ChapterPDF Available

Possession and modification—a perspective from Canonical Typology

Authors:

Abstract

The chapter addresses a set of semantic, syntactic, and categorial criteria for canonical attributive modification and canonical inalienable possession. Canonical attributive modification is expressed by a dedicated word class (adjective) denoting a property concept, while canonical possession is inalienable possession of a relation noun (kin term, (body) part). Irina Nikolaeva and Andrew Spencer argue that modification-by-noun and alienable possession constructions can be analysed as non-canonical variants of canonical modification and possession. In many languages the same morphosyntax is used to express various combinations of the four constructions and the authors explore some of the typological variation in terms of selective violations of canonical properties. © editorial matter and organization Dunstan Brown, Marina Chumakina, and Greville G. Corbett 2013. © the chapters their several authors 2013. All rights reserved.
9
Possession and modificationa
perspective from Canonical Typology
IRINA NIKOLAEVA and ANDREW SPENCER
9.1The link between possession and modification
1
There are two very common ways in which a lexical noun head in a nominal phrase
(NP/DP) can be (loosely speaking) expanded by a dependent. In the first we have
some attributive modifier, (the) tall girl, and in the second we have a possessor
expression, (the) womans daughter, the girls hand. The semantics and the nature of
the dependency are rather different in the two cases. An attributive modifier
typically (we will say, canonically) denotes a gradable property concept, which is
normally represented semantically as a one-place predicate: tall(x). The meaning of
an attributively modified noun is the subset of the denotations of the head noun and
the class denoted by the property concept (that is, a tall girl denotes an entity which
is simultaneously a girl and a tall entity, i.e. of which the expression [tall(x) &girl(x)]
holds).
Inalienably possessed nouns such as hand or daughter are transitive, that is, they
are semantically two-place predicates: daughter-of(x, y),hand-of(x, y), where the y
argument stands for the entailed possessor. A possessive expression headed by a
relational noun, such as the womans daughter,the girls hand, denotes the head
noun in its relation to the possessor argument, which we can represent informally as
hand(x, girl), daughter(x, woman). Ordinary common nouns such as tree,woman
are usually treated as one-place predicates in semantics, with an argument structure
that can be represented as tree(x),woman(x). Such nouns do not entail the existence
1
Some of this material has been presented to audiences at the Linguistics Association of Great Britain,
Kings College London, September 2007, the International Conference on Adjectives, Université de Lille 3,
1315 September 2007, the 7th meeting of the Association of Linguistic Typology, Paris, 2528 September
2007, as well as the Creating Infrastructure For Canonical Typology, University of Surrey, 910 January
2009. We are grateful to the editors and to an anonymous referee for helpful comments. Part of this work
was conducted under grant 119393 to Andrew Spencer from the Arts and Humanities Research Council, to
whom we express our gratitude.
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 207 16.7.2012 1:42pm
of a possessor argument, though one can be provided, of course, in which case the
possessed noun is an argument in a certain semantically unspecified two-place
relation as discussed below.
Clearly, the two types of construction, attributive modification and (inalienable)
possession, share the property of being noun-headed but are otherwise rather
different in type. This difference is generally reflected in the morphosyntax of the
constructions. Attributive modification is normally expressed by a dedicated lexical
class of adjectives whose members may show special morphosyntax, specifically
agreement in features such as gender, number, or case. In some languages, the
adjective heads a kind of relative clause, as in the Nilotic languages, DhoLuo
(Stafford 1967) and Lango (Noonan 1992). A number of typological studies have
been devoted to adjectives (Bhat 1994; Wetzer 1996; Dixon and Aikhenvald 2004),
though it has to be said that attributive modification plays a Cinderella role
compared to predicative use (the main focus of Wetzer 1996, for instance). This is
slightly odd, given that the attributive use of adjectives seems to be primary
(cf. Croft 1991).
A variety of typological studies of possessive constructions have been published.
Many of them deal principally with the notion of possession in general, including
possessive predication (of the kind Mary has a daughter), though studies such
as Koptjevskaja-Tamm (2003a, 2003b) do deal specifically with the kind of NP-
structure we are concerned with here. Possessive constructions may be head-marked
or dependent-marked (in the sense of Nichols 1986). Head-marking often takes the
form of agreement morphosyntax, in which the possessed noun (possessee, posses-
sum) is marked with pronominal features of the possessor. This is familiar from a
number of Uralic and Altaic languages. However, in other cases the head is marked
by appearing in a special form. In Maltese, for example, (Koptjevskaja-Tamm 2003a,
647) the noun (iz-)zij-a (the) aunttakes the construct form zijt as in zijt Pawlu
Pauls aunt. Dependent marking is typically by case: the default or definitional
function of the case labelled genitivein descriptive grammars is to express the
possessor in a possessive construction, though often a language lacks a specific
dedicated case with just this function alone and uses some more general oblique
case for the purpose.
2
Another typical strategy is to use an adposition, such as
English of, or simple juxtaposition of the possessor and the possessed noun. A
double-marking pattern involves both possessive agreement and a kind of genitive
as, for example, in the Turkish definite izafet.
2
It is very common to find the term genitiveused as a synonym for possessivein the literature, even
when the language concerned has no case system. We will distinguish carefully between the terms
possessive constructionand possessorand the term genitive, reserving the latter for clearly defined
case systems.
208 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 208 16.7.2012 1:42pm
Given the rather different syntactico-semantic dependencies between head and
dependent in modification and possession constructions it is not surprising that
languages should choose to express tall girl and womans daughter differently. And
yet when we examine the structure of nominal phrases and nominal dependents
more closely we find there is a close typological connection between the two sorts of
construction. Languages sometimes adopt strategies for the two constructions that
are very similar if not identical. Perhaps surprisingly there have been relatively few
detailed studies of the morphosyntactic encoding strategies used to signal either
attributive modification or possession. An important exception is the work of
Koptjevskaja-Tamm (particularly, Koptjevskaja-Tamm 1997,2000,2001,2002,
2003a, 2003b, 2004) and it will be obvious that her research has had a significant
influence on our thinking about these issues.
The close relationship between the expression of possession and modification
is revealed by the well-known ezafefamily of constructions in Persian and
other Iranian languages. In a typical example of this strategy there is an
uninflecting marker (in Persian -(y)e) which comes between the possessor
and possessed nouns and which in some languages may not even be affiliated
morphosyntactically with either constituent. This marker, however, is also used
to signal the dependency between an attribute and the head. The ezafe family
of constructions demonstrates clearly that there is some deeper link between
the two constructions than would be apparent from examining other language
types.
A very clear relationship can be seen between the possessive construction and
attributive modification in certain languages in which the possessive construction
acquires the agreement morphosyntax of typical adjectival constructions in that
language. We consider two such cases here, the agreeing genitive postposition
construction found in a number of Indo-Aryan languages (for instance, Hindi-
Urdu), and the possessor cliticin Albanian, both of which have been consistently
misanalysed in the literature.
3
In Hindi-Urdu there is a set of adjectives which inflect, agreeing with the modified
noun in number (singular/plural), gender (masculine/feminine), and case (direct/
vocative/oblique). The basic pattern of agreement for the direct and oblique cases is
shown in (1):
3
We depart from Koptjevskaja-Tamm (2003a) in not relating these constructions directly to the Bantu
a-of-associationpossessive construction. This is a construction, widespread in Bantu languages, in which
a possessor is marked by a preposed -aformative which agrees in noun class with the possessed noun.
However, in Swahili, for instance, the possessor agreements are expressed with the pronominal set of
agreements, found with various pronouns, determiners, and verbs, and not with the agreement set
reserved for the rather restricted class of adjectives. The parallel between the possessive construction
and attributive modification is therefore not necessarily exact.
Possession and modificationa perspective from Canonical Typology 209
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 209 16.7.2012 1:42pm
(1) Hindi adjective agreement
direct Sg acchaa lar
!kaa acchii lar
!kii
Pl acche lar
!ke acchii lar
!kiyãã
oblique Sg acche lar
!ke acchii lar
!kii
Pl acche lar
! acchii lar
!kiyõ
good boy(s)’‘good girl(s)
The agreement is morphosyntactic in the sense that an inflecting adjective shows
agreement even if the modified noun is itself indeclinable (e.g. ghar house).
Thus far the description is uncontroversial. However, Hindi-Urdu also has a set of
postpositional clitics which signal grammatical functions. In traditional Western
descriptions as well as recent theoretical discussion these have been treated as case
markers (Masica 1991). One of these alleged case markers is the clitic postposition
kaa. This takes as its complement the possessor NP and it then agrees with the
possessed noun for gender (masc/fem), number (sg/pl), and case (direct/vocative/
oblique) in exactly the manner of an inflecting adjective such as acchaa:(kaa
masculine singular direct, ke all other masculine forms, kii all feminine forms).
This means that the kaa postposition effectively serves to turn the possessor noun
phrase into an attributive modifier phrase:
4
(2) Hindi genitivepostposition kaa: (McGregor 1995,9; see also Payne 1995)
a. us strii kaa bet
·aa
that woman kaa.m.sg son
that womans son
b. us strii ke bet
·e
that woman kaa.m.pl son.pl
that womans sons
c. us strii ke bet
·e kaa makaan
that woman kaa.m.obl.sg son.obl.sg kaa.m.dir.sg house(m)
that womans sons house
d. us aadmii kii bahnõ kaa makaan
that man kaa.f sister.obl.pl kaa.m.dir.sg house(m)
that mans sistershouse
Notice that sonin (2c) is marked as oblique. All postpositions, including kaa, take
the oblique form of their noun complement. Hence, the word bet
·aa sonhas to
appear in the oblique case form, bet
·e, when it is the complement of kaa in (2c). This
4
Note that we concur with Payne (1995) in denying that kaa is a derivational element which creates
adjectives. But that doesnt mean we agree with Payne in concluding that the kaa postposition is some
kind of genitive case marker.
210 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 210 16.7.2012 1:42pm
means that the kaa of us strii kaa that womansin (2c) has to take the obl.m.sg
form ke. A similar pattern is found in a number of other Indo-Aryan languages, such
as Punjabi, Nepalese, and others. Clearly, the constructions illustrated here show
that possession is marked by means of an adjectival strategy in the sense that it
exhibits the same pattern of attributive agreement. Treating formatives such as the
Hindi-Urdu kaa postposition as a case marker is simply an error (see Otoguro 2006,
Spencer 2005a, and Spencer and Otoguro 2005 for more detailed exemplification of
this point).
Albanian adopts essentially the same strategy as Hindi for expressing possession.
A very convenient concise summary of the relevant grammar is provided by
Newmark (1998, xxviif, esp. xxix, xlivf] and also Morgan (1984). The system is
described in detail in Buchholz and Fiedler (1987,201), Morgan (1984), and New-
mark (1998, xxix). Albanian nouns inflect for number, case, and definiteness and
fall into several inflectional classes broadly correlated with masculine or feminine
gender. Adjectives inflect for gender/number and fall into two groups, articulated
and unarticulated. The unarticulated adjectives do not agree in the features of
definiteness or case. The articulated adjectives are immediately preceded by a clitic
(or possibly affix), which Newmark calls the proclitic attributive article. This article
marks agreement with the head noun for gender, number, case, and to some extent
definiteness in a rather complex patterning, the essence of which is the following: i
for masculine nominative singular, efor other nominative/accusative forms, for
feminine oblique, and elsewhere. For definite noun phrases there are two
patterns of agreement depending on whether the modifier is the first in the string
of modifiers (proximal declension) or not (distal declension). Essentially what we
find is a form neutralization in which e/së from the proximal set are replaced by
in the (default) distal set. Examples for definite nouns are shown in (36), adapted
from Zymberi (1991,104):
(3) Nom djali i mirë dhe i sjellshëm
Acc djalin e mirë dhe të sjellshëm
Obl djalit të mirë dhe të sjellshëm
boy art good and art polite
the good and polite boy
(4) Nom djemtë e mirë dhe të sjellshëm
Acc djemtë e mirë dhe të sjellshëm
Obl djemve të mirë dhe të sjellshëm
boys art good and art polite
the good and polite boys
Possession and modificationa perspective from Canonical Typology 211
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 211 16.7.2012 1:42pm
(5) Nom vajza e mirë dhe të sjellshëm
Acc vajzën e mirë dhe të sjellshëm
Obl vajzës mirë dhe të sjellshëm
girl art good and art polite
the good and polite girl
(6) Nom vajzat e mirë dhe të sjellshëm
Acc vajzat e mirë dhe të sjellshëm
Obl vajzave të mirë dhe të sjellshëm
girls art good and art polite
the good and polite girls
We make no apologies for labouring the details of this system. The point is that
exactly the same patterning, in all its byzantine complexity, is found with possessive
constructions. The possessor noun acquires the proclitic attributive article, which
then agrees with the possessed head in exactly the manner just discussed:
(7) Albanian genitive(adapted from Zymberi 1991,53f.):
a. një djalë i një fshati
indf boy.nom.sg.indf art.m.nom indf village.obl.sg.indf
a boy of a village
b. studentët i shkollës
student(m).nom.sg.def(m)art.m.nom school(f)obl.sg.def
the (male) student of the school
c. studentes së kolegjit
girl.obl.sg.def art.f.obl college.(m)obl.sg.def
(to) the (female) student of the college
As can be seen from these examples the article agrees with the possessed (head)
noun, not with the possessor. If the article were to agree with fshati,shkolës,kolegjit
it would have to take the form .
In all cases of attributive modification or the possessor construction the proclitic
attributive article is associated syntactically with the dependent, that is, the adjective
or the possessor phrase, as in traditional descriptions. For attributive modifiers this
can be seen particularly clearly, from a variety of tests. First, the articles are repeated
on conjoined modifiers/possessors and dont seem to be able to take wide scope (this
can be seen from (3) above). This indicates pretty clearly that they form a constituent
with the modifier, not the head. Second, when an articulated adjective is used
predicatively it brings its article with it (Morgan 1984,238):
212 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 212 16.7.2012 1:42pm
(8) a. Djali ësht [i zgjuar]
boy.def is [art intelligent]
The boy is intelligent.
b. *[Djali i] ësht zgjuar
[boy.def art] is intelligent
Third, when an adjective is modified by a degree modifier the article remains
immediately adjacent to the adjective (9):
(9)një vajzë shumë e dashur (*e shumë dashur)
indf girl very art lovable (indf very lovable)
a very lovable girl
Plank (2002,165) provides evidence that nouns modified by nouns show the same
constituent structure, given here in (10):
(10) a. Akademia e Shkenca-ve të Shqipëri-së
academy.def art sciences-obl art.pl Albania-obl
the Academy of Albanian Sciences
b. Akademia e Shkenca-ve e Shqipëri-së
academy.def art sciences-obl art.sg Albania-obl
Albanian Academy of Sciences
From these examples we can see that the article takes a different form depending on
whether it is construed with sciences((10a), , plural agreement) or with Albania
((10b), e, singular agreement). In (10a) the constituent structure is [Akademia
[e Shkencave [të Shqipërisë]]]. In (10b) the constituent structure is [[[Akademia
[e Shkencave]] [e Shqipërisë]]]. Clearly, in (10a) we have an instance in which a
modifying possessor is itself modified, forming a minimal pair with (10b). The point
is that in (10b) the article which agrees with the head noun Akademia is not adjacent
to that noun but appears as a prefix to the noun Shqipërisë.
Example (10) also illustrates the important point that the genitive-marked NP
behaves like a noun phrase in the syntax and not like an adjective phrase. On both of
the construals of (10) a possessor-marked noun is modified as a noun, namely by
means of the possessive construction, and not as an adjective. Further examples can
be found in Buchholz and Fiedler (1987,418).
The Albanian possessor construction, like the Hindi-Urdu kaa construction, has
traditionally been mislabelled as a genitive case, a misapprehension that frequently
reappears in modern discussion of the language (despite the clearly enunciated
warnings to the contrary in Newmark 1957). However, the Albanian construction,
like its Indo-Aryan congener, has nothing to do with case (cf. also Koptjevskaja-
Possession and modificationa perspective from Canonical Typology 213
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 213 16.7.2012 1:42pm
Tamm 2003a, 660665) and demonstrates the inherent connection between posses-
sion and attributive modification.
In this chapter we take a closer look at these affinities and investigate them from
the viewpoint of Canonical Typology. This will lead us to ask a number of general
questions about possession and modification and what makes them at once similar
and distinct. The goal of the chapter is to identify canonical and less canonical
instances of possession and modification, and cross-linguistic regularities in their
expression. As far as we can tell, however, there has been no systematic study of the
relationships between the whole range of constructions lying between attributive
modification and inalienable possession.
9.2Canonical properties
9.2.1Types of canonical property
We begin by considering canonical possessionand canonical modification. Our
first assumption is that there is a family of constructions in which an object-denoting
category (noun) is the head and some other category is in a dependent relation to
that head. This dependency takes the form either of a possessive construction or an
attributive construction. The way we will proceed is to discuss a number of prop-
erties of the constructions under study and their component concepts (such as
nounor modification) and identify those properties we regard as canonical (we
will sometimes use the term criterionas a synonym for canonical property). These
criteria will take the form of absolute statements (a noun denotes a concrete object)
or contextually defined statements (the relation between possessor and possessed is
permanent/individual-level). The criteria we suggest refer to two domains: seman-
tics and syntax. We further divide the broadly syntactic properties into three groups:
external syntax, internal syntax, and categoriality. The three-way categorization is
mainly for expository convenience; we leave it to future research to determine
whether there are any strong correlations between these categories and canonical
properties.
External syntax refers to the distribution of the whole phrase within the wider
syntactic environment in its relation to other elements, in particular, the head.
A specific instance of external syntax is the relationship between the possessor
NP/DP and other types of phrase with a similar function, as in PossSyn1below.
Internal syntax refers to the relations between words/phrases within the smallest
relevant phrase (the possessor or the modifier, in this case). Categoriality has to do
with word-class membership, but of course we will have to rely on some character-
ization of canonical word classes. Preliminary observations about the canonical
properties of nouns and verbs are made in Spencer (2005b). That paper proposes a
Principle of the Morpholexically Coherent Lexicon, in which all lexical classes have
214 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 214 16.7.2012 1:42pm
uniquely identifiable morphological, syntactic, and semantic properties. For instance,
a noun is a word denoting a physical object which expresses the argument of a verb,
while an adjective denotes a gradable property in the form of an intransitive predicate,
serving as the attributive modifier of a noun (cf. also Croft 1991). The purpose of
that paper is to explore the various ways in which morpholexical coherence is
compromised by actual lexical entries. However, we can take the idealized properties
as canonical.
To some extent our claims about syntactic criteria are at the mercy of specific
model-particular assumptions, but we will try to be as neutral as we can in this
regard, while at the same time trying not to exclude specific models of grammar.
One area where we can hardly avoid controversy though is where we try to talk
about the argument structure of a noun. The notion of argument structure lies on
the interface between semantics and syntax and it is very unclear which side of that
divide it should lie on. However, we think there are enough clear cases to allow us to
identify canonical criteria even here. We will make a number of claims about what
we think the canonical syntactic patterns are.
In principle, we might also identify criteria for the encoding strategies, that is the
morphosyntactic exponents typically used to express those constructions. For in-
stance, we might speculate that possessive agreement morphology on a head noun is
a canonical way of expressing possession, while we might suggest that it is canonical
for an adjective to agree in gender, number, case, definiteness, or whatever with its
head noun. However, encoding strategies are heavily language dependent and their
study, while interesting for typology generally (as will become clear in section 4), is
tangential to our present concerns.
9.2.2Canonical possession
It is sometimes thought that the prototypical possessive relationships include part/
whole, kinship, and legal ownership/disposal (Langacker 1991,169, Langacker 1995,
59, Koptjevskaja-Tamm 2001,961). These are indeed the most frequent (and in this
sense prototypical) possessive meanings. However, the distinction between these
three meanings, on the one hand, and other possessive meanings, on the other, is
rarely grammaticalized. On the other hand, many languages do have unambiguous
encoding of the alienable/inalienable distinction (Chappell and McGregor 1996).
As has been repeatedly noted in the literature (among others Chomsky 1970;
Jackendoff 1977; Partee 1997; Partee and Borschëv 2003), the nature of this distinc-
tion largely determines the semantics (and perhaps the syntax) of possessives. We
therefore prefer to treat alienables and inalienables differently.
There is no apparent way of grouping part/whole, kinship, and ownership to-
gether in semantic terms, while the alienable/inalienable distinction may be ex-
pressed as the opposition between relational vs. non-relational nouns. In this
Possession and modificationa perspective from Canonical Typology 215
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 215 16.7.2012 1:42pm
chapter we will rely heavily on Barkers(1995) insights in assuming that there is a
(relatively) clear semantic distinction to be drawn between alienable and inalienable
possession, specifically at the level of nominal argument structure representations.
Following much previous work, Barker (1995) distinguishes between lexical(or
intrinsic) possession and extrinsicpossession, which roughly correspond to in-
alienableand alienablepossession, respectively. For the former the semantic
interpretation is determined by the meaning associated with the head noun, which
provides a determinate specification of the nature of the relation between the
possessor and the possessed. Thus, the phrase Johns wife biases an interpretation
of inalienable possession (the woman to whom John is married). In principle, this
does not preclude the possibility that the expression could denote somebody elses
wife, whom John has selected as a winner in a competition for wives, but such an
interpretation is constrained to occur in very special pragmatically stretched con-
texts and does not sound completely natural. This unique interpretation is ensured
by the fact that the nominal wife is relational. Relational nouns such as kin terms,
meronyms (part-of term, for instance, body parts), topological nouns, or nouns
denoting an inherent property are semantically dependent and expected to occur in
possession constructions, although the entities classified as inalienable are known to
vary considerably from language to language.
5
The important point is that in each
language the choice is determined lexically: it has to do with the semantics (argu-
ment structure) of the possessed noun rather than style, the speakers construal of
the situation, or other similar factors (cf. Nichols 1992,120).
So, the notion of inalienable possession refers to those nouns, or those nouns in
certain usages, whose semantics effectively defines them as two-place predicates:
lxly[head(x,y)], where the yvariable ranges over possible possessors (somebodys
head). The representations for Marys head/daughterwill therefore be roughly
as in (11):
(11) a. lxly[head(x,y)](mary)
b. lxly[daughter(x,y)](mary)
This entails the existence of the possessor argument. For instance, body parts imply
the existence of a body, as do other part-terms in partwhole relationships (such as
the corner of a box). Similarly, kin terms imply their related kin terms: mother "
child,sister "sister and so on. In many languages with possessive agreement
morphology such words cannot be used without their possessor agreements (see
Nichols and Bickel 2005).
5
As is well known, there is no clear semantic basis for the alienable/inalienable opposition in a cross-
linguistic sense, nor any universally applicable hierarchy of what counts as inalienable (Chappell and
McGregor 1996,89 Heine 1997). Moreover, there may be different degrees of inalienability: some languages
have several possessive classes. We will ignore these distinctions for present purposes.
216 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 216 16.7.2012 1:42pm
Lexical possessives contrast in Barkers theory with extrinsic (or alienable) pos-
sessives involving non-relational nouns, e.g. Johns book. As mentioned above, a
non-relational common noun is normally a one-place predicate: we can assume that
bookhas the argument structure lx[book(x)] or some such. In cultures that have
an institution of legal ownership the expression Johns book will often be paraphra-
sable as something like the book which John owns. However, this is by no means
the only possible interpretation. In many languages this expression can also refer to
the book that John wrote, the book that he stole, the book he always talks about, the
book he wants to buy, or the book that he has just been given to review as a class
assignment. The possessive relation here is not narrowly determined on the basis of
inherent semantic properties of the possessed noun and, consequently, it allows for
various readings. Although there is often a bias for an ownership interpretation,
consistent with a paraphrase using the verb own, other relations are also inferable.
That Johns book, unlike Johns wife, is semantically ambiguous is due to the fact that
the noun book is not intrinsically relational. Its semantic structure is such that there
is no exclusive candidate for elaboration by the possessor nominal. Whatever the
precise factors and mechanisms might be, the interpretation of the intended relation
between the possessor and the possessee is extrinsic to the lexical semantics of
possessed noun.
The twofold classification of possessives into lexical/intrinsic/inalienable and non-
lexical/extrinsic/alienable raises the question of which type is semantically primary.
We take inalienable possessives to instantiate canonical possession. Taylor (1989,
1995,1996) suggested that the semantics of possessives is not amorphous but instead
has an internal structure, which does not depend on a single definitional criterion.
The possessive relationship is best regarded as a cluster of several independent
properties (simplified after Taylor 1996,340):
(12) The possessive relationship
a. The possessor is a specific human being;
b. The possessed is an inanimate entity, usually a concrete physical object;
c. The relation is exclusive, in the sense that for any possessed entity, there is
usually only one possessor;
d. The possessor has exclusive rights of access to the possessed;
e. The possessed is typically an object of value, whether commercial or
sentimental;
f. The possessors rights of access to the possessed are invested in him
through a special transaction;
g. The possession relation is long term, measured in months and years, not
in minutes or seconds;
h. The possessor is typically located in the proximity of the possessed.
Possession and modificationa perspective from Canonical Typology 217
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 217 16.7.2012 1:42pm
Inalienable possession is characterized by essentially all the properties listed above. It
involves an exclusive asymmetric long-term relation and physical proximity between
two entities; for each possessee there is only one possessor who has the right to make
use of the possessee, and the possessor is normally an individuated human being (cf.
Heine 1997,2001). Quite probably humans are the most frequent possessors in type
and token corpus frequency measures (see Vergnaud and Zubizarreta 1992 and the
papers in Coene and dHulst 2003). It seems intuitively reasonable to say that
humans are also canonical possessors, however, nothing in our discussion hinges
on whether they are. Finally, it follows from the categorization in (12) that we take
inalienable possession to be a permanent property of a noun head. If Marys head
refers to her own head then the property Mary has a headis surely a permanent, i.e.
individual-level property (Carlson 1977) of Mary, whereas if it refers, say, to a
chicken head she happens to have it is more likely to be a temporary, stage-level
possession. In sum, we believe that Taylors prototypical properties characterize
canonical (inalienable) possession, so this may well be a case of a prototypical
situation also being canonical.
Moving broadly from most semantic/conceptualto most syntactic, we propose
the four groups of canonicity criteria for possession.
Semantics:
PossSem1: The relation between possessor and possessee is permanent (i.e. an
individual-level rather than stage-level predication)
PossSem2:Possessors are humans
PossSem3:Possessees are relational nouns (kin terms/meronyms . . . ) and hence
are two-place predicates
External syntax:
PossSyn1:(In languages with a specifier system), possessors occupy a spec(ifier)
position (and therefore are canonically in paradigmatic opposition to
other spec elements)
Internal syntax:
PossSyn2:(In languages with a specifier system), possessors can take their own
specifiers (other possessors and determiners) and attributive modifiers
Category:
PossSyn3:Possessors are (canonical) nouns, hence, show independent number
opposition, take attributive modifiers as adjuncts, determiners (and
possessors) as specifiers
PossSyn4:The possessed entity is a canonical noun
218 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 218 16.7.2012 1:42pm
PossSyn5:The possessed noun is the semantic and syntactic head of the con-
struction (and the possessor expresses its argument)
The syntactic criteria seem to us to be fairly uncontroversial. One notion that may
require explanation is specifier. We broadly follow Sag, Wasow and Bender (2003)
in taking a specifier of a nominal to be a generalization of the notion of determiner,
that is, a syntactic satellite to a noun head which is not a complement or modifier
and which (canonically) serves to fix referential properties. The intuition we wish to
capture here is that canonical possessors are canonical nouns, and canonical nouns
have the potential to serve as referential expressions. In languages whose syntax
includes a specifier system this means that canonical nouns will take a specifier.
Moreover, by PossSyn1in a canonical system with specifiers we expect possessors to
be in complementary distribution with determiner elements, and not, say, with
qualitative adjectives (indeed, we find it hard to even imagine such a possibility).
These criteria are not entirely independent. PossSyn2follows in large part from
the fact that possessors are canonically nouns (PossSyn3), which in turn follows
from PossSem2and PossSem3. Since possessors are nouns they will lexically head a
nominal phrase (NP or DP), hence, PossSyn2. However, we have not found a way of
deriving PossSyn1: although possessed nouns are canonically relational (PossSem3)
this does not mean that they have to have specifiers: the argument of a noun can
equally be expressed by means of an adpositional or case-marked complement, and,
indeed, has to be in a language which lacks a nominal specifier system.
9.2.3Canonical modification
Following a long typological tradition (e.g. Dixon 1991; Croft 1990) we take the basic
type of modifier to be a word denoting some gradable property concept which
characterizes a physical object. Canonical modifiers are words denoting gradable
properties of size, shape, age, and the like. But we differ from Gil (2005) in that we do
not take colour terms to represent a canonical type of attribute (even if in some sense
colour might be a prototypical attribute). Colour terms tend to behave differently
from other attributes. They are often grammaticalized from nouns and can some-
times retain nominal properties lacking in true adjectives. This is even found in
English: for instance, in archaic or poetic registers we can say things like a sky of
blue, a coat of green, a wine of deepest red, and so on. These constructions are
unavailable to non-colour terms: *a sky of bright,*a coat of warm,*a wine of sweet.
Typologists agree that gradable property concepts are frequently expressed by a
separate word category of adjective (see Croft 1991 for typical discussion). Even
in languages in which the adjective is a minor or peripheral category, its exemplars
typically express prototypically gradable properties such as size or colour (Dixon
1991). We therefore propose that it is canonical for an adjective to denote a gradable
Possession and modificationa perspective from Canonical Typology 219
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 219 16.7.2012 1:42pm
property (even though in those languages that have such a category, the adjective can
normally be used in a variety of functions). At the same time, we maintain with
many others, including Bhat (2000), Croft (1991), and Spencer (1999) that attributive
modification is the principal grammatical function realized by adjectives. The
predicative use of an adjective, or of any category other than a (finite) verb, is
secondary, as is secondary predication as in She painted the door blue.
Canonical modifiers are one-place predicates whose only argument is identified
with that of the head noun.
6
In the standard semantic analysis, the meaning of
attributive modification by gradable property words such as the big house is set
intersection. The denotation of the adjective identifies a certain subset in the
denotation of the head noun by specifying which house is meant. Since modification
narrows the concept associated with the head noun, it serves the purpose of
classifying the respective entity. Distributionally, then, an adjective-noun group is
identical to a common noun in isolation. We will assume the standard semantics of
set intersection (taking both nouns and adjectives to be one-place predicates for the
sake of simplicity) (cf. Siegel 1980; Higginbotham 1985; Larson and Segal 1995 and
many other references). Thus, a typical instance will be (13):
(13) the big house: the (house \big) ¼lx [big(x) & house(x)]
In other words, the big house is an entity that belongs both to the class of houses and
the class of big things (modulo syncategorematicity).
The standard syntax of attributive modification is adjunction or juxtaposition of
attribute and head. The adjective itself does not determine phrasal properties in any
way. Where a language has a specifier system we take it to be canonical for attributes
to be distinguished from specifiers by iterativity: canonically it is possible to have an
indefinite number of attributive modifiers but it is non-canonical to have more than
one specifier. (Recall that canonicaldoes not mean prototypical, so the fact that
multiple attributes are generally rare is not relevant here.) For languages which have
a wealth of adjective-oriented morphosyntactic devices a canonical attribute will, for
instance, agree with the head for typical features (especially gender, but also number
and perhaps case, definiteness, and other nominal categories).
Canonical properties of attributive modifiers:
Semantics:
ModSem1:The modified element denotes a physical object
ModSem2:Modifiers denote gradable property concepts and hence one-
place predicates
6
A good deal of discussion of adjective semantics centres around puzzles such as alleged criminal,
good violinist, and so on. We take such syncategorematic adjectives to be non-canonical and not relevant
to our discussion.
220 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 220 16.7.2012 1:42pm
External syntax:
ModSyn1:Modifiers are adjuncts to noun heads
Internal syntax:
ModSyn2: Modifiers do not take specifiers and other (attributive) modifiers
Category:
ModSyn3: Modifiers are (canonical) adjectives
ModSyn4: The modified word is a canonical noun
ModSyn5: The modified word is the semantic and syntactic lexical head of
the construction (and the modifier the dependent)
We can summarize the syntactic difference between canonical modifiers and pos-
sessors schematically as in (14):
(14) Canonical modifiers and possessors compared
Possessors Modifiers
Lexical class N A
Syntactic distribution Spec, NP Adjunct
9.3Intermediate (non-canonical) constructions
We now explore two sets of constructions headed by lexical nouns, one a type of
non-canonical modification, modification-by-noun, and the other non-canonical
possession, namely, alienable possession. In this section we summarize the way
that the two intermediate construction types are related to each other and to our
two canonical types.
9.3.1Modification-by-noun
Given our characterization of canonical modifiers in section 9.2.3, non-canonical
modifiers denote non-gradable properties such as circumstances (time, place, man-
ner), function, material, events, and the like. It is possible for languages to modify
nouns with reference to such entities, even if they have purely noun-like denotations
(such as concrete count nouns) and belong to the lexical class of nouns. This means
that categorially non-canonical modifiers are words which canonically denote a
referential object rather than words which denote a property. We therefore refer
to this type as modification-by-noun.
7
According to Koptjevskaja-Tamm (2000,
7
There does not seem to be a standard term for this semantic phenomenon. It has been referred to
as non-anchoring relationin Koptjevskaja-Tamm (2000,2004) and other work, and as specification
in Heine (1997,1567), but the latter term is also used in defining phrase structure relations, so we will
not use it.
Possession and modificationa perspective from Canonical Typology 221
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 221 16.7.2012 1:42pm
2004), a dependent in the modification-by-noun relation (the modifying noun)
serves to classify, describe, or qualify the class of entities denoted by the head,
whereas Heine (1997,157) states that it refers to the same general entity as the
specifiedand characterizes the head noun by narrowing down the range of possible
referents that may qualify as the specified. These descriptions effectively characterize
intersective modification: the modified expression denotes a subset of the entities
denoted by the head noun.
A typical instance of modification-by-noun is Germanic-style noun-noun com-
pounding. In a regular endocentric compound of the type London bus or coffee table
the first noun serves as the attributive modifier of the head noun. Such modifying
nouns may sometimes take their own attributive modifiers, but they are similar to
adjectives in that they fail to take nominal specifiers such as definite articles (*a [the
capital] bus vs. a London bus). In English it is especially difficult to determine
whether the modifier London in London bus is categorially a noun or an adjective
because there is little relevant morphology which distinguishes the two categories,
but modifying nouns in non-lexicalized compounds have a distribution which is
rather similar to that of adjectives. For instance, we can coordinate such nouns with
adjectives: tribal and clan allegiances (see also Huddleston and Pullum 2002,
Chapter 5, section 14.4, on what they call the composite nominal construction).
The semantics of noun-noun compounding is a matter of debate. There are two
approaches, which we can think of as the semantic approach and the pragmatic
approach. Proponents of the semantic approach try to identify a small set of
primitive semantic relations that hold between head and modifier, such as location,
instrument, part-of, and so on. A good example of this approach is Levi (1978).
Proponents of the pragmatic approach argue that the relation between the two
nouns can be any pragmatically appropriate, contextually determined relation.
Downing (1977) provides a wealth of instances of compounds which can be inter-
preted in a whole host of ways depending on the precise context. For instance, bike
girl could mean girl who comes to work on a bike,girl who is standing next to a
bike, or whatever. When we look at nonce formations it is difficult to avoid the
conclusion that only the pragmatic account is adequate, because its so easy to
concoct or attest examples which defy any analysis in purely semantic terms. Even
fixed lexicalized expressions can sometimes be impossible to define in terms of
simple semantic primitives. For example, speed camera means photographic appar-
atus for taking images of drivers exceeding the speed limit. We therefore assume
that the semantic contribution of the noun-noun compound construction is to
predicate an unspecified, contextually determined relation, <, between the two
nouns (cf. AllensR-relation (1978) and many subsequent authors).
222 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 222 16.7.2012 1:42pm
Many languages do not allow a noun to modify another noun directly and so we
find a variety of morphosyntactic encoding strategies being deployed in order to
express dependencies of the kind Noun1 which bears some relation to Noun2.A
popular strategy is to turn a noun into a word which has the grammatical properties
of an adjective. We will call such denominal adjectives relational adjectives.
8
The
pureform of relational adjective is one in which the derived adjectival form has
exactly the same range of denotations as the original (undetermined) noun. In other
words, a relational adjective is derived from the base noun by a process of trans-
position (Beard 1995) which alters the morphosyntax of the word but does not
introduce any additional predicate to the semantic representation. Relational adjec-
tive formation is simply a way of bringing a noun into line with the grammar of the
language so that it can serve as a syntactic modifier. In this sense, the construction
relational adjective þnounsubserves effectively the same function as English
noun-noun compounding, and may be synonymous to it, cf. the English expressions
prepositional phrase/preposition phrase.
English relational adjectives do not provide good models of the type of construc-
tion we are interested in. They are generally borrowings from Romance languages or
Greek and compete with other devices for realizing modification-by-noun, notably
with noun-noun compounding. Moreover, English relational adjectives seldom have
a purely relational semantics, but often acquire additional nuances of meaning or
restrictions on meaning. In other languages, however, especially those lacking the
compounding strategy, a relational adjective will typically involve no additional
semantics whatsoever. In Russian, which lacks Germanic-type noun-noun com-
pounding, relational adjectives are derived from nouns with the suffix -n-. Like
canonical qualitative adjectives, they normally precede the head and agree with it in
case, number, and gender, although their syntactic behaviour does differ in part
from the behaviour of qualitative adjectives: relational adjectives in Russian do not
normally occur as predicates and lack so-called short forms. A typical example is
knižnyj magazin book shop, where kniž- is the palatalized allomorph of the root
knig-bookfound before suffixes such as -n-, while -yj is an agreement affix.
The meaning of relational adjectives in -n- is as general as in English noun-noun
compounds. For example, knižnyj usually means having books, with booksand
zvezdnyj >zvezda startypically means starry, having stars, with stars. However, it
is easy to find examples with lexicalized uses, just as we can find compounds with
fixed interpretations. For instance, stennoj derived from stena walldoes not mean
having a wall, with a wallbut located on a wall(relation of place), zamočnyj
derived from zamok locktypically means used for a lock(relation of purpose),
moločnyj derived from moloko milkusually means made of milk(relation of
8
Though the reader is warned that the term relational adjectiveis sometimes used in other senses, too.
Possession and modificationa perspective from Canonical Typology 223
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 223 16.7.2012 1:42pm
material), and nočnoj derived from nočnightmeans in the night(time relation).
But even with examples such as these we find alternative uses. The particular
interpretation depends on the semantics of the base noun and general knowledge
of the world. For instance, moločnyj can occur in expressions such as moločnyj
kombinat/zavod milk factory,moločnaja dieta milk diet,moločnyj saxar lactose
(literally milk sugar), and moločnyj brat brother through ones wetnurse. While
zvezdnyj usually has the proprietive meaning of with/having starsas in zvezdnoe
nebo starry sky, with other head nouns it can have other meanings, cf. zvezdnye
vojny star wars,zvezdnyj god sidereal year, and zvezdnaja karta star chart(see
also other examples cited in Mezhevich 2002).
Thus, the semantics of Russian relational adjectives in -n- is essentially empty:
they express some pragmatically defined relation between the head noun and its
modifier.
9
Overall, then, the range of functions and uses of relational adjectives
corresponds very closely to the range of functions and uses of attributively modify-
ing nouns in noun-noun compounds.
In sum, in modification-by-noun a head noun is modified by a word which itself
has the (canonical) denotation of a noun. This means that the modifier is non-
canonical in a categorial sense. This type of modification involves a non-canonical
use of an object-denoting word (a noun or some denominal category) and has to be
to some extent syntactically non-canonical simply by virtue of the fact that attribu-
tive modification is a non-canonical grammatical function for such words to realize.
A noun used as an attributive modifier will lose some of its canonical noun
properties, in particular, its referentiality. This is particularly clear in languages
with a specifier system: attributively used nouns can sometimes head their own
modifiers but do not normally take their own specifiers. For proponents of the DP-
hypothesis, for instance, this means that what gets turned into an attribute is an
expression of category N or NP, not DP.
Whether modification-by-noun is syntactically non-canonical in other respects is
not so clear. If a language permits a noun to modify another noun by means of
syntactic adjunction then perhaps we can say that the structure is syntactically a
canonical modification structure. (Matters are made more complex here by the fact
that there is relatively little agreement about the nature of adjunction in any case,
and in the case of attributive modification in particular.) However, this seems to us
unlikely. In practice, noun-noun compounding, if syntactic at all, lacks much of the
syntactic freedom of genuine attributive modification. Indeed, a canonical definition
9
Although the distinction between relational and qualitative adjectives is primarily semantic, the
boundary between them is vague in part because many relational adjectives may acquire a metaphorical
qualitative meaning, cf. serdečnaja boleznheart diseaseand serdečnyj čelovek warm-hearted person
(literally heart person). The former is interpreted as involving a relation between the notions diseaseand
heart, and the latter expresses a quality of the modified noun and does not contain a direct reference to
heart.
224 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 224 16.7.2012 1:42pm
of compounding might well take a compound to be a type of word rather than a type
of phrase. One could argue that the canonical semantic purpose of a compound is to
serve as a name for some permanently or temporarily nameworthy entity (see Dahl
2004,252,256, Booij 2009, and Spencer 2011).
Modification-by-noun is also semantically non-canonical, in that it has to be
mediated by a pragmatically determined relation. Although it appears in some
cases that the relation between modifier and modified is semantically specified (a
bookshop would be odd if it did not sell books, but much odder if it were not even a
shop), in many cases the nature of the relationship between the modifier noun and
the head noun is not determined by the meanings of the parts and is instead
contextually or pragmatically defined.
9.3.2Alienable possession
The second intermediate construction type, alienable possession, is illustrated by
expressions such as Johns book. In languages such as English there hardly seems to
be any morphosyntactic difference between this and inalienable possession (e.g.
Johns wife) though of course in many languages the two types are sharply distin-
guished. But even when alienable possession looks superficially like inalienable
possession they differ in terms of argument structure of the possessed noun. We
have argued above that if the possessum is a relational noun, the possessive relation
is the relation implied by the possessum, and this corresponds to inalienable
possession. But if the possessum is non-relational, there will be no way for the
head noun to determine the semantic relation between the two nouns. Alienable
possession is defined over a noun lacking the lexical, intrinsic argument structure of
a relational noun. It often (always?) has semantic range extending beyond the core
semantic notions of kin relation, body part, part-whole, or ownership, and in fact
the precise relationship has to be determined by the context and pragmatics, rather
like a noun-noun compound in English.
10
For instance, as mentioned above, an
expression like Johns book means essentially no more than the book which bears
some relation to John. It does not make a great deal of sense semantically to think of
these interpretations as possession in a strict sense, despite the fact that this is
commonly the type of term used to describe these constructions. The only reason
for referring to them as possession is because the formal means of expressing these
meanings are often the same as those for expressing ownership.
The question is how non-canonical possessive meanings arise and how a single
construction comes to express such a profusion of semantic relations. We have seen
10
Kay and Zimmer (1976) argue that the possessive (in their terms, Genitive) NP is simply a
metalinguistic instruction to the hearer that there is some kind of relation between the possessor and
the possessee. The hearer automatically supplies the appropriate interpretation in a manner that makes
sense, given the context in which they appear.
Possession and modificationa perspective from Canonical Typology 225
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 225 16.7.2012 1:42pm
that Taylor describes the semantics of possession as the prototypical properties listed
in (12) above. Since the prototype is a complex cluster of properties, it permits
deviations. Possessive constructions involving non-relational nouns are open to
multiple interpretations which differ greatly with regard to their relative degree of
prototypicality in the sense that the more relevant properties characterize the
relation, the more it resembles the inalienable prototype. For example, when Johns
book is taken to mean the book that John is reading at the moment (but which does
not belong to him)properties (12c), (12d), (12g), and perhaps (12f). do not hold.
Deviations are restricted by context-based conditions and can be described with
reference to various parameters, such as control and time (Heine 1997). The context
here may be characterized in terms of the speakers encyclopaedic knowledge of the
usual relation(s) that obtain between the relevant entities, as well as the specific
discourse situation that might mediate and modify this relation.
Some languages exhibit several alienable constructions signalling different kinds
of semantic relationships between the possessor and the possessed. For instance, in
Lonwolwol so-called direct possessive marking occurs when the possessed item is an
inalienable object, which includes most kinship terms, most body parts, and other
parts of wholes (data from Paton 1971):
(15) a. hela-k b. v«ra-m
brother-1sg hand-2sg
my brother’‘your hand
In contrast, alienable nouns can occur with one of the five indirect possessive hosts
which denote the intended use of the possessum, and act as host to the possessive
pronouns: liquid possessive host, container possessive host, fire possessive host,
vessel possessive host, and the general possessive host used for nouns that do not
belong to other categories. In this construction the possessive host itself receives the
agreement marker, for instance:
(16)a-k m«l«h
edible-1sg food
my food
When the possessor is a lexical noun we see two constructions. With inalienable
possession we see the direct construction but with the possessor juxtaposed to the
possessum. With alienable possession the third person agreement marker attaches to
one of the five possessive hosts:
(17) a. hela-n vant«n
brother-3sg man
a/the mans brother
226 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 226 16.7.2012 1:42pm
b. w«ma-n bülbülan
water liquid-3sg friend
the water of the friend
Which of the five hosts the alienable possessor noun occurs with does not depend on
its own meaning, but rather on the nature of the relation that holds between the
possessor and the possessed:
(18) a. ha-k ol
general-1sg coconut
my copra (my coconut used as copra)
b. a-k ol
edible-1sg coconut
my coconut (for eating)
c. ma-k ol
liquid-1sg coconut
my coconut (for drinking)
What we have here is grammaticalization of a number of specific semantic relations
within the very general semantically underdetermined relation between the alienable
possessor and possessed.
11
It is important then (as many have pointed out) that the notion of alienable
possession should not be equated with any notion of a legalownership. Ownership
may well be the frequent interpretation in cultures with notion of personal property
and, like inalienable possession, it involves more or less all relevant properties. That
is why it is often taken to be one of the prototypical possessive relationships.
However, it is very easy to override that interpretation. We doubt that ownership
(under whatever construal) is in some sense a canonical interpretation for alienable
possession. A more subtle interpretation, valid for many languages, might be that
ownership is a (perhaps language-specific) unmarked default interpretation of
alienable possession.
12
On the other hand, it is unclear what other type of interpret-
ation could be canonical for alienable possession. The interpretation of alienable
possession thus seems to proceed along much the same lines as modification-by-
noun, provided we factor out those instances in which ownership is implied.
11
We are grateful to Dunstan Brown for comments which clarified our thinking about examples such
as these.
12
Taylor (1989,681) argues that when the possessor is in focus (This is johns car), as well as in some
other special contexts in English, the ownership reading is the only available interpretation of the alienable
possessive relation. Barker (1995) and Storto (2004) provide further evidence that (legal) ownership is an
unmarked interpretation of extrinsic possessives.
Possession and modificationa perspective from Canonical Typology 227
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 227 16.7.2012 1:42pm
There are two points to make about the syntax of alienable possession.
First, alienable possession is non-canonical as a syntactic construction because the
possessed noun (head) is a non-canonical, non-relational possessum, hence, the
possession relation is not well defined. The possessive meaning is obtained by
effectively allowing coercionof a possessor argument on a noun which does not
really have that argument structure (indeed, this is exactly how alienable possession
is expressed in Lexical Functional Grammar, see Bresnan 2001,293). It may well be
that we would wish to say that a non-relational noun (such as girl or tree as opposed
to leg or branch) is canonical as a noun, but such a noun is not canonical as a
possessed noun.
There are, in fact, several ways in which a noun can be non-canonical as a
possessed noun. The most important are listed in (19):
(19) Non-canonical types of possessed noun
non-relational object tree, girl
event journey
deverbal/deadjectival noun destruction, arrival, sincerity
While simple event nouns such as journey do not seem to be much different from
any other non-relational noun, the overtly deadjectival or deverbal nouns interact
with possessive constructions in a rather complex way, as is well known from the
copious literature on argument structure realization in such constructions. Typically,
the possessive construction is used with such derived nominals to express one or
other of the base lexemes arguments: the enemys destruction of the city, Marys
arrival, the Presidents sincerity are all directly relatable to the enemy destroyed the
city,Mary arrived, and the President is sincere. The fact that such nouns have an
implied argument structure brings them closer to the canonical possessed noun, but
they do not give rise to canonical possessive constructions. Even if we were to
identify true relational nouns such as kin terms with deverbal nominals, thereby
satisfying PossSem3by fiat (a dubious manoeuvre), the latter would still almost
always violate criterion PossSem1.
The second important point about the syntax of alienable possession is that it is
similar to inalienable possession in that the possessor is canonically a fully fledged
NP/DP, and not just a bare noun. This makes alienable possessors crucially different
from the nominal modifiers discussed in section 9.3.1, even though in both cases we
have a one-place predicate and some kind of semantically underdetermined relation
between the dependent and the head. We have seen that modification-by-noun is
just that, a relation between two nouns, while alienable possession is a relation
between a head noun and a noun phrase. It is because alienable possession estab-
lishes a relation between a NP/DP and a head noun that it can so readily assume the
same form of morphosyntactic expression as inalienable possession.
228 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 228 16.7.2012 1:42pm
9.4The morphosyntax of possession-modification
in typological perspective
There exist languages which take our four-way typology seriously and adopt four
different encoding strategies, though this is not especially frequent. One example is
Lele (NigerKongo, data from Frajzyngier 2001), which exhibits the alienable/inali-
enable opposition. The inalienable construction is head marked: the head noun
follows the dependent and bears possessive agreement that indicates the gender of
the possessor (20a, b). With alienable possession the word order is reversed: the
dependent precedes the head. The dependent is followed by marker kV agreeing
with the head in gender (20c):
(20) a. kùrmbàlo cày
chief head.3m
chief s head
b. kùrmbàlo kamday
chief wife.pl3m
chief s wives(Frajzyngier 2001,66)
c. grà cànigé kèy
dog Canige gen.3m
Caniges dog(Frajzyngier 2001,70)
In contrast, nominal modification is achieved by simple juxtaposition of two
nouns involving no agreement and no linking elements, as shown in (21) (Frajzyngier
2001,75):
(21) a. kàrà túgú
people village
village people
b. wèlè kàsùgù
day market
market day (¼Sunday)
c. Godu tamá
monkey woman
monkey woman
Property-denoting words used as attributive modifiers take one of two forms
(Frajzyngier 2001,88). Either we see a postnominal attribute agreeing in number
and gender with the head (22a,b), or we see a postnominal non-agreeing verbal form
(a deverbal noun, 22c):
Possession and modificationa perspective from Canonical Typology 229
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 229 16.7.2012 1:42pm
(22) a. tamá tùnya
woman(F) tall.f
tall woman
b. kama jàlwa
waters cold.pl
cold waters
c. kùlbá bóré
cow be.white.vn
white cow
Other languages which belong to this type are Tuvaluan (Besnier 2000) and Hoava
(Davis 2003). In a number of other languages the alienability opposition is absent but
modification-by-noun and attributive modification are clearly distinct from posses-
sion, so our four construction types are rendered by three encoding strategies. This is
the case in Turkish and Kolyma Yukaghir (Maslova 2003).
At the other end of the spectrum, Chukchi (Chukotko-Kamchatkan) illustrates a
language type in which all four constructions under consideration are expressed by
essentially the same encoding strategy, in this case an adjectival strategy. Chukchi has
distinct morphology for deriving both relational and possessive adjectives and
both types of adjective are fully productive (Koptjevskaja-Tamm 1995; Skorik 1961,
225268 on possessive adjectives, 268280 on relational adjectives; NB both these
sections come in the Nounschapter of Skoriks grammar). The possessive
adjectives are formed with the suffix -(n)in(e)/-(n)en(a) (for plural human possessors
-rgine/-rgena) as in (23a), while the relational adjectives are formed with -kin(e)/-ken
(a), as in (23b).
13
(23) Possessive and relational adjectives in Chukchi
a. Possessive adjectives in -(n)in(e)/-rgin(e)
etlegenfatheretleg-in walefathers knife
?ett?e dog?ett?-in renren(the/a) dogs food
tenec?enflowertenec?-in wetwetthe leaf of the/a flower
b. Relational adjectives in -kin
emnuntundraemnun-kin gennik tundra animal
l?elensummerl?elen-kin ewir?ensummer clothing
Possessive adjectives in Chukchi tend to be used for kin relations, part-whole
relations, ownership, and other relations cross-linguistically described under the
13
The relational adjective strategy is not incompatible with the compounding strategy. Chukchi has
noun-noun compounding as well, but relational adjective formation is very productive and less prone to
lexicalization and in this sense is the default way to achieve modification-by-noun.
230 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 230 16.7.2012 1:42pm
rubric of prototypical possession, and are regularly found with inanimate nouns, too.
The relational adjectives, however, are found mostly with inanimate nouns and are
used for general pragmatically determined relations (but with some overlap with
possessive adjectives, e.g. for partwhole relations). Inanimates then can take either
the possessive suffix or the relational suffix. Skorik (1961,2689) offers the minimal
pair reproduced in (24a, b):
(24) a. weem-in p
ec?e
c
?e
n
river-poss.adj current
the river(s) current
b. weem-kine-t
wekwə-t
river-rel.adj-pl rock-pl
the rocks in the river
Skorik (1961,249,268) explicitly states that with inanimates, possessive morphology
entails inalienable possession (for us, a kind of canonical possession), while relational
morphology entails alienable possession. On the other hand, animates seldom take
relational adjective suffixes, even to express alienable possession. For them the pos-
sessive forms express both alienable and inalienable possession. Chukchi then has a
possessive suffix for nouns which are prototypically able to own things, but possessive
morphology has been co-opted for expressing a prototypically possessive relationship
with respect to inanimates, namely, inalienable possession (particularly, partwhole).
Possessive adjectives are the principal way of expressing NP-internal possession in
this language. The example below is taken from Dunn (1999,148153).
14
It shows that
the possessive form retains its nominal category, in that it can take its own possessor.
(25)Jare-n uweqəc-in ətləgən
Jare-poss.adj.3sg.abs husband-poss.adj.3sg.abs father-e-3sg.abs
[He was] Jares husbands father(Dunn 1999,149)
The possessor can also be modified by an adjective. The adjective is incorporated
when unfocused. For instance, the analytic phrase n-ilgə-qin qora
N
əADJ-white-
ADJ reindeer,white reindeerfocuses to some extent on the adjective white,
whereas the incorporated form elgə-qora
N
əwhite-reindeeris neutral with respect
to information structure (Skorik 1961,103,429). With the (circumfixal) comitative
and coordinative cases incorporation is obligatory, as seen in (26):
15
(26)qolet
?ə
ljo>muri nivinimuri ga-npənacg-ərg-ena-akka-ta
next.day we hunted com-old.man-pl-poss.adj-son-com
The next day we hunted with the sons of the old men.
14
In the morpheme glosses estands for an epenthetic schwa which cannot be affiliated to any
particular morpheme.
15
Plural number is not usually expressed in oblique cases in the class of nouns to which ekk-son
belongs. The vowel alternations are the result of regular vowel harmony.
Possession and modificationa perspective from Canonical Typology 231
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 231 16.7.2012 1:42pm
In (27) (Dunn 1999,150) we see the possessive form -njiw-in of the noun -njiw-
uncle, modified by means of an incorporated adjective ?eqe bad:
(27)ənr?aq ənqen ?eqe-njiw-in ekke-t¼?m
then that bad-uncle-poss.adj.3sg.abs son-3pl.abs¼emph
lejwələət jet-g?e-t ecgi
walk-e-nmlz-e-3pl.abs come-pfv-3pl as.soon.as
ənqen ?era-m
N
əl-at-
NN
o-g?a-t
that.3sg.abs race-announce-e-inch-pfv-3pl
Then that bad uncles sons came, they walked there, as soon as they heard
about the race.
On the other hand, possessive and relational adjectives are trueadjectives in the
sense that they show attributive agreement with the head, as shown in (28):
(28) a.
nelwəl?ə-kine-k
?aacekə-k
herd-rel.adj-loc youth-loc
at the youth from the herd(i.e. belonging to)
b. mirg-ine-t kupre-t
grandfather-poss.adj-pl net-pl
grandfathers nets
c. mirg-ine-te kupre-te
grandfather-poss.adj-ins net-ins
with grandfathers net
Chukchi then employs the same adjectival strategy for all construction types. This is
also true of Hindi and Albanian, except that in Chukchi the modifying noun is
transposed into an adjective lexically, whereas in Hindi and Albanian the entire
noun phrase is turned into an agreeing attributive modifier by a particle or post-
position which bears the agreement morphology. Although Hindi and Albanian do
have noun-noun compounds with attributive meaning, these are apparently not as
productive as those, say, of the Germanic languages (as Kachru (2006,114) explicitly
points out for Hindi). In Albanian the possessive construction would appear to be
the default, productive way of achieving modification-by-noun. Buchholz and Fie-
dler (1987,219221) identify twenty-five uses for the adnominal possessive, including
subject-like functions (the answer of the pupil,the author of the article), object-
like functions (the defence of the fatherland), picture-noun constructions, partitive-
type constructions (a salad of tomatoes,the number of listeners), and so on. The
kaa-strategy in Hindi is also a productive way of expressing modification-by-noun
and is regularly used to render a wide range of types (the following examples are
taken from Dymshits 1986,58: time, puchle saal kii fasal last year kaa harvest,last
years harvest; material, lakr
!ii kaa ghar wood kaa house,wooden house, quality;
232 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 232 16.7.2012 1:42pm
laal rang kaa kapr
!aa red colour kaa material,red-coloured material,khušii ke
aasuu joy kaa tears,tears of joy; quantity, tiin saal kaa baccaa three years
kaa child,three-year-old child,paac rupaye kii kitaab five rupees kaa book,
five-rupee book,bii kilomiit
!ar kaa raastaa twenty kilometer kaa road,twenty-
kilometer road; function, bait
!hne kii jagah place kaa sitting,place for sitting,
nahaane kaa paanii bathing kaa water,bathing water,silaaii kii mašiin sewing
kaa machine,sewing machine. In addition, the kaa construction expresses the
subject, object, instrument, and other arguments of deverbal nominals in expressions
of the form the arrival of the delegation,the director of the school,preparation for
the journey,a feeling of confidence, and so on.
When different morphosyntactic strategies are used for different construction
types, modification-by-noun occupies a pivotal point in the typology of the posses-
sion-modification family of constructions. Languages very often co-opt either the
adjectival strategy or the possession strategy to express such a meaning, but as far as
we can tell, there has been very little discussion in the literature about precisely these
relationships.
The type that involves two major strategies, one for possession and another for
modification, is very widespread. For example, in Taleshi alienable and inalienable
possession is expressed by the number-dependent obliquemarker on the depen-
dent noun.
16
(29) a. Huseyn-i ka b. merdak-un ka
Huseyn-obl.sg house men-obl.pl house
Huseyns house’‘mens house
c. palang-i pust
panther-obl.sg skin
the skin of the panther
Other semantic types employ the uninflectable linking clitic -a. In (30) we show
examples of nominal modification expressing various semantic relations, whereas
(31) shows attributive modification:
(30) a. palang-a pust b. s
ə
ng-a ka
panther-lkr skin stone-lkr house
panther skin’‘stone house
c. ner-a palang
male-lkr panther
male panther
16
The Taleshi data are obtained from Gerardo De Caro.
Possession and modificationa perspective from Canonical Typology 233
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 233 16.7.2012 1:42pm
(31) a. xub-a âdam b. qašang-a
kela
good-lkr man beautiful-lkr girl
a/the good man’‘a/the beautiful girl
In some languages modifying nouns are not turned into adjectives by an overt
derivational process, but show canonical adjectival properties such as attributive
agreement. For instance, in Tundra Nenets (Uralic) the possessive construction with
lexical possessors is dependent-marked: the possessor takes the genitive expressed by
-hin the singular and -qin the plural on non-possessed nouns, or a variety of other
affixes on possessed nouns. The possessive construction renders the usual alienable
and inalienable meanings:
(32) a. nyíbya-h mal8b. puxacya-h pad8ko
needle-gen end woman-gen bag
the end of the needle’‘the womans bag
In contrast, qualitative and nominal modification is expressed by prenominal juxta-
position. The semantic difference between these two constructions has been studied
in detail by Tereščenko (1967) and Tereščenko (1973,219ff.) and by Nikolaeva (2002),
where it is argued that the dependent genitive expresses the relationship between two
referents, while the dependent nominative is non-referential and expresses a prop-
erty. Consider the following contrast:
(33) a. ti-h ya
reindeer-gen soup
reindeerssoup, soup for the reindeer
b. ti ya
reindeer soup
reindeer soup (soup made of reindeer meat)
Other examples of nominal modification are sax8r xidya sugar cup, cup used for sugar
and myercya pyi windy night. The dependent nominative shows optional number
agreement with the head. In this it does not differ from qualitative adjectives, cf.:
(34)a.nyarawa-q lo
N
key8-q b. serako-q lo
N
key8-q
copper-pl button-pl white-pl button-pl
copper buttons’‘white buttons
A roughly similar pattern obtains in a Numic language, Tümpisa (Panamint)
Shoshone, described by Dayley (1989), where genitives do not agree, whereas adjec-
tives and modificational nouns show case agreement with the head.
In contrast, a language such as Finnish uses its possessive construction with
genitive case as its principal productive strategy for modification-by-noun (there
are also noun-noun compounds in Finnish but their productivity is not as great as
234 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 234 16.7.2012 1:42pm
in, say, Germanic languages). Possession is expressed by putting the possessor in the
genitive case in -n:Presidenti-n nimi the Presidents name,kirjo-j-en sisältö book-
pl-gen content,the content of the books(Karlsson 1987,89). Karlsson explicitly
notes that the genitive strategy is the principal modification-by-noun strategy when
he says Genitive expressions <...>are typical to Finnish; in many European
languages the corresponding forms are preposition or adjective structures or com-
pound nouns.He then cites examples including:
(35) a. Helsingi-n yliopisto
Helsinki-gen university
Helsinki University
b. Niemise-n perhe
Nieminen-gen family
the Nieminen family
c. ruotsi-n kieli-n opettaja
Swede-gen language-gen teacher
the Swedish language teacher
Finnish attributive adjectives are a clearly distinguishable lexical category and they
agree with their head noun in number and case (there is no grammatical gender). Other
languages which belong to this type are, for example Swedish and Hausa. In these
languages the possessive encoding strategy is employed for modification-by-noun.
Finally, we find languages in which only non-canonical alienable possession (but
not canonical inalienable possession) assimilates in some sense to the class of
modifiers. For instance, in Miya (Chadic) the primary strategy for attributive
modification involves postnominal adjectives agreeing in gender and number with
the head. Miya further opposes the so-called directand indirectgenitives (Schuh
1998,246f.). Direct genitives indicate inalienable possession and involve tonal alter-
nations but no agreement-sensitive elements.
(36) a. báa Vàziya b. átín laahə
father Vaziya nose jackal
Vaziyas father’‘jackals nose
Indirect or linkedgenitives have the following structure: head nounagreeing
linkerdependent noun. The linker is syntactically and phonologically attached
to the dependent and expresses gender and number agreement. The construction
is ambiguous between alienable possession and modificational interpretations, as
illustrated in (37) and (38), respectively.
17
(37) a. mb!
ergu na Vaziya b. t!
emakwíy niy Vaziya
ram lkr.m Vaziya sheep.pl lkr.pl Vaziya
Vaziyas ram’‘Vaziyas sheep
17
In these examples variations in the form of the masculine singular linker na/n are determined
phonologically by the quality of the following segment.
Possession and modificationa perspective from Canonical Typology 235
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 235 16.7.2012 1:42pm
(38) a.
dlángər
naakám b. ts!
erak!
etaamúku
animal lkr.m.house standing lkr.f.sun
domestic animal’‘midday
Thus, both non-canonical constructions (modification-by-noun and alienable pos-
session) are expressed by the adjectival strategy.
In Yamphu (Tibeto-Burman), as described in Rutgers (1998), the genitive in mi(n)
is used for inalienable possession (although it is not entirely clear what counts as
inalienable in this language). The genitive does not agree. But other types of
adnominal dependent host an attributivizing morpheme which agrees with the
head in number: m(a) in the singular and h(a) in the plural. It can be used on
attributive adjectives (albeit optionally).
(39)a.utthri;ma akma b. utthri;ha u
N
white.sg.attr pig white.pl.attr flowers
white pig’‘white flowers
Attributivizers are also found on nouns if they combine with postpositions and oblique
cases. These elements express various relationships between the head and dependent
noun, for example, the similative or sociative. This exemplifies modification-by-noun.
(40) a. pasadokma hæ
N
child.sim.sg.attr you
you who are like a child
b. akoknu
N
ha yamiji
load.soc.pl.attr person.nsg
people with a load, load people
In addition, the plural attributivizer is present on the so-called possessivecase in -æ
(the singular attributivizer is omitted in this instance). The possessive case expresses
a certain belonging together(Rutgers 1998,70), which can be understood as some
kind of possessive relation but the nature of the relationship need not be so strict.
This corresponds to our understanding of semantically undetermined alienable
possession. Alienable possession is illustrated in (41):
(41)a.Sittambæ khimbe b. ma;ksæ gotthabe
Sittambæ.poss house.loc bear.poss shed.loc
in the house of Sittamb’‘in the shed of the bear
We can say that examples (40) and (41), i.e. nominal modification and alienable
possession, exemplify the same encoding strategy: the dependent agrees with the
head and hosts an additional marker which specifies the exact nature of the semantic
relationship between the head and the dependent. In our examples it is similative,
sociative, or possessive(i.e. general pragmatically defined relation). The point is
236 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 236 16.7.2012 1:42pm
that, just like in Miya, this strategy differs from non-agreeing inalienable genitives
and exhibits canonical modificational morphosyntax (attributive concord).
9.5The canonical perspective on possession-modification
We began our study with the observation that constructions which realize attributive
modification and (inalienable) possession share the property of being headed by a
noun, embodied in our virtually equivalent criteria ModSyn4/ModSyn5and Poss-
Syn4/PossSyn5. We developed a set of semantic, syntactic, and categorial criteria
for canonical attributive modification and canonical inalienable possession and
argued that modification-by-noun and alienable possession constructions are non-
canonical variants of these canonical constructions. Thus, we can take a canonical
instance of possession, such as Johns head or the branch of the tree and progressively
render it less canonical by violating increasingly more of the canonical criteria. We
can replace the relational term headwith a non-relational term such as hator
knifeand obtain an instance of alienable possession, Johns hat/knife. Similarly, we
can replace the possessum branchwith a term such as ceremonyto obtain a phrase
denoting, say, a religious ceremony associated with a particular tree: the ceremony of
the tree/the tree ceremony. The less canonical the possessor and possessum the less
the construction resembles what we would normally think of as possession. In
an expression such as tree ceremonywhat we have is modification-by-noun.
Conversely, we can take the notion of attributive modification and render the
modifier progressively less canonical. Semantically non-canonical modifiers
will include non-gradable concepts (dead rat,prime number) and syntactically
non-canonical modifiers will include those which are derived from verbs, such as
participles or relative clauses (a broken vase/the vase which John broke). When the
syntactically non-canonical modifier is a noun we get modification-by-noun: tree
ceremony.
By considering these criteria we have seen one possible way of appreciating the
(perhaps unexpected) unity of the four relevant constructions, as well as the differ-
ences between them. We saw that in many languages the same morphosyntax is used
to express various combinations of the four constructions and we have explored
some of the typological variation in terms of selective violations of canonical
properties. If our analysis is correct, there should be no idealway of expressing
intermediate construction types, modification-by-noun, and alienable possession.
All languages will have to make compromises in order to express these relations.
Thus, the adjectival strategy (relational adjective) used for expressing the modifica-
tion-by-noun construction in Russian, Taleshi, and Nenets can be seen to deploy the
morphosyntax of canonical modification even though the modifier is non-canonical
from the point of view of categoriality. On the other hand, where modification-by-
noun is expressed using a possessive strategy (as in Finnish, Swedish, or Hausa) the
Possession and modificationa perspective from Canonical Typology 237
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 237 16.7.2012 1:42pm
modifier, being a noun, does not belong to the canonical category for modifiers, and
therefore the construction assumes some of the properties of nominal dependents.
We further saw that alienable possession differs somewhat minimally from canon-
ical possession in terms of the semantics of the possessed noun, and so it is often
expressed by the same encoding strategy. But it is still a distinct construction type,
which is reflected in languages with an alienability opposition. Moreover, there are
languages like Miya or Yamphu which treat alienable possession as a kind of
modification (as opposed to fully nominalinalienable possession), reflecting the
fact that this type is not canonical possession.
Other than our own recent studies (Nikolaeva and Spencer 2009) we know of no
research which addresses directly the question of the relationship between all four of
these grammatical relationships or construction types, even though a number of
typological investigations have been devoted to individual types. Moreover, it is not
even clear how current conceptions of typology would address this question, since the
hierarchies and scales they generally propose do not naturally lead to the stepping-
stone effect which links adjectival modification and possession. In contrast, a feature of
the canonical approach to typological description is that the canonical criteria
appealed to are often very general and hence apply to a whole range of construction
types. Some of the criteria involve statements about category membership and about
basic syntactic and semantic relations. The result is that it becomes easier for us to see
overlaps and commonalities between apparently disparate construction types: we can
say exactly which canonical properties are violated in each particular case.
The Canonical Typology perspective also allows us to gain a better insight into
disputatious cases where analysts are at odds over how to describe certain phenom-
ena because they fail to fit closely enough into canonical (or indeed, prototypical)
construction types. A rather striking instance of this is the case of the Indo-Aryan
and the Albanian possessive constructions, discussed in section 9.1. Perhaps the
majority of authors who discuss possessive constructions in these languages speak of
agenitive case. If this description were taken seriously it would make it very
difficult for typologists to arrive at a clear understanding of what might be meant
by case system. By factoring out the components of the construction we can see that
these languages are simply more permissive than most in allowing the form of the
possessor phrase to deviate from the canonical expression of a noun phrase. And
Chukchi is even more permissive, in the sense that it deploys the morphological
system to create adjectives from nouns to render all semantic types of adnominal
dependents.
In sum, we believe that the canonical perspective allows us to look at familiar facts
in a new and possibly more illuminating light. The advantage of Canonical Typology
is that it gives us the tools to examine the precise relationship between attributive
modification and possession in those areas where the boundaries are blurred by
attested linguistic constructions.
238 Irina Nikolaeva and Andrew Spencer
OUP UNCORRECTED PROOF REVISES, 16/7/2012, SPi
Dunstan Brown / Canonical Morphology and Syntax 09_Brown_Ch09 REVISES page 238 16.7.2012 1:42pm
... This relation can be, for example, a part-whole, a kinship, an ownership or a control relation, but it can also express much looser association (see e.g. Seiler 1983: 4;Heine 1997: 3-10, 33-40;Partee 1997;Partee & Borschev 2003;Herslund & Baron 2001: 1-4;Nikolaeva & Spencer 2013;Myler 2016: 3-4;Ortmann 2018). ...
... The semantic nature of POSS, however, is different from the relations introduced by relational nouns. POSS roughly corresponds to what Barker (1995;), Partee (1997, Partee & Borschev (2003), Ackerman & Nikolaeva (2013), and Nikolaeva & Spencer (2013) call R. These authors emphasise that, in contrast to relational nouns, the possessive predicate introduced by non-relational nouns is semantically underspecified. ...
... If inalienable possession is a canonical possessive relation, as proposed by Nikolaeva & Spencer (2013) within the framework of Canonical Typology (Brown & Chumakina 2013;Corbett 2015;Bond 2019), the generalisation in (40) can be thought of as an implicational relation on canonical and less canonical types of possession: if a language allows a less canonical possessive relation between two SR pivots to license SS-marking, it will also allow more canonical possessive relations to do so. Recall that we understand as "inalienable" as "part-whole" in the context of SR. ...
Article
Full-text available
Some languages with switch-reference use same-subject markers in structures where the internal possessor of one subject corefers with another subject, but the subjects do not corefer with each other. We analyse such patterns as a type of non-canonical switch-reference (Stirling 1993; de Sousa 2016) and show that languages differ in what types of possessive relations license same-subject marking. Languages that allow alienable possessive relations in switch-reference also allow inalienable relations to license same-subject marking, but not 'vice versa'. In addition, alienable, but not inalienable possessive relations, must be morphosyntactically expressed when licensing same-subject marking. Adopting a modified version of Stirling’s (1993) approach, we derive these implicational relations from anaphoric conditions licensing non-canonical switch-reference.
... A rich and extensive survey of the strategies of adnominal possession in the languages of Europe is provided byKoptjevskaja-Tamm (2003).Nikolaeva & Spencer (2013) andSpencer (2013: 348-356) offer a recent insightful discussion of possession and modification, laying out the possibilities in a typologically diverse range of languages. ...
Preprint
Full-text available
Linguistics, and typology in particular, can have a bright future. We justify this optimism by discussing comparability from two angles. First, we make explicit some of the logical underpinnings of typological sciences, linguistics included, which we believe are worth reminding ourselves of. Second, we give a brief illustration of comparison, and particularly measurement, within modern typology.
Chapter
Competition takes many forms. A newly identified type of competition involves the featural specification of one of the competitors as a key factor. In the particular instance treated here, whether a given item has a competitor depends on its number (and sometimes its person). We focus on the use of the genitive case versus adjective-like forms in possessive expressions (broadly understood). The data come primarily from the Slavonic languages, where a surprising original system of possessive pronouns competing with personal pronouns has played out rather differently through the family. We find a variety of outcomes, from conservative to highly innovative, with some instances of competitors settling into different niches.
Article
The three most common strategies used to modify a head noun, namely through a possessive, an adjective and a relative clause construction feature in Siwi the use of the preposition n . Its presence is obligatory in the possessive constructions, but only present before an adjective or a relative clause in some contexts, depending on the level of restriction that the speaker wants to place on the head noun. The aim of the article is to describe the use and function of n in all three contexts of noun modification in Siwi and present supplementary data that helps the understanding of the global function of this preposition.
Article
Full-text available
Linguistics, and typology in particular, can have a bright future. We justify this optimism by discussing comparability from two angles. First, we take the opportunity presented by this special issue of Linguistic Typology to pause for a moment and make explicit some of the logical underpinnings of typological sciences, linguistics included, which we believe are worth reminding ourselves of. Second, we give a brief illustration of comparison, and particularly measurement, within modern typology.
Article
Full-text available
This study investigates the pragmatic referentiality and semantic modifiability of incorporated nouns. While some researchers argue that incorporated nouns have a referential function, others claim that they are not used to refer. Similarly, some hold that incorporated nouns are modifiable, whereas other researchers maintain that they cannot be modified. In order to tease apart these conflicting views, the present study systematically investigates the cross- and intra-linguistic variation regarding the referentiality and modifiability of incorporated nouns. A pre-defined set of criteria for the identification of referentially vs. non-referentially used nouns and modifiable vs. non-modifiable nouns, taken from Functional Discourse Grammar, is applied to incorporated nouns in a sample of 21 languages. The results show variation between referentially used modifiable nouns, non-referentially used modifiable nouns and non-referentially used non-modifiable nouns, both across and within languages. In addition, referentially used modifiable incorporated nouns and non-referentially used non-modifiable incorporated nouns appear to occur independently of each other, such that the conflicting perspectives on the referentiality and modifiability of incorporated nouns may be related to differences between studies and theoretical approaches in the languages they focus on. Moreover, incorporated non-referentially used modifiable nouns are only found in languages that also show incorporated referentially used modifiable nouns, which suggests that two independent incorporation processes should be distinguished: the incorporation of modifiable nouns and the incorporation of non-modifiable nouns.
Article
Full-text available
We present three case studies of the distribution of adjective + head noun (‘adjective’) vs. head noun + noun-genitive (‘genitive’) constructions based on datasets extracted from the Russian National Corpus. Each case study focuses on a different set of non-head referents: case study 1 examines non-heads that are country names (like ‘Norway’ as in norvežskij N vs. N Norvegii), case study 2 looks at non-heads that refer to leaders (like ‘president’ as in prezidentskij N vs. N prezidenta), and the focus of case study 3 is non-heads that are person names (like ‘Petja’ as in Petina N vs. N Peti). Head nouns in all three datasets were annotated for the same set of nine semantic categories representing an Individuation Hierarchy. This hierarchy accounts for only some of the patterns that we see across the case studies. Other patterns can be explained in terms of: ‘uniqueness’, which favors the genitive construction when the head noun is a unique entity; ‘salience’, which favors the genitive construction when the non-head is more salient than the head noun; and ‘obligatoriness’, which favors the genitive construction when the head is a relational noun that presupposes a specific non-head.
ResearchGate has not been able to resolve any references for this publication.