
noun. That is, we did not observe any “English-style” constructions such as
mein Mamas bademütze ‘my mommy’s bathing cap’. This is quite surpris-
ing as children DO obviously overgeneralise -s. Moreover, most of the
nouns that the children in our sample correctly use as -s-PRs could be
modified in the target language because kinship terms and even proper
nouns may be combined with determiners or some other modifiers in most
varieties of spoken German – even though they do not HAVE TO. And in-
deed all children in our sample that produce -s-PRs provide modifiers for
the nouns they use as -s PRs when they appear in other contexts: Andreas
affixes -s to papa ‘daddy’ when it is an unmodified PR (as in (8c) above),
but he combines this noun with a possessive pronoun in other contexts (as
in (13a)). Hannah only uses her own name in an -s possessive construction,
but at the same time she mostly combines her name with determiners in
other contexts, see e.g. (13b). Mathias only uses unmodified proper nouns
with -s, but he produces combinations of these nouns with determiners in
other contexts (e.g. (13c)). Leonie produces -s overgeneralisations from
stage III on, but the PR nouns she uses with -s appear without a determiner,
though they are used with modifiers in other contexts, as in (13d). The PR
nouns that appear with an overgeneralized -s marker in stage III and IV,
occur with determiners during the same stage, as in (13e). Svenja also com-
bines -s with nouns that she otherwise uses with articles and other modifiers,
see e.g. (13f), including the common noun junge ‘boy’, which she incor-
rectly affixes with -s. Moreover, Svenja is one of those children that already
show a contrast between the -s construction and the possessive construction;
and we find minimal pairs with an unmodified proper name or kinship term
in the -s construction and the same name with a modifier in the preposi-
tional construction of (13g, h).
170 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
restrict the use of -s to proper names and kinship terms and overgeneralise -s
to modified proper names and kinship terms. We argue that children who
hear -s with single nouns would only assume that -s can also be combined
with more complex noun phrases if they found positive evidence for this in
the input. Thus, children acquiring English would learn to use -s with com-
plex noun phrases because they hear such combinations, whereas German
children would never have any evidence that would lead them to extend the
use of -s in this way. This means that German children do not have the
means to produce any adnominal possessive constructions with modified
PR nouns before they have acquired prepositional or genitive constructions.
This would leave them only one option: to treat the common PR nouns they
want to use as if they were proper names. Note that treating a common
noun like a proper name is not that uncommon for nouns with animate ref-
erents. For instance, we know quite a few adults who use words such as
baby and cat as proper names. Thus, children’s overgeneralisations are not
completely outside the limits of the target language, children are simply
stretching these limits when they are faced with a dilemma: violating the
syntactic constraint on -s or violating the requirement to use a determiner or
other modifier with a count noun. Once prepositional or genitive construc-
tions are acquired, there is no pressure to affix -s to common nouns that
require a modifier.
The idea that children need positive evidence to assume that -s can be
used for modified PRs can be captured straightforwardly: one could simply
assume that children’s unmarked expectation would be that bound mor-
phemes will concern the word to which they are attached, rather than the
phrase. One could also capture this idea in feature-based underspecification
models of morpho-syntactic development – such as the one assumed by
Eisenbeiß (2003; see Eisenbeiß 2009 for a summary). In such models,
grammatical features are only integrated into morphological representations
if there is positive evidence in the input that requires them. For instance,
children only integrate number features into lexical entries for nouns when
they are confronted with contrasts between singular and plural forms. The
difference between a simple noun and a complex noun phrase is captured by
additional features for phrasal projections in different syntactic frameworks
(see e.g. Grimshaw 1994). Thus, one could assume that English children
include such a phrase-feature in their input specification for ’s because they
find positive evidence in their input that supports this. In contrast, German
children do not find such positive evidence and thus do not extend the range
of elements that can be affixed by -s.
172 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
We already observed that the syntactic factors that determine the choice of
possessive construction in German are not relevant for English, where fac-
tors such as animacy play a crucial role. So far, no published study seems
to have systematically varied all the factors discussed in the literature on
adult English, but Armon-Lotem et al. (2005) carried out an elicited pro-
duction study with English-speaking children (3;2–6;3). These children used
’s with count nouns but not with mass nouns; and they used ’s with animate
PRs 90% of the time to encode a part-whole relation (the cowboy’s arm),
but less than 50% of the time for inanimate PRs (the tractor’s wheel).
With respect to the acquisition of the target word order, some studies
explicitly mention that possessive pronouns are correctly positioned (see
e.g. Ruff 2000 for German) and we are not aware of any reports that chil-
dren who acquire the languages under study ever incorrectly position the
possessive pronoun after the PR noun. Similarly, studies of German -s pos-
sessives, English ’s-possessives, and Japanese no-possessives report that
children show the PR-initial target word order from the beginning (Brown
Learning to encode possession 173
1973; Clahsen, Eisenbeiß and Vainikka 1994; Clancy 1985; Eisenbeiß 2000;
Radford 1990; Radford and Galasso 1998; Ruff 2000). Both observations
also hold for our German data. As we mentioned above, the PR typically
follows the PM in German prepositional possessive constructions; and pre-
positional constructions where the PR precedes the PM are highly marked.
However, of the 13 prepositional possessive constructions that we have
found in our German data, 3 exhibit the marked PR-initial order (15a–c)
and one has a fronted prepositional phrase with a wh-element (15d):
(15) a. aber das is nich von wurst die pelle. (Carsten / 3;6)
but that is not of sausage the skin.
‘But that is not the skin of the saussage’
b. ob deiner auch von diese malers das anspitzt.
whether yours also of these crayons that sharpen
‘(Let’s see) whether yours also sharpens that (tip) of these crayons?’
(Carsten / 3;6)
c. das is vo von de von de Sascha fahrrad. (Svenja/13 / 3;2)
that is of of the of the Sascha bike.
‘That is Sascha’s bike.’
d. von wem hast du die schuhe da an? (Svenja/08 /3;0)
of whom have you the shoes there on?
‘Whose shoes have you got on, there?’
Note that the examples where the prepositional PR phrase precedes the PM
come from different children and corpora and similar examples can easily
be found even in a cursory search of the German child corpora in the
CHILDES database (MacWhinney 2000); see e.g. the following examples
from the corpus provided by Wagner (1985):
(16) aber das sind von Pfe[#] von der Reitschule Pferde.
but that are of hor[#] of the riding:school horses.
‘But that are the horses of the riding school.’ (Roman /9;2)
For Greek, which exhibits two different orders for PM and PR for non-pro-
nominal adnominal possessive constructions, we cannot draw any firm con-
clusions about the early stages, due to a scarcity of early data (see Eisenbeiß
2003 for a discussion of Marinis 2002, 2003). For Hebrew, where the PR
follows the PM in constructions with the marker shel, Berman (1985) and
Armon-Lotem (1998) argue that children produce possessive constructions
174 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
with the non-target-like order PR < PM before they start to use the correct
order and then finally the posssssive marker shel.
Taken together, these observations tentatively suggest a tendency to
place the PM before the PR, independently of the target language and con-
struction. However, further research is required in this domain as the find-
ings for Hebrew are based on a rather small set of examples and thus have
to be taken with caution. Moreover, more systematic corpus searches for
prepositional constructions in German child language are required.
Our discussion so far suggests that children acquire the adnominal posses-
sive constructions of their target language incrementally, but show early
sensitivity to the constraints on the use of these constructions in their target
language. Now, we will have a closer look at the order in which children
start to encode different types of possessive relationships. Table 2 gives an
overview of the the types of PRs in adnominal possessive constructions (in-
cluding Carsten’s three dative possessive constructions, which were not
included in Table 1).
As can be seen in Table 2, all PR (pro)nouns in stages I and II refer to
the children themselves or to other people. In stage III, we find a non-
human, but animate PR as in (10a), but inanimate PRs only appear once in
Svenja’s data and four times in Carsten’s data from stage IV as in (17) and
(18). Note, however, that in example (18), where the referent of the posses-
sive pronoun is a vehicle, this PR is construed as having “animate” proper-
ties.
Note that 2 of the 10 utterances that encode two different possessive rela-
tions come from Andreas (stage III) and all the others occur in stage IV.
One might attribute this to the fact that the combination of two possessive
relations in the same utterance requires a certain sentence length. However,
some of these sentences are actually only 4 or 5 words long; and in each of
stages I and II, we find more than a hundred utterances that are longer than
3 words (see e.g. (6a)). Thus, the appearance of combinations such as (20)
in stage III might not be an artifact of increasing sentence length. It is at
least in line with the general observation that children in stages III and IV
extend the range of PRs and possessive relations from ownership relations
with human PRs via body part relations to part-whole relations for inani-
mate objects.
4.3. Summary
Our analysis of the German child data and our literature review lead to the
following generalizations:
178 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
5. Predicative possession
“symbiotic” relationship between two entities that only receive their seman-
tic interpretation by virtue of one another: a possessive relation is “a relation
between two entities, a PR and a PM, such that one, the PR, is seen as being
in some way related to the other, the PM, as having it near or controlling it”
(Baron, Herslund, and Sørensen 2001: 2). However, there is no PR without a
PM and a PM is not a PM without a PR. Possessive relations share this
property with locative constructions and experiencer constructions: some-
thing is only conceived of as a Location if something is located there and
this element is only considered to be a Locatum if it has a Location. Simi-
larly, an Experiencer requires a Stimulus to experience something, just as a
stimulus cannot be construed as a Stimulus unless it causes a sensation in
an Experiencer. By contrast, Agents do not necessarily need Patients for
their activities – we can talk about people who kick, dance, and write with-
out ever specifying who they kick or what they dance or write.
Given this inherent co-dependency between the entities involved in a
possessive, locative or experience relationship, it is difficult to rank these
entities according to their agentivity or their control over the relationship –
i.e. with respect to the variables that (co-)determine which of them is lin-
guistically realized as the topic or grammatical subject of a construction.
Hence, it is unsurprising that we can observe a large amount of inter- and
intra-language variability in the syntactic realization of Experiencers and
Stimuli, Locata and Locations, and PRs and PMs (see e.g. Chappel and
McGregor 1996; Heine 1997; Seiler 1983). An Experiencer can be realized
as topic or subject (Most farmers fear angry roosters) or it can be encoded
as an object (Angry roosters frighten most farmers) – and the same is true
for the Locatum (The pencils are (lying) in the box vs. The box contains
some pencils).
Similarly, cross-linguistic studies of predicative possession construc-
tions frequently make a distinction between HAVE-constructions, where
the PR is realized as topic and subject (I have/own a car), and BELONG-
constructions, where the PM appears in this role (The car belongs to me.
The car is mine/ Peter’s; see e.g. Baron, Herslund, and Sørensen 2001 for
overview). Some authors have claimed that this distinction can be observed
in all human languages (Heine 1997: 33), though this is controversial.
HAVE and BELONG constructions do not just differ with respect to the
mapping of arguments onto grammatical roles; they can also differ with
respect to the definiteness of their arguments. In particular, the PM is typi-
cally defninite in German BELONG constructions, but indefinite in German
HAVE constructions (Heine 1997: 30f., see e.g. Das Huhn gehört mir ‘The
chicken belongs to me’ vs. Ich habe ein Huhn). Note, however, that it is
Learning to encode possession 181
Table 4. Types of PM Noun Phrases co-occurring with haben ‘have’ and gehören
‘belong’
Note that all of the utterances with haben are target-like with respect to
definiteness and with respect to the mapping of arguments onto grammati-
cal roles. By contrast, we find mapping problems with gehören. Svenja
never uses target-like dative marking for the PR. In example (21b), she
184 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
For gehören, we observe more than just mapping problems. The data in
Table 4 also suggest that locative BELONG is a pre-cursor for possessive
BELONG: of the 3 children who use gehören more than once, only Svenja,
the most advanced child, uses both type of constructions. Recall, however,
that she shows mapping problems. She also uses the proximity preposition
bei ‘by’ in locative BELONG-constructions and not the target preposition
zu ‘to’, which is directional, as shown by (23a,b). These utterances are
somewhere between a locative and a possessive construction as they seem
to focus on placing an object in the proximity of an animate PR. In Anne-
lie’s data we only find 5 cases of the locative variant (24); and Leonie uses
the locative variant once in stage III and 6 times in stage IV (25a), but only
produces one instance of possessive gehören (25b).
Location + Predication
Child Location Predication Possession Total
Possession + Possession
Ann 24 7 1 – – 32
Han – 1 – – – 1
I Leo – 2 1 – – 3
Mat 10 8 2 – – 20
Total 34 (61%) 18 (32%) 4 (7%) 0 (0%) 0 (0%) 56
Ann 29 15 1 – – 45
Han 2 – – – – 2
II Leo 1 1 – 1 – 3
Mat 12 3 – – – 15
Total 44 (68%) 19 (29%) 1 (2%) 1 (2%) 0 (0%) 65
And 82 34 10 3 1 130
Ann 31 22 1 4 – 58
Han 1 1 – – – 2
III
Leo 5 3 2 2 – 12
Mat 6 16 – 1 – 23
Total 125 (56%) 76 (34%) 13 (6%) 10 (4%) 1 (<1%) 225
Ann 42 43 6 16 – 107
Car 64 135 21 15 1 236
Han 20 29 1 4 – 54
IV Leo 83 83 6 4 3 179
Mat 34 53 9 8 1 105
Sve 275 227 7 34 1 544
Total 518 (42%) 570 (47%) 50 (4%) 81 (7%) 6 (<1%%) 1225
Total 721 (46%) 683 (43%) 68 (4%) 92 (6%) 7 (<1%) 1571
5.3. Summary
6. External possession
EPCs are very common in German and occur in everyday language ad-
dressed to young children (see Neumann 1995; Appendix A, A23). Here,
the PR is typically realized as a dative-marked “extra” argument that is not
subcategorized by the verb. This extra dative argument refers to the PR of
the entity that is the Patient of the Action, while this PM is encoded as the
direct accusative object (die Haare ‘the hair’). This possessive relation
makes it possible that the dative noun phrase is integrated into the construc-
tion as an argument with all the morpho-syntactic characteristics of a sub-
190 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
The dative PRs in these constructions are not adjuncts with a semantic case
that is solely determined by their thematic role and unaffected by any syn-
tactic processes. Rather, we observe the same syntactically determined case
alternations for dative PRs (as in (32)) as for indirect dative objects of
three-place verbs (as in (33)). In so-called recipient passive sentences with
the auxiliary kriegen/bekommen ‘get/become’ (e.g. (32b, 33b)), the dative
noun phrase of the corresponding active sentence (e.g. (32a, 33a)) carries a
nominative marker:
Thus dative PRs behave differently from dative adjuncts with semantic
case, e.g. the so-called ethical datives in (34), which express the attitude the
referent of this noun phrase has to the action encoded in the verb. In con-
trast to dative PRs, ethical dative adjuncts do not exhibit any case alternation
in recipient passives. That is, the ethical dative pronouns keep their dative
marking, as in (34b, 35b) while external PRs appear with nominative mark-
ing in recipient passives, as in (35b) – just as indirect dative objects, as in
(34b). Moreover, ethical datives can be combined with both indirect objects,
as in (34), and dative PRs, as in (35), which suggests that they occupy a
different syntactic position, while indirect objects and dative PRs have the
same status.
(36) a. Sue legt dem Pferd den Sattel auf den Ruecken.
Sue lays the-DAT horse the-ACC saddle on the back.
‘Sue puts the saddle on the horse’s back.’
b. Sue legt den Sattel auf den Ruecken von dem Pferd.
Sue lays the-ACC saddle on the back of the horse
‘Sue puts the saddle on the horse’s back.’
mentioned: (i) as an extra argument, (ii) within the same noun phrase as the
Patient or (iii) both. Table 6 provides an overview.
Total
Total
Total
Both
Both
Both
EPC
EPC
EPC
IPC
IPC
IPC
And – 2 – 2 – 2 – 2 – 4 – 4
Ann – – – – – – – – – – – –
Han – – – – – – – – – – – –
III
Leo – – – – – – – – – – – –
Mat – – – – – – – – – – – –
Total – 2 – 2 – 2 – 2 – 4 – 4
Ann – 2 – 2 – 1 – 1 – 3 – 3
Car 1 12 – 13 – 4 1 5 1 16 1 18
Han – – – – – – – – – – – –
IV Leo – 2 – 2 1 1 – 2 1 3 – 4
Mat – 5 – 5 – – – – – 5 – 5
Sve – 13 1 14 1 1 2 1 14 1 16
Total 1 34 1 36 2 7 1 10 3 41 2 46
Total 1 36 1 38 2 9 1 12 3 45 2 50
(37) jetz hau ich dir aber aber ganz xxx den popo.
now hit I you-DAT PART PART really xxx the-ACC botty.
‘Now I will really hit you on the botty.’ (Svenja/15 / 3;2)
Clearly, further studies of external possession with older children are nec-
essary. Moreover, in order to raise the number of potential contexts for ex-
ternal possession contexts, elicitation stimuli or games might be needed
(see e.g. the picture book stimulus by Eisenbeiß and McGregor 1999).
We have obtained some preliminary evidence from an elicitation study
with Japanese and 20 German and Japanese children (2–6 yrs; Eisenbeiß
and Matsuo 2003, 2005). This study made use of the puzzle tasks where
children are asked to describe events depicted on a puzzle board to obtain
puzzle pieces with matching pictures. The individual pictures on the puzzle
board differ minimally from each other, so that children have to express the
differences verbally in order to clearly identify the puzzle piece they want.
For the external possession study, three different types of events with af-
Learning to encode possession 195
fected body parts were depicted and the pictures differed with respect to the
participants and body-parts, see Figure 1.
In order to obtain the desired puzzle piece, children had to mention Agents,
Patients and those body parts that were affected by the action. The German
target sentences with a dative PR involved three different constructions,
one with a nominative subject and a prepositional phrase for the body part
as in (40), one with a nominative subject and a direct accusative object as in
(41), and one with a subject, a direct accusative object and a prepositional
phrase for the body part as in (42).
(42) Der Junge legt dem Pferd den Sattel auf den Ruecken
the boy puts the-DAT horse the-ACC saddle on the back
‘The boy puts the saddle on the horse’s back.’
An initial analysis of the data showed early adaptations to the target lan-
guage: Japanese children consistently realized the possessive relationship
noun-phrase-internally, whereas German children used EPCs in more than
80% of the utterances that referred to a body part and its PR. However,
many utterances produced in the elicitation game were not target-like. When
children produced adnominal possessive constructions instead of external
PR constructions (which are not strictly ungrammatical, just dispreferred),
they sometimes exhibited the -s overgeneralizations to single PR nouns that
we described above, leading to non-target-like utterances such as (43).
When they attempted to produce external PR constructions, children often
do not manage to realize the PR as a case-marked dative noun phrase, but
use a prepositional phrase instead, as in (44). Note that these utterances are
not always target-like. If they use the preposition von ‘of’ for the PR, they
should use it within an adnominal possessive construction, following the
PM. Instead, they sometimes position the von-phrase to the left of the PM,
sometimes with other elements in between. This leads to utterances that are
not acceptable for the adult speakers we consulted, as in the case of (44a,b).
We also found utterances where the child used two prepositional phrases
with directional prepositions – one for the body part and one for its PR, as
in (44c). Thus, both were mentioned as the endpoint of the caused motion
event, but in a parallel fashion. This is also not an option in adult German.
Note that the deviations from the target we observed in the elicitation
games do not seem to be an artifact of the elicitation technique. Recall that
we found -s overgeneralizations and fronted prepositional phrases for PRs
in different sets of corpus data. We will argue that the deviations from the
target in these utterances result from difficulties in the intergration of pos-
sessive and Agent-Patient relations. This is supported by the observation that
they were more likely to use adult-like external possessive constructions for
two-place verbs such as bite than for three-place verbs such as put.
6.3. Summary
Clearly, more studies are required for EPCs. However, our analysis of predi-
cative possessive constructions in the German child data has led to some
preliminary generalizations and insights:
When we compare the results for adnominal, predicative and external pos-
sessive constructions, we observe incremental development in all three do-
mains: the constructions of the target language emerge step by step. At the
same time, the range of functions that are encoded by possessive construc-
tions increases over time. For adnominal possessive constructions, the range
of relations expands from prototypical possessive relations via body part
relations to relationships between objects and their parts. In the case of
predicative possessive constructions, we can observe a primacy of locative
relations in early stages, which is in line with the typological literature on
possession.
In addition to incremental development, we found initial lexical restric-
tions for both possessive -s-markers and possessive pronouns in adnominal
possessive constructions. As discussed above, both incremental develop-
ment and initial lexical restrictions are explicitly predicted by structure-
building approaches as well as usage-based approaches. By contrast, cap-
turing such a developmental path would require additional assumptions in a
full-competence approach, for instance, reference to the interaction between
linguistic development and the learning of cultural norms for possessive
relations and negotiations about them.
While children may not use the full range of forms and functions in the
early two-word stage, they seem to adapt to the core properties of their target
language early on. For instance, children overgeneralise the German -s-
Learning to encode possession 199
predicate remaining
utterances D-contexts +D- D+N
formula combinations
% of
child Filea age MLU n anal. stage n +D % n tokens types per file
+D
2;4– 2,01–
1–2 651 473 I 71 33 46 16 48 17 #1: 7, #2: 7
2;5 2,11
3 2;6 2,53 438 340 II 83 35 42 5 14 30 #3: 25
Ann
4 2;7 2,61 589 490 III 93 70 75 2 3 68 #4: 56
2;8– 2,54–
5–6 744 674 IV 133 125 94 9 7 116 #5: 37, #6: 47
2;9 3,07
2;0– 1,18–
1–2 552 442 I 33 19 58 14 74 5 #1: 3, #2: 2
2;1 1,23
2;2– 1,23–
3–4 499 355 II 50 10 20 6 60 1 #3: 1, #4: 3
Han 2;3 1,38
5 2;4 1,65 96 54 III 10 7 70 0 0 7 #5: 5
2;6– 2,45– #6: 23, #7: 47,
6–8 831 538 IV 136 131 96 0 0 131
2;8 2,85 #8: 31
Learning to encode possession 203
predicate remaining
utterances D-contexts +D- D+N
formula combinations
% of
child Filea age MLU n anal. stage n +D % n tokens types per file
+D
1;11 1,57–
1–2 576 341 I 27 12 44 8 67 4 #1: 2, #2: 2
–2;0 1,67
2;1– 1,60–
3–4 659 414 II 32 11 34 7 64 4 #3: 3, #4: 1
2;2 1,66
Leo 2;3– 1,86– #5: 10, #6: 26,
5–8 1587 1166 III 183 131 72 21 16 110
2;5 2,08 #7: 26, #8: 17
9– 2;6– 2,12– 3157 2462 IV 545 499 92 33 7 466 #9: 36, #10: 38,
15 2;11 3,06 #11: 58, #12: 54,
#13: 42,
#14: 47, #15: 52
#9: 0, #10: 2,
#11: 0, #12: 0,
9– 2;3– 1,25–
1210 901 I 178 22 12 0 0 22 #13: 2, #14: 0,
17 2;9 2,11
#15: 3, #16: 2,
#17: 10
18 2;11 2,62 123 115 II 18 4 22 0 0 4 #18: 4,
Mat
19– 2;11 2,62– #19: 21, #20:
264 243 III 67 58 87 0 0 58
21 –3;0 2,65 n.a., #21: 23
#22: 37, #23:
22– 3;1– 2,24– 6, #24: 23,
804 719 IV 204 184 90 0 0 184
27 3;6 3,51 #25: 27, #26:
35, #27: 28
Andb 1 2,1 2,44 2344 1450 III
b
Car 1 3,6 4,22 2314 1795 IV
a
The column “file” provides the number of the respective recording. The column MLU
shows the mean length of utterance for this recording. The columns under “utterances”
provide the total number of utterances and the total number of analyzable utterances (i.e.
utterances that were intelligible and not simple yes/no answers or formulas such as hallo
‘hello’). Under “D-contexts” are: (i) the number of contexts in which an adult native
speaker would have produced a D-element, i.e. a nominal function word (determiner, pos-
sessive pronoun or quantifier), (ii) the number of overt D-elements and (iii) the percentage
of D-contexts where a D-element was used. The column predicate+D-formula shows how
many of the D-elements that children produced were found in potentially formulaic com-
binations with a small set of high-frequency predicates (e.g. das-is-ein-X ‘that-is-a-X’, die-
mama ‘the mommy’). The following columns show how many tokens and different types
of D-element-noun combinations remained after we excluded the predicate+D formulas
from the total number of D-elements in D-contexts.
b
For the cross-sectional data from the advanced stages no quantitative analyses of predi-
cate+D formula and D+N combinations were carried out as both Andreas and Carsten
used D-elements with broad ranges of nouns and predicates in obligatory contexts.
204 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
Acknowledgements
We would like to thank the children and the families who provided us with
data and Harald Clahsen for letting us use the Lexlern data. We are grateful
to Bill McGregor, Martin Henson and Bettina Landgraf for valuable com-
ments on earlier drafts and to Anette Rosenbach and Nikola Koch for many
discussions about possessive constructions. The research reported in this
article was supported by a grant from the Research Promotion Fund, Uni-
versity of Essex.
References
Eisenbeiß, Sonja
2000 The acquisition of the Determiner Phrase in German child language.
In The Acquisition of Syntax: Studies in Comparative Developmental
Linguistics, Marc-Ariel Friedemann and Luigi Rizzi (eds.), 26–62.
London: Longman.
Eisenbeiß, Sonja
2003 Merkmalsgesteuerter Grammatikerwerb. Doctoral dissertation, Uni-
versity of Düsseldorf. (Available online at http://docserv.uni-
duesseldorf.de/servlets/DerivateServlet/Derivate-3185/1185.pdf.)
Eisenbeiß, Sonja
2006 Documenting child language. In Language Documentation and De-
scription. Vol. 3, Peter K. Austin (ed.), 106 –140. London: SOAS, The
Hans Rausing Endangered Languages Project.
Eisenbeiß, Sonja
2007 The lexical learning hypothesis. Essex Research Reports in Linguis-
tics 54: 1– 4. (To appear in: The Cambridge Encyclopedia of the Lan-
guage Sciences.)
Eisenbeiß, Sonja
2009 Generative approaches to language learning. Linguistics 47 (2): 273–
310.
Eisenbeiß, Sonja and Ayumi Matsuo
2003 External and internal possession – a comparative study of German
and Japanese child language. Poster presented at the 28th Annual
Boston University Conference on Language Development, Boston
University, USA.
Eisenbeiß, Sonja and Ayumi Matsuo
2005 Eliciting language production data from young children. Presentation
at the Xth International Congress for the Study of Child Language,
Berlin, Germany.
Eisenbeiß, Sonja and William B. McGregor
1999 The circle of dirt. Ms. Max-Planck-Institute for Psycholinguistics,
Nijmegen.
Eisenbeiß, Sonja and Ingrid Sonnenstuhl
2007 Transcription conventions for the L-family corpus. Ms. University of
Essex.
Goldberg, Adele
1995 Constructions. Chicago: University of Chicago Press.
Goldberg, Adele
2006 Constructions at Work: the Nature of Generalization in Language.
Oxford: Oxford University Press.
Grimshaw, Jane
1994 Minimal projection and clause structure. In Syntactic Theory and
First Language Acquisition: Cross-linguistic Perspectives, Barbara
Lust, M. Suner and J. Whitman (eds.), 75–83. Hillsdale, N.J.: Law-
rence Erlbaum Associates.
208 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
Guasti, Maria Teresa, Anne Christophe, Brit van Ooyen and Marina Nespor
2001 Pre-lexical setting of the head: Complement parameter through pros-
ody. In Approaches to Bootstrapping: Phonological, Lexical, Syntac-
tic and Neurophysiological Aspects of Early Language Acquisition,
Vol. 1, Jürgen Weissenborn and Barbara Höhle (eds.), 231–248. Amster-
dam /Philadelphia: Benjamins.
Harbert, Wayne
2007 The Germanic Languages. Illustrated Edition. Cambridge: Cambridge
University Press.
Heine, Bernd
1997 Possession: Cognitive Sources, Forces, and Grammaticalization. New
York: Cambridge University Press.
Hyams, Nina M.
1996 The underspecification of functional categories in early grammar. In
Generative Perspectives on Language Acquisition. Empirical Find-
ings, Theoretical Considerations and Crosslinguistic Comparisons,
Harald Clahsen (ed.), 91–127. Amsterdam/ Philadelphia: Benjamins.
Gerhard Jäger and Annette Rosenbach
2006 The winner takes it all – almost. Cumulativity in grammatical varia-
tion. Linguistics 44 (5): 937–971.
Kiparsky, Paul
1982 From cyclic phonology to lexical phonology. In The Structure of
Phonological Representations. Part 1, Harry van der Hulst and Norval
Smith (eds.), 131–175. Dordrecht: Foris.
König, Ekkehard and Martin Haspelmath
1998 Les constructions à possesseur externe dans les langues d’Europe. In
Actance et Valence dans les Langues d’Europe, Jack Feuillet (ed.),
525– 606. Berlin and New York: Mouton de Gruyter.
Küntay, Aylin and Dan I. Slobin
1996 Listening to a Turkish mother: some puzzles for acquisition. In Social
Interaction, Social Context, and Language, Dan I. Slobin, Julie Ger-
hardt, Amy Kyratzis and Jiansheng Guo (eds.), 265–286. Mahwah,
NJ: Erlbaum.
Lust, Barbara
1994 Functional projection of CP and phrase structure parametrization: an
argument for the strong continuity hypothesis. In Syntactic Theory
and First Language Acquisition: Cross-Linguistic Perspectives. Vol. 1,
Barbara Lust, Margarita Suner and John Whitman (eds.), 85–118.
Hillsdale, NJ: Erlbaum.
MacWhinney, Brian
2000 The CHILDES Project: Tools for Analyzing Talk. 3rd Edition. Vol. 2:
The Database. Mahwah, NJ: Erlbaum.
Marcus, Gary F.
1993 Negative evidence in language acquisition? Cognition 46: 53–85.
Learning to encode possession 209
Marcus, Gary F., Steven Pinker, Michael Ullman, Michelle Hollander, John T. Rosen
and Fei Xu
1992 Overregularization in language acquisition. Monographs of the Society
for Research in Child Development. Vol. 57 (Serial No. 228), 1–165.
Marinis, Theodoros
2002 Acquiring the possessive construction in Modern Greek. In Proceed-
ings of the 1999 GALA Conference, Ingeborg Lasser (ed.), 57–80.
Frankfurt: Lang.
Marinis, Theodoros
2003 The Acquisition of the DP in Modern Greek. Amsterdam /Phila-
delphia: Benjamins.
Miller, Max
1976 Zur Logik der frühkindlichen Sprachentwicklung. Stuttgart: Klett.
Mills, Anne E.
1985 The acquisition of German. In The Crosslinguistic Study of Language
Acquisition. Vol. 1, Dan I. Slobin (ed.), 141–254. Hillsdale, NJ: Erl-
baum.
Neumann, Dorothea
1995 The dative and the grammar of body parts in German. In The Gram-
mar of Inalienability: A Typological Perspective on Body Part Terms
and the Part-Whole Relation, Hilary Chappell and William B.
McGregor (eds.), 745–779. Berlin /New York: Mouton de Gruyter.
Payne, Doris L and Immanuel Barshi (eds.)
1999 External Possession. Amsterdam /Philadelphia: Benjamins.
Penner, Zvi, Manuela Schönenberger and Jürgen Weissenborn
1994 The acquisition of object placement in early German and Swiss
German. Linguistics in Potsdam 1: 93–108.
Penner, Zvi and Jürgen Weissenborn
1996 Strong continuity, parameter setting and the trigger hierarchy: On the
acquisition of the DP in Bernese Swiss German and High German. In
Generative Perspectives on Language Acquisition. Empirical Find-
ings, Theoretical Considerations and Crosslinguistic Comparisons,
Harald Clahsen (ed.), 161–200. Amsterdam/Philadelphia: Benjamins.
Peters, Ann M. and Lise Menn
1993 False starts and filler syllables: ways to learn grammatical mor-
phemes. Language 69: 742–777.
Phillips, Colin
1996 Order and structure. PhD thesis, Cambridge, MA: MIT.
Pinker, Steven
1984 Language Learnability and Language Development. Cambridge, MA:
Harvard University Press.
Pinker, Steven
1989 Learnability and Cognition: The Acquisition of Argument Structure.
Cambridge, MA: MIT Press.
210 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
Radford, Andrew
1990 Syntactic Theory and the Acquisition of English Syntax: The Nature
of Early Child Grammars of English. Oxford: Blackwell.
Radford, Andrew
1996 Towards a structure-building model of acquisition. In Generative
Perspectives on Language Acquisition. Empirical Findings, Theoreti-
cal Considerations and Crosslinguistic Comparisons, Harald Clahsen
(ed.), 43–90. Amsterdam /Philadelphia: Benjamins.
Seiler, Hansjakob
1983 Possession as an Operational Dimension of Language. Tübingen:
Narr.
Shibatani, Masayoshi
1994 An integrational approach to PR raising, ethical datives, and adversa-
tive passives. BLS 20: 461– 486.
Slobin, Dan I., Melissa Bowerman, Penelope Brown, Sonja Eisenbeiß and Bhuvana
Narasimhan
in prep. Putting things in places: developmental consequences of linguistic
typology.
Stenzel, Achim
1994 Case assignment and functional categories in bilingual children:
Routes of development and implications for linguistic theory. In Bi-
lingual First Language Acquisition: French and German Grammati-
cal Development, Jürgen M. Meisel (ed.), 161–208. Amster-
dam/Philadelphia: Benjamins.
Stephany, Ursula
1997 The acquisition of Greek. In The Crosslinguistic Study of Language
Acquisition. Volume 4, Dan I. Slobin (ed.), 183–333. Mahwah, NJ:
Erlbaum.
Stephany, Ursula
1998 A crosslinguistic perspective on the category of nominal number and
its acquisition. In Studies in the Acquisition of Number and Diminu-
tive Marking, Steven Gillis (ed.), 1–23. (Antwerp Papers in Linguis-
tics 95) Antwerpen: Universitaire Istelling Antwerpen.
Tomasello, Michael
1998 One child’s early talk about possession. In The Linguistics of Giving,
John Newman (ed.). Amsterdam/Philadelphia: John Benjamins.
Tomasello, Michael
2003 Constructing a Language: A Usage-Based Theory of Language Ac-
quisition. Cambridge, MA: Harvard University Press.
Tomasello, Michael
2006 Acquiring linguistic constructions. In Handbook of Child Psychology,
D. Kuhn and R. Siegler (eds.), 1–101. New York: Wiley.
Tsunoda, Tasaku
1995 The possession cline in Japanese and other languages. In The Gram-
mar of Inalienability: A Typological Perspective on Body Part Terms
and the Part-Whole Relation, Hilary Chappell and William B.
McGregor (eds.), 565–630. Berlin /New York: Mouton de Gruyter.
van der Linden, E. and A. Blok-Boas
2005 Exploring possession in simultaneous bilingualism: Dutch/French
and Dutch/Italian. In EUROSLA Yearbook, S. Foster-Cohen, M. d. P.
García-Mayo and J. Cenoz (eds.), 103–135. Amsterdam /Philadelphia:
Benjamins.
212 Sonja Eisenbeiß, Ayumi Matsuo and Ingrid Sonnenstuhl
Wagner, Klaus R.
1985 How much do children say in a day? Journal of Child Language 12:
475– 487.
Wexler, Kenneth
1998 Very early parameter setting and the unique checking constraint: a
new explanation of the optional infinitive stage. Lingua 106: 23–79.
Plain vs. situated possession in Czech:
A constructional account
Mirjam Fried
1. Introduction
The focus of this paper is the kind of possession that can be identified
broadly as a time-stable relation that is presupposed (not asserted) between
two entities, a possessor (PR) and a possessum (PM). It is well known that
this relation can be expressed in a number of ways even in a single lan-
guage, let alone cross-linguistically; what still remains to be worked out in
sufficient detail is the exact nature of the variation and the relationships
among the variants. My purpose here is (i) to identify the factors that help
differentiate between syntactically distinct expressions of this kind of pos-
session, and (ii) to propose a way of representing the patterns in a network
of grammatical constructions organized around their shared features. The
illustrative material comes from authentic Czech usage, both written and
spoken, as attested in the Czech National Corpus (CNK).1
Heine (1997: 143) labels the presupposed, time-stable possession as “at-
tributive possession”, so categorized in contrast to “predicative possession”,
and this categorization includes characteristic differences in syntax as a
crucial criterion: phrasal syntax for the former, clausal for the latter. This
distinction, though, is too general to give us a realistic picture of the variety
of possessive expressions. The non-clausal expressions tend to come in
several distinct formal variants which represent a rather diverse set, both in
their form and their function. I will examine a subset of such variants in
Czech and offer a functionally and cognitively oriented analysis that is mo-
tivated by three general questions: (i) What kind of linguistic knowledge is
necessary for native-like production and comprehension of these possessive
patterns? (ii) How can the speakers’ knowledge and understanding be ade-
quately represented? And (iii) can the representation help us make more
precise claims about attributive possession as a linguistic category, with
implications beyond accounting for the Czech facts?
1
The material used here comes from roughly 400 million-word written corpora
(SYN2000, SYN 2006PUB), supplemented by a sample from Czech language
and literature (LITERA, SYNEK), and a 2.5 million-word spoken corpora (PMK,
BMK, ORAL2006).
214 Mirjam Fried
2
The distribution of these two forms is conditioned morphologically: the post-
nominal genitive is obligatory if the PR is morphologically neuter (1b), or a
multiword NP, or a plural noun, or is modified by a relative clause. In other
cases, the agreeing pre-nominal form is the neutral choice.
3
There are additional variants, including a morphological adjective as an expres-
sion of PR status. I will not be concerned with these marginal cases here but the
general approach provides ways for incorporating such patterns as well.
4
A note on presenting the data: when the surrounding text is necessary for clearer
understanding, it will be enclosed in curly brackets {} and left without interlinear
glossing.
5
Abbreviations in examples: ACC ‘accusative’, AP ‘affected PR’, AUX ‘auxiliary’,
DAT ‘dative’, F ‘feminine’, GEN ‘genitive’, GP ‘genitive PR’, IMP ‘imperative’,
INF ‘infinitive’, INS ‘instrumental’, LOC ‘locative’, M ‘masculine’, N ‘neuter’,
NEG ‘negative’, NOM ‘nominative’, PRES ‘present tense’, PST ‘past tense’,
RF ‘reflexive’, PR ‘possessor’, SG/PL ‘singular/plural’, and SPRL ‘superlative’.
Plain vs. situated possession in Czech: A constructional account 215
b. Post-nominal
Ohneme noiky dítte v kolenou
bend.PRES.1PL leg.ACC.F.PL child.GEN.SG.N in knee.LOC.PL
‘We’ll bend the baby’s legs at [the] knees’ [SYN2000;34388742]
My primary concern will be the relationship between (1) and (2); reference
to the Implicit PR variant is necessary mainly as a background against
which the explicit patterns can be studied. The example in (3) is simple:
even the minimal context given here is enough to suggest that the owner of
the lost glasses is the subject referent, rather than some other person. The
preferred interpretation has to do with what we know about glasses as
common personal possessions. Other times the implicitness allows a greater
range of possible interpretations, especially if the sentence contains more
than one potential candidate for the PR status, including direct discourse
participants, or if the PM is something other than a body part. Speakers
know to infer the appropriate configuration based on conventional expecta-
tions about possible possessive relationships vis-à-vis particular context.
However, if it is the case that a possessive relation can be (and in Czech
very often is) left implicit, then we have to ask what reasons speakers might
have for choosing the explicit options and particularly, what – if anything –
conditions the choice of (1) vs. (2).
I will be concerned only with the second (easier) question here, pursuing
the hypothesis that the differences should revolve primarily around the
(in)alienability of the PM and the affectedness of the PR as two independ-
ent, competing factors; the hypothesis is motivated by the patterning found
in many other languages. However, the analysis will demonstrate that the
interaction between affectedness and (in)alienability is systematically more
complex than this: specifically, in cases of conflict, affectedness takes
precedence over inalienability in licensing the AP form. This leads, among
216 Mirjam Fried
6
Only these two schemas in Heine’s typology are relevant here. Unlike, for ex-
ample, Russian or Romanian, Czech does not have any locative-type patterns
[PM (is) at PR] expressing possession.
Plain vs. situated possession in Czech: A constructional account 217
for/to PR]. Both schemas are said to correlate with permanent inalienable
possession. Neither of these broad schemas thus helps explain the fact that
the Czech GP and AP have very different distributions in actual discourse
and that the difference in form between (1) and (2), as well as some special
syntactic constraints associated with (2) but not with (1), follow from a
number of co-occurring properties: the inherent meaning of the PM and the
PR, the involvement of the PR in the event expressed by the clause, verb
semantics, and information flow. I will also suggest that the presence or
absence of some of these properties correlates with particular speech situa-
tions.
The two forms will thus be best treated as grammatical constructions in
the sense of Construction Grammar (see especially Fillmore 1989; Croft
2001; Fried and Östman 2004), and I will also draw on the notion “cognitive
frame” (Fillmore 1982) for incorporating the possessive relation in the con-
structional representations. I will argue that each possessive variant consti-
tutes a semantically and pragmatically distinct pattern – a conventional
cluster of semantic, pragmatic, and morpho-syntactic properties – and that
each pattern represents a functional prototype within a network of possibili-
ties for expressing attributive possession.
The paper is organized as follows. The semantic and pragmatic proper-
ties of the PR and the PM are discussed in sections 2 and 3, respectively,
addressing the issue of a possessive prototype. Section 4 studies the interac-
tion between the possessive relation and the structure and meaning of the
sentence in which it occurs. All of this comes together in section 5, in which
I present the patterns (1) and (2) as two grammatical constructions that oc-
cupy overlapping domains within the functional space of attributive posses-
sion. I illustrate the ways in which the constructions in the network may
interact both with each other and with the possessive prototype, suggesting
a more systematic account of the ways in which the prototype can be ex-
tended to more peripheral instances. Section 6 briefly concludes the paper.
Other times, though, only one form is available. For example, the AP in
(5a) cannot be replaced by GP (5b) and the examples in (6) demonstrate the
reverse:
Similar effects have been noted for various languages (e.g. O’Connor 1994;
Croft 1985; Berman 1982; Chappell and McGregor 1995; Manoliu-Manea
1995; Payne and Barshi 1999), including Czech (e.g. Zimek 1960; Pit’ha
1992; Fried 1999a, 2008), and the subsequent discussion will show that the
contrast has to do both with the involvement of the PR in the reported event
and the relative communicative prominence of the PR and PM in a given
piece of discourse.
Plain vs. situated possession in Czech: A constructional account 219
2.1. Affectedness of PR
The relationship exemplified in (5) illustrates the fact that inalienable pos-
session overwhelmingly prefers AP in Czech, to the point of leaving it as
the only expressive option in some cases (i.e., if the speaker has a reason to
mention the PR explicitly in the first place, instead of opting for Implicit
PR). The case in (6), then, appears to be an ‘exception’ since the only pos-
sibility there is GP, even though the PM is a body part like in (5).7 How-
ever, this apparent contradiction is fully motivated and I will revisit it in
sections 4 and 5. First let us consider a less categorical situation: examples
where AP appears to be replaceable by a GP form. The expressions in (a)
are corpus attestations of AP, the (b) examples are their GP counterparts
(constructed).
7
Whatever observations are made in this paper about the behavior of body parts
as the PM, they apply to all specific candidates for body part status. Czech does
not make any grammar-coded distinctions between different types of body parts.
220 Mirjam Fried
The issue in (7)–(9) is not the incompatibility between the PM and the GP
form, as is the case in (5). The problem in (7)–(9) is the degree to which
the PR is involved in the depicted events. The example in (7) is the most
flexible one: GP (7b) implies that the conductor punched some tickets that
belonged to us but were not necessarily related to our riding the train. The
AP (7a), in contrast, presents the tickets as necessary for the ride, we were
holding them in our hands, handed them over, and then took them back
from the conductor, to hang onto them. The GP form is in principle possi-
ble but somewhat odd, given what we know about the usual ways in which
passengers and conductors interact. A GP form in (8) would be even more
problematic since its implication would be contradicted by the subsequent
coordinated clause: GP (8b) would be felicitous whether the cop is present
or not when the speaker is squatting in front of the cop’s door; the automatic
interpretation actually would be that the cop is not around. Yet the point of
the speaker’s bet is that he will be simultaneously having a conversation
with the cop. The contextual incompatibility is still stronger in (9): GP (9b)
would be possible whether or not the PR was alive during the currency re-
form, while AP (9a) is felicitous only if the PR was alive. The AP empha-
sizes this state of affairs, which makes it the only coherent choice for the
follow-up about the consequences the PR suffered (selling off property to
make ends meet).
To summarize, the two forms display a systematic division of labor that
can be related to Bally’s (1995/1926) notion of personal domain or (subjec-
tively applied) indivisibility. GP expresses plain possession in the broadest
sense, where the concept of indivisibility plays no role. By contrast, AP casts
the possession relation as something that is relevant to the PR in a particular
way, as something in his sphere of interest beyond just the fact of being
owned. AP signals that the PR is being affected (positively or negatively)
Plain vs. situated possession in Czech: A constructional account 221
by something that affects the PM (cf. also Chappell and McGregor 1995).
In this light, it is not surprising that certain types of possessa strongly prefer
AP, in contrast to GP: the tighter the possessive relationship, the greater the
chance that manipulating the PM will directly affect the PR. Body parts are
the extreme on the continuum since they are truly inseparable from the PR,
and hence the unacceptability of (5b). The GP form can only be interpreted
in one way: the ears are not attached to the PR’s body (are not, that is, bona
fide body parts) but are some ear-like objects, physically detached from any
bodies.
This extreme restriction applies only to the pre-nominal GP, though; as
shown in (1b), the post-nominal genitive is sometimes attested with body
parts and the form maintains the true body part reading. We thus have to
conclude that the relative flexibility between allowing both AP and GP ex-
tends all the way to the top of the possessibility hierarchy. Nevertheless, a
difference in form (AP vs. GP) always correlates with a shift in interpreta-
tion, no matter what the PM. For example, the choice of GP with a body
part in (1b) is motivated by the type of discourse and the focus of attention.
This token is taken from instructions about how to use a rectal thermometer
to take a baby’s temperature. The communicatively relevant issue here is
how to manipulate the baby’s position so that the person succeeds in what
needs to be done; the author of the instructions is not concerned with what
(dis)comfort it may bring to the baby, which would be the only possible
reading if an AP form were chosen. A similar analysis applies to example
(4), in which a kinship relation is cast in a different light depending on
whether AP or GP is used. The GP in (4b) would imply that the speaker
was estranged from, or at least not very close to, his father and therefore
unaffected by his death, in contrast to (4a), which unequivocally implies that
the father’s death had tangible consequences, as is clear from the subsequent
text.
Two generalizations emerge from these facts: (i) inherent (in)alienability
is not a good predictor of AP vs. GP encoding (as also noted explicitly by
Bally 1926/1995; Chappell and McGregor 1995) and (ii) the PR in the AP
form plays a special role in the event expressed by the predicate, while no
special status is associated with the GP variant.
This semantic freedom suggests that the GP pattern simply marks a very
general relationship between two entities, which can be conceptualized as
denoting a unit of sorts. This conceptualization accommodates possession
(including inalienable) in the narrow, experiential sense as well, but the
form evidently is not restricted to encoding truly possessive relations. The
semantic and functional breadth is corroborated by the fact that the same
form is found not only in combinations such as (11), all expressing (at least
loosely understood) possession, but also in various common ‘genitive’
224 Mirjam Fried
The AP pattern is clearly different from GP in that AP does not cover the
syntactic function shown in (12); we cannot create an AP paraphrase of
(12). However, paraphrasing the expressions in (11) is possible, at least in
principle. Thus, considering the AP examples in (7)–(9) above and (13) be-
low, AP might not appear to present a dramatically different picture from
GP with respect to the kinds of possessa it permits, along the full possessi-
bility hierarchy:
It has been noted in the external PR research that external PRs tend to co-
occur with possessa at the high end of the possessibility hierarchy, with dif-
ferent cut-off points in different languages (Payne and Barshi 1999; Payne
1997a, b). We have now seen that the Czech AP roughly follows the same
scale of preferences but it is worth checking this general tendency against
the semantic range attested in actual discourse. The sample in (2), (7)–(9),
Plain vs. situated possession in Czech: A constructional account 225
(14) a. things that are part or features of self (body parts; name, title;
speech; life; doubt, memory, intention, self-confidence, right to
decide, etc.)8
b. members of ‘family’, understood broadly as a culturally established
unit of shared domestic life (kinship relations; pets and other do-
mestic animals)
c. garments and their parts
8
This group can also be conceptually further decomposed into more specific
categories, some of which are also suggested by Tsunoda (1995): parts of body,
representations of self, cognitive and speech qualities, manifestations of self-
awareness, etc. Such subcategorization, however, leaves the main point of the
generalization in (14a) for Czech unaffected, since these additional conceptual
distinctions seem to have no grammatical reflexes.
226 Mirjam Fried
This list makes it evident that there is no blanket prohibition on the usage of
abstract nouns. Moreover, the attested combinations cannot be simply re-
placed with GP without changing their meaning, as we have seen in section
2. It is important to note, though, that the abstract concepts that co-occur
with AP tend to come from particular semantic domains and are experien-
tially based: they have to do with mental or physical states and cognitive
capacities typical of human beings (14a) or with personal and social rituals
(14f). The latter can be easily extended further into more specialized con-
texts, in which a possessive construal can apply to abstract concepts that are
inherent in various types of social institutions in general, such as political
organization and public life (14g), and where the PR is, therefore, a collec-
tive, not an individual. Granted, the categories in (14f–g) do not represent
the same sense of ownership/possession as the classes in (14a–e) and as
such are somewhat removed from the central, prototypical definition of
possession. Nonetheless, they are not the only source of abstract nouns in
the AP pattern and so they do not invalidate the general observation about
the possessibility of abstract entities.
On the other hand, we also find a number of both abstract and, espe-
cially, concrete entities that do not seem to fit easily into any of the catego-
ries in (14) at all: branka ‘goal/score [in soccer]’, branká ‘goalie’, pacient
‘[hospital] patient’, tramvaj ‘street car’, devo ‘wood’, smlouva ‘contract’,
etc. Taken out of context, these items seem entirely random and might sug-
gest, yet again, that the AP pattern is not that different from GP in the range
of possessa it permits. However, ‘context’ is the operative word here; these
items are always invoked and interpreted within specialized contexts (e.g.,
medical care, transportation, commerce), sometimes even in distinct types
of discourse (sports reporting), which frame possessive relations in ways
specific to those contexts and establish the conditions that allow the PR to
be cast as affected in particular ways. No such contextual framing is re-
quired for the use of the GP variants. And it is also worth pointing out that
Plain vs. situated possession in Czech: A constructional account 227
The corpus data thus confirm the conclusions of previous studies, namely,
that non-affective verbs generally require a GP form. The following exam-
ples further illustrate the prohibition on APs with such verbs, in contrast to
the attested GP, both with transitive (17) and active intransitive (18) predi-
cates:
Plain vs. situated possession in Czech: A constructional account 229
9
In terms of grammatical roles, the dative complements could be classified as
indirect objects, with the understanding that a two-place predicate in Czech can
encode its arguments using the pattern [subject – indirect object], e.g. pomoci
‘help’ in (23). Since it is more informative to refer to these complements either
through their case form (always dative) or their semantic role status, I choose
not to label them in terms of grammatical roles here. The point is that Czech da-
tive NPs are primarily motivated semantically, not syntactically in terms of
grammatical roles (cf. also Fried 1994).
Plain vs. situated possession in Czech: A constructional account 233
This connection has been explored in detail elsewhere (Fried 1999a, b) and
I will not revisit it here beyond summarizing that the AP shares the special
affectedness with other dative-marked roles (experiencer, recipient, benefi-
ciary/maleficiary) and particularly with the usage shown in (24), which I
label Dative of Interest (DI). Both DI and AP require the referent to be hu-
man and in both cases, the dative NP is an extra element not required by the
valence of the verb. However, the datives are incorporated into the sentence
as if they were full-fledged arguments. But AP adds a particular feature that
is absent in the other dative roles, including DI: the presupposed possessive
relation between the referent of the dative and something else in the sen-
tence. This is an important point that bears on the issue of framing posses-
sive relations in terms of broad cognitive schemas, such as the Genitive vs.
Goal-based schemas suggested in Heine’s (1997) typology. It is of course
true that at some very abstract level, all of the datives, including AP, relate
to the concept of goal-ness. But it is also very saliently the case that AP is
distinct from other goal-expressing roles and that the distinction is not just
a matter of interpretation but has systematic reflexes in semantic constraints
and syntactic behavior. In the following section, I will suggest a way of re-
organizing the conceptual space of attributive possession so that we can
capture more accurately the relationships – commonalities as well as con-
trasts – between the relevant semantic categories.
The syntactic behavior only confirms that GP and AP cannot be taken simply
as two alternative and fully comparable expressions of attributive possession.
To capture the essence of the distinction, we can label the patterns as Plain
234 Mirjam Fried
For our purposes here, it is not crucial to dwell on all the additional details
of this general frame (such as, perhaps, listing a set of preferred properties
for each frame element). I will simply take (25) as a minimal way of repre-
senting speaker’s conceptual understanding of prototypical possession,
which then is shaped into different instantiations by elaborating on the spe-
cific characteristics of the PR and/or the PM. Crucially, though, this shaping
requires reference to both form and meaning, not just one or the other. The
conceptual prototype organized in the background frame does not, by itself,
say anything about the morphological or syntactic requirements associated
with the morphosyntactic strategies for expressing the possessive relation.
Those involve additional layers of constraints and I will argue that the best
way to capture the nature of those expressive strategies is to treat each pattern
as a grammatical construction in the sense of Construction Grammar, i.e. as
a conventionally expected association between the elements of this frame
and their linguistic expression.
and these have to do with integrating the possessive relation with the se-
mantic and syntactic structure contributed by the head verb.
First off, it is the AP construction as a whole, not the head predicate,
that supplies the link to the possessive relation, through the statement
[frame POSSESSION]), which brings along the PR and the PM and all the
background knowledge associated with the representation in (25). Second,
the constructional val(ence) statement in the outer box captures the fact that
the syntactic and semantic properties of the argument supplied by the head
verb are constrained by the requirement that its event role be semantically
non-agentive (thus ensuring the prohibition on the PM as a transitive or ac-
tive intransitive subject). The AP construction does not specify anything
about the PM’s semantic or syntactic role (both of these are determined by
the semantics of the head verb). However, the co-indexing (#2) between the
head verb, the constructional valence, and the POSSESSION frame indicates
that whatever this argument is in terms of its semantic and syntactic role in
the sentence, it will be interpreted as the PM. Third, the construction also
gives the PR an independent information-structure status relative to the
PM; this must be specified as the construction’s prag(matic) property. It
follows from this feature that both the PR and the PM are subject to articu-
lating regular information structure relations, which operate independently
of this particular construction (cf. the discussion in section 2.2); put differ-
ently, both the PR and the PM can appear as either a topical or a focal ele-
ment, independently of each other.
The rest of the representation contains features that are shared across AP
and DI, as is indicated by the inherit statement at the top; all the features
that come from this relationship are printed in gray, to show that these
specifications are not unique to the AP construction (strictly speaking, the
inherit statement would be sufficient and all the remaining gray-colored
information need not be spelled out). These features include the following.
First, the construction is syntactically a verb-based pattern (the syn(tax)
statement at the top). Second, the construction itself has additional valence
requirements, namely, the PR (#1) is in the dative (case DAT) and is inter-
preted as other datives of interest (expressed through the rel(ation) statement
specifying the semantic role, labeled ). And finally, it is the construction
as a whole that carries the overall meaning of Situated Possession, spelled
out in the sem(antics) statement, as a combination of the inherited DI se-
mantics and the POSSESSION frame.
Plain vs. situated possession in Czech: A constructional account 237
Affected Possessor
inherit Dative-of-Interest
syn [ cat v, lex + ]
prag [ 'greater discourse independence of Possessor vis--vis Possessum' ]
sem [ 'circumstances described by the predicate have significant consequences
for the interested party (#1), whose referent is not in control of the event' ]
frame POSSESSION
FE #1 Possessor [human]
FE #2 Possessum [ ]
Genitive Possessor
syn [cat n] inherit Modification
prag ['restrict reference of the noun (#2) by property exrpessed in #1']
sem frame POSSESSION
FE #1 Possessor [human]
FE #2 Possessum [ ]
10
Additional constructions expressing attributive possessive relations (such as
adjectival, coming-into-possession patterns based on transfer, etc.) can be incor-
porated into this network, once their features are properly worked out.
240 Mirjam Fried
event partic.
Plain possession (GP)
GEN/poss. morphol. inalienable < alienable
The network thus captures important facts about the way constructions may
interact and the ways in which linguistic expressions may stretch the prop-
erties of those constructions. I will now briefly comment on three such ex-
amples, all involving the AP construction.
As already noted, a systematic account of possession always faces the
question of what should count as possession and how inclusive or non-
inclusive we should be in defining the PR and the PM. Even with the rela-
tively permissive prototype assumed in this paper (compared to Taylor’s),
we still have to account for examples such as (27), in which the PM does
not easily fit the categories suggested for AP in (14):
We could declare that (28) is not a case of possession but a simple part-
whole relation. The advantage of such an analysis would be that it would
preserve the concept of possession as an experiential gestalt, specific to
human beings. This way we would also avoid the danger inherent in any
prototype-based analysis: in order to account for every new deviation, we
could, in principle, keep relaxing the prototype ad infinitum, which then
amounts to justifying just about anything as an instance of the same con-
cept, and the prototype loses its coherence as a tool of systematic analysis.
However, a categorical exclusion of examples such as (28) leaves unan-
swered an obvious similarity between them and the AP pattern, both for-
mally (the ‘whole’ being in the dative) and in the overall affective interpre-
tation: when the roots die, the tree is certain to die as well. In order to
account for these and similar extensions, the constructional approach offers
an alternative that allows us to incorporate the full range of deviations that
may arise in actual discourse, while at the same time preserving the posses-
sive prototype as the conventional semantic basis. The AP construction ex-
pects the prototype – represented in the frame – to hold, but of course the
match between the prototype and the lexical fillers of the constructional
slots will not be always perfect, stretching the prototype to varying degrees.
In (28), the stretching concerns the semantics of the PR, but at the same
time, it is very close to the prototype in two ways: (i) the PM is (construable
242 Mirjam Fried
as) inalienable in the same way body parts are and (ii) the mutual relation-
ship between the whole and its part is fully compatible with the affective
meaning of the AP construction: the whole is affected because its constitu-
tive part is affected. It is also important to note that GP, shown in (29), does
not evoke the situated possession reading but stays purely at the level of a
part-whole relationship. In (29), the implication may very well be that the
trees will somehow make it anyhow; in any case, (29) is not concerned with
the fate of the trees, it is about the fate of the roots only.
I would argue that this difference is due precisely to the fact that the GP
construction is not associated with a special meaning; the form can cover
both possessive and non-possessive relations, and consequently cannot im-
pose a possessive reading so easily on combinations that deviate from the
prototype in a radical way (such as presenting an inanimate entity as an
owner of anything). By contrast, if the semantic and pragmatic properties of
the AP construction as a whole (especially the affectedness of the PR) are
invoked we have a principled way of explaining what allows the stretch
into domains in which we do not have real PRs but only a very specific
(and tight) part-whole relationship. In other words, it is the use of the AP
construction in encoding a (close) part-whole relation that allows a personi-
fication reading, i.e. a conceptualization which mimics a relationship be-
tween an animate PR and a body part.
We could say that both of these cases (27)–(28) illustrate scenarios in
which the AP construction, as a conventional grammatical pattern in its
own right, facilitates manipulations of the possessive frame, sometimes also
in an interaction with the closely related, but more general, PART-WHOLE
frame.11 But the AP construction may also attract other patterns in the net-
work and pull them into an AP reading because of certain shared construc-
tional properties. Here I have in mind the issue of intransitive agents and
their potential for compatibility with an AP interpretation, in an apparent
11
I have not provided the details of this frame or its place in the network, mostly
for reasons of keeping the representation uncluttered and easily readable, given
the focus of this paper (pure possession). But it is obvious that it must be part of
a more complete representation of this functional space, particularly in working
out the Genitive-related domain.
Plain vs. situated possession in Czech: A constructional account 243
12
While I have not run across such an example in the corpus, I can think of usages
such as (i). The verb semantics found in these combinations is still waiting for
more careful research.
(i) Syn se jim opíjel.
son.NOM.SG.M RF 3PL.DAT drink.to.excess.PST.SG.M
‘Their son drank heavily [to their worry/embarrassment].’
244 Mirjam Fried
mate participants and the action of one (the police) clearly is intended to
have consequences for the other (ostensibly mice, figuratively for the in-
habitants):
We have to conclude that examples such as (30) are more plausibly analyzed
as instances of the DI construction, such as exemplified in (24) and (31), in
which any kind of verb, including all semantic types of intransitives, can be
used, and which expresses a situation with an indirect effect on the ‘inter-
ests’ of an animate entity. DI of course overlaps with the AP construction
to a great extent; outside of not having the possessive dimension, the DI
only differs in that it places absolutely no constraints on the verb semantics.
It is not a stretch, then, for a DI token to invite an AP reading, provided that
certain features of that token coincide with a particular narrow set of fea-
tures of the AP construction.
This finally brings us to the significance of the three items that are under-
lined in Figure 3. The features human on the PR and inalienability on the
PM are central to the notion of experientially defined possession. We can,
therefore, expect that if an inherently non-possessive expression (such as
DI) invites a possessive reading, it can be only at the level of the core pos-
sessive properties. But satisfying these two features of the possessive frame
does not, by itself, guarantee a successful AP interpretation. The PR must
also be construable as affected by the event expressed by the verb, which is
a central property of the AP construction. This is, of course, related to the
lexical meaning of the verb, which is constrained in AP, but that restriction
is evidently not as rigid as the requirement that, whatever the verb, the PR
must come out as an affected entity. Some verbs with active semantics are
inherently better equipped for such a stretch (e.g. verbs of removal) than
others (e.g. pracovat ‘work’), and that is what accounts for the relative
(un)availability of DI tokens for an AP interpretation.
The important point is that all these shifts, whether they involve exten-
sions of the possessive prototype into broader semantic domains or, on the
contrary, attracting tokens of non-possessive constructions, can all be ex-
plained by appealing to the same cluster of properties (the core features of
Plain vs. situated possession in Czech: A constructional account 245
the possessive prototype associated with AP) and to the AP pattern as a con-
structional gestalt: the situated possessive meaning of the whole construc-
tion is the ‘glue’ that holds all these seemingly disparate uses together. We
could thus think of the three underlined items as having a privileged status
within this network: they constitute the set of features that are instrumental
in various partial shifts, giving rise to and at the same time constraining
novel usages.
6. Conclusions
Perhaps the central – and inherently thorny – issue in sorting out possessive
expressions in any language is the question of how we define the concep-
tual category to begin with. The present analysis is based on the notion of
possessive prototype understood as an experiential gestalt, which takes the
PR to be necessarily an animate (human) entity and the PM to be placed
somewhere along the possessibility hierarchy, without stating categorically
what may or may not count (universally) as a PM. Based on the corpus at-
testations of the Czech patterns, we can draw at least two conclusions about
possessibility. (i) Rather than relying on purely linguistic categories, such
as (in)animacy, concreteness, control, etc., possessibility is best defined in
terms of culturally determined clusters of concepts and expectations about
what is conventionally construed as possessible; the conventional under-
standing can be then extended to cases where possessive construal is condi-
tioned by the type of discourse or genre. And (ii) different grammatical
forms expressing possession may interact with the possessibility hierarchy
in different ways.
These generalizations are based on a close study of two syntactic patterns
that both express possession as a time-stable and presupposed relation. The
analysis has established that the patterns are not equivalent semantically or
pragmatically and, therefore, cannot be treated simply as structural variants
of a single possessive schema. Each pattern encodes a distinct conceptuali-
zation of a possessive relation, compatible with different communicative
contexts: plain possession is expressed by a Genitive PR (GP), which shares
certain features with non-possessive genitives; situated possession – an idio-
syncratic combination of possession and affectedness – is expressed by da-
tive-marked Affected PR (AP), which shares certain features with non-
possessive dative-marked roles. It follows that an adequate representation of
the speakers’ understanding of these patterns requires reference to several
layers of information: semantic and structural limits on the PM (AP), affect-
246 Mirjam Fried
Acknowledgements
Data
References
Bally, Charles
1995 The expression of concepts of the personal domain and indivisibility
in Indo-European languages. In Chappell, Hilary and William B.
McGregor (eds.), 31–61. [First puplished in 1926.]
Berman, Ruth A.
1982 Dative marking of the affectee role: data from Modern Hebrew. He-
brew Annual Review 6: 35–59.
Chappell, Hilary and William B. McGregor (eds.)
1995 The Grammar of Inalienability: A Typological Perspective on Body
Part Terms and the Part-Whole Relation. Berlin /New York: Mouton
de Gruyter.
Croft, William
1985 Indirect Object ‘Lowering’. Berkeley Linguistic Society 11: 39–51.
Croft, William
2001 Radical Construction Grammar. Syntactic theory in typological per-
spective. Oxford: Oxford University Press.
Dane
, Franti
ek
1974 Functional sentence perspective and the organization of the text. In
Papers on Functional Sentence Perspective, F. Dane
(ed.), 106 –128.
Praha: Academia.
Fillmore, Charles J.
1982 Frame Semantics. In Linguistics in the Morning Calm, The Linguistic
Society of Korea (ed.), 111–137. Seoul: Hanshin.
Fillmore, Charles J.
1989 Grammatical Construction Theory and the familiar dichotomies. In
Language Processing in Social Context, R. Dietrich and C. F. Grau-
mann (eds.), 17–38. Amsterdam: North-Holland /Elsevier.
Firbas, Jan
1966 On defining the theme in functional sentence perspective. Travaux
linguistiques de Prague 2: 267–280.
Fried, Mirjam
1994 Grammatical functions in case languages. Berkeley Linguistic Society
20: 184 –193.
Fried, Mirjam
1999a From interest to ownership: a constructional view of external posses-
sors. In External Possession, Doris L. Payne and Immanuel Barshi
(eds.), 473–504. Amsterdam /Philadelphia: John Benjamins.
Fried, Mirjam
1999b The ‘free’ datives in Czech as a linking problem. In Annual Work-
shop on Formal Approaches to Slavic Linguistics 7, K. Dziwirek, H.
Coats, and C. Vakareliyska (eds.), 145–166. Ann Arbor: Michigan
Slavic Publications.
248 Mirjam Fried
Fried, Mirjam
2008 Dative possessor: delimiting a grammatical category based on usage.
In Gramatika a Korpus / Grammar and Corpora, F. tícha and M.
Fried (eds.), 51–64. Praha: Academia.
Fried, Mirjam and Jan-Ola Östman
2004 Construction Grammar: a thumbnail sketch. In Construction Grammar
in a Cross-Language Perspective, Mirjam Fried and Jan-Ola Östman
(eds.), 11–86. Amsterdam: John Benjamins.
Grepl, Miroslav and Petr Karlík
1998 Skladba spisovné etiny [Syntax of literary Czech]. Praha: Votobia.
Heine, Bernd
1997 Possession: Cognitive Sources, Forces, and Grammaticalization.
Cambridge: Cambridge University Press.
Manoliu-Manea, Maria
1995 Inalienability and topicality in Romanian: pragma-semantics of syn-
tax. In Chappell, Hilary and William B. McGregor (eds.), 711–743.
O’Connor, Mary Catherine
1994 The marking of possession in Northern Pomo: Privative opposition
and pragmatic inference. BLS 20: 387–401.
Payne, Doris L.
1997a The Maasai external possessor construction. In Essays on Language
Function and Language Type, Joan Bybee, John Haiman and Sandra
A. Thompson (eds.), 395–422. Amsterdam/ Philadelphia: John Ben-
jamins.
Payne, Doris L.
1997b Argument structure and locus of affect in the Maasai External Pos-
session Construction. BLS 23 [Special Session on Syntax and Se-
mantics in Africa]: 98–115.
Payne, Doris L. and Immanuel Barshi (Eds.)
1999 External Possession. Amsterdam /Philadelphia: John Benjamins.
Pit’ha, Petr
1992 Posesivní vztah v etin. Praha: AVED.
Sgall, Petr, Eva Haji
ová and Jarmila Panevová
1986 The Meaning of the Sentence in its Semantic and Pragmatic Aspects.
D. Reidel Publishing Company.
Taylor, John R.
1989 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford:
Clarendon Press.
Tsunoda, Tasaku
1995 The possession cline in Japanese and other languages. In Hilary
Chappell, and William B. McGregor (eds.), 565–630.
Zimek, Rudolf
1960 K chápání posesivnosti. Rusko-
eské studie, Sborník Vysoké
koly
pedagogické v Praze, Jazyk a literatura II, 131–156.
Attributive possessive constructions in Oceanic
Frantisek Lichtenberk
1. Introduction
(1) ara-gu
name-1SG:POSS
‘my name’ (Lichtenberk 1983a: 282)
1
Oceanic is a subgroup within Austronesian. Oceanic languages are spoken in
mainland New Guinea and neighbouring islands, Island Melanesia, Polynesia
and Micronesia, but not all of the indigenous languages of New Guinea, Island
Melanesia and Micronesia are Oceanic. For a detailed overview of the Oceanic
family see Lynch et al. (2002).
2
Besides the Leipzig Glossing Rules, the following abbreviations are used in
glossing the examples: CONSTR – construct; NONSG – non-singular; NUM –
numeral marker; PC – paucal; REAL – realis; SV – stem vowel; THC – thematic
consonant.
The glossing conventions are – by and large – those of the sources. In some
cases the glosses have been adjusted for the sake of uniformity. The inclusive
forms are not considered here a subtype of the first person but a category of its
own (Daniel 2005; Lichtenberk 2005a), hence the absence of specification of
person. Stress marking has been omitted from the Manam examples. The Toqa-
baqita data come from my own field notes.
250 Frantisek Lichtenberk
Language Location
’Ala’ala Papua New Guinea
Anejom Vanuatu
Araki Vanuatu
Bali-Vitu Papua New Guinea
Banoni Papua New Guinea
Cèmuhî New Caledonia
Erromangan Vanuatu
Fijian (Standard) Fiji
Gapapaiwa Papua New Guinea
Hawaiian Hawai’i (Polynesia)
Hoava Solomon Islands
Houaïlou New Caledonia
Iaai New Caledonia
Kairiru Papua New Guinea
Kilivila Papua New Guinea
Kokota Solomon Islands
Kosraean Kosrae (Federated States of Micronesia)
Kwaio Solomon Islands
Labu Papua New Guinea
Lenakel Vanuatu
Lenkau Papua New Guinea
Lolovoli Vanuatu
Lou Papua New Guinea
Manam Papua New Guinea
Mussau Papua New Guinea
Nalik Papua New Guinea
Niuean Niue (Polynesia)
Paamese Vanuatu
Pohnpeian Pohnpei (Federated States of Micronesia)
Pukapukan Pukapuka (Polynesia)
Puluwatese Puluwat (Federated States of Micronesia)
Rotuman Rotuma (Fiji)
Tamambo Vanuatu
Tobati Papua (western New Guinea, Indonesia)
Toqabaqita Solomon Islands
Ulithian Ulithi, Fais (Federated States of Micronesia)
Vinmavis Vanuatu
Wayan Fiji
Zabana Solomon Islands
252 Frantisek Lichtenberk
Map 1. The boundaries of the Oceanic subgroup. (From: The Oceanic languages, John Lynch, Malcolm Ross and Terry Crowley, 2002, p. 5
[Map 1.2 there], Richmond, Surrey: Curzon. Reproduced by permission of Taylor & Francis Books UK.)
Attributive possessive constructions in Oceanic 253
Lolovoli
(4) hava-da
family-NONSG(INCL):POSS
‘our family’ (Hyslop 2001: 169)
Lolovoli
(5) no-da hala
POSS.CLF-NONSG(INCL):POSS visitor
‘our visitor’ (Hyslop 2001: 180)
3
Considering the elements that carry possessive suffixes in indirect possessive
constructions to be (possessive) classifiers is the usual approach adopted in recent
descriptive work. However, Palmer and Brown (2007) argue that in the Kokota
language and possibly some other languages those elements are, in fact, “generic
nouns” (p. 208), and that it is these nouns that head possessive constructions.
However, this analysis is not without problems. Here, the standard analysis is
retained, considering those elements to be classifiers. Since the present study is
concerned primarily with the semantic/pragmatic properties of Oceanic posses-
sive constructions, the question of whether in a given language those elements
form a morphosyntactic category of their own or whether they are perhaps a
subcategory within the category of nouns is not of primary importance here.
254 Frantisek Lichtenberk
With very few exceptions, the possessive affixes are suffixes, as in (1)–
(5) above. A few languages have possessive prefixes. However, the prefixes
normally exist in addition to possessive suffixes and their use is restricted
in various ways. For example, in Western Fijian dialects the possessive pre-
fixes are used only in direct possessive constructions and only when the
PM–PR relation is other than kinship, such as ‘my blood’ (Geraghty 1983).
It should also be noted at this point that the “possessive” affixes have, in
some languages, functions other than indexing the PR in a possessive con-
struction. For example, in Toqabaqita the same set of suffixes is used with
one class of transitive verbs to index the direct object, and with certain
verb-phrase internal particles to index the subject.
The PR may be encoded by a noun phrase, in which case the typical pat-
tern is for the PR noun phrase to be cross-referenced by means of a posses-
sive affix either on the PM noun if the possessive construction is of the di-
rect type, or on the possessive classifier if the possessive construction is of
the indirect type (see section 2.1.2 for discussion of cross-referencing). Ex-
amples (6) and (7) from Hoava illustrate:
The ordering of the expressions of the PM, the PR and the possessive clas-
sifier varies from language to language: for example, (classifier) PM (PR)
(Hoava), (PR) PM (classifier) (Manam), PM (classifier) (PR) (Anejom).
With respect to PR phrases that are lexical (rather than pronominal; but see
the Vinmavis example (24) in section 2.1.3), three basic types of cross-
referencing can be distinguished: full, partial and construct. In full cross-
referencing the possessive affix cross-references the PR both for person and
for number. Besides singular and plural, many Oceanic languages also have
a dual number, and some also have a trial or paucal number. A singular/
plural/dual/paucal system with full cross-referencing is found in (Standard)
Attributive possessive constructions in Oceanic 255
Fijian. This is illustrated in the two pairs of examples in (8)–(11) for the
singular and the paucal numbers. In each pair, the same possessive suffix is
used whether or not there is a PR phrase present.
(12) ’i’i-na
tail-3SG:POSS
‘its tail’ (Keesing 1985: 113)
(13) falai-ga
head-3PL:POSS
‘their heads’ (Keesing 1985: 113)
(In [17] the possessive suffix -n on nade ‘breast’ indexes the baby’s mother,
not the baby; the breast is “a source of milk to be sucked by the baby”,
hence the possessive classifier lida, “used with nouns whose referents are
things one sucks the juice out of, but without consuming the flesh in any
way” [Lynch 2000: 60].)
The same construct suffix is used when the PR noun phrase is plural.
The construct suffix is also used on the PM noun in direct possessive con-
structions:
(20) retpo-n
wife-3SG:POSS
‘his wife’ (Crowley 1998: 52)
(23) netal-n
leg-3SG:POSS
‘his/her leg’ (Crowley 2002: 642)
Similar use of the third person singular possessive suffix as a construct suffix
with plural independent pronouns as PR phrases is found in Lou and Lenkau
(Ross 1988: 332).
In some languages independent personal pronouns cannot form a PR
phrase. The PR can be indexed only by means of a possessive affix. Bali-
Vitu4 is one such language: example (26a) without a pronominal PR phrase
is grammatical, while (26b) with a pronominal PR is not:
4
Ross (2002a) calls the language “Bali-Vitu”, but van den Berg and Bachet say
in a grammar of Vitu that “Ross’s sketch is primarily a description of the Bali
variety” (2006: 2).
Attributive possessive constructions in Oceanic 259
Next we will briefly consider some possessive systems that are different
from the typical pattern in more significant respects.
Within the basic A-O constrast, each Polynesian language has a number of
subtypes of possessive construction. In the Hawaiian examples above, the
possessive markers are formally prepositions. In another type of construc-
tion, the possessive markers are fused with an article; see (29) and (30) from
Pukapukan:
There are some other types of possessive system found in Oceanic, but as
these will not figure in the discussion of the semantics/pragmatics of pos-
sessive constructions, they are mentioned here only very briefly. In some
languages prepositions are used in possessive constructions, the comple-
ment of the preposition being the PR noun phrase. Discussion of possessive
prepositions in some Oceanic languages can be found in Hooper (1985).
The usual case is for possessive prepositions to be part of a larger system of
possessive constructions. Possessive prepositions are found in Polynesian
languages (see section 2.2 for Hawaiian), where they are part of the overall
A–O possessive contrast and have a classificatory function, not unlike the
possessive classifiers found in the typical Oceanic pattern.
Finally, there are a few Oceanic languages that have only one basic type
of possessive construction: there is no contrast comparable to those between
direct and indirect, suffixing and bare, or A and O types of construction.
Labu is one such language (Siegel 1984), and Tobati is another (Donohue
2002). Niuean, a Polynesian language, also has basically just one type of
possessive construction, although, according to Massam and Sperlich
(2000), traces of the Polynesian A–O possessive contrast do exist there.
3.1. Introduction
We can now turn our attention to the vast majority of Oceanic languages
that have possessive systems of multiple construction types and subtypes,
and consider their use. Is their use governed strictly lexically, each noun
having to be specified for which type of possessive construction it occurs in
the PM position? Or are there some general patterns that determine the use
of the various types? Once again, given the large number of languages in-
volved, the answer is not simple. There are clear patterns, but there are also
exceptions to these patterns. In this section and in section 4, the focus will
be on the patterns; the existence of exceptions will be considered in section
5. The basic, overall pattern is that the choice of a possessive construction
depends on the relation between the referents of the PM and the PR
phrases. Since the notion of relation is central to the understanding of how
the systems of Oceanic possessive constructions (normally) operate, some
discussion of the notion is in order.
262 Frantisek Lichtenberk
There are (at least) three different types of the notions of “relation” and “re-
lational” that are relevant to the present discussion. One is the idea that
possessive constructions are relational. There are two entities which stand
or are put in a certain relation to each other, one as PM and the other as PR.
As is well known, the number of possible relations expressed by possessive
constructions, while not open-ended, is quite large. Besides ownership (PR
owns PM: ‘my money’, the money that belongs to me), some other relations
are: use (PR uses PM without necessarily owning it: ‘my bus’, the bus I
will take), control (PR has control over PM without owning it or using it:
‘my office’, the office I am in charge of), manufacture (PM is made by PR:
‘my cake’, the cake I baked), kinship (PM is a kin of PR’s: ‘my sister’), part
of a whole (PM is part of PR: ‘my head’, the head which is part of my own
body), and many others (see, for example, Langacker 1995: 56–57). This I
take to be an uncontroversial sense of the notion of relation.
Another sense of the terms “relation” and “relational” has to do with the
fact that certain concepts are inherently relational. Nouns that express in-
herently relational concepts are sometimes referred to as “relational nouns”,
such as mother (see, for example, Barker 1995; Partee 1997; Partee and
Borschev 2003). When inherently relational nouns occur in the PM position
in a possessive construction, the type of relation usually expressed involves
inalienable possession, as in my mother. When a noun that is not inherently
relational occurs in the PM position, the type of relation usually expressed
involves alienable possession, as in my knife. Relational nouns tend to
strongly favour a certain kind of relation between the PM and the PR. The
relation is intrinsic to the meaning of a relational noun. Barker (1995) also
uses the term “lexical possession” for this. Thus with my mother (with the
core meaning of mother) there is an intrinsic relation of kinship. With non-
relational nouns, on the other hand, there is typically no intrinsic relation
between the PM and the PR, and a variety of relations are freely available.
Thus with my knife the relation may be one of ownership (the knife I own),
use (the knife I use without owning it), manufacture (the knife I made), etc.
In such cases, the relations can be said to be “extrinsic” (Barker 1995). The
notions of inalienable and alienable possession and intrinsic and extrinsic
relations will be relevant in later discussion. I take the notions of inherently
relational concepts and relational nouns also to be uncontroversial.
It is the third sense of “relation(al)” that has enjoyed some controversy
in Oceanic linguistics with respect to possessive constructions. This is the
Attributive possessive constructions in Oceanic 263
Paamese
(33) vati-n
head-3SG:POSS
‘his/her head’ (Crowley 1996: 389)
Toqabaqita
(34) suul-a fa qota
juice-3SG:POSS CLF areca.nut
‘juice of an areca nut (being chewed)’
(fa is a “numeral” classifier used in noun phrases referring to fruit
and certain other entities, not a possessive classifier)5
5
As (34) shows, the use of the classifier fa is not dependent on the presence of a
numeral in the noun phrase.
Attributive possessive constructions in Oceanic 265
B. Natural bodily products that emanate from the PR’s body, and other
products of physical bodies: for example, ‘tears’, ‘sweat’, ‘urine’, ‘faeces’,
‘semen’, ‘voice, sound’ (produced by the PR), ‘breath’, ‘smell/scent’ (ex-
uded by the PR); and also ‘shadow, shade’ (cast by the PR), ‘reflection (of
the PR, e.g. in water), ‘picture’ or some other representation of the PR, all
of which are often part of a polysemy:
Manam
(35) boro tae-di
pig faeces-3PL:POSS
‘pigs’ excrements’ (Lichtenberk 1983a: 279)
Tamambo
(36) nunu-ku
photo/reflection/picture/shadow-1SG:POSS
‘my photo/reflection/picture/shadow’ (a likeness of me)
(Jauncey 1997: 229)
C. Entities, matter on the surface of the PR’s body. Included here are
concepts such as sores, dirt, tattoes, clothing (especially, though not
necessarily, when being worn by the PR), and parasites such as lice:
Banoni
(37) kipi-na-i moono
dirt-3SG:POSS-ART girl (Lynch and Ross 2002: 445,
‘the girl’s dirt’ from Lincoln 1976)
Lolovoli
(38) tatai-ne
tattoo-3SG:POSS
‘her tattooes’ (Hyslop 2001: 171)
Kilivila
(39) nano-gu
mind-1SG:POSS
‘my mind’ (Senft 1986: 45)
266 Frantisek Lichtenberk
Lolovoli
(40) domi-mu
thought-2SG:POSS
‘your thoughts’ (Hyslop 2001: 172)
Nalik
(41) a nounau-naande
ART shape-3PL:POSS
‘their shape’ (also ‘their interest in something’) (Volker 1998: 130)
Puluwatese
(42) yii-e-mw
age-SV-2SG:POSS
‘your age’ (Elbert 1972: 283)
Toqabaqita
(43) qi ninima-ku
at beside-1SG:POSS
‘beside me’
Lolovoli
(44) Lo tagu-i bongi gai-vesi ….
LOC behind-CONSTR day NUM-four
‘After four days ….’ (Hyslop 2001: 176)
Kilivila
(45) ina-si
mother-3PL:POSS
‘their mother’ (Senft 1986: 140)
Attributive possessive constructions in Oceanic 267
Iaai
(46) ihumwi-
friend-3SG:POSS
‘his/her friend’ (Ozanne-Rivierre 1976: 157)
Manam
(47) udi tanom-a-di
banana plant-NMLZ-3PL:POSS
‘the planting of the bananas’ (Lichtenberk 1983a: 251)
Toqabaqita
(48) riki-la-na wane baa
look.at-NMLZ-3SG:POSS man that
‘that man’s appearance/look/mien’ (i.e., the way the man appears to
be to others, how others see him) (lit.: ‘that man’s looking-at’)
Kokota
(49) mereseni-na mheke
medicine-3SG:POSS dog
‘medicine for dogs’ (Palmer 2002: 506)
Paamese
(50) Inau nakanian s so-k.
1SG 1SG:REAL:eat self-1SG:POSS
‘I ate by myself.’ (Crowley 1996: 407)
268 Frantisek Lichtenberk
Manam
(51) Rube-gu u-yalale.
alone-1SG:POSS 1S:REAL-go
‘I went alone, by myself.’ (Lichtenberk 1983a: 307)
6
Geraghty (1983) calls the i- prefix a “preformative”.
Attributive possessive constructions in Oceanic 269
Example (54) illustrates the drink classifier and (55) the general one:
(54) na me-mun
t
There are languages that have only one possessive classifier. In one respect,
such forms are unlike possessive classifiers in other languages, because
they do not contrast with other classifiers. Nevertheless, the term “classi-
fier” is retained here for two reasons. First, systems with single classifiers
are historical reductions of systems with multiple classifiers. And second,
constructions with a simple classifier do contrast with direct possessive
constructions, without a classifier.
270 Frantisek Lichtenberk
Leenhardt (1932) gives the form as dov xivu; the representation d v
i-vu is
7
details of the use of the two constructions vary from language to language.
Wilson (1982) has put forward an “Initial Control Theory” to account for
the uses of A- and O- possession in Polynesian: the choice of a possessive
construction is determined by whether or not the PR has control over the
initiation of the relation. If the PR does have control, A-possession is used;
if the PR does not have control, O-possession is used. This is illustrated for
Hawaiian in the next pair of examples. In (63) the PR initiates the relation
by having the child, and so A-possession is used. On the other hand, in (64)
the PR does not initiate the relation to his parent, and so O-possession is
used.
A more complex classifier system exists in Lenakel, which has a food classi-
fier, a drink classifier, a plant classifier (for plants planted by the PR), a
general classifier, and a classifier for locations (occupied by the PR). The
latter classifier is only optional, and the general one may be used instead.
The location classifier is illustrated in (66):
In some languages the original classifier system has undergone great ex-
pansion. This is the case in most Micronesian languages. (Lee 1975: 111)
gives a list of 19 “commonly used classifiers” for Kosraean. The list includes
classifiers for transportation; land and shelters; plants; tools, pets and toys;
drink; several classifiers for food; several classifiers for kinship relations;
and several classifiers for decorations. And there is a general classifier. Ex-
ample (67) contains the tool/pet/toy classifier:
For Pohnpeian, Rehg (1981) gives a list of 21 classifiers, which is not ex-
haustive. In fact, Rehg (1981: 179) says that “how many [possessive classi-
fiers] there are in Ponapean [Pohnpeian] is difficult to determine”.
Quite a few of the classifiers in the Micronesian languages are transpar-
ently related to nouns, and one finds cases of repeaters (Aikhenvald 2000),
where a classifier is used with a noun from which it has developed through
grammaticalization. The classifier and the source noun have identical or
very similar forms. For example, Ulithian has a vehicle classifier of the
form waa, which is used when the PM serves as a means of transportation,
such as a ship, a bicycle, a plane or a canoe, for the PR. The word for ‘ca-
noe’ too is waa:
(69) nunu-ku
photo/reflection/picture/shadow-1SG:POSS
‘my photo/reflection/picture/shadow’ (a likeness of me)
(Jauncey 1997: 229)
The different senses of a noun may also call for the use of different posses-
sive classifiers, as in (71) and (72) from Araki. In (71), where the sense of
the PM noun is ‘pig’, it is the “economic possession” classifier pula- that is
used, while in (72), where the sense of the same PM noun is ‘pork’, it is the
food classifier ha- that is used:
(71) pula-ku po
POSS.CLF-1SG:POSS pig
‘my pig (I breed)’ (François 2002: 100)
(72) ha-ku po
POSS.CLF-1SG:POSS pig
‘my piece of pork (to eat)’ (François 2002: 100)
274 Frantisek Lichtenberk
And in (73) and (74) from Lolovoli there is a contrast between the food and
the drink possessive constructions, depending on whether the PM noun has
the sense of citrus fruit to eat or citrus juice to drink:
And in (77)–(79) from Manam there is a three way contrast with the noun
‘head’ in the PM position, and in each case the sense of the noun is that of a
body part. However, in (77) the head is part of the PR’s own body and so
the direct construction is used; in (78) the head is food for the PR and so
the alimentary classifier construction is used; and in (79) the head is neither
part of the PR’s own body nor food for him/her, and it is the general classi-
fier construction that is used:
(77) paana-gu
head-1SG:POSS
‘my head (part of my body)’ (Lichtenberk 1983a: 302)
Attributive possessive constructions in Oceanic 275
Fluidity is found even if the PM referent is one and the same entity but is
conceptualized differently. As shown in (75) and (76) above from Fijian,
the noun ‘mango’ can occur in the PM position with the food or the drink
classifier. However, one and the same mango can serve a purpose other
than being for the PR’s consumption, for example as something to be sold,
in which case the general possessive classifier is required, and the distinc-
tion between a mango as food and a mango as “drink” disappears:
(81) baligo-gu
grass.skirt-1SG:POSS
‘my grass-skirt (when I am wearing it)’ (Lichtenberk 1983a: 301)
In the examples just given, the PR is (or may be) one and the same person.
There is a different kind of fluidity, where one and the same entity is re-
ferred to by means of different possessive constructions because of different
perspectives due to different PRs. For example, one and the same woman
may be one person’s wife and another person’s sister. In Kosraean, two dif-
ferent classifiers are used:
276 Frantisek Lichtenberk
Kosraean, which has a large set of possessive classifiers (section 3.4), has
several kinship classifiers. Example (83) contains the classifier for mothers
and wives, and (84) one of the classifiers for siblings.
Fluidity having to do with PR perspective is common in the opposition
between passive and active possession. In passive possession the PR is a
Patient, Theme or Stimulus in the relevant situation (section 3.3), and cor-
respondingly in active possession the PR is an Agent or Experiencer. One
and the same state of affairs can be encoded from the perspective of the
Patient/Theme/Stimulus or that of the Agent/Experiencer. This is the case in
(85) and (86) from Fijian, with the food/passive and the general classifiers,
respectively:
In the Fijian examples the formal contrast is between two indirect construc-
tions. In (87) and (88) from Manam the contrast is between a direct con-
struction and an indirect construction. The PM in the direct construction is
a nominalization of the verb nanari-t-a ‘tell a story about’.
(87) nanari-t-a-a-gu
tell.story-THC-TRANS-NMLZ-1SG:POSS
‘my story (story about me)’ (Lichtenberk 1983a: 303)
while terms for persons in authority over ego require an indirect alienable
construction”, for example, ‘spouse’ and ‘mother’, respectively.
In some languages the exceptional treatment of some kinship terms is
due to lexical replacement. In such cases, terms that are relatively new to the
language do not (necessarily) select the direct construction. For example, in
Toqabaqita the term for ‘mother’ occurs in the direct construction but the
term for ‘father’ does not:
(89) thaina-ku
mother-1SG:POSS
‘my mother’
8
The Proto Oceanic reconstructions are from Lynch et al. (2002).
Attributive possessive constructions in Oceanic 279
between him and the body parts, in contrast to the formerly inalienable rela-
tionship between the animal and the body parts when it was still alive.”
Note, however, that the indirect construction is also used when the internal
organ is part of the PR’s, for example, an animal’s, own body, and so the
implication in what Crowley says is that the use of the indirect construction
expressing the alienable nature of the relation between an animal’s internal
organ and the butcherer has been extended to the relation when the organ is
part of the PR’s own body. Be that as it may, the fact remains that the noun
that refers to a bird’s crop occurs in the direct construction, not the indirect
one, and there is no obvious explanation for that. Outside of the area of in-
ternal organs, there is no explanation for the fact that the noun for ‘shin’
occurs in the indirect construction although the noun for ‘lower leg’ occurs
in the expected direct construction.
A different kind of exceptional behaviour of body-part terms is found in
Toqabaqita. Such nouns occur in the direct/suffixing construction unless they
are in the scope of a modifier other than the PR, such as a verb,9 a numeral
or a demonstrative. Compare (91), where the noun ‘eye’ occurs in the direct/
suffixing construction, and (92) and (93), where the same noun occurs in the
bare/non-suffixing construction:
(91) maa-ku
eye-1SG.POSS
‘my eye(s)’
9
In Toqabaqita, verbs can directly modify nouns.
280 Frantisek Lichtenberk
ated vis-à-vis the PR; it is viewed basically as an aspect of the PR. On the
other hand, through specification the PM is individuated, given more iden-
tity with respect to the PR, and there the same construction is used that
serves to express alienable possession, where the PM is always individu-
ated with respect to the PR.
In the present study, the Oceanic possessive systems have been character-
ized as basically relational in nature: exceptions apart, the choice of the
type of possessive construction depends on the nature of the PM–PR rela-
tion. This corresponds to point (i) in the quote from Pawley and Sayaba.
Their point (ii) has to do with fluidity. Fluidity is, of course, evidence of the
relational nature of the system: one and the same noun occurs in the PM
position of different types of possessive construction depending on the na-
ture of the relation between the PM and the PR. Pawley and Sayaba say that
according to the relational hypothesis “for any noun, speakers have some
choice of possessive marker, constrained only by their imaginations or be-
lief systems” (see the quote above; emphasis added here). I am not aware of
any such strong version of the relational analysis, according to which any
noun in a given language exhibits, in principle, fluidity. (See also further
below on fluidity and rigidity.)
Importantly, Pawley and Sayaba’s point (i) is independent of their point
(ii). Fluidity is not necessary for a system to be relational. Even if no noun
exhibited fluidity, where each noun occurred in only one type of possessive
construction and so could be said to belong in a certain noun class, the over-
all system could still be relational if each noun class was defined by the na-
ture of the relation between the PM and the PR. An unstated assumption in
Pawley and Sayaba’s approach is that noun classes are essentially arbitrary.
This, however, is not the case. As Aikhenvald (2000: 21) points out: “There
is always some semantic basis to the grouping of nouns into classes, but
languages vary in how much semantic transparency there is.” And: “In lan-
guages with purely semantic assignment the class of a noun can be inferred
from its meaning.” (p. 22). Thus, noun classes are semantically motivated to
various degrees, even though in a noun-class system “[e]ach noun … belongs
to one (or occasionally more than one) class(es)” (Aikhenvald 2000: 21).
Classifier systems (for example, numeral-classifier systems) too are based
on semantic/pragmatic principles. Sometimes there is fluidity, but there are
also exceptions: “The choice of a classifier is usually semantically trans-
parent; in some cases, however, the semantic link between a noun classifier
and a noun is not obvious.” (Aikhenvald 2000: 82).
It is not lack of fluidity that counts as an exception. Only those nouns
that require a possessive construction that goes against the general pattern
are exceptional. Such exceptions apart, the Oceanic possessive systems are
basically relation-based, regardless of the degree of fluidity they permit.
One more remark on fluidity is in order. Descriptions of Oceanic lan-
guages that do comment on fluidity (and most of them do), say explicitly or
imply that certain nouns do not have fluidity, that they can occur only in one
282 Frantisek Lichtenberk
on the other hand, there is often no such strongly salient, context-stable re-
lation, and correspondingly the referents of one and the same PM phrase
can stand in various relations to their PRs. This was illustrated in the dis-
cussion of fluidity in section 4, in particular by the “mango” examples (75),
(76) and (80) from Fijian: a mango as an item of food, or as an item of
“drink” or as property to be sold. In alienable possession, the interpretation
of the relation between a PM and its PR is highly variable across contexts.
Possessive classifiers specify more closely the nature of the PM–PR re-
lation. The development of possessive classifiers for alienable possession is
well motivated because of the variability in the PM–PR relation. On the
other hand, in inalienable possession there is no such strong motivation be-
cause of the presence of a highly salient, default relation between the PM
and the PR. The default interpretation of linguistic constructions needs no
overt marking (cf. Haiman 1985; Dixon 1994; Croft 2001 [2002]). On the
other hand, there is motivation for there being an overt marker of a non-
default interpretation. For possessives, this is illustrated by the next pair of
examples from Manam. The default interpretation of ‘X’s skin’ is for the
skin to be part of the PR’s own body, as in (94), where the direct construc-
tion is used. However, if the intended interpretation is not the default one,
this is signalled by means of a classifier, as in (95), in this case the alimen-
tary classifier:
(94) usi-gu
skin-1SG:POSS
‘my skin (the skin of my body)’ Lichtenberk, field notes
In the experiment the subjects were presented with several sets of pos-
sessive noun phrases, some of which had inherently relational nouns as
PMs, for example his children, and others had nouns as PMs that are not
inherently relational, such as her cookies. For each stimulus the subjects
were asked to give one interpretation of the relation between the PM and
the PR. The results of the study convincingly demonstrated the existence
both of a PM effect and of a PR effect.
The PM effect has to do with the fact that the relational nouns as PMs
elicited a restricted range of interpretations of the PM–PR relations, while
the non-relational nouns elicited a broader range of interpretations. Fur-
thermore, with the relational nouns there was always one interpretation that
was clearly dominant, while such strong dominance was not found with the
non-relational nouns. Thus, the interpretation of his children was in terms
of a kinship relation, while her cookies elicited a variety of interpretations:
the cookies she owns, the cookies she made, the cookies she bought, the
cookies she will eat.
The PR effect has to do with the fact that the nature of the PR had a
greater effect on the interpretation of the PM–PR relation with PMs that are
not inherently relational, while with the inherently relational nouns as PMs
the interpretations were quite stable. Thus, for example, both for the sol-
dier’s legs and for the general’s legs the interpretation was uniformly that
of the legs being part of the soldier’s or the general’s own body. On the
other hand, for the soldier’s regiment the dominant interpretation was that
of the regiment the soldier is a member of, while for the general’s regiment
the dominant interpretation was that of the regiment the general is in charge
of. The PR effect is stronger with non-relational nouns than with relational
nouns because with non-relational nouns there is typically no intrinsic, sali-
ent relation between the PM and the PR. It is true that a certain kind of PR
is likely to favour a certain kind of interpretation, but such contextual fac-
tors are absent or attenuated with “neutral” PRs, such as possessive deter-
miners, for example, her cookies.10
Although the study was done on English, the assumption is that very
much the same cognitive factors operate in other languages and that they
operated in the history of Austronesian, when the system of possessive
classifiers began to develop. Obviously, a system of possessive classifiers
does not have to develop (after all, they are not common in the languages of
10
William B. McGregor has suggested (pers. comm., 11 August 2007) that the
gender of the PR might be relevant in some cases; cf. her cookies and his coo-
kies. The study discussed here did not take gender into account.
Attributive possessive constructions in Oceanic 285
the world), but if they do develop they are more likely to develop, at least
initially, for alienable possession, with PM nouns that are not inherently
relational, than for inalienable possession, with PM nouns that are inher-
ently relational. (Some Micronesian languages do have possessive classifiers
for different kinship categories, as mentioned and illustrated for Kosraean
in section 4 above), but those languages have large sets of possessive clas-
sifiers, which are later developments, postdating the emergence of the first,
restricted set.)
With very few exceptions, Oceanic languages have more than one type of
attributive possessive construction. In the typical system, there is a distinc-
tion between a direct possessive construction, where the PM noun carries
affixes that index the PR, and more than one subtype of indirect construc-
tion, where the possessive affixes are attached to a possessive classifier.
The direct construction type is strongly associated with inalienable posses-
sion, where there is an intrinsic link between the PM and the PR. There are,
however, also language specific exceptions where certain PMs take an indi-
rect construction rather than the direct one.
With some exceptions, the choice of a possessive construction depends
on the nature of the relation between the PM and the PR. This relational
nature of the choice of possessive construction is particularly strongly evi-
denced by fluidity in the possessive systems. Sometimes such fluidity is due
to the polysemy of a noun, but in some cases there is fluidity without poly-
semy, and the choice of a possessive construction depends on the pragmatics
of the situation (for example, whether or not an article of clothing is being
worn at the relevant time).
When there is no fluidity, one can say that the given noun belongs in a
certain class (the classes being established on the basis of the type of pos-
sessive construction used), but such classes are themselves by and large
semantically/pragmatically grounded in the nature of the PM–PR relation.
To say that the Oceanic possessive systems are relational in nature and
semantically/pragmatically motivated does not mean that the choice of a
possessive construction is always predictable, even disregarding genuine
exceptions. As pointed out in section 3.4, the type of possessive construc-
tion that a noun selects may be based on metonymy. Thus, in Manam the
noun for ‘garden’ selects the alimentary (food and drink) possessive classi-
fier, because gardens are places where food is grown (Lichtenberk 1983a).
286 Frantisek Lichtenberk
However, not in all languages does the noun ‘garden’ select the food or ali-
mentary classifier. For example, in Anejom it selects the general classifier
or the customary-possession classifier, even though the language has a food
classifier (Lynch 2000); and in Zabana it selects the general classifier, even
though the language has an alimentary classifier (Fitzsimons 1989). Meton-
ymy is language/culture specific. A system may be sematically/pragmati-
cally motivated, but that does not mean that everything is predictable.
In Oceanic, possessive classifiers are used (again with some exceptions)
to express alienable possession. The development of possessive classifiers
for alienable possession was motivated by the fact that with nouns that are
not inherently relational there is typically no highly salient, context-stable,
relation between the PM and the PR. A classifier specifies more closely the
type of the relation. There are specific classifiers, such as food and drink.
There is also a general classifier (provided a language has more than one
classifier), which only signifies that the relation is not any one of the more
specific types. In inalienable possession, where the PM noun is intrinsically
relational, there is normally a highly salient, context-stable kind of relation
between the PM and the PR, and the development of classifiers there is
much less motivated.
While the development of a system of possessive classifiers for alien-
able possession was motivated, it is also a fact that the original system of
(at least) three classifiers has been simplified in some languages or has dis-
appeared altogether. While cognitive factors may motivate the existence of
a grammatical construction or contrast, they do not determine their exis-
tence, and such factors may be overridden by other kinds of development.
Acknowledgement
References
Aikhenvald, Alexandra Y.
2000 Classifiers: A Typology of Noun Categorization Devices. Oxford:
Oxford University Press.
Barker, Chris
1995 Possessive Descriptions. Stanford: CSLI Publications.
Attributive possessive constructions in Oceanic 287
Leenhardt, Maurice
1932 Documents Néo-Calédoniens. Paris: Institut d’Ethnologie, Université
de Paris.
Lichtenberk, Frantisek
1983a A Grammar of Manam. Honolulu: University of Hawai‘i Press.
Lichtenberk, Frantisek
1983b Relational classifiers. Lingua 60: 147–176.
Lichtenberk, Frantisek
1985 Possessive constructions in Oceanic languages and in Proto-Oceanic.
In Andrew Pawley and Lois Carrington (eds.), 93 –140.
Lichtenberk, Frantisek
2005a Inclusive–exclusive in Austronesian: an opposition of unequals. In
Elena Filimonova (ed.), 261–289.
Lichtenberk, Frantisek
2005b Inalienability and possessum individuation. In Zygmunt Frajzyngier,
Adam Hodges and David S. Rood (eds.), 339–362.
Lichtenberk, Frantisek, Jyotsna Vaid and Hsin-Chin Chen
2004 Alienable/inalienable possession in grammar and in the lexicon: evi-
dence from possessive phrase interpretation in English. Paper pre-
sented at the 4th International Conference on the Mental Lexicon,
University of Windsor, 30 June – 3 July 2004.
Lincoln, Peter C.
1976 Describing Banoni, an Austronesian language of Southwest Bougain-
ville. Unpublished PhD thesis, University of Hawai‘i.
Lynch, John
1973 Verbal aspects of possession in Melanesian languages. Oceanic
Linguistics 12: 69–102.
Lynch, John
1978 A Grammar of Lenakel. Canberra: Pacific Linguistics.
Lynch, John
1982 Towards a theory of the origin of the Oceanic possessive construc-
tions. In Amran Halim, Lois Corrington and S.A. Wurm (eds.), 243–
268.
Lynch, John
2000 A Grammar of Anejom. Canberra: Pacific Linguistics.
Lynch, John
2001 Passive and food possession in Oceanic languages. In Andrew
Pawley, Malcolm Ross and Darrell Tryon (eds.), 193–214.
Lynch, John and Malcolm Ross
2002 Banoni. In John Lynch, Malcolm Ross and Terry Crowley (eds.),
440–455.
Lynch, John, Malcolm Ross and Terry Crowley (eds.)
2002 The Oceanic Languages. Richmond, Surrey: Curzon.
290 Frantisek Lichtenberk
Nichols, Johanna
1992 Linguistic Diversity in Space and Time. Chicago and London: Uni-
versity of Chicago Press.
Ozanne-Rivierre, Françoise
1976 Le Iaai: Langue Mélanésienne d’Ouvéa (Nouvelle-Calédonie). Pho-
nologie, Morphologie, Esquisse Syntaxique. Paris: Société d’Études
Linguistiques et Anthropologiques de France.
Palmer, Bill
2002 Kokota. In John Lynch, Malcolm Ross and Terry Crowley (eds.),
498–524.
Palmer, Bill and Dunstan Brown
2007 Heads in Oceanic indirect possession. Oceanic Linguistics 46: 199 –
209.
Partee, Barbara H.
1997 Genitives – a case study. Appendix B to Theo M. V. Janssen (1997),
Compositionality. In Johan van Benthem and Alice ter Meulen (eds.),
464–470.
Partee, Barbara H. and Vladimir Borschev
2003 Genitives, relational nouns, and argument-modifier ambiguity. In
Ewald Lang, Claudia Maienborn and Cathrine Fabricius-Hansen
(eds.), 67–112.
Pawley, Andrew
1973 Some problems in Proto-Oceanic grammar. Oceanic Linguistics 12:
103–188.
Pawley, Andrew and Lois Carrington (eds.).
1985 Austronesian Linguistics at the 15th Pacific Science Congress. Can-
berra: Pacific Linguistics.
Pawley, Andrew, Malcolm Ross and Darrell Tryon (eds.).
2001 The Boy from Bundaberg: Studies in Melanesian Linguistics in Hon-
our of Tom Dutton. Canberra: Pacific Linguistics.
Pawley, Andrew and Timoci Sayaba
1990 Possessive-marking in Wayan, a Western Fijian language: noun class
or relational system? In Jeremy H. C.S. Davidson (ed.), 147–171.
Rehg, Kenneth L. (with the assistance of Damian G. Sohl)
1981 Ponapean Reference Grammar. Honolulu: University of Hawai‘i
Press.
Attributive possessive constructions in Oceanic 291
Rivierre, Jean-Claude
1980 La Langue de Touho: Phonologie et Grammaire du Ce muhi
(Nou-
velle-Calédonie). Paris: Société d’Études Linguistiques et Anthro-
pologiques de France.
Ross, Malcolm
1988 Proto Oceanic and the Austronesian Languages of Western Melane-
sia. Canberra: Pacific Linguistics.
Ross, Malcolm
2002a Bali-Vitu. In John Lynch, Malcolm Ross and Terry Crowley (eds.),
362–386.
Ross, Malcolm
2002b ’Ala’ala. In John Lynch, Malcolm Ross and Terry Crowley (eds.),
347–361.
Ross, Malcolm
2002c Mussau. In John Lynch, Malcolm Ross and Terry Crowley (eds.),
148–166.
Ross, Malcolm, Jeff Siegel, Robert Blust, Michael A. Colburn and W. Seiler (eds.).
1984 Papers in New Guinea Linguistics no. 23. Canberra: Pacific Linguis-
tics.
Salisbury, Mary C.
2002 A grammar of Pukapukan. Unpublished PhD thesis, University of
Auckland.
Schütz, Albert J.
1985 The Fijian Language. Honolulu: University of Hawai‘i Press.
Senft, Gunter
1986 Kilivila: The Language of the Trobriand Islanders. Berlin / New York:
Mouton de Gruyter.
Siegel, Jeff
1984 Introduction to the Labu language. In Malcolm Ross, Jeff Siegel,
Robert Blust, Michael A. Colburn and W. Seiler (eds.), 83–157.
Sohn, Ho-min and B. W. Bender
1973 A Ulithian Grammar. Canberra: Pacific Linguistics.
Taylor, John R. and Robert E. MacLaury (eds.).
1995 Language and the Cognitive Construal of the World. Berlin /New
York: Mouton de Gruyter.
van den Berg, René and Peter Bachet
2006 Vitu Grammar Sketch. Ukarumpa, Papua New Guinea: Summer In-
stitute of Linguistics.
Volker, Craig A.
1998 The Nalik Language of New Ireland, Papua New Guinea. New York:
Peter Lang.
Wilson, William H.
1976 The O and A possessive markers in Hawaiian. Unpublished MA the-
sis, University of Hawai‘i.
292 Frantisek Lichtenberk
Wilson, William H.
1982 Proto-Polynesian Possessive Marking. Canberra: Pacific Linguistics.
Wivell, Richard
1981 Kairiru grammar. Unpublished MA thesis, University of Auckland.
Possessive clauses in East Nusantara, the case of
Tidore
1. Introduction
The former occur typically on verbs, and the latter typically on nouns. But
neither do so exclusively. It is not the case, therefore, that the distribution
of the prefixes is given by the word class of the host. Of course, the likeli-
hood that the semantics of a verb call for an agent or undergoer participant
rather than a possessor participant may be greater and at the same time pos-
sessor participants will be linked most typically with possessed goods ex-
pressed in nominal predicates (cf. Taylor 1989, 1996 on the prototypical
properties of possession as an experiential gestalt, which typically involves
a possessum that is a ‘specific concrete thing (usually inaninmate)’.) Yet
these are facts of life rather than language, and indeed both types occur, as
will be demonstrated. Clearly the gloss ‘A’ for ‘actor’ may be somewhat
misleading as it would appear to exclude undergoer arguments in intransi-
tive predications, but these, too, align with transitive A:
1
The following abbreviations are used: A – actor, first argument of actor predica-
tion; CAUS – causative; CTRF – centrifugal; – CTRP – centripetal; DIST – distal;
EMPH – emphatic; F – feminine; GEN – generic; H – human; INAL – inalienable;
LOC –locative; M – masculine; N – neutral; NH – non-human; NM – noun marker;
NOM – nominalisation; O – object; PL – plural; POS – possessor, first argument of
possessive predication; PRED – predicate; PROX – proximal; S – subject; SG – sin-
gular; U – unmarked (feminine+non-human); and <> – unsegmented complex.
Possessive clauses in East Nusantara, the case of Tidore 295
tive possession the possessum and the possessor are in a predicational con-
figuration in which either may be the predicate or the argument (see also
the introduction to this volume). It has been claimed that all languages dis-
tinguish these two configurational relations (Heine 1997: 26). Tidore appears
to be a counter-example.
One of the most curious examples showing that virtually anything may be
the predicate requires a little explanation. In example (7), the interjection
296 Miriam van Staden
joo is used as the main predicate in a sentence. There are two forms for
‘yes’, formal joo and informal oe. During conversations, listeners are ex-
pected to signal to the speaker that they are still listening by saying joo! or
oe! at regular intervals. In the example below, the speaker comments on my
own acquisition of this discursive practice. The modifier saki ‘tasty’ is also
used to express satisfaction with non-food items and bahaya ‘dangerous’ is
a loan from Malay that is used to express ‘extreme’ qualities:
Galela
(8) a. A-wi-dohu i-lamo
3N.S-3SG.M.O-foot 3N.S+3SG.M.O-big
‘His foot is big.’
2
Van Baarda calls these forms “participles”.
Possessive clauses in East Nusantara, the case of Tidore 297
b. A-wi-dohu i-la-lamo
3N.S-3SG.M.O-foot 3N.S+3SG.M.O-RED-big
‘His big foot.’
The Northeast Halmahera languages not only have inflected nouns and ad-
jectives, but even number expressions with human referents require a
predicative construction with an inflected numeral:
The possessor noun phrase may be internally complex. It may, for instance,
consist of a possessive construction:
Like most East Nusantara languages, but unlike some of the related North
Halmahera languages, e.g. Tobelo (Holton 2003), Tidore distinguishes alien-
able from inalienable possession. The latter unsurprisingly includes the ex-
pression of body parts, kinship terms, locative nouns and expressions of
part-whole relations (Chappell and McGregor 1996), but also items that are
‘baptized’, i.e. ceremonially attached to a possessor, such as ‘boats’,
‘houses’, and ‘names’ (van Staden 2006). The group of alienables include
notions of non-permanent or incidental ownership as in ‘my bag’, and asso-
ciation, such as ‘his photograph’ (the photograph in which he appears). The
difference between the two is that in the case of inalienable possession the
possessive marker may in all cases be reduced to the 3NH marker ma-
whereas in alienable possession, the possessive prefix must cross-reference
person and number of the possessor. The formal difference between alien-
able and inalienable possession is thus slight and I consider them to be
variations on the possessive construction, rather than different construction
types.
The possessive construction as a whole is typically an argument in a
clause. It may then be preceded by a preposition:
As this example shows it can be quantified also. Whether the quantifier has
scope over the possessum only or over the entire construction is ambigu-
ous. Tidore has no opposition between ‘their three children’ and ‘three of
their children’. Adjectival modification of the possessive construction has
so far not been attested in discourse; although in elicitation it is deemed
possible:
(19) ?Ona na-ngofa jang rimoi.
3PL 3PL.POS-child beautiful one
‘Their beautiful child; a beautiful child of theirs.’
We will return to modification and quantification in section 5.1, in which
implications for the analysis of possessive constructions are discussed.
An important difference between the Topic construction and the other two
is that the prosodic contour of the Topic construction shows a rise in pitch
on the Topic followed by a sharp drop, indicated by the slash in the lan-
guage line, whereas the other two have a regular declarative clause contour.
Although they are similar in form and appear to have a similar meaning,
sema ‘be’ and soma ‘additional participant’ must be distinguished. In most
contexts, they are in complementary distribution:
3
The other North Halmahera languages have a morpheme re (Sahu, Visser and
Voorhoeve 1987) or de (e.g. in Galela, van Baarda 1895) which, like se in Tidore,
functions both as a conjunction and as a preposition on (some) oblique argu-
ments. It is also found in expressions of possession, for example in Galela where
it may only be used with inanimate possessors ‘to indicate that it has something’
(van Baarda 1895: 246):
(i) O kurisi de ma gogocoho-ka
NM chair with 3SG.POS armrest-DIR
‘The chair with armrests.’ or ‘The chair has armrests.’
In this use it resembles the Tidore preposition soma. However, there is no regular
sound correspondence between Sahu r, Galela d and Tidore s. In fact, according
to Wada (1980: 503), Tidore should have had y instead.
302 Miriam van Staden
other North Halmahera languages. Instead, some of the related North Hal-
mahera languages have an existential predicate ka, which in Tidore is found
only as a bound morpheme in the locative predicates (e.g. ka-tai ‘be located
seawards’, ka-tau ‘be located upwards’, etc.) and in the negative existential
predicates ka-ua ‘be not’, ka-yang ‘be not yet’ and ka-rewa ‘be no more’.
The fourth construction type involves the bound morpheme due ‘posses-
sion(s)’ that takes the position of the possessum. It is considered a ‘dummy’
possessum element since it does not occur outside the possessive construc-
tion. In function it resembles the predicative possessor constructions, as
found in English when the possessor is the predicate and the possessum its
argument (‘the red one is mine’) or vice versa (‘mine is the red one’). This
construction type will be referred to as ‘generic possession’, after the ‘ge-
neric’ element due. But Tidore has no set of independent possessive pro-
nouns that could function as the predicate. Instead, the equivalent in Tidore
involves non-verbal predication with two NPs, one of which is a possessive
construction built on a ‘dummy’ possessum due. This possessive construc-
tion may be the argument of the construction, as in (25), where the predicate
gives the identity of the possessed entity, or it may be the predicate, as in
(26):
The two sentences answer different questions. In the first example, (27), the
utterance is a response to the question which one she owns, while in the
second, (28), it is an answer to the question who owns (or should own, as in
this case) the entity:
Heine’s study of possession led him to conclude that in all known lan-
guages clear differences show up between ‘attributive possession’ and
‘predicative possession’ (1997: 26). In the former, the possessor and pos-
sessum are generally contained in a noun phrase and the possessor func-
tions as an ‘attribute’ of the possessum; in the latter, the relation between
the possessum and the possessor is expressed in a predication. In the former
possession would be presupposed, in the latter possession is asserted. In
terms of Hengeveld and Mackenzie’s (2008) framework, following Dik
(1989, 1997), it could be said that in the former case the possessor further
restricts the potential set of referents designated by the possessum (from
‘all possible entities X’ to ‘the entities X belonging to Y’) while in the latter
a relation of possession is predicated of the possessor and possessum argu-
ments. The expression of both types may rely on the same conceptual struc-
ture and even on the same linguistic form, but ‘clear differences’, according
to Heine, may always be found. Tidore is an exception. The evidence pre-
sented here shows that one construction is used whether possession is pre-
supposed or not. The possessive construction in Tidore is a type of predica-
tive possession. But this predication is regularly embedded as an argument
of another predicate in which case the entire construction may used referen-
tially. This is the typical case where possession is presupposed, i.e. where
an attributive possessive construction would be expected. In brief, the
structure is: [Y pos-X] for ‘Y has X’, which is embedded as an argument
into [[Y has-X] Verb] ‘Y’s X Verbs’.
304 Miriam van Staden
4
Yes /no questions in Tidore are formed by adding ‘or no’ to the questioned
clause. In conjunction with aspectual modifiers rai ‘already’ and moju ‘still’,
their negative counterparts are used, so rai bolo yang ‘yet or not yet’ (as in ex-
ample (33)), moju bolo rewa ‘X still or no more’.
Possessive clauses in East Nusantara, the case of Tidore 305
In discussions with native speakers, these sentences are never felt to be in-
complete or to require a context as is often the case with non-clausal utter-
ances.
It is notable that when used as an independent utterance the possessive
construction typically has a modal or aspectual modifier. In the context of
Tidore this may be an important fact since historically the modal and aspec-
tual modifiers are intransitive verbs, taking as their only argument the event
description that is in their scope. In some of the related North Halmahera
languages they still carry person marking like other verbs, and even in Ti-
dore some of the modal and aspectual modifiers regularly function as main
predicates, e.g. rewa ‘not anymore’ (cf. example (31) above):
In the first reading it is the possessive relation that is predicated, but in the
second possession of sons is presupposed and it is just number that is predi-
cated. The former reading, without the presupposition, is expected when the
predicate is the possessum; the latter reading is expected if the possessive
construction is a referential phrase and the predicate is duga rimoi ‘only
one’.5 A similar situation arises with numerals:
Tidore has no distinction between ‘one of her children’ and ‘her one child’.
In the first reading the numeral has scope over the entire possessive con-
struction (cf. the paraphrase ‘as for her children, one is crying’), and in the
second it scopes over just the possessum (cf. she has one child; it cries).
5
A reading in with naunau as the predicate is excluded here on prosodic grounds
because there is no drop in pitch after i-ngofa and on syntactic grounds because
in that case it could not have be followed by duga rimoi ‘only one’. But it is in
theory possible to say:
(ii) Una i-ngofa [nau-nau.]pred
3SG.M 3SG.M.POS-child RED-male
‘His child is a boy.’
Possessive clauses in East Nusantara, the case of Tidore 307
Perhaps the most striking property of the possessive construction is that the
possessum behaves like a verbal predicate with respect to the placement of
the negative focus particle kama. Tidore negative particles ua ‘not, no’,
rewa ‘not anymore’, ifa ‘don’t’, yang ‘not yet’ all occur in clause final po-
sition. When focus is on the main predicate, a particle kama may directly
precede it. This applies equally to verbal and nominal predicates:
In this last example, the noun kolano ‘king’ predicates a property (function,
profession) of some person (recall example 5). Because the negative parti-
308 Miriam van Staden
The focus particle cannot precede noun phrase arguments, as (44) shows,
but in clauses containing a possessive construction kama may occur before
the possessum, as in (45) to (47):
Similar evidence comes from the type of complement that loan conjunction
untuk takes. As we saw earlier, beneficiaries may be expressed through the
possessive construction. An alternative to this construction is the use of a
purpose clause introduced by the Malay loan word untuk ‘for’. In Standard
Indonesian this is a preposition directly marking the beneficiary, as in (50).
In Tidore, the complement of untuk cannot be a noun phrase, hence the un-
grammaticality of (51), but must be clausal, as in (52):
Indonesian
(50) Apa surat ini untuk saya?
what letter this for 1SG
‘Is this letter for me?’
phrase, as in (55) (see section 6 for more cases of analogy in structure be-
tween Tidore and North Moluccan Malay): 6
This demonstrates that the possessive construction groups with clauses and
not with noun phrases.
6
In formal contexts Standard Indonesian may interfere with North Moluccan Malay.
Possessive clauses in East Nusantara, the case of Tidore 311
The fact that adjectives and verbs occur in the possessive construction in a
syntactic position typically associated with nouns, but not as heads of noun
phrases, may be explained by the predicative rather than referential function
that the head of the possessive construction has. The reason that the vast
majority of possessums are nominal is semantic rather than structural: what
is possessed is typically a physical object expressed by a noun.
There is no need to assume nominalisation in these cases. In actor
clauses, the main predicate is typically verbal, although nominal and adjec-
tival predicates are also attested, while in possessor-clauses the main predi-
cate is typically nominal, although both adjectival and verbal predicates
also occur. Contrast this to the nominalisation process involving the prefix
ma (also for 3NH.POS): 7
Here, there is evidence of derivation, since the ma-prefixed stem takes actor
marking:
7
Again in North Moluccan Malay we find the literal equivalent:
(iii) Mina ma- jang sado! Tidore
de pe cantik sampe! North Moluccan Malay
3SG(.F) POS beautiful until
‘She is such a beauty!’
312 Miriam van Staden
So far the arguments supported the analysis of the Tidore possessive con-
struction as basically ‘clausal’. However, although historically this may be
the case, and synchronically also the construction shows clausal properties,
there are several ways in which possessive constructions differ from actor-
clauses and these could be taken as counter-arguments against this clausal
analysis.
First of all, unlike typical actor clauses, possessive constructions do not
take aspectual or temporal modifiers when functioning as arguments of
verbs. And as independent utterances they are never modified by modal aux-
iliaries such as dadi ‘be able to’, or maya ‘want to’:
5.3. Evaluation
The various arguments that were presented show that Tidore lacks a dis-
tinction between an attributive and predicative possessive construction. The
possessive construction has clausal properties also when possession is pre-
supposed. The result is that in Tidore there is a split in the system of argu-
ment marking on the predicate: transitive predicates cross-reference the
first argument by means of a ‘subject’ or ‘actor’ prefix, as do intransitive
predicates that describe actions, states, experiences, etc., except those
predicates that give a possessive relation, because these take a ‘possessor’
first argument. This gives the regular oppositions illustrated in the introduc-
tion, and repeated here:
314 Miriam van Staden
contains a possessive ligature that is derived from the verb punya ‘have’:
pu in Ambonese Malay (Tjia 1997; van Minde 1997), pe in Irian and North
Moluccan Malay. Where the varieties of Eastern Indonesian Malay differ is
whether punya may be used in full in attributive possessive constructions,
and whether this construction is also used to express benefaction. The
claims in this section apply only the variety spoken on Tidore, although
following native speaker practice it is referred to as North Moluccan Malay
which would cover all of the North Moluccas.
Where languages are in prolonged contact, the systems may converge to
a greater or lesser extent. The diachronic process underlying such syntactic
convergence of systems has been referred to as ‘metatypy’ (Ross 1996,
1997, 2001, 2006). In metatypy ‘the syntax of one of the languages of a bi-
lingual speech community is restructured on the model of the syntax of the
speaker’s other language’ (Ross 2006). The result is very similar ways of
speaking in languages that are in contact. In the North Moluccas, in particu-
lar in Ternate, Tidore and Taba, this has occurred on a large scale. We find
it for instance in the use of spatially loaded terms in the regional Malay va-
rieties (Bowden 2005). The North Moluccan Malay expressions appear to
be literal copies from the indigenous varieties:
8
Note that Tidore has distinct morphemes for ‘sea’ (ngolo), moving seawards
(hoo) and seaward location (=tai), where Malay has only lao ‘sea’, showing that
Tidore was the model for North Moluccan Malay rather than vice versa.
9
In North Moluccan Malay this locative corresponds to the locative preposition
in Standard Indonesian; in Tidore the locative is a predicate deriving prefix.
316 Miriam van Staden
In the beneficiary expressions, too, the choice is between the full verb and
the reduced form:
In predicative possession only the verb may not be reduced, giving a formal
distinction between the two functions. And in generic possession also the
full form of the verb is obligatory, but the similarity in structure to Tidore is
again striking:
The question that this parallel raises is whether perhaps the Tidore con-
struction was reanalysed as clausal on the basis of its similarity to the Malay
possessive construction with its obvious clausal history. But the opposite is
also possible. Perhaps the North Moluccan Malay construction evolved un-
der the influence of the clausal constructions in the indigenous languages of
the area. A third possibility is of course that the two happen to be similar
synchronically, but have independent histories. There are two arguments
against the first scenario. First, clausal possessive constructions are a char-
acteristic of all North Halmahera languages as will be demonstrated in the
next sections. Furthermore, such constructions are found throughout East
Nusantara. Second, the North Moluccan Malay possessive construction has
the possessor-possessum order which is unusual for an Austronesian lan-
guage. Malay dialects spoken in western Indonesia have the usual posses-
sum-possessor order. But this order is common for both Papuan and Austro-
nesian languages in East Nusantara. It appears that the indigenous Papuan
languages influenced the Austronesian languages in this area. If the clausal
analysis in Tidore is a case of metatypy, we would need to assume that
North Moluccan Malay first took the order possessor-possessum from lan-
guages like Tidore, and then introduced a clausal construction to express this
relation of possession. And this construction was then subsequently bor-
rowed back into Tidore. Complex scenarios like these should not always be
ruled out, but it would seem more likely that the influence has been the other
way round. The North Moluccan Malay construction is based on the indige-
nous languages of the area in which it is spoken. For the same reason it is
unlikely that the Malay construction developed completely independently
from the indigenous languages of the area. In the following two sections the
expression of possession in East Nusantara as the linguistic area that Tidore
forms part of, and in the North Halmahera Family will be explored in order
to examine to what extent the Tidore construction is a product of its neigh-
bours or ancestors.
7. East Nusantara
East Nusantara is roughly the area east of Bali and Sulawesi up to and in-
cluding the Bird’s Head of Papua and perhaps even further East along the
North coast of New Guinea (van Staden and Reesink 2008; Klamer et al.
2008; Klamer 2003; Baird 2002). This area is characterized historically by
high mobility of people and intense contact between people and languages.
It is a complex patchwork of languages. Here different branches of Aus-
318 Miriam van Staden
Figure 1. East Nusantara. Papuan languages are found in Papua and in the encircled
areas; Austronesian languages are spoken elsewhere in the archipellago.
Map adapted from Klamer, et al. (2008)
Possessive clauses in East Nusantara, the case of Tidore 319
Biak
(73) … ko-swar min ko-ve=s-i
1PL.INC-love member 1PL.INC-POS=3PL .ANIM-SPC
‘… we love our neighbours.’ (Van den Heuvel 2006: 437)
Buru
(74) Kawasan p-em-nake-k geba rua ute tinge […]
head CAUS-STAT-3SG.POSS-k person two DAT 3SG […].
‘The village head put two people at his disposal […]’
More generally the possessor affixes correspond to either the subject or the
object markers, also when no verbal element is found. For instance in May-
brat the possessor is expressed by a subject prefix in inalienable possessive
constructions:
Maybrat
(75) Fnia m-ao.
woman 3U-foot
‘The woman’s foot.’ (Dol 1999: 149)
And other Papuan languages of the Bird’s Head such as Meyah (Gravelle
2004: 171), Moi (Menick 1995: 65), Mpur (Odé 2002: 62) and Hatam
(Reesink 1999: 49) are identical.
The Southern Bird’s Head language Inanwatan has two highly similar sets
of prefixes for inalienable possession (de Vries 2004: 29) and object marking
(de Vries 2004: 36) although the two differ in certain morpho-phonological
properties. As in the North Halmahera languages, the possessive markers in
Inanwatan have optional arguments that the object markers lack.
In Adang and Abui, two Alor-Timor-Pantar languages, we find that per-
son prefixes marking possessor on nouns also cross-reference objects, though
not subjects. In Adang only a particular subset of verbs is inflected:
Adang
(76) N-e na-fel mi habu.
1SG-GEN 1SG-ear COMP wide
‘My ears are wider (e.g. than yours)’ (Haan 2001: 134)
Adang
(77) Sa na-tan.
3SG.NOM 1SG-ask
‘S/he asked me.’ (Haan 2001: 46)
Possessive clauses in East Nusantara, the case of Tidore 321
But in Abui there are even two sets of possessor prefixes that correspond to
two sets of second argument markers. The prefix for inalienables in (78a)
corresponds to the ‘patient’ marker in (78b), and the prefix for alienables in
(79a) corresponds to the ‘locative’ category in (79b), which includes bene-
ficiaries and ‘themes’ and which may be thought of as locations (Kratochvíl
in progress).
Abui
(78) a. Na-min
1SG.INAL-nose
‘My nose.’
b. Me na-dak-e.
come 1SG.PAT-clutch-IPFV
‘Come and hold me tight.’
Abui
(79) a. Ne-wil.
1SG.AL-child
‘My child.’
b. Simon ne-l to-ha-loi.
Simon 1SG.LOC-give DISTR.REC-3ii.PAT-chase
‘Simon chases me.’
And further east in the Cenderawasih Bay inalienable nouns in Yawa have a
prefix identical to the undergoer prefixes used on transitive verbs and stative
verbs.
Yawa
(80) In-aneme.
1SG.OBJ-hand
‘My hand.’ (Jones 1986; see also Klamer et al. 2008)
In the final part of this chapter I compare the Tidore possessive construc-
tion to the closely related languages of the North Halmahera family. The
first obvious question is whether a clear distinction between attributive and
predicative possession can be made in these languages, whether the posses-
sive constructions in these languages, too, appear to be clausal. But the
other North Halmahera languages are also interesting because they appear
to be rather conservative compared to Tidore in a number of respects, in-
cluding the systems of person cross-referencing on both verbal predicates
and possessum noun phrases. It is argued here that these systems of cross-
referencing have some important clues to a possible account of the present-
day rather unique distinction between actor prefixes and possessor prefixes
in Tidore.
The North Halmahera family is usually presented as consisting of ten
languages.10 However, six of these are so closely related that they may be
considered dialects of one language ‘Northeast Halmaheran’ (Voorhoeve
1988): Galela, Tobelo, Tobaru, Pagu, Loda, Modole, and similarly Ternate
and Tidore are dialects of one language (cf. Figure 2). The Northeast Hal-
mahera languages are the most archaic in terms of retention of sounds from
proto-Halmaheran (Wada 1980), SOV word order and postpositions, use of
articles and noun class markers, the spatial deictic system, and also the oc-
currence of object prefixes, which in Tidore as well as in Ternate and West-
10
The data for this section were taken from various sources, most of them dating
from the early 20th century, but some more recent studies are also available.
Where the reports are in disagreement I give specific references. For details on
Galela see Van Baarda (1908) and Shelden (1998, 1991), on Tobelo see Hueting
(1936) and Holton (2006, 2003), on Pagu Wimbish (1991) and on Tabaru Fortgens
(1928). For comparative studies see in particular Van der Veen (1915); Capell
(1975); Voorhoeve (1988, 1987, 1987); and Wada (1980). I have omitted from
the presentation some details on phonetic realization, e.g. differences between
long and short vowels and stress where they are non-pertinent to the present
discussion. Galela has a distinction between a dental /d/ and a retroflex //
which are represented by the single grapheme d in this paper. I have changed
the glosses of some of the examples to align the terminology or I have added
glosses where the original sources (usually the older ones) do not give them.
Also, most older sources as well as some of the more recent ones (e.g. Wimbish)
give the verbal markers and the possessive markers as separate words, but there
is enough morpho-phonological evidence to warrant their presentation as pre-
fixes in all North Halmahera languages.
Possessive clauses in East Nusantara, the case of Tidore 323
Makian (van der Veen 1915) have been lost. The possessive constructions in
all the North Halmahera languages show strong resemblance, although
again the Northeast Halmahera languages appear to be more conservative
than Tidore.
The first question is to what extent the possessive constructions in the other
North Halmahera languages have clausal properties like Tidore. Although
this issue has not been addressed for any of the languages in the literature,
text material shows that the possessive construction may be used predica-
tively in at least several different North Halmahera languages, e.g. in Tobelo:
324 Miriam van Staden
Pagu
(82) ma kolan awi-ngoak mosoles dauk
ART king 3M.POSS- child bachelorette there:south
m-o-matetengo, ami-penjagaan o macan de o garuda …
3SG.F-SU-one 3SG.F.POSS-guard ART tiger and ART eagle
‘The king’s only daughter is there, and she has a tiger and an eagle
for guards.’ (Wimbish 1991: 136)
(83) Awi-lamo.
3SG.M.POS-big
‘His bigness; how big he is!’ (van Baarda 1908: 38)
The Northeast Halmaheran languages and Sahu all have pronoun invento-
ries that are very much like the pronouns in Tidore and Ternate. The per-
sonal pronouns are without exception historically morphologically complex
as shown in Table 1 where the roots of the pronouns are underlined:
singular plural
Tidore Tobelo Tidore Tobelo
1 EXC ngo-ri ngo-hi ngo-m ngo-mi
1 INCL – – ngo-ne ngo-ne
2 ngo-na ngo-na ngo-n ngi-ni
3M u-na u-nanga
3F mi-na mi-nanga o-na o-nanga
3 NH e-na e-nanga e-na e-nanga
The suffix na(nga) found with the third person pronouns has its origin in the
demonstratives, cf. re-na(nga) ‘here’ and ge-na(nga) ‘there’. It may be typed
as a ‘deictic pointer’. First and second person have a prefixing element ngo-
that also occurs before women’s names or kin terms in all North Halmahera
languages, e.g. ngo-Desi ‘Desi’ (see e.g. van der Veen 1915: 194; van Staden
2000: 92; Visser and Voorhoeve 1987: 53). In Sahu, ngo is in opposition
with a marker a before nouns referring to men, but this is not found in the
other languages. Its presence on first and second person pronouns could be
explained if ngo- is considered a kind of vocative element used to refer to
those present in the discourse. However, the noun markers ngo- (and a- in
Sahu) is not used before proper names or kinship terms when addressing a
person (Visser and Voorhoeve 1987: 36).
Where Tidore and Ternate (cf. Hayami-Allen 2001: 136–137) are rather
different from the other languages is that they lack a set of possessive pro-
nouns, which can be formed morpho-syntactically in the other members of
the North Halmahera family.
In the Northeast Halmahera languages as well as in Sahu the personal
pronouns may be preceded by a specific marker to give what appear to be
possessive pronouns:
326 Miriam van Staden
Pagu
(84) Oli to-una ai-kadu dumoi, w-a-gogon-oka
so POS-3SG.M 3SG.M.POS-sack one 3SG.M-NH-hide-NFUT
‘So he hid his sack (lit. so his sack, he hid it).’ (Wimbish 1991: 158)
Tobelo
(85) To-munanga ami-gakana nenanga?
POS-3SG.F 3SG.F.POS-knife this
‘Is this her knife?’ (Hueting 1936: 353)
Sahu
(86) Wala ge t-a Salaka
house that POS-NM Salaka
‘That house is Salaka’s’ (Visser and Voorhoeve 1987: 54)
To start with the first question on the relation between locative preposi-
tions te and toma in Tidore and possessive to, it is indeed tempting to see
the two as related. This would then suggest an instance of locative-to-
possessive grammaticalisation (Heine 1997). However, the status and his-
tory of such locative prepositions in the North Halmahera languages is un-
clear. Sahu has a preposition toma11 but its distribution has not been de-
scribed and it is not in opposition with te but with ra (John Severn, pers.
comm.). A form to occurs infrequently in Tabaru (Fortgens 1928: 389–390)
to derive verbs with a locative component, e.g. toobiri to arrive by night,
and Fortgens suggests that the many names of languages beginning in to
(Togutil, Tobelo, Tabaru and even Tidore, regularly pronounced as Todore)
may also find their origin in a to- prefixed root. But signs of a locative
preposition to are completely absent in other Northeast Halmahera lan-
guages such as Pagu, and are generally unexpected in SOV languages.
In relation to the second question, pronouns preceded by to may occur
dependently and independently. In other words, they may occur before
noun phrases as apparent attributes (my, your, etc.), or they may constitute
noun phrases on their own (mine, yours, etc.) as in the following example:
Tobelo
(87) to-ngohi ahi-tiwi nenanga? koali, to-ngohi.
POS-1SG 1SG.POS-money this no POS-1SG
‘Is this my money? No, (it’s) mine.’ (Hueting 1936: 353)
In all the North Halmahera languages that have to/tV it also occurs on the
question word particle, as in (88):
Tobelo
(88) to-nago-nanga manga-tau nenanga?
pos-who-3P 3P.POS-house this
‘Whose house is this?’ (Hueting 1936: 355)
The marker is, however, not obligatory – at least not in all languages and in
all contexts. What precisely determines its presence is still not entirely clear.
In Galela it is absent in contrastive expressions and after conjunctions or
adverbs (van Baarda 1908: 70), but not in Pagu and Loda. And in Pagu it is
11
Visser and Voorhoeve (1987) do not mention it in their sketch grammar although
it occurs in one of the examples in contrast to re the preposition for company /
instrument (cf. Tidore se) (p.56).
328 Miriam van Staden
Pagu
(89) … wa-make nage de ami lomang-oka.
3M.SG.A:3NH.O-see that with 3SG.F.POS name-N.FUT
‘… he saw that it had her name (lit. he saw it already with her name).’
(Wimbish 1991: 133)
Pagu
(90) Ei ngoi ai-ngoak iwa.
hey 1SG 1SG.POS-child gone
‘Hey, I don’t have any children.’ (Wimbish 1991: 135)
Pagu
(91) To-ngoi naga alat-oka.
POS-1SG ‘is’ tool-NFUT
‘my tool is already here.’ (lit. mine is tool-already)
(Wimbish 1991: 135)
Tobelo
(92) Nenanga ma-hangaji ai-ja? Koali,
this 3NH.POS-district.head 3SG.M.POS-fishing.net NEG,
to ma-uku.
POS 3NH.POS-hukum
‘Is this the district head’s fishing net? No, the hukum’s.’
(Hueting 1936: 353)
Pagu
(93) Ma kolan awi-ngoak mosoles dauk
ART king 3SG.M.POS-child bachelorette there:south
m-o-matetengo, ami-penjagaan o macan de o garuda.
3SG.F-S-one 3SG.F.POS-guard ART tiger and ART eagle
‘The king’s only daughter is there, and she has a tiger and an eagle
for guards.’ (Wimbish 1991: 136)
Summarising, all North Halmahera languages except for Tidore and Ternate
have a marker to that occurs before the personal pronouns, and at least in
some languages it also occurs before possessor noun phrases. In most lan-
guages the distribution of to is restricted at least in part by the occurrence
of adpositions and adverbs, and in Sahu discourse considerations also play
a role in its distribution. It appears then that to does not so much derive
possessive pronouns as mark the semantic role of the possessor in the pos-
sessive construction. This ties in with a predicative analysis for these lan-
guages also. The marker distinguishes possessor ‘first arguments’ from
other first arguments as a distinct semantic role in the clause, giving the
possessive relation as a special event type.
But is there further support for predicative possession also in the other
North Halmahera languages? And how does the possessive construction in
these languages shed any light on the Tidore situation? In order to address
this question it is necessary to examine the possessive prefixes and the sub-
ject and object markers in the North Halmahera languages in some detail.
8.3. Possessive and other person prefixes in the North Halmahera languages
The possessive prefixes in the North Halmahera languages all bear close
resemblance. The differences that we do find are, however, rather signifi-
cant. Also the differences in the systems of verbal cross-reference marking
between Tidore on the one hand and the Northeast Halmaheran languages
on the other is important and may shed light on the curious synchronic
situation in Tidore. In this section I explore the possible origin of the pos-
sessive markers in Tidore as related to the object cross-reference markers
that are found in the Northeast Halmaheran languages but not in Tidore.
Table 2 gives an overview of the possessive prefixes in a number of North
Halmahera languages. The bracketed elements are optional, often charac-
teristic of informal, spoken language.
330 Miriam van Staden
possessive prefixes
Tidore Tobelo Tabaru Galela Sahu
singular
1 ri ahi ai (a)i (a)ri
2 ni ani (a)ni (a)ni (a)ni
3M i ai (ani) wi (a)wi ai
3F mi ami (a)mi (a)mi (a)mi
3 NH maa ma ma ma ma
plural
1 EXC mi mia mi(a) minga amia minga
1 INCL na nanga nanga nanga nanga
2 ni nia ni(a) ninga ania ninga
3 H.PL na manga manga manga ma(nga)
3 NH ma ma ma ma ma
a
ma is incongruous in all aspects. It is likely that it has its origins outside the pos-
sessive paradigm, possibly as a stative verb marker (see Holton 2006) and then
saw its distribution widened to include relations of associated and relational
possession (and reflexivity). This use then developed into the distinction be-
tween human plural and non-human possessors, which is otherwise absent in the
paradigms of person affixation.
Table 3. Pronouns, subject and object markers in the North Halmahera languages
8.3.1. Prefix a-
Pagu
(94) t-o-sano
1SG-SU-ask
‘I ask (something)’ (Wimbish 1991: 46)
Pagu
(95) t-a-sano
1SG-NH-ask
‘I ask a specific question’ (Wimbish 1991: 46)
This means that if we disregard the inclusive form and consider the endings
in -o- to be separate morphemes, only the aberrant first person singular sub-
ject prefix is different from the object prefix and the possessive marker. So
then the question is whether the possessive prefix would be derived from
the object cross-reference marker or whether it could have derived directly
from the pronominal roots.
Yet there is another link to the object cross-referencing paradigm. In the
North Halmahera languages, the possessive markers typically had a prefix a-
and it is possible that this prefix is the same a- as the one occurring before
the object markers when these refer to referents that are non-human, directly
affected by the verb and specified or definite. One reason to assume this is
that at least according to Wimbish’ account, the Pagu language does not
allow for the possibility of having human denoting objects that are affected
by the event, and the other Northeast Halmaheran languages appear to be
similar in this respect. In the following paragraphs, I will try to make a
plausible case that the semantic role of possessor could be just this: a human
object that is affected by the event. If that is the case, then the generalisation
would be that all affected objects take a and all unaffected objects have o,
but how does that explain that there are no transitive verbs that have af-
fected human objects? The hypothesis is then that there is a restriction on
participant structure which stipulates that each predicate may have only one
argument that is actively involved in the event, either as an agent or as a
patient. This means that if there is a human agent, the patient cannot be an
affected human. But if there is no agent, then this leaves the possibility of
having an affected human object. And this is the possessive construction: a
predication without a subject, but with an affected human object.
The argumentation is as follows. First, we follow Wimbish and assume
that the subject prefixes are indeed historically bi-morphemic, consisting of
the pronominal roots followed by o in intransitives of when the second argu-
ment is either human or not affected by the predicate (“oblique”) as in Table 4.
334 Miriam van Staden
singular plural
person ‘intransitive’ fused person pluralizer ‘intransitive’ fused
root marker form root marker form
1EXC h/ri o to mi ngaa o mi/miyo
1INC n/f/w/p/h o fo
2 ni o no ni nga o ni/niyo
3M u/w o wo
i/y nga o yo/i/ya
3F mi o mo
3NH e o yo
a
nga is the form of the pluralizer that survives today, but it is actually more likely
that it was ka or /a, as evidenced by the various languages that have a /k/ in the
3pl object prefixes.
For the irregular plural forms we assume that the old pluralizer, now lost in
the subject prefixes but still visible in the possessive markers in several
languages, prevented the kind of fusion that took place with the singular
forms, explaining their irregular endings in –i, as in Tidore, and the forms
miyo and niyo in the Northeast Halmahera languages. These subject markers
followed by o are again followed by the pronominal roots cross-referencing
the object when this is human. Fusion of subject and object prefix takes
place in a number of cases, reducing for instance the 3NH object pronoun e
to ‘zero’. In the case of a non-human affected object (direct object), the
marker is not o but a. Otherwise the process is identical, giving an identical
fused paradigm with the one presented in Table 4. This account gives a ‘de-
fective paradigm’ as in Table 5 as it appears to exclude the possibility of
affected human arguments:
-o- -a-
single argument + –
unaffected non-human object + –
unaffected human object + –
affected non-human object – +
affected human object ? ?
Possessive clauses in East Nusantara, the case of Tidore 335
12
Sheldon (1991) discusses only miya and niya, but Van Baarda also mentions the
other two forms.
336 Miriam van Staden
9. Conclusion
References
Allen, W. Sidney
1964 Transitivity and possession. Language 40 (3): 337–343.
van Baarda, M. J.
1895 Woordenlijst Galelareesch-Hollandsch. Met Ethnologische Aanteek-
eningen, op de Woorden, die Daartoe Aanleiding Gaven. Den Haag:
Koninklijk Instituut voor Taal-, Land- en Volkenkunde van Neder-
lands Indië.
van Baarda, M. J.
1908 Leidraad bij het Bestudeeren van ’t Galela’sch Dialect, op het Eiland
Halmaheira. Den Haag: Nijhof.
Possessive clauses in East Nusantara, the case of Tidore 337
Baird, Louise
2002 A grammar of Keo: an Austronesian language of East Nusantara.
PhD thesis, Australian National University.
Bowden, John
2005 Language contact and metatypic restructuring in the directional sys-
tem of North Maluku Malay. Concentric: Studies in Linguistics 31 (2):
133–158.
Capell, Arthur
1975 The West Papua Phylum: General, and Timor and the areas further
West. In New Guinea Area Languages and Language Study. Volume
1: Papuan languages and the New Guinea Linguistic Scene, Stephen
A. Wurm (ed.), 667–716. Canberra: Pacific Linguistics.
Chappell, Hilary and William B. McGregor (eds.)
1996 The Grammar of Inalienability: A Typological Perspective on Body
Part Terms and the Part-Whole Relation. Berlin /New York: Mouton
de Gruyter.
Dik, Simon C.
1989 The Theory of Functional Grammar. Part I: The Structure of the
Clause. Dordrecht: Foris.
Dik, Simon C.
1997 The Theory of Functional Grammar. Part I: The Structure of the
Clause. Second, revised edition, edited by Kees Hengeveld. Berlin /
New York: Mouton de Gruyter.
Dol, Philomena H.
1999 A Grammar of Maybrat: A Language of the Bird’s Head, Irian Jaya,
Indonesia. PhD thesis, Leiden University.
Foley, Willam A.
1986 The Papuan Languages of New Guinea: Cambridge Language Sur-
veys. Cambridge: Cambridge University Press.
Fortgens, J.
1928 Grammaticale aantekeningen van het Tabaroesch, Tabaroesche volks-
verhalen en raadsels. Bijdragen tot de Taal-, Land- en Volkenkunde
84: 300 –544.
Gravelle, Gilles
2004 Meyah: an east Bird’s Head language of Papua, Indonesia. PhD thesis,
Vrije Universiteit.
Grimes, Charles E.
1991 A grammar of Buru. PhD thesis, Australian National University.
de Groot, Casper
1983 On non-verbal predicates in Functional Grammar: the case of pos-
sessives in Hungarian. In Advances in Functional Grammar, Simon
C. Dik (ed.), 93–122. Dordrecht: Foris.
Haan, Johnson Welem
2001 The grammar of Adang: a Papuan language spoken on the Island of
Alor East Nusa Tenggara – Indonesia. PhD thesis, University of Sydney.
338 Miriam van Staden
Hayami-Allen, Rika
2001 A descriptive study of the language of Ternate, the Northern Moluc-
cas, Indonesia. PhD thesis, University of Pittsburgh.
Heine, Bernd
1997 Possession: Cognitive Sources, Forces and Grammaticalization.
Cambridge: Cambridge University Press.
Hengeveld, Kees
1992 Parts of speech. In Layered Structure and Reference in a Functional
Perspective, Michael Fortescue, Peter Harder and Lars Kristoffersen
(eds.), 29–55. Amsterdam: John Benjamins.
Hengeveld, Kees and J. Lachlan Mackenzie
2008 Functional Discourse Grammar. Oxford: Oxford University Press.
Holton, Gary
2003 Tobelo. Languages of the World /Materials, Vol. 328. Munich: Lincom
Europa.
Holton, Gary
2006 The relational noun marker in Tobelo (Northeast Halmaheran). Tenth
International Conference on Austronesian Linguistics (10-ICAL).
Puerto Princesa City, Palawan, Philippines.
Hueting, A.
1936 Iets over de spraakkunst der Tobeloreesche taal. Bijdragen tot de
Taal-, Land- en Volkenkunde 94 (3): 295–407.
Jelinek, Eloise and Fernando Escalante
1988 ‘Verbless’ possessive sentences in Yaqui. In In Honor of Mary Haas:
From the Haas Festival Conference on Native American Linguistics,
William Shipley (ed.), 411–429. Berlin/New York: Mouton de Gruyter.
Jones, Linda K.
1986 The question of ergativity in Yawa, a Papuan language. Australian
Journal of Linguistics 6: 37–55.
Klamer, Marian A. F.
2003 ‘Report’ constructions in Kambera (Austronesian). In Reported Dis-
course: A Meeting Ground for Different Linguistic Domains, Tom
Güldemann and Manfred von Roncadorm (eds.), 323–340. Amster-
dam: John Benjamins.
Klamer, Marian A. F., Ger P. Reesink and Miriam van Staden
2008 East Nusantara as a Linguistic Area. In From Linguistic Areas to
Areal Linguistics, Pieter Muysken (ed.), 95–149. Amsterdam /Phila-
delphia: Benjamins.
Kratochvíl, Frantiek
2007 A grammar of Abui: A Papuan Language of Alor. PhD Thesis, Leiden
University.
Possessive clauses in East Nusantara, the case of Tidore 339
Lichtenberk, Frantisek
1985 Possessive constructions in Oceanic languages and in Proto-Oceanic.
In Austronesian Linguistics at the 15th Pacific Science Congress,
Andrew Pawley and Lois Carrington (eds.), 93–140. Canberra: Pacific
Linguistics.
Lichtenberk, Frantisek
2002 The possessive-benefactive connection. Oceanic Linguistics 41 (2):
339–474.
Margetts, Anna
2004 From implicature to construction: emergence of a benefactive con-
struction in Oceanic. Oceanic Linguistics 43 (2): 445–468.
Menick, Raymond
1995 Moi: a language of the West Papuan Phylum. a preview. In Tales from
a Concave World: Liber Amicorum Bert Voorhoeve, Connie Baak,
Dick van der Meij and Mary Bakker (eds.), 55 –73. Leiden: Projects
Division, Department of Languages and Cultures of South-East Asia
and Oceania.
van Minde, Don
1997 Malayu Ambong: Phonology, Morphology, Syntax. Leiden: CNWS.
Odé, Cecilia
2002 A sketch of Mpur. In Languages of the Eastern Bird’s Head, Ger P.
Reesink (ed.), 45–107. Canberra: Pacific Linguistics.
Reesink, Ger P.
1998 The Bird’s Head as a Sprachbund. In Perspectives on the Bird’s
Head of Irian Jaya, Indonesia, Jelle Miedema; Cecilia Odé and Rien
A.C. Dam (eds.), 603–642. Amsterdam: Rodopi.
Reesink, Ger P.
1999 A Grammar of Hatam. C–146. Pacific Linguistics. Canberra: ANU.
Reesink, Ger P.
2000 West Papuan languages: roots and development. Paper presented at
Papuan Pasts, Canberra.
Reesink, Ger P. (ed.)
2002 Languages of the Eastern Bird’s Head. Canberra: Pacific Linguistics.
Ross, Malcolm
1996 Contact-induced change and the comparative method: cases from
Papua New Guinea. In The Comparative Method Reviewed: Regular-
ity and Irregularity in Language Change, Mark Durie and Malcolm
Ross (eds.), 180–217. New York: Oxford University Press.
Ross, Malcolm
1997 Social networks and kinds of speech community events. In Archae-
ology and Language, Roger M. Blench and Matthew Spriggs (eds.),
209–261. London: Routledge.
340 Miriam van Staden
Ross, Malcolm
2001 Contact-induced change in Oceanic languages in North-West Mela-
nesia. In Areal Diffusion and Genetic Inheritance, Alexandra Y.
Aikhenwald and Robert M.W. Dixon (eds.), 134–166. Oxford: Ox-
ford University Press.
Ross, Malcolm
2006 Metatypy. In The Encyclopaedia of Language and Linguistics, Vol. 8,
Keith Brown (ed.), 95–99. New York: Elsevier.
Schmidt, Pater Wilhelm
1895 Die Sprachlichen Verhältnesse von Deutsch-Neuguinea. Zeitschrift
für Afrikansiche, Ozeanische und Ostasiatische Sprachen II (4).
Shelden, Howard
1991 Galela pronominal verb prefixes. In Papers in Papuan Linguistics,
Tom Dutton (ed.), 161–175. Canberra: Pacific Linguistics.
Shelden, Howard
1998 Transitivity and Galela pronominal reference. In SIL Electronic
Working Papers SILEWP 1998-005. (http://www.sil.org/silewp/
1998/005/) Dallas, Texas: SIL International.
van Staden, Miriam
2000 Tidore: A linguistic description of a language of the North Moluccas.
PhD thesis, Leiden University.
van Staden, Miriam
2006 The body and its parts in Tidore, a Papuan language of Eastern Indo-
nesia. Language Sciences 28: 323–343.
van Staden, Miriam and Ger P. Reesink
2008 Serial verb constructions in a linguistic area. In Serial Verb Construc-
tions in Austronesian and Papuan Languages, Gunter Senft (ed.),
17–54. Canberra: Pacific Linguistics.
Taylor, John R.
1989 Possessive genitives in English. Linguistics 27: 663–686.
Taylor, John R.
1996 Possessives in English: An Exploration in Cognitive Grammar. Ox-
ford: Oxford University Press.
Tjia, Johnny
1997 Verb serialisation in Ambonese Malay. MA thesis, University of
Oregon.
Van den Heuvel, Wilco
2006 Biak. Description of an Austronesian language of Papua. PhD thesis,
Vrije Universiteit.
van der Veen, Hendrik
1915 De Noord-Halmahera'se taalgroep tegenover de Austronesiese talen.
Leiden: van Nifterik.
Possessive clauses in East Nusantara, the case of Tidore 341
Verhaar, John W. M.
1995 Toward a Reference Grammar of Tok Pisin: An Experiment in Corpus
Linguistics. Oceanic Linguistics Special Publication, Vol. 26. Hono-
lulu: University of Hawai’i Press.
Visser, Leontine E. and C. Lambertus Voorhoeve
1987 Sahu-Indonesian-English Dictionary and Sahu Grammar Sketch.
Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en
Volkenkunde, Vol. 126. Dordrecht: Foris.
Voorhoeve, C. Lambertus
1987 The Non-Austronesian languages in the North Moluccas. In Hal-
mahera dan Raja Empat sebagai Kesatuan majemuk, E.K. M. Masi-
nambow (ed.). Jakarta: Lembaga Ekonomi dan Kemasyarakatan Na-
sional, Lembaga Ilmu Pengetahuan Indonesia.
Voorhoeve, C. Lambertus
1987 The masked bird: Linguistic relations in the Bird’s Head area. In
Peoples on the Move, Paul Haenen and Jan Power (eds.), 78–101.
Nijmegen: Centre for Australian and Oceanic Studies.
Voorhoeve, C. Lambertus
1988 The languages of the North Halmahera Stock. Papers in New Guinea
Linguistics 26: 181–209. Canberra: Pacific Linguistics.
de Vries, Lourens J.
2004 A Short Grammar of Inanwatan, an Endangered Language of the
Bird’s Head of Papua, Indonesia. Canberra: Pacific Linguistics.
Wada, Yuiti
1980 Correspondence of consonants in North Halmahera languages and
the conservation of archaic sounds in Galela. In The Galela of Hal-
mahera: A Preliminary Survey, Naomichi Ishige (ed.), 497–529.
Senri Ethnological Series, Vol. 7. Osaka: National Museum of Eth-
nology.
Wimbish, Sandra Gay
1991 An Introduction to Galela through the Analysis of Narrative Dis-
course. Ann Arbor: University of Michigan Press.
Wurm, Stephen A.
1982 Papuan Languages of Oceania. Ars Linguistica, Vol. 7. Tübingen:
Gunter Narr Verlag.
Possessive expressions in the Southwestern Amazon
1. Introduction
The Southwestern region of the Amazon river basin, which in the present
article is understood to be the region that covers the Bolivian and Brazilian
sides of the Guaporé, Mamoré and Upper Madeira rivers, is one of the
world’s linguistically most diverse places. The Southwestern Amazon, here
also referred to as the Guaporé region, harbours representatives of seven
different linguistic stocks, including Arawak, Chapacura, Macro-Jê, Nam-
bikwara, Pano, Tacana and the majority of the branches of the Tupi linguis-
tic stock. Furthermore twelve language isolates or unclassified languages
are spoken in the region. The majority of the approximately 50 languages of
the region are highly endangered with extinction, about half of them having
fewer than 50 speakers and a third ten or fewer. During the past two dec-
ades, a considerable number of initiatives to document and describe these
languages have fortunately been taken.
This article is meant to give a preliminary survey of the different ways
adnominal possession is expressed across a sample set of languages from the
Southwestern Amazon (see Map 1). In section 2 I will sketch the types of
adnominal possessive expression found in Kwaza, a language isolate spoken
by approximately 25 people. In section 3 I will discuss the similarities with
possessive expressions in other languages of the region: Aikanã (isolate),
Arikapú (Macro-Jê), Baure (Arawak), Kanoê (isolate), Latundê (Nambik-
wara), Mekens (Tupí) and Wari’ (Chapacura). The final section 4 contains a
table summarising the different possessive strategies. In view of the fact
that the languages discussed here belong to different families, and even in-
clude three isolates, explanations for certain similarities are probably found
in areal diffusion.
This article deals mainly with adnominal possession, which can be de-
scribed as a grammatical construction involving a modifying noun and a
head noun that expresses a possessive relationship between a possessor
(PR) and a possessum (PM), respectively. It contrasts with predicative pos-
session and external possession. Predicative possession constructions in-
volve the verb ‘to have’, or another verbal copula, and express a possessive
344 Hein van der Voort
Kwaza, which has been studied and described since the mid 1990s by the
author (e.g. van der Voort 2004), is a morphologically complex language
with a variable word order, although there is a clear tendency towards head-
final structures. The grammatical categories of Kwaza are verbs, nouns,
adverbs and particles. Most of the grammatical complexity is found in ver-
bal morphology, whereas nominal morphology is relatively simple. How-
ever, the highly productive nominalisation possibilities enable the creation
of morphologically very complex nouns. The morphology is mainly suffix-
ing. The verb is obligatorily inflected for person and mood, although mor-
phological ellipsis is attested under specific circumstances. The morpho-
logical structure of the verb is: root + optional derivation + obligatory
person inflexion + obligatory mood inflexion, as shown in this example: 1, 2
1
The Kwaza consonants /c/ and /x/ are pronounced as IPA retracted [t] and [s]
respectively. The vowel /y/ is pronounced somewhere between IPA [] and [].
Main word stress is marked by an apostrophe ['] preceding the stressed syllable.
346 Hein van der Voort
(1) cari-'nã-da-ki
shoot-FUT-1S-DEC
‘I will kill (it, e.g. the game animal).’
The verb that is unmarked for person is interpreted as having a third person
subject. Overt expression of arguments as pronouns or other independent
constituents is not obligatory. Nouns are not obligatorily inflected, there is
no nominal number marking, and the animate object case is not always ap-
plied. The distinction between inclusive and exclusive person reference is
found in both the pronominal and the bound person marking paradigms.
There are several basic moods, including declarative, interrogative, volitive,
imperative, negative imperative, etc. There are several moods for subordi-
nated adverbial clauses and there is a cosubordinated mood for clause
chains. Different nominalisers are used for attributive and complement
clauses. In the following subsections I will sketch the different possessive
strategies in Kwaza and analyse their morphosyntactic properties.
2
The following abbreviations are used: ART – article, AUX – auxiliary, BEN –
benefactive (verbal derivation), BER – beneficiary, C – coreferent, CAU – causa-
tive, CD – directional classifier, CL – classifier, COL – collective, COMIT –
comitative, COP – copula, CSO – cosubordination, DEC – declarative, DEM –
demonstrative, DET – detrimental, DR – directional, DS – different subject, F –
feminine, FUT – future, IMPF – imperfective, INCL – inclusive, INFL – Wari’
‘inflection’, INT – interrogative, LOC – locative case, M – masculine, N – neuter,
NOM – nominaliser, OBL – oblique case, P – plural, PM – possessum; POS –
possessive, POT – potential, PR – possessor; PROX – proximate, REF – referen-
tial, REFL – reflexive, S – singular, VOL – volitive. The numerals 1, 2, and 3
indicate respectively, first person singular, second person, and third person. The
hyphen (-) indicates a morphemic boundary; the equal sign (=) indicates compo-
sition or a clitic boundary; and the period (.) separates semantic units in a port-
manteau morpheme.
Possessive expressions in the Southwestern Amazon 347
The only exception to the pattern described here is the first person inclusive
pronoun txana, in the possessive use of which omission of the morpheme
-dy- is preferred, i.e. txana-h
‘ours (INCL)’ rather than txana-dy-h
.
Headless possessive constructions, in which an explicit PM is absent,
are frequently attested in Kwaza:
3
Reverse constituent order is also possible.
348 Hein van der Voort
(7) 'si-dy-h
I-POS-NOM
‘mine’
(8) *si-dy-ki
I-POS-DEC
‘It is mine.’
If the element -h
really does have a nominalising function under all cir-
cumstances, example (8) would be grammatical. In the following subsec-
tions, I will discuss the elements -h
(§ 2.2.1, § 2.2.2) and -dy- (§2.2.3), in
order to better understand the combination -dy-h
.
Possessive expressions in the Southwestern Amazon 349
4
Even though the relatively rare modifier-noun word order could help support an
analysis of (9) as a (classifying) compound, I chose to regard it as an attributive
construction since, in contrast to verbal compounds, full nominal compounds
are very rare in Kwaza. Furthermore, Kwaza classifiers are suffixes and cannot
occur as free elements (van der Voort 2004: 128ff.).
350 Hein van der Voort
(14) ha-ha-'ro-ki
clean-clean-CL:vessel-DEC
‘He washes pans.’
(17) aky-'h
two-NOM
‘two (ones)’
352 Hein van der Voort
The following examples show the contrast between a predicate and a head-
less relative clause that is nominalised by a specific classifier:
(18) a'xy-dy-'xa-tsy-tse
house-CAU-2S-POT-DEC
‘You are going to make a house.’
Since the classifier -xy in this example has the specific semantic content
‘house’ (it is indeed etymologically related to the noun a'xy ‘house’), its
referent can be omitted without loss of semantic content, similar to headless
relative clauses and the pro-drop possibility of predicates.5 Compare the
following to example (2), with which no difference in meaning was attested:
(21) tawi'wi-dy-xy
Tavivi-POS-CL:house
‘Tavivi’s house’
5
Overt pronominal expression of arguments was attested to have a disambiguating
or emphatic effect. The pragmatic consequences of full versus reduced expression
of the PM have not been investigated.
Possessive expressions in the Southwestern Amazon 353
(24) o'lu-dy-'kai
curassow-POS-CL:leg
‘leg of a curassow bird’
Many body-part classifiers are of this type and probably just as many are
not. Furthermore, a number of classifiers that have a very general referential
scope and that are not etymologically related to a full noun can also occur
with the empty root e-. The following examples illustrate the classifiers -ro
‘container, vessel, cup’ and -ri ‘flat object’:
(26) e-'ro
Ø-CL:vessel
‘container’, ‘radio’ 6
6
The second interpretation of this example is an illustration of (the frequently
attested) lexicalisation of classifiers in combination with other elements.
354 Hein van der Voort
In view of the special position that possessive -dy- occupies in the general
make-up of Kwazá nominal morphology, it is necessary to consider other
morphemes with which it is homophonous. There is another form -dy- that
most likely originates in the possessive morpheme. It is used in a fixed
combination with the adverbial nominaliser -nãi, with the specific meaning
‘language of’. This construction is adverbial and no possessed head can be
identified on the semantic level, as the following examples show:
7
In some rare cases the omission of the possessive morpheme may not seem to
change the meaning, as in ari-'ri ‘the tapir’s liver’. The structure of ari-'ri is
fundamentally different from (28), however, since it (ari-'ri) represents a clas-
sified nominal comparable to a nominal compound. A similar alternative for ex-
ample (24), *olu-'kai, was rejected by the consultant.
Possessive expressions in the Southwestern Amazon 355
One of the reasons to consider -dy- as originally possessive here is its simi-
lar behaviour with regard to the first person inclusive pronoun. The expres-
sion txana-nãi ‘our language’ is preferred to txana-dy-nãi. Remember that a
similar preference with regard to the canonical possessive was mentioned
in the paragraph under example (3).
An additional morpheme that looks suspiciously like the canonical pos-
sessive is the comitative case suffix -dyn
. Consider example (31):
The comitative belongs with the beneficiary -du, instrumental -ko and loca-
tive -na to the semantic cases of Kwaza; these do not mark obligatory
grammatical arguments but optional “oblique” arguments. There are both
formal and semantic similarities with the canonical possessive construction,
but there are also important differences. Even though there could be an
etymological relationship with the possessive morpheme -dy-, and perhaps
with the verbal derivational reflexive morpheme -n
-, the comitative mor-
pheme -dyn
is not further analysable synchronically. And even though in
certain languages the comitative can mark a PM (e.g. McGregor 2001: 81;
Stassen 2006: 771) nothing like that has been observed in Kwaza.
Finally, the possessive morpheme -dy- is homophonous with the causa-
tive/benefactive morpheme -dy- and with the different subject marker -dy-.
It is possible that they are etymologically related, but it is difficult to dem-
onstrate this empirically. Being mainly verbal suffixes, their distribution is
to a large extent complementary with that of possessive -dy-. However,
their very different specific functions, combined with their very limited and
specific distributions are sufficient reason to consider of each of these items
synchronically as separate morphemes.
8
In Kwaza as well as in various other languages of the region, the words for ‘jag-
uar’ and ‘dog’ are identical.
356 Hein van der Voort
(32) mã-tja'te
mother-3.POS
‘his mother’
The possessive morpheme -tjate does not form part of a paradigm and it
refers exclusively to a third person PR. It has no other uses and there are no
other head-marking possessive strategies to express different persons. The
morpheme -tjate is not further analysable. It may be a borrowed morpheme,
as will be discussed in (§3.1.3).
Overt expression of the dependent PR is rare in the alternative posses-
sive construction, but it is attested:
Herslund and Baron (2001: 14) discuss the phenomenon that languages make
use of different adnominal possessive constructions to distinguish different
kinds of possessive relationships, such as alienable vs. inalienable. Rijkhoff
(this volume) mentions a special definiteness effect for third person PR
markers in Uralic and Turkic languages. In Kwaza, however, no such dis-
tinctions were attested between the alternative and the canonical possessive
in the third person. The following contrasted examples are semantically
identical:
(34) kanwã-tja'te
canoe-3.POS
‘his canoe’
(35) '
-dy-h
ka'nwã
he-POS-NOM canoe
‘his canoe’
Possessive expressions in the Southwestern Amazon 357
(36) tsu'ty-si'ki
-dy-'h
'bu-ki
head-CL:skin he-POS-NOM put-DEC
‘He uses his own hat.’
In the following example, the object can refer either to the wife of the third
person subject or to the wife of another third person:
(39) h
dj-tja'te-dy-h
eto'hoi
older.brother.female-3.POS-POS-NOM child
‘child of older brother of female’9 (lit. ‘her older brother’s child’)
9
This expression was equated by the informant with the more general kinship
terms kore ‘son of brother or sister’ and koretay ‘daughter of brother or sister’.
358 Hein van der Voort
(42) e-'kai-tja'te
Ø-CL:leg-3.POS
‘his leg’
Another analysis of the dummy root e- relates it to the verb e- ‘to have’,
with which it is homophonous. The verb e- ‘to have’ is used to express
predicative possession in Kwaza, as in:
This verb occurs also in specific habitual constructions, where it can be lit-
erally interpreted as an abstract predicative possessive:
The relationship between the verb ‘to have’, possession and aspect has been
discussed in typological work by among others Benveniste (1960), Heine
(1997) and Seiler (1977).
To a very limited extent, the verb e- ‘to have’ is also used as an existen-
tial verb. The normal existential verb in Kwaza is a- ‘to be’, which in addi-
tion functions in another specific habitual construction. As it happens, there
is also a semantically empty adverb formative root a-. This dummy root is
used to create adverbs out of bound verbal derivational directional markers.
Since the two roots e- and a- both can function as semantically empty for-
mative dummy roots and as semantically very abstract verb roots, their par-
allel distributions suggest a relationship between their different functions.
Further discussion of the dummy roots would take us too far from the pre-
sent topic and merits a separate article. The reason for mentioning their
parallelism here is to show that the empty noun formative e- and the verb
root e- ‘to have’ should probably be considered as the same element.
The dummy root e- or i- is also encountered (without the meaning ‘to
have’) in Kanoê (§3.2), Latundê (§3.3), Baure (§ 3.4) and other languages
of the Guaporé region. It was discussed as an areal feature in Crevels and
van der Voort (2008: 167–168) and van der Voort (2005: 397–398).
The Aikanã (also referred to as Massaká, Tubarão or Huari) are the tradi-
tional neighbours of the Kwaza. Their language is probably also an isolate,
360 Hein van der Voort
although there are some lexical and structural similarities with its neigh-
bouring languages. Aikanã is spoken by about 150 persons. Since the mid
1980s the language was studied and was partially described by the Brazilian
linguist Ione Vasconcelos (2003) and by an American team of linguists led
by Leanne Hinton (ed. 1993), and it is presently being studied by the author.
Aikanã is a morphologically highly complex language. Like Kwaza, most
of the morphological complexity resides in the verb. Furthermore it has
similar classifying and directional suffixes, valency-changing suffixes, a
distinction between future and non-future tense, person and mood inflexion,
animate object case marking, and no nominal number inflexion. Unlike
Kwaza, but similar to another Rondônian isolate Kanoê (§ 3.2), Aikanã has
several different verbal inflexion classes, some of which require argument
prefixes rather than suffixes. Also, it lacks an inclusive-exclusive distinc-
tion.
With respect to possessive expressions, Aikanã shows both differences
and similarities to Kwaza. Unlike Kwaza, Aikanã has a (partially transpar-
ent) set of specific possessive pronouns. Some of the possessive pronouns
contain the element -z, which probably originates from the productive
beneficiary case suffix that also has a genitive function.10 Furthermore, a
specific verbal benefactive inflexional paradigm exists for beneficiary and
possessive objects. In the following table the different Aikanã pronouns and
benefactive inflexions are listed:
10
Aikanã /z/ is usually pronounced as IPA [] and in nasal environments some-
times as [n]. The vowel /ü/ is pronounced as IPA [y].
Possessive expressions in the Southwestern Amazon 361
(48) 'h
z 'kapedi'ka
your shoulder
‘your shoulders’
(51) h
z-'je pürü-'ka-pa-ua-'
your-OBL work-1S-CL:big-2S.BEN-DEC
‘I worked for you.’
(52) ü'?ü-ka-ua- h
z-'je kaw-me-i-'za
keep-1S-2S.BEN-DEC your-OBL eat-2S-NOM-COL
‘I kept for you things for you to eat.’
11
It is unclear whether the suffix -z is related either with the Kwaza possessive
suffix -dy- or with the Kwaza beneficiary suffix -du.
Possessive expressions in the Southwestern Amazon 363
In example (56) the possessive pronoun is used as a verb root. The example
could be interpreted literally as: ‘This is mine for me.’. In (57) a personal
pronoun forms the root of the predicate: ‘This is (it) for me.’. There is a
similar difference between (58) and (59). In the following set of examples,
the first expression is less common, since, as the consultant explained, a
wife is a human being:
364 Hein van der Voort
(62) wãwã'?
'txütxü-ku-
child my-1S.BEN-DEC
‘The child is mine.’
(63) wãwã'?
'kari-ku-
child that-1S.BEN-DEC
‘That is my son/daughter.’
(64) kura-de'ri
husband-3.POS
‘her husband’
This alternative possessive is attested very frequently, but again, it only exists
for a third person PR and it does not seem to be determined by features of
alienability or animacy. Among the languages of the Guaporé region, only
Possessive expressions in the Southwestern Amazon 365
personal possessive
1S aj ña, jã
2S m
pja
3S oj ojo
1P ajte jato
2P m
te pjato
3P ojte ojoto
12
The Kanoê vowel /y/ is pronounced as IPA [].
366 Hein van der Voort
Just as in Aikanã, PRs that contain the possessive morpheme (-o) represent
morphologically independent nouns and no further marking is required of
either the PR or the PM.
Kanoê does not have an alternative possessive construction like the
Kwaza (cf. §2.3) and Aikanã (cf. §3.1.3) expressions that involve just the
third person singular.
Similar to Kwaza (as in examples (23), (26), (27) and (42)), Kanoê has a
large subset of classifiers that can be turned into full nouns by the semanti-
cally empty noun formative root i-. The empty root occurs mainly with ‘in-
alienable’ items such as body part classifiers. In possessive constructions,
these classifiers build equivalent alternative expressions. Either the classi-
fier combined with the empty root functions as an independent PM, as in
example (67), or the empty root is omitted and the classifier is attached as a
bound morpheme to the PR, as in example (68).
(68) ytse-o-katsi
tree-POS-CL:root
‘root of tree’ (Bacelar 2004: 100)
(69) ytse-katsi
tree-CL:root
‘tree root’ (Bacelar 2004: 100)
When used in combination with the oblique case suffix -ni, the Kanoê pos-
sessive suffix -o has a beneficiary (70) or comitative (71) sense. This was
attested only with the first and second person singular personal pronouns
and it notably does not involve possessive pronouns:
Lakondê and Latundê are dialects of a language that belongs to the northern
branch of the small Nambikwara language family. The language, hence-
forth referred to as Latundê, has about 20 speakers (there is only one sur-
viving speaker of the Lakondê dialect). It has been described by Telles
(2002a), on whose work this section is based. Like many languages of the
Guaporé region, Latundê is a morphologically complex language. The pos-
sessive constructions of Latundê, which are also discussed in a separate
article by Telles (2002b), are different from those in the language isolates
discussed above. Latundê has a full paradigm of possessive prefixes, which
are attached to the nominal PM. The prefixes are reduced versions of the
personal pronouns, listed in Table 3 below:
Table 3. Latundê personal pronouns and possessive prefixes (from Telles 2002a:
150, 156 and 2002b: 158)
personal possessive
1S ta ja ta -
1P nh nh-
2 wa ja wa -
3S hãja hãj-, nã-, ã-
3P a a wja a w-, ã-
a
The possibility of distinguishing between the singular and plural of the third per-
son possessive in Latundê does not exist in Lakondê (Telles 2002a: 157).
(72) wa-sih-te
2.POS-house-REF
‘your house’ (Telles 2002a: 156)
13
In spontaneous speech the referential suffixes also play a not yet fully understood
role in the marking of pragmatic status (Telles: 2002a: 215).
Possessive expressions in the Southwestern Amazon 369
(73) ta-najn-kin
n-te
1S.POS-head-CL:round-REF
‘my head’ (Telles 2002a: 163)
(75) kejajn-tawn-te
peccary-tail-REF
‘peccary tail’ (a kind of tail) (Telles 2002a: 159)
(76) kejajn-ã-tawn-te
peccary-3.POS-tail-REF
‘the tail of the peccary’ (Telles 2002a: 159)
Again, there are alternative ways to express possession, since there may be
a reason to consider the third person possessive morpheme ã- as a special
case, different from the others. Note that, in addition to full nouns, Latundê
370 Hein van der Voort
(77) ta-ni-tu
1S.POS-CL:hemispheric-REF
‘my pan’ (Telles 2002a: 194)
(78) ta-ã-ni-tu
1S.POS-3.POS-CL:hemispheric-REF
‘my pan’ (Telles 2002a: 194)
(80) ta-nãn-kah-tãn
1.POS-Ø-CL:long-IMPF
‘it was mine (my bow)’ (Telles 2002a: 161)
14
Although Latundê has a small set of classifiers, they have a similar distribution
and nominalising potential as in Kwaza. Some classifiers are even identical in
form with those in Kwaza, Kanoê and Aikanã, which is probably due to areal
diffusion (van der Voort 2005: 395–397).
Possessive expressions in the Southwestern Amazon 371
(81) ta-nãn-saw-te
1.POS-Ø-CL:liquid-REF
‘my liquid (e.g. coffee)’ (Telles 2002a: 162)
(82) *ã-nãn-te
3.POS-Ø-REF
‘his (e.g. hand)’ (Telles 2002a: 162)
Baure is a language that belongs to the southern branch of the Arawak lan-
guage family. It is spoken by about 60 elderly people in the Bolivian low-
lands. It has been described by Danielsen (2007), on whose work this section
is based. Like most languages of the region, the morphology of Baure is
rather complex and shows polysynthetic traits. It has several characteristics
in common with other non-Arawak languages, such as classifiers and direc-
tionals. There is no inclusive-exclusive distinction in the person reference
system. There is a gender distinction for the third person singular. The sys-
tem is based on person proclitics that are attached both to verbs and nouns.
When attached to verbs, the proclitics refer to a subject argument, when
attached to nouns they refer to a PR. In addition, there is a possessive suffix
-no that is homophonous with a nominaliser.
As Danielsen (2007: 119) explains, two different kinds of nouns are dis-
tinguished with regard to the expression of possession. Optionally possessed
nouns receive a person proclitic and usually the derivational possessive suf-
fix -no, as shown by the following examples:
(83) yaki
fire
‘fire’ (Danielsen 2007: 87)
(84) ni=yaki-no
1S=fire-POS
‘my fire’ (Danielsen 2007: 87)
372 Hein van der Voort
(85) e-ser
Ø-tooth
‘a tooth of someone’ (Danielsen 2007: 120)
(86) ni=ser
1S=tooth
‘my tooth’ (Danielsen 2007: 120)
(87) kove’
dog
‘dog’ (Danielsen 2007: 124)
15
Danielsen (2007: 120) mentions the existence of several lexicalised combinations,
such as -waki ‘palm of the hand’, ewaki ‘forked branch of tree’.
16
Consequently, these nouns are not attested with person proclitics, except in certain
lexicalised combinations.
Possessive expressions in the Southwestern Amazon 373
As a bound noun root, the generic noun per ‘domestic animal’ can function
as a classifier. Consequently, the above construction is very similar to the
possessive classifier constructions discussed for Gran Chaco languages by
Fabre (2007) and Messineo and Gerzenstein (2007) and the ‘indirect’ pos-
sessive construction described by Lichtenberk (this volume) for Oceanic
languages. The generic noun per also occurs in other constructions than the
above.
Note that possessive NPs may also involve possessive pronouns, albeit
rarely, as shown in example (91).
17
Baure // represents IPA []. The glide /y/ is pronounced as IPA [j].
18
The root -wer ‘house’ belongs to a small set of nouns that cannot undergo deriva-
tion. Therefore, the possessive suffix -no cannot be applied here, even though the
referent might be regarded as alienable.
374 Hein van der Voort
(92) ni=ay-ino-wo=pi=ro
1S =desire-BEN-COP =2S =3SM
‘I wish it for you.’ (Danielsen 2007: 16)
Wari’ (also known as Pakaa Nova) is a language that belongs to the small
Chapacura language family of western Brazil and northern Bolivia. It has
close to 2000 speakers and it is described in Everett and Kern (1997), on
whose work this section is based. Wari’ has a grammaticalised gender sys-
tem, like certain other languages of the region. The language is syntacti-
cally rather than morphologically complex. Person marking is realised
through particles that Everett and Kern refer to as nominal and verbal in-
flexional clitics (NICs and VICs). Inclusive and exclusive first person plural
and third person feminine and masculine are distinguished. According to
Everett and Kern (1997: 311) there are no possessive pronouns. The prag-
matically unmarked constituent order of Wari’ is VOS.
Possessive constructions in Wari’ involve PR-marking on the PM, like
in Latundê. The possessed noun agrees through an inflexional clitic with
person, number and third person gender of the PR. There are three different
Possessive expressions in the Southwestern Amazon 375
The third construction is limited to a specific set of kinship terms that re-
quire the verbal ‘inflection’ prefix co-. The kinship term always has a first
person singular possessive interpretation. It is followed by a VIC because,
as explained by Everett and Kern (1997: 375–377), it has undergone zero-
derivation from an original verbal clause. The literal interpretation of (96)
is therefore quotative: ‘Xijam he (says) “my younger sibling”’. The con-
struction is grammaticalised and not transparent to the speakers of Wari’.
Also in other parts of Wari’ grammar, the quotative is used in other func-
tions than to quote direct speech, such as future tense.
Note that in all constructions an independent noun referring to the PR
can be omitted if understood from context, as in e.g. xiri-con ‘his house’
(cf. example 94), and in:
19
Wari’ has two noun classes. The class of inalienable nouns is characterised by
the fact that the citation form ends in -xi ‘1P.INCL’. The class of possessable nouns
does not have this characteristic (Everett and Kern 1997: 3). From the examples
below it appears that the noun ‘house’ belongs to the inalienable class, whereas
the noun ‘child’ belongs to the possessable class.
20
Wari’ /x/ represents IPA [(t)∫].
376 Hein van der Voort
(97) wina-hu’
head-2P.POS
‘your head’ (Everett and Kern 1997: 236)
(99) e-pisa
2S-liver
‘your liver’ (Galucio 2001: 32)
Possessive expressions in the Southwestern Amazon 377
If there is no overt PR noun, third person PRs are expressed by the prefix i-
before consonants and by the prefix s- before vowels:
(101) i-piso
3S-foot
‘his foot’ (Galucio 2001: 35)
(102) s-anp
3S-head
‘his head’ (Galucio 2001: 36)
When alienable nouns occur possessed, this is indicated in the same way as
the inalienable situation, except for a subclass of nouns which require pre-
fixation of a t.21 The word ek ‘house’ belongs to this subclass:
(103) o-tek
1S-house
‘my house’ (Galucio 2001: 33)
A third set of nouns does not accept person prefixes, but needs to be juxta-
posed to an inflected obligatorily possessed (inalienable) noun of general
semantic content that functions as a classifier. The word apara ‘banana’
belongs to this subclass:
21
This t is not an affix. As Galucio (2001: 32) observes, its occurrence is neither
phonologically nor morphologically predictable, and its prefixation to certain
nouns under possession represents a regularity in the lexicon.
378 Hein van der Voort
(106) *o-apara
1S-banana
‘my banana’ (Galucio 2001: 33)
(107) i-tek
3S-house
‘his house’ (Galucio 2001: 76)
(108) se-tek
3S.C-house
‘his (own) house’ (Galucio 2001: 76)
Arikapú is a language that belongs to the small Jabutí branch of the Macro-
Jê language family. It is spoken by only one elderly person. The language
has been studied by the author since 2001, and it was recently classified in
cooperation with Eduardo Ribeiro (Ribeiro and van der Voort 2010). Its
traditional neighbouring languages belong mainly to the Tuparí family. Like
the Chapacura and Tupí languages, Arikapú is morphologically a rather
simple language although there are some nominal and verbal suffixes. Per-
son marking is realised by agreement prefixes, that can be used in addition
to personal pronouns. The distribution of person marking obeys an ergative
pattern.
Arguments are indicated through the application of person prefixes on
the verb. Possession is indicated by the same prefixes, when applied to PM
nouns, referring to the PR:
Possessive expressions in the Southwestern Amazon 379
(109) a-kan
2-hat
‘your hat’
(110) i-n
'r
1S-house
‘my house’
(113) ta-n
'r-n ta-n't
3-place-LOC 3-sleep
‘He’s sleeping in his own house.’
The prefixes for first person singular (i-) and first person plural (txi-)22 are
also used as third person or impersonal prefixes, especially in citation forms.
These two prefixes seem to function often as a dummy PR with inalienable
concepts as well as a dummy argument with verbs. When the PR or the
third person object is expressed by a full noun, the impersonal prefixes are
absent. The following examples contrast these constructions:
(115) i-'k
1S-skin
‘(its) paper’, ‘my skin’
(116) patxi='k
tobacco=skin
‘cigarette paper’
(117) 'txi-ku'ju
1P-wing
‘our/one’s/its wing’
(118) 'aro='kuju
guan=wing
‘wing of a guan bird’
Example (115) and (117) show that the absence of a (pro-)noun may allow for
grammatical ambiguity, since in principle, various interpretations of the per-
son prefixes are possible.
Alternatively, examples (116) and (118) could perhaps be analysed as at-
tributive compounds rather than as possessive constructions, or maybe even
as classifying derivations, since the forms k ‘skin’ and kuju ‘wing’ were
never attested as free nouns.
Unlike many other Macro-Jê languages (Rodrigues 1999: 191) and unlike
Baure (§ 3.4) and Mekens (§ 3.6), Arikapú does not have generic classifying
nouns to distinguish between alienable and inalienable possession.
22
Arikapú /x/ represents IPA [].
Possessive expressions in the Southwestern Amazon 381
4. Final remarks
The languages discussed in this article form only a small subset of the entire
number of languages of the Southwestern Amazon, and they mainly repre-
sent the language diversity on the Brazilian side of the Guaporé River.
Kwaza is taken as a standard of comparison and is presented in more depth
than the other languages, because of the sources, data and experiences
available to me. The purpose of this article is to provide a first impression
of the different structural possibilities for the expression of adnominal pos-
session in the Southwestern Amazon region, and to explore possible areal
linguistic relationships.
Two main types of adnominal possessive strategies were encountered,
both expressed in a range of different varieties. Type I is characterised by
the occurrence of a general possessive element that is attached to the PR. It
forms the standard possessive construction in Aikanã, Kanoê and Kwaza. It
could be considered as a ‘genitive’ construction, but I have not used this
term in order to avoid possible association with case marking, which is dif-
ferent in these languages. Type II is characterised by the attachment of PR
agreement morphemes or person markers to the PM. This is the standard
possessive construction in Arikapú, Baure, Latundê, Mekens and Wari’. An
overview of the different characteristics of possessive expressions is pre-
sented in the Table 4:
Aikanã
Kwaza
Kanoê
Baure
Wari’
Sakel (2004) on Mosetén, which all are Bolivian isolates, and Guillaume
(2008) on Cavineña, a Tacanan language. Furthermore, work on other
members of the language families represented here, in particular the Tupí
languages, needs to be considered. The study of this corpus may shed addi-
tional light on the extent of areal diffusion of specific possessive strategies
recurring in this article, or may lead to the discovery of others.
Finally, possession in other parts of South America deserves investiga-
tion. Especially the Amazonian languages still suffer from a lack of atten-
tion in general typological work, whereas their impressive genealogical di-
versity makes them an important testing ground for universal aspects of
grammar. Some of these languages employ rare or unique possessive
strategies, like the abovementioned isolate Movima, in which inalienable
and predicative possession are expressed by reduplication (Haude 2006:
238ff.).
Acknowledgements
The Kwaza, Aikanã and Arikapú data in this article are from my own field
work in Rondônia during extended periods between 1995 and 2009. With-
out the help of my language consultants Mario Kwazá, Luiz Aikanã, Manoel
Aikanã and Nazaré Arikapú this work would not have been possible. I also
want to acknowledge the Netherlands Organization for Scientific Research
(NWO) for having generously financed my study of Kwaza, Aikanã and
Arikapú during many years. In addition, I would like to express my deep
gratitude to the indigenous communities in Rondônia for their hospitality. I
am furthermore very grateful to Willem Doelman for providing an excellent
map. Finally, I owe many thanks to Swintha Danielsen, Mily Crevels, Rik
van Gijn, Olga Krasnoukhova, Bill McGregor, Pieter Muysken, Jan Rijkhoff
and Leon Stassen for their valuable comments. Data from other languages
cited here are from the work by colleagues, who do not necessarily agree
with my analyses, and I apologise for sometimes having strayed from theirs.
All errors are mine.
Possessive expressions in the Southwestern Amazon 385
References
Dryer, Matthew S.
2007 Noun Phrase structure. In Language Typology and Syntactic De-
scription, Volume II: Complex Constructions, Timothy Shopen (ed.),
151–205. Second edition. Cambridge: Cambridge University Press.
Everett, Daniel L. and Barbara Kern
1997 Wari’: The Pacaas Novos Language of Western Brazil, London:
Routledge.
Fabre, Alain
2007 Morfosintaxis de los clasificadores posesivos en las lenguas del Gran
Chaco (Argentina, Bolivia y Paraguay). UniverSOS. Revista de Len-
guas Indígenas y Universos Culturales 4: 67–85.
Galucio, Ana Vilacy
2001 The morphosyntax of Mekens (Tupi). PhD thesis, University of Chi-
cago.
Gijn, Rik van
2006 A grammar of Yurakaré. PhD thesis, Radboud Universiteit Nijmegen.
(Avail. on http://webdoc.ubn.ru.nl/mono/g/gijn_e_van/gramofyu.pdf)
Gil, David
2001 Escaping Eurocentrism: fieldwork as a process of unlearning. In Lin-
guistic Fieldwork, Paul Newman and Martha Ratliff (eds.), 102–132.
Cambridge: Cambridge University Press.
Guillaume, Antoine
2008 A Grammar of Cavineña. Mouton Grammar Library 44. Berlin /New
York: Mouton de Gruyter.
Haude, Katharina
2006 A Grammar of Movima. PhD thesis, Radboud Universiteit Nijmegen.
(Avail. on http://webdoc.ubn.ru.nl/mono/h/ haude_k/gramofmo.pdf)
Heine, Bernd
1997 Possession: Cognitive Sources, Forces, and Grammaticalization.
Cambridge Studies in Linguistics 83. Cambridge: Cambridge Uni-
versity Press.
Heine, Bernd and Tania Kuteva
2002 World Lexicon of Grammaticalization. Cambridge: Cambridge Uni-
versity Press.
Herslund, Michael and Irène Baron
2001 Introduction: Dimensions of possession. In Baron et al. (eds.), 1–25.
Hinton, Leanne (ed.)
1993 Aikana Modules: A class report on the fieldnotes of Harvey Carlson.
Typescript, 158 pp. Berkeley: University of California.
Koptjevskaja-Tamm, Maria
2006 Possession, Adnominal. In Keith Brown (ed.), 765–769.
Possessive expressions in the Southwestern Amazon 387
Kroeker, Menno
2001 A descriptive grammar of Nambikuara. International Journal of
American Linguistics, 67 (1): 1–87.
Lichtenberk, Frantisek
this vol. Attributive possessive constructions in Oceanic.
McGregor, William B.
2001 The verb HAVE in Nyulnyulan languages. In Baron et al. (eds.), 67–98.
Messineo, Cristina and Ana Gerzenstein
2007 La posesión en dos lenguas indígenas del Gran Chaco: toba (guay-
curú) y maká (mataguayo). Línguas Indígenas Americanas (LIA-
MES) 7: 61–79.
Nichols, Johanna
1992 Linguistic Diversity in Space and Time, Chicago: University of Chi-
cago Press.
Payne, Doris L. and Immanuel Barshi (eds)
1999 External Possession. Typological Studies in Language 39. Amster-
dam / Philadelphia: John Benjamins.
Ribeiro, Eduardo and Hein van der Voort
in press Nimuendajú was right: The inclusion of the Jabuti language family
in the Macro-Jê stock. International Journal of American Linguistics
76 (4).
Rijkhoff, Jan.
this vol. On the co-variation between form and function of adnominal posses-
sive modifiers in Dutch and English.
Rodrigues, Aryon Dall’Igna
1986 Línguas brasileiras: Para o conhecimento das línguas indígenas, São
Paulo: Edições Loyola.
Rodrigues, Aryon Dall’Igna
1999 Macro-Jê. In Robert M.W. Dixon and Alexandra Y. Aikhenvald (eds.),
165–206.
Sakel, Jeanette
2004 A Grammar of Mosetén. Mouton Grammar Library 33. Berlin /New
York: Mouton de Gruyter.
Seiler, Hansjakob
1977 On the semantico-syntactic configuration ‘Possessor of an act’. In
Sprache und Sprachen: Gesammelte Aufsätze, Hansjakob Seiler (ed.),
169–186. München: Wilhelm Fink.
Stassen, Leon
2006 Possession, Predicative. In Keith Brown (ed.), 769 –773.
Stassen, Leon
2009 Predicative Possession. Oxford Studies in Typology and Linguistic
Theory. Oxford: Oxford University Press.
388 Hein van der Voort
Telles, Stella
2002a Fonologia e gramática Latundê/ Lakondê. PhD thesis, Vrije Univer-
siteit Amsterdam.
Telles, Stella
2002b Construções possessivas em Latundê. In Línguas Indígenas Brasilei-
ras: Fonologia, gramática e história, Ana Suelly Arruda Câmara Cab-
ral and Aryon Dall’Igna Rodrigues (org.), 157–163. Atas do I. Encontro
Internacional do GTLI, tomo II. Belém: Editora Universitária UFPA.
Vasconcelos, Ione
2003 Aspectos fonológicos e morfofonológicos da língua Aikanã. PhD
thesis, Maceió: Universidade Federal de Alagoas.
Voort, Hein van der
2004 A Grammar of Kwaza. Mouton Grammar Library 29. Berlin /New
York: Mouton de Gruyter.
Voort, Hein van der
2005 Kwaza in a comparative perspective. International Journal of Ameri-
can Linguistics, 71 (4): 365–412.
Voort, Hein van der
2006 Construções atributivas em Kwazá. Boletim do Museu Paraense
Emílio Goeldi (Ciências humanas), 1 (1): 87–104. (Also available on
http://marte.museu-goeldi.br/seer/index.php/boletimhumanas/article/
viewFile/5/30)
Possession in the visual-gestural modality:
How possession is expressed in British Sign Language
1. Introduction
All known languages have some way of expressing possession, and signed
languages are no exception. The primary question we pose in this chapter
is: Is possession expressed differently in signed languages due to the use of
the visual-gestural (rather than the aural-oral) modality, or are patterns re-
lating to possession essentially the same for signed and spoken languages?
Our chapter begins with background about British Sign Language (BSL),
followed by an overview of the pronominal system of BSL. We then move
on to look at attributive and predicative possession in BSL and finally an
adjectival predicate of predisposition in BSL closely related to the posses-
sive pronoun. Although this chapter is primarily an overview of possession
in BSL, we will also include observations on other signed languages, such
as American Sign Language (ASL), where applicable.
BSL is the natural signed language used by the deaf community in the
United Kingdom. BSL has its own phonological, morphological and syn-
tactic structure. Phonologically, lexical signs in BSL are made up of four
primary parameters: handshape, movement, location (i.e. place of articula-
tion) and palm/finger orientation. Every sign is specified for each of these
parameters, which are phonologically contrastive (i.e. minimal pairs can be
identified with each parameter).
Clearly, BSL is neither simply an elaborated gestural system nor a manual
code based on English. Nonetheless, because BSL is a minority language
within the United Kingdom, there is strong and constant contact between
BSL and English and thus BSL does borrow elements from English. Having
said that, we will point out instances where there are possessive construc-
tions within signed languages which are known borrowings from the sur-
rounding spoken language.
390 Kearsy Cormier and Jordan Fenlon
The examples in this chapter come from a variety of sources. Some are
from or based on elicited data, some are from broadcast television pro-
grammes, some are from an annotated corpus of BSL stories and fables
(Woll, Sutton-Spence, and Waters 2004), and some are from naturalistic
video-recorded conversation. In all cases, examples are from or based on
signing from both native and non-native signers. To be representative of the
community, it is important to include data from both native and non-native
signers, since only a very small percentage of the British Deaf community,
generally considered to be 5–10%, are native signers born to deaf parents,
and thus non-native signers (deaf with hearing and usually non-signing par-
ents) make up the vast majority of the typical deaf population.1
One of the striking characteristics about signed languages is the way that
signers use the space around them for referential purposes. Signers refer to
a person/object physically present within a discourse situation simply by
pointing to him/her/it. For referents who are not present, the signer estab-
lishes a location in space for the referent by pointing to that location. These
pointing signs acts as pronouns and are glossed here as PRO-1 (first person)
or PRO-non1 (for non-first person) – see Figures 1a and 1b.2 The first person
singular pronoun consists of a point to the signer’s own chest.
1
The proportion of native signers in the UK is generally considered to be ap-
proximately 5–10%, following statistics reported in Australia and the United
States (Deaf Society of New South Wales 1998; Mitchell and Karchmer 2004).
2
Following conventions in the sign language literature, glosses are given for signs
using all caps. Signs which require more than one English word for translation
are glossed with English words separated by hyphens (e.g. ICE-CREAM).
Fingerspelled words are indicated with hyphens in between letters (e.g. J-O-E) –
see also footnote 9. Pronouns are glossed as PRO-1 for first person and PRO-
non1 for non-first person. Possessive pronouns are glossed as POSS-1 for first
person and POSS-non1 for non-first person. For ease of exposition, the dual and
number-incorporated pronouns will be notated with a simple English gloss in-
stead (e.g. TWO-OF-US, THREE-OF-THEM). Within examples, when indexing
different non-first person locations, PRO-2/POSS-2 is used for forms which index
the location of the addressee. Indices such as a, b, c (e.g. PRO-a, POSS-a) are
used to index non-addressed participants. Unless otherwise noted, where indices
occur with nouns (e.g. BOX-a), the noun sign may have been established in that
locus previously in the discourse, or some sign within the example establishes
Possession in the visual-gestural modality 391
3.1. Person
There has been a fair amount of debate within the sign language literature
regarding the issue of person marking on pronouns in signed languages. In
the early days of sign language research, researchers assumed a three-
person system analogous to those found in spoken languages (Friedman
1975; Klima and Bellugi 1979; Padden 1983, 1990). A three-person system
is problematic, however, because there is no listable set of location values
in the signing space to which a non-first person pronoun may point, for ad-
dressee or non-addressed participants. That is, there is no single location
that may act as a morpheme to indicate second or third person. To address
this issue, some researchers have taken the view that sign language pro-
nouns do not exhibit person marking at all and that locations associated
with pronouns instead act as variables (‘loci’) whose content comes from
discourse (Cormier, Wechsler, and Meier 1999; Lillo-Martin and Klima
1990). More recently, most sign linguists subscribe to a two-person system
as proposed by Meier (1990) for ASL. Such an analysis recognises the ‘lis-
tability problem’ (Rathmann and Mathur 2002) of multiple second / third
person location values while at the same time recognising the special status
the noun at that locus. Where repetitions of movement occur, the number of path
movements is indicated with ‘+’ (e.g. POSS-non1++ has two movement paths).
In English translations of examples, for ease of exposition, different gendered
pronouns (e.g. he vs. she) are used to distinguish different referents, though BSL
pronouns do not mark gender.
392 Kearsy Cormier and Jordan Fenlon
of first person, for which there is only one location (the signer’s chest). This
two-person system has been assumed by other researchers for ASL and
other signed languages (Engberg-Pedersen 1993; Farris 1994, 1998; Liddell
2003), including BSL (Cormier 2007), and will be assumed in this chapter
as well.3
3.2. Number
3
Not all sign language researchers subscribe to a two-person analysis for signed
languages – e.g. Berenz (2002) argues for a three-person system for Brazilian
and American Sign Languages, as do Alibasic, Ciciliani and Wilbur (2006) for
Croatian Sign Language.
Possession in the visual-gestural modality 393
The first person plural pronoun PRO-1pl is produced with an index hand-
shape at the signer’s chest but, instead of pointing directly at the signer’s
chest, in the plural form the index finger traces a small circular movement
just in front of the chest. This form is the least indexic of all the pronouns –
i.e. the other pronouns point to the location(s) associated with their refer-
ents. The first person plural pronoun only indexes (points to) the signer’s
chest and does not index the other referents.
394 Kearsy Cormier and Jordan Fenlon
Figure 4a. PRO-1pl (citation form) ‘we’ Figure 4b. PRO-1pl (displaced) ‘we’
(excluding someone salient in the
discourse)
4
A similar system of exclusive pronouns has also been identified in ASL – for
more see Cormier (2005, 2007).
Possession in the visual-gestural modality 395
Exclusive forms were identified with first person plurals (PRO-1pl) and
with first person number-incorporated plurals (3/4/5-OF-US). Because the
first-person dual pronoun indexes the location of each referent, no gram-
matical exclusive form is proposed for the dual pronoun, only plurals and
number-incorporated forms.
Pro-drop (with either the subject or object pronoun) is common in BSL and
other signed languages, particularly with singular pronouns, when the ref-
erent is retrievable from context, as in (4). With plural pronouns, studies of
ASL and Australian Sign Language (Auslan) have shown that first person
plural pronouns are dropped more often than first person singular pronouns
(Schembri and Johnston 2006; Wulf, Dudis, Bayley, and Lucas 2002). With
double agreement verbs (a type of verb in which the start and end location
of the verb or the direction in which the verb is facing reflects the subject /
source and object/goal of the verb, respectively), pro-drop is usually the rule
rather then the exception, particularly with objects, as in (5).
5
Word order in BSL like other signed languages is quite variable on the surface.
Grammatical processes like topic marking make many different word orders
possible. There have been attempts by sign language researchers to posit a basic
default order from which other orders are derived. Such researchers have
claimed that the basic word order for ASL is SVO (Fischer 1975; Neidle, Kegl,
MacLaughlin, Bahan, and Lee 2000). There has not been as much work on word
order in BSL but it appears that BSL is also, underlyingly at least, SVO, although
see Deuchar (1983) for a different perspective.
396 Kearsy Cormier and Jordan Fenlon
Pronoun copy is also common, whereby the pronoun in situ is then copied
and produced again at the end of the clause as in (6).
4. Possession in BSL
The singular and plural possessive pronouns in BSL behave very similarly
to the singular and plural personal pronouns, respectively. The primary dif-
ference is the handshape. While the personal pronoun PRO uses an ‘index’
handshape (extended index finger similar to the handshape used in a point-
ing gesture), the possessive pronoun POSS uses a fist handshape with the
thumb alongside the index finger or wrapped across the other fingers. For
the first person form, the back of the fingers contact the signer’s chest, and
for the non-first person form the back of the fingers face outward away
from the signer, toward the location associated with the referent.
4.1.1. Number
The singular forms are shown in Figures 5 and 6. Singular possessive pro-
nouns representing the possessor (PR) may occur on their own with the
possessum (PM) noun (e.g. POSS-a CAR ‘his car’) or in apposition with a
personal pronoun (e.g. PRO-a POSS-a CAR ‘his car’). (See § 4.1.2 for the
use of POSS with nominal PRs.)
Plural possessive pronouns use the same handshape as the singular posses-
sive forms. The non-first person plural possessive pronoun POSS-non1pl
‘their’ is very similar to the pronoun PRO-non1pl ‘them’ (with a sweeping
motion across the locations associated with the referents) but uses a fist
handshape, as shown in Figure 8. The first person plural possessive pro-
noun POSS-1pl ‘our’ as shown in Figure 7 is the same as the first person
plural pronoun (PRO-1pl ‘we’: circular motion at the chest) but has a fist
handshape; the back of the fingers remain facing the signer during the
production of the sign.
398 Kearsy Cormier and Jordan Fenlon
Although these plural forms do exist and are used, it is also possible (and
quite common) for BSL signers to use the singular possessive instead of the
plural with plural referents, both in first and non-first person, for collective
plurals, as shown in (9a). If the number of referents is not already known or
not clear in context, then a plural personal pronoun or other quantifier may
accompany the singular possessive; in this case, again, the pronoun/quanti-
fier generally precedes the possessive but can occur afterwards. In (9b) and
(9c), the non-first person possessive is directed towards the same general
location as the quantifier/pronoun which occurs in the same utterance.6 In
(9d) and (9e), the singular first person possessive is, as always, directed
towards the signer’s chest. If a plural pronoun or quantifier is used in com-
bination with a possessive, the data and intuitions from deaf informants in-
dicate that the possessive must be singular and cannot be plural, as shown
in (9f).
6
With non-first person forms, singular collective possessive pronouns are directed
towards a location in space that appears approximately equidistant between the
multiple locations associated with the referents (if dual or plural), or the same
general location of the pronoun (if a number-incorporated pronoun).
7
This particular example is an emphatic use of the possessive, as described in
§ 4.1.5.
400 Kearsy Cormier and Jordan Fenlon
(11) POSS-a POSS-b POSS-1 LAND PRO-a PRO-b PRO-1 WILL SELL
‘His land, her land and my land’/‘The land that each of us owns will
be sold.’
Possession in the visual-gestural modality 401
With a nominal PR, the PR, possessive pronoun and PM are placed in ap-
position as in (13). The possessive pronoun usually follows the PR, such
that the typical order is PR POSS PM.8 This same ordering is identified for
nominal PRs in ASL, Croatian Sign Language, and Austrian Sign Lan-
guage (Pichler et al. 2008).
8
It is possible for the possessive pronoun to precede the possessor in both BSL
and ASL (e.g., POSS-non1 JOHN CAR NICE ‘John’s car is nice’). MacLaughlin
(1997) shows that this construction (POSS PR PM) can only be used in ASL
when the possessor in question has been established previously (i.e. it cannot be
used with indefinite possessors) and that this construction is strongly preferred in
topic position. According to Neidle, Kegl, MacLaughlin, Bahan and Lee (2000),
“the proper analysis of this construction remains something of a mystery” (p.
182). We only identified one token of this ordering in our corpus of BSL data;
by far the most frequent ordering was PR POSS PM.
402 Kearsy Cormier and Jordan Fenlon
First person plural possessive pronouns in BSL also have an exclusive form,
similar to the exclusive personal pronouns described in § 3.3. The exclusive
form for BSL shown in Figure 9 below is the same as the citation form as
in Figure 7 above but is displaced to the signer’s left or right side, and this
form may exclude any salient referent.
9
Fingerspelling systems within signed languages are based on the written alphabet
of the surrounding hearing community and are used for various purposes, in-
cluding proper names and other concepts for which a native lexical sign may not
exist. Auslan is historically related to BSL; Auslan, BSL and New Zealand Sign
Language are generally considered to be dialects of the same language (Johnston
2003; McKee and Kennedy 2000), and all three languages share a common two-
handed fingerspelling system.
Possession in the visual-gestural modality 405
ASL also has a similar affixal ’S possessive marker. Pichler et al. (2008)
note than although this form was likely borrowed from Signed English sys-
tems (sign systems which are not natural signed languages but are based on
English morphology and syntax), it is judged acceptable by ASL signers,
particularly in utterances with multiple possessors, as in FATHER ’S
BROTHER ’S WIFE. This marker (glossed by Pichler et al. as APOS-
TROPHE-S) consists of a modified form of the letter -S- from the finger-
spelling system used by ASL signers.
Unlike English, these affixal possessive markers are not obligatory for
expressing possession with nominal PRs in Auslan and ASL, and are used
along with other types of possession marking as described in this chapter.
To our knowledge, the use of two distinct forms marking possession (where
one encodes inalienable possession and the other marks alienable posses-
sion, as in (17) above) has not been identified in any other signed language
in a possessive construction without an overt nominal PR. Use of the per-
sonal pronoun instead of the possessive pronoun has been noted for some
other signed languages – specifically, ASL, Croatian Sign Language, and
Austrian Sign Language – but the claim with these signed languages is that
the personal pronoun PRO may optionally be used instead of POSS in any
possessive construction, not just with inalienable nouns (Pichler et al. 2008).
In addition to the use of possessive pronouns, BSL can also associate PMs
with PRs by changing the location of some noun signs directly within the
signing space. The primary criterion for such spatial marking of noun signs
is that the place of articulation of the sign must be the neutral space in front
of the signer, not a location on the body; the sign may then be located at a
particular location in space associated with some referent. It is not entirely
clear if this spatially marks PMs only for possession (‘my house’, ‘your
house’) or if the marking is at particular locations which just so happen to
correspond to locations associated with PRs (‘the house here/associated
408 Kearsy Cormier and Jordan Fenlon
with me’, ‘the house there/associated with you’). In (20a) and (20b), the
sign HOUSE is produced at two different locations. Note the first person
pronoun PRO-1 is used as a possessive in (20a), as in (18b) above, and thus
the spatial marking of HOUSE at the first person locus is redundant, while
in (20b) both tokens of the pronoun PRO-1 refer to the subject of the clause,
not to the PR, so the only indication of the PR here is the non-first person
spatial marking of the noun HOUSE.
The lexical verb HAVE in BSL (shown below in Figure 12) can be used to
indicate possession. The PR, if overtly expressed, acts as subject of the sen-
tence. HAVE (glossed here as HAVEposs) can be used with any type of al-
ienable possession, as in (21).
Possession in the visual-gestural modality 409
As with many other languages, the same lexical item HAVE which is used
for possession can also be used for existence (Heine 1997), glossed in (22)
as HAVEex.
BSL also has signs for negative possession and negative existence. For ne-
gation of possession, the sign HAVE-NEGposs is phonologically related to
the sign HAVE, with what Brennan (1992) described as a negative affix
occurring on some BSL signs (forearm rotation accompanied by an opening
of the hand; see Figure 13). This sign is used for possession only (23a), not
existence (23b). There are other variants of this sign also meaning ‘not
have’, including one which begins with two 5-hands palm down and ends
with both hands pronated so that the palms are facing upwards.
410 Kearsy Cormier and Jordan Fenlon
4.2.2. BELONG
d. *1-BELONG-a
‘That box is mine’
In (26), the referents ‘book’, ‘teacher’, and ‘bedroom’ may have been set
up in space already within the discourse, or they may not have been. If they
have not, the sign BELONG establishes the loci for each. In example (26c),
either the location of the box has already been established in space previ-
ously in the discourse, or the box is physically present during the utterance.
The verb BELONG falls into the category of verbs in BSL known as
agreement verbs, described in § 3.4 above, in which the verb begins at the
location associated with the subject and ends at the location associated with
the object. Agreement verbs exist in all known signed languages. However,
a predicative possession form like BELONG in BSL which acts as an agree-
ment verb has not to our knowledge been identified in any other sign lan-
guage outside of the British Sign Language family.
BSL also has a compound sign which has combined the adjectival predicate
of predisposition with the sign ALWAYS to resulting in an adverbial form
meaning ‘typically’, glossed in (29) as ALWAYS-TYPICAL and shown in
Figure 18. In ALWAYS-TYPICAL, the repeated movement of the predis-
positional marker is lost. This pattern of phonological reduction of move-
ment in the surface form is common in compound signs in signed languages
(Klima and Bellugi 1979; Sutton-Spence and Woll 1999).
In this chapter we have seen that British Sign Language exhibits many of
the same patterns of possession as spoken languages and in similar ways.
BSL has both attributive and predicative possession. BSL exhibits patterns
consistent with those in spoken languages with inalienable possession. The
lexical item HAVE is used for both possession and existence, which is an-
other pattern seen in many spoken languages. BSL also has a split between
HAVE and BELONG (foregrounded PR vs. foregrounded PM) shared by
many spoken languages.
When we turn to looking at differences across signed languages regard-
ing possession, these are generally the types of differences that we find
across spoken languages – i.e. those features which are subject to parametric
variation. Some languages like Australian Sign Language (Auslan), ASL and
English have a possessive clitic which attaches to nouns; some languages
like Danish (and, it seems, present-day BSL) do not. Some spoken lan-
guages mark inalienability while others do not; there is evidence that BSL
and some other signed languages mark inalienability (optionally at least)
though it seems in different ways.
So it is clear there are many similarities between possession in signed
languages and spoken languages. The differences seem to be strongest when
considering the locative nature of many BSL signs. Across spoken lan-
guages, there is a strong link between possession, existentials and location
Possession in the visual-gestural modality 417
(Freeze 1992; Heine 1997; Herslund and Baron 2001).10 Specifically, in many
spoken languages, the PM in a possessive construction is encoded as subject
while the PR is encoded as a locative (e.g. dative) argument. In signed lan-
guages, signs such as pronouns, some nouns, agreement verbs, demonstra-
tives, etc. are inherently locative. As we have seen in this chapter, many
signs denoting (and/or grammaticised from) possession – e.g. possessive pro-
nouns, nouns which can be spatially located, BELONG as an agreement verb,
the predispositional markers BE-TYPICAL-OF and ALWAYS-TYPICAL –
are spatially modified. This use of space is a natural product of the visual-
gestural modality and is perhaps the most striking difference between signed
and spoken languages when it comes to possession (and also pronominal/
nominal reference and agreement in general), and is a good reason to con-
sider signed languages when looking at the marking of possession across
different languages.
Indeed, typological studies which aim to cover phenomena across a
wide variety of languages often neglect to include signed languages. This
oversight is unfortunate because signed languages have much to offer lan-
guage typologists. The features and categories that all known spoken and
signed languages have in common are candidates for universal grammar,
while those categories that differ across signed and spoken languages – par-
ticularly those which are thought to be universal amongst spoken languages –
can help us tease apart those features which are truly inherent to human lan-
guage versus those which may not be universal after all. In this chapter we
have seen that while possession is likely a semantic category that is ex-
pressed in all human languages, language modality certainly does shape the
way in which it is expressed.
Acknowledgements
10
Though see Payne (this volume) for arguments against the “possession is loca-
tion” view.
418 Kearsy Cormier and Jordan Fenlon
References
Cormier, Kearsy
2007 Do all pronouns point? Indexicality of first person plural pronouns in
BSL and ASL. In Visible Variation: Comparative Studies on Sign
Language Structure, Pamela M. Perniss, Roland Pfau and Markus
Steinbach (eds.), 63–101. Berlin /New York: Mouton de Gruyter.
Cormier, Kearsy, Stephen Wechsler and Richard P. Meier
1999 Locus agreement in American Sign Language. In Lexical and Con-
structional Aspects of Linguistic Explanation, Gert Webelhuth, Jean-
Pierre Koenig and Andreas Kathol (eds.), 215–229. Stanford, CA:
CSLI Press.
Deaf Society of New South Wales
1998 Hands up NSW: A Profile of the Deaf Community of NSW. Sydney:
Deaf Society of New South Wales.
Deuchar, Margaret
1983 Is British Sign Language an SVO language? In Language in Sign: An
International Perspective on Sign Language, Jim Kyle and Bencie
Woll (eds.), 69 –76. London: Croom Helm.
Engberg-Pedersen, Elisabeth
1993 Space in Danish Sign Language. Hamburg: Signum Press.
Farris, Michael A.
1994 Sign language research and Polish Sign Language. Lingua Posna-
niensis 36: 13–36.
Farris, Michael A.
1998 Models of person in sign languages. Lingua Posnaniensis 40: 47–59.
Fenlon, Jordan and Kearsy Cormier
2006 Inalienable possession in British Sign Language. Paper presented at
the Ninth International Conference on Theoretical Issues in Sign Lan-
guage Research, Universidade Federale de Santa Catarina, Floriano-
polis, SC, Brazil, 6–9 December 2006.
Fischer, Susan D.
1975 Influences on word order change in American Sign Language. In
Word Order and Word Order Change, Charles N. Li (ed.), 1–25.
Austin: University of Texas Press.
Freeze, Ray
1992 Existentials and other locatives. Language 68: 553–595.
Friedman, Lynn
1975 Space and time reference in American Sign Language. Language 51
(4): 940–961.
Heine, Bernd
1997 Possession: Cognitive Sources, Forces, and Grammaticalization.
Cambridge: Cambridge University Press.
420 Kearsy Cormier and Jordan Fenlon
seisen@essex.ac.uk f.lichtenberk@auckland.ac.nz
Ingrid Sonnenstuhl
Institute for Language and Information
University of Düsseldorf
Universitaetsstrasse 1
40225 Düsseldorf
Germany
ish@phil-fak.uni-duesseldorf.de
Index of subjects
constructional networks, 213, 217, 238– existence, 3–4, 80, 109, 111, 115–116,
240, 246 133, 138, 181, 409–411, 416
context, 6, 20–21, 23, 113, 226–227, 250, existential constructions, 4, 82, 108n,
283, 286 115, 128n, 133–134, 136–138, 300
coreference (see also discourse status, existential predicates/verbs, 108, 120,
co-referential), 17 301–302, 359, 409–411
corpora, 13n, 27, 44n, 45, 134, 154, existential schema, 216
202, 213, 390 experiential gestalt, 216, 241, 245, 294
extension
dative arguments, 192, 194, 197, 232 in child language, 174, 178–179, 188
dative constructions, 192, 239 metaphorical, 114n, 138
dative marking, 156–157, 183, 189– prototype, 244, 246
191, 195–196, 198, 214, 222, 227, external possession constructions (EPCs)
231–-232, 236, 239, 245 (see also affected possessors), 2, 5–
dative object, 190–192 6, 144, 153–154, 189–192, 199, 214,
Dative of Interest (DI), 6, 233, 236–237, 216, 230, 344
239 in child language, 192–198
dative possessor, 214, 222, 227, 231–232, external possessor – see external posses-
235–236, 245 sion construction
definite, 13–28 passim, 37–47 passim,
54–55, 57, 59, 79–89 passim, 93, flexible parts-of-speech, 296–297
181–182, 187, 332 fluidity (of classifiers), 7, 250, 263, 273–
dependency reversal, 59, 76n 276, 281–282, 285
direct possessive construction, 6–7, 253– focus, 222–223
254, 256, 258–260, 263–267, 269– focus markers, 304, 307–308
270, 273, 276–277, 279, 282, 285 frame, 22, 24, 234
direct prefixes, 121 cognitive, 217
discourse context, 3, 15n, 19, 22, 30–32, interpretive, 234
37, 42, 47–48, 240 frame participant, 235, 237, 239
discourse status Functional Grammar, 52
anchored, 27, 41–44, 47 functional prototype, 217
coreferential, 27, 28 –31, 36, 379, functional space, 217, 238, 246
382–383
given, 3, 13, 17–20, 26–28, 41, 47 genitive
inferable, 25–27, 36–47 compound, 369
initial mention, 15, 24, 26, 28n, 46 construction, 6, 14, 56, 108, 150, 156,
new, 3, 13, 17–20, 24–28, 31, 37–38, 162, 167, 171, 178, 381
41–42, 45–47 non-determiner/non-referential, 63, 75
non-initial, 24, 46 possessor, 214, 222, 238, 245
text referent, 27, 32–36, 41 of quantification, 77
discourse functions, 20, 47–48 -s, see -s genitive
double object construction, 299 schema, 217, 233
dummy, 9, 300, 302, 335, 358–359, 370– syntactic – see syntactic genitive
371, 380 goal schema, 217, 233
Index of subjects 427