Part

Published on June 2016 | Categories: Documents | Downloads: 59 | Comments: 0 | Views: 345
of 31
Download PDF   Embed   Report

circumbaltic patrtitive

Comments

Content

1

‘A Piece of the Cake’ and ‘a Cup of Tea’: Partitive and
Pseudo-Partitive Nominal Constructions in the CircumBaltic Languages
Maria Koptjevskaja-Tamm

0. Introduction
‘A piece of the cake’ and ‘a cup of tea’ are both examples of what are traditionally called ‘partitive
constructions’ in linguistics. On closer inspection, however, we see that only in ‘a piece of the cake’ are
we really talking of a PART of something rather than an AMOUNT of some substance, as we do in ‘a cup
of tea’. In this paper, ‘a cup of tea’ will therefore be called a pseudo-partitive construction. The paper is
thus a cross-linguistic, or rather an areal typological study of constructions such as those illustrated in
(1) and (2) for Finnish, Russian, Swedish and English (seen in the translations of the other examples) 0.
(1)

Partitive nominal constructions (PCs)
a. Finnish
pala
tästä
hyvästä
kakusta
bit:NOM this:ELATgood:ELAT cake:ELAT
‘a bit of this good cake’
b. Russian
čaška
ètogo
vkusnogo
čaja
cup:NOM this:GEN.SG.M delicious:GEN.SG.M tea:GEN
‘a cup of this good tea’
c. Swedish
en
kopp
av
detta
goda
te
a:SG.COM
cup
of
this:SG.N good:DEFtea
‘a cup of this good tea’
(2) Pseudo-partitive nominal constructions (PPCs)
a. Finnish
säkki
perunoita
sack:NOM potato:PRTV.PL
‘a sack of potatoes’
b. Russian
čaška
čaja/čaju
cup:NOM tea:GEN/tea:PRTV
‘a cup of tea’
c. Swedish
en
kopp te
a:SG.COM cup tea
‘a cup of tea’

Although examples like those above are often referred to as partitive constructions, here I will follow
the usage adopted in modern syntactic and semantic theories and discriminate between partitive (such as

2
in ex. (1)) and pseudo-partitive (such as in ex. (2)) (nominal) constructions/ NPs. The reasons for this
terminological distinction will be given in Sections 1 and 2.
As (1) and (2) illustrate, there is considerable variation between languages in the grammatical
marking of the substance-denoting expression in partitive and pseudo-partitive constructions, ranging
from case inflections (as in (1a, b)) to prepositions (as in (1c)) to zero marking (as in 2c). It is this
variation that will be the topic of this paper. Also, if we compare the two examples given for each of the
languages, we will see another interesting difference: while Russian and English use more or less the
same structure in (1) and (2) (involving the genitive case and the preposition of respectively), Finnish and
Swedish distinguish between the two (by choosing between the elative and the partitive case in the case
of Finnish and by using or not using the preposition av in Swedish). This kind of morphosyntactic
variation will constitute the main focus of the present paper, which will thus aim at answering the
following question:


What is the structure of partitive and pseudo-partitive constructions across languages in general and in
the circum-Baltic languages in particular?

This question is discussed both synchronically and diachronically. The data from a large number of
European languages form the typological background for the study, whereas the circum-Baltic languages
(Finnish, Estonian, Lithuanian, Latvian, Russian, Belarusian, Polish, German, Swedish and Danish) are
analyzed in a much more detailed way, in particular with respect to the changes in their construction
types.
Finally, the synchronic and diachronic facts are used to draw conclusions about possible
grammaticalization sources for partitive and pseudo-partitive constructions in general. To my knowledge,
no previous typological work has been carried out for this domain.
The order of presentation will be as follows. Section 1 attempts to sort out the terminological mess in
connection with the term “partitive“, while Section 2 sets the stage for the study itself, by defining the
main terms used in it and formulating its main goals. Section 3 gives a fairly detailed account of the
situation in two Finnic languages, Finnish and Estonian, which is then used as the basis for more general
hypotheses about the structure and development of PCs and PPCs. These generalizations are discussed in
Section 6 in connection with the Indo-European circum-Baltic languages, the data on which are presented
in Sections 4 (Balto-Slavic) and 5 (Germanic). Finally section 7 places the circum-Baltic languages in a
broader typological pan-European context.

1. Various traditional usages of the term “partitive“
The term “partitive“ is one of those many other traditional grammatical terms, which have developed
several, not necessarily related, meanings. Grammatical tradition knows of “partitive usages of cases (e.g.
of genitive)“, “partitive case“, “partitive article“ and “partitive construction/NP“. In this section I will
merely list these usages of “partitive“ – mainly to avoid possible misunderstandings as to what the paper
actually looks at and what it leaves out. “Partitives“ (usages, cases, articles or constructions) have to do
with reference to:






parts of physical objects.
subsets of definite (super)sets,
definite quantification, and
indefinite quantity.

Thus, within the Indo-Europeanistic tradition, the term “partitive“ is normally associated with case
semantics, primarily in relation to the genitive case. Discussions and lists of meanings attributed to
genitives frequently include “partitive (meanings/uses of) genitives“. These terms, in their turn, are used
quite inconsistently in different works. Thus, “partitive genitive“ may cover:


reference to body-parts and “organic“ parts of objects: ‘the roof of the house’, ‘the middle of the
street’, ‘the lion’s head’ (cf. Pitkänen 1979: 220 - 222, Herslund: 1997)

3


reference to a set from which a subset is selected by means of various non-verbal words (cf.
Brugmann & Delbrück 1909: 597 - 599, Behaghel 1923: 485 - 498, Smyth 1956: 315 - 317, Wessén
1970: 25; another term for these usages is genetivus totius), e.g.:
- adjectives in the superlative degree: ‘the best among the Troyans’,
- numerals: ‘three of the boys’,
- quantifier nouns: ‘a section of the barbarians’, ‘an amphora of that good wine';



definite quantification, i.e. indication of the kind of entity that is quantified by a nominal quantifier, a
numeral, a quantifying adjective, etc. (e.g. Brugmann & Delbrück 1909: 597 - 599, Behaghel 1923:
485 - 498; another term for these usages is genetivus generis): ‘an amphora of wine’, ‘dozens of
soldiers';



indefinite quantity, i.e. reference to “partial objects“ of certain verbs (‘to eat’, ‘to drink’ etc.),
normally alternating with accusatives (cf. Behaghel 1923:575 - 578, Smyth 1956:320 - 325), e.g. in
Classical Greek, Gothic and Old High German, Sanskrit and Balto-Slavic.

Note that as we proceed along this list, the original idea of partiality is growing more and more
bleached. Thus, subsets of definite supersets may be considered as their parts only at a certain level of
abstraction. For the other two cases – ‘an amphora of wine’ and ‘to eat some bread’ – the term “partitive“
is in fact fairly misleading. More specifically, such constructions do not refer to a “part“ in any
reasonable sense, since there is no well-defined “whole“ to which it could relate.
In this paper we will mainly be interested in constructions expressing the second and the third
meanings. These two meanings and the corresponding constructions are, interrelated in various
interesting ways, including grammaticalization, as will be shown in this paper. Such considerations
motivate the relatively recent term “pseudo-partitive“, launched by Selkirk (1977).
For Finno-Ugrists, “partitive“ primarily refers to the partitive case which may be distinguished in the
nominal paradigm of Finnish, Estonian and a few other mainly Finnic languages. As we will see, the
partitive case can sometimes function as the Finnic equivalent to the partitive genitive. The “partitive
article“ occurs mainly in studies of French.
Partitive constructions have attracted much attention in modern syntactic and semantic theories
(cf. e.g. the papers in Hoeksema 1996), which also talk about the semantic notion of partitivity (Barwise
and Cooper 1981, de Hoop 1992) and partitive as an Abstract Case in the sense of GB-theory (Belletti
1988).
Now, given this terminological confusion, I will try to define the object of the present study in more
precise terms.

2. Pseudo-partitive vs. partitive nominal constructions (PPCs
vs. PCs)
2.1. General: a tentative definition
Let us repeat here the English examples introduced above which illustrate partitive and pseudopartitive nominal constructions (hence PCs and PPCs respectively):
(1)

(2)

Partitive nominal constructions (PCs)
a. a cup of that good tea
b. a pile of Mary’s books
Pseudo-partitive nominal constructions (PPCs)
a. a cup of tea
b. a pile of books

4
Both partitive and pseudo-partitive nominal constructions are noun phrases consisting of two
nominals, one of which is a quantifier (cup, pile). Although the same quantifiers may appear in both types
of NPs, their role is different:


Partitive nominal constructions involve a presupposed set of items referred to by one of the nominals
(‘that good tea’, ‘Mary’s books’); and the quantifier indicates a subset which is selected from it.



In a pseudo-partitive nominal construction the same word merely quantifies over the kind of entity
(‘tea’, ‘books’) indicated by the other nominal.

The two main parts of PCs and PPCs will be called Measure vs. Substance in the rest of this paper.
The two constructions differ thus primarily with respect to the referentiality and, in particular, the
specificity of their Substance component: in PCs it receives a specific interpretation, while it is nonspecific in PPCs. However, as we will see below, there is no watertight borderline between PPCs and
PCs, and I will sometimes talk about the semantic space of nominal quantification to refer to the
meanings which are covered both by PCs and PPCs.
2.2. The main question of the study
The term ‘a partitive construction’, as it is most often used, actually covers all cases which have to do
with a selection of a subset from a superset, and thus includes also NPs with numerals and other
quantifiers such as (3) and (4):
(3)
(4)

a.
b.
a.
b.

some of that good tea
five of Mary’s books
some tea
five books

What sets ex. (1) and (2) apart from these latter is the nominal nature of the quantifier. Cup and pile
are real nouns, which, at least outside their quantifier usages, share inflectional and syntactic behaviour
with other nouns. In this sense, ex. (1) and (2) resemble other NPs consisting of two nominals, e.g. those
in ex. (5):
(5)

a. a map of England
b. a student of physics

For the purpose of this study, I have deliberately chosen to focus on constructions with nominal
quantifiers. PCs as understood here are also close to constructions expressing (organic) parts of objects
and body-parts, such as a corner of the room, the main difference being the emphasis on quantification in
the former case1. In fact, the examples in (1) and (2) look like typical cases of adnominal attribution: the
Measure nominal has an attribute introduced with the preposition of, which is thus the construction
marker in English PCs and PPCs.
The English situation is by no means universal, as is illustrated by the following Armenian
examples:
(6)

Armenian (Natal'ja Kozinceva p.c.)
a. Partitive
mi
gavath
ayd hamov surtch-ic
one cup:NOM that good
coffee:ABL
‘one cup of that good coffee’
b. Pseudo-partitive
mi
gavath
surtch
one cup:NOM coffee:NOM
‘one cup of coffee’

5
We can see that in Armenian


there is a significant difference between PCs and PPCs: the construction marker in the former is the
ablative case ending on the Substance, whereas the latter involves juxtaposition (there is thus no overt
construction marker);



both constructions differ from other NPs, in which nominal attributes preceding the head are normally
marked with the syncretic genitive/dative case (ex. 7):

(7)



(8)

Armenian: nominal attribution (Natal'ja Kozinceva p.c.)
surtch-i
gavath
coffee-GEN/DAT cup:NOM
‘a coffee cup’
and, finally, the Substance in PCs is marked with the ablative case, which is typically used to mark a
point of departure and various types of sources in expressions with the general meaning of movement
and separation (ex. 8 below).
Armenian: ‘FROM’ (Fairbanks & Stevick 1975: 44)
im
yehphayramerikha-ic
e gal-is
my brother-DEF America-ABL is come-PART.IMPF
‘My brother comes from America’
These observations lead us to the main question of the present study:



What means are used to express the relationship between the Measure and the Substance in PCs and
PPCs across languages in general and in the Circum-Baltic languages in particular?

2.3. Nominal quantifiers vs nouns, numerals and numeral classifiers
The expected cross-linguistic variation in the internal structure of PPCs and PCs stems from the
nature of nominal quantifiers, or rather, from the conflict between their origin and functions:

 they are nouns, 
but

 they are used in functions which are fairly atypical for nouns
Thus, typically, nouns are used for reference, or, in slightly different terms, for introducing and
manipulating participants in a discourse (cf. e.g. Croft 1991, Hopper & Thompson 1984). The primary
function of nominal quantifiers, on the other hand, is to create a unit of measure which may further be
counted (Croft 1994:162, Bisang forthc.). This intermediate character of nominal quantifiers accounts
for their double similarity, both with typical nouns and with typical quantifiers, e.g. numerals 2. The
semantic connection with numerals is especially evident with count nouns, which can be quantified both
by the process of counting (ten boys, eighty stones) and by the process of measuring (a group of boys, a
cartload of stones). “Pair“ is a good example of a word which is used both for counting and measuring:
on the one hand, a pair of shoes is a new unit of measure as compared to shoes; on the other hand, pair
has a precise numerical value – a pair of shoes consists of exactly two shoes. In the next sections we will
see numerous examples of nominal quantifiers which show various degrees of association with or
alienation from nouns and numerals in their inflectional and syntactic behaviour. Somewhat
schematically we can represent this double-sided nature of nominal quantifiers in the following way:
originate as

function as

6
typical nouns<-------------

nominal quantifiers ------------------>

typical quantifiers

It has been repeatedly stated that there is a semantic distinction between nominal quantifiers /
measures and numeral classifiers in such languages as Vietnamese, Chinese and Japanese. Measures
“create“ units to be counted for those entities that either do not come in “natural units“ (like mass nouns),
or come in “different units“ (cf. six bunches of carrots, two rows of trees and three fronds of a palm).
Classifiers, on the other hand, actualize the semantic boundaries of a given count noun by designating its
“natural unit“, e.g. in Hmong ib tug neeg ‘one classifier person’ or ib rab riam ‘one classifier knife’
(Bisang forthc.:6 - 8, Croft 1994:162 - 163). In practice there is no sharp border between the two. And
although European languages do not normally resort to numeral classifiers, comparable examples do
occur in them. Thus, in Finnish the word kappale ‘piece’ may be used in counting individuals, as in kaksi
kappaletta poikia ‘two (pieces of) boys:PRTV.PL’ (Alho 1992: 7), and the same concerns the word
styck(e) ‘piece’ in Swedish; such examples, however, are very rare, in contrast to the case of genuine
numeral classifiers which are more or less obligatory in numeral constructions. Pair is again a good
example of a measure word which has certain classifier-like uses in combination with nouns of the
pluralia tantum kind: thus, even though scissors and trousers do come in natural units and are countable,
the corresponding nouns have to be accompanied with pair when occurring in numeral constructions
(three pairs of trousers).
Semantically, the class of measure nominals/quantifier nouns is quite heterogeneous, and different
classifications have been suggested in linguistic works (e.g. Eschenbach 1993). The following nonexhaustive list presents the major semantic subtypes of measure nouns:








Conventionalized measures: a litre of milk, a kilo of apples
Abstract quantity nouns: a large amount of apples
Containers: a cup of tea, a pail of apples
Fractions / parts: a slice of bread, a quarter of an hour, a large section of students
Quantums (for mass nouns): a lump of sugar, a drop of milk
Collections (for count nouns): a group of students, a herd of sheep
Forms (both for mass and count nouns): a pile of sand / bricks, a bouquet of roses

We would thus expect that, even within one and the same language, there may be a certain degree of
variation in the structure of (P)PCs matching these semantic differences. (Pseudo-)partitive meanings can
also be occasionally attributed to some expressions which normally lack this interpretation. Thus,
correspondences to both a coffee cup and a cup with coffee in language after language have been reported
to be sometimes used in the meaning of ‘a cup of coffee’. In this paper, however, we will restrict
ourselves to major, standard patterns of (P)PCs across languages.
There are also very interesting connections between constructions with nominal quantifiers and
quantifying constructions with words like “many“, as well as between (P)PCs and verbal total vs. partial
objects. These, however, will be more or less left out in the present paper.

3. Finnic
3.0. General
In Finnish and Estonian the semantic space of nominal quantification is divided between two main
constructions, which differ in the case-marking of the Substance: the partitive vs elative case. The
synchronic division of labour between the two in Finnish is analyzed in Section 3.1.; this analysis
underlies the hypothesis about the grammaticalization path from separative /ablative-like constructions to
PCs to PPCs suggested in Section 3.2. PPCs in Finnish and, especially, in Estonian also share a number
of peculiarities with numeral constructions, which, in turn, invites further generalization about possible
grammaticalization connections between the two, as suggested in Section 3.3.
3.1.

Finnish: constructions with nominal quantifiers

7
In Finnish, PPCs and PCs involve partitive case marking on the Substance (ex. (9a-b)). In PCs the
Substance can sometimes be marked with the elative case (ex. (9c)). In both instances, the Substance
normally follows the Measure (but cf. below, ex. 12):
(9)

Finnish (Päivi Juvonen p.c.)
a. Osta
säkki
perunoita!
buy:IMP.2SG sack:NOM potato:PRTV.PL
‘Buy a sack of potatoes!’
b. Anna minulle
pala
tätä
give me:ALL
bit:NOM this:PRTV

hyvää
kakkua
good:PRTV cake:PRTV /

/

litra
tuoretta
maitoa-si.
litre:NOMfresh:PRTV milk:PRTV-2SG.POSS
‘Give me a bit of this good cake’ or ‘a bit of this good kind of cake /
a litre of your fresh milk.’
c. Anna minulle
pala
tästä
hyvästä
give me:ALL
bit:NOM this:ELATgood:ELAT
kakusta / litra
tuoreesta
maidosta-si
cake:ELAT /litre:NOM fresh:ELAT milk:ELAT-2SG.POSS
‘Give me a bit of this good cake / a litre of your fresh milk.’
Such NPs differ considerably from typical instances of nominal attribution where attributes precede
their head and are marked in the genitive case, e.g. poja-n tuoli - ‘boy-GEN chair’ (‘a/the boy’s chair’).
Dependents to non-quantifier nouns are never marked with the partitive case and only occasionally with
the elative case (cf. Christen this volume). On the other hand, as we will see in section 3.2., constructions
with nominal quantifiers show significant similarities with numeral constructions.
Even in the presence of definite determiners, partitive-marked NPs do not necessarily refer to a
specific set or a specific quantity, but often receive the ‘kind’-interpretation (Alho 1992 and p.c.). Ex.
(9b) above can be interpreted both as ‘a bit of this good cake’ and as ‘a bit of this kind of cake’ and is
thus ambiguous between a PPC and a PC. The next example, however, allows only the ‘set’-interpretation
(and is, thus, clearly a PC due to the context from which it comes (Päivi Juvonen p.c.):
(10)

sit
se
osti
ison
kasan
then (s)he bought:3SG big:GEN.SG armful:GEN.SG
niitä
ilmapalloja
this:PRTV.PL balloon:PRTV.PL
‘Then he bought a big armful of these balloons.’ (Päivi Juvonen p.c.)

Elative complements to quantifier nominals allow only the ‘set’-interpretation: thus, ex. (9c) with
‘your fresh milk’ is most appropriate when there is a definite quantity of fresh milk as opposed to, say,
yesterday’s milk (Päivi Juvonen p.c.). Leino (1993) suggests that whenever there is a choice between a
partitive-marked and an elative-marked complement to a quantifier, the former seems to be the neutral
choice for simply mentioning a quantity of the substance, while the latter somehow involves a part-of
operation, treating the substance as a predefined entity in some way. Fig. 1 shows the combinability of
various quantifier nominals with elative-marked and/or partitive-marked complements in Finnish. Of
these words, the word osa ‘part’ is particularly often used, both with a concrete meaning and with a more
general quantifying meaning (‘some’) and shows the greatest degree of alternation between elativemarked and partitive-marked complements:
(11)

Osa
part:NOM.SG

poik-ia/
boy-PRTV.PL

poj-ista
meni koti-in
/ boy-ELAT.PLwent:3SG home:ILL

‘Some of the boys went home.’ (Johanna Laakso p.c.)
<ADD FIG. 1 SOMEWHERE HERE>

8

In other words,


elatives combine most naturally with words referring to parts of a whole (‘half’, ‘a third’, ‘part’,
‘slice’, ‘bit’).
Apart from combinatorial restrictions, there are other restrictions on the use of elatives in PCs which
have to do with the overall meaning of the sentence, i.e.


the most natural contexts for elative-marked complements to nominal quantifiers are those providing
a fairly concrete interpretation: ‘a part comes /is taken away or separated from the whole in a way
which has a clear impact on the whole’.

Thus, while ex. (9c) is quite acceptable, the elative counterpart to ex. (11) would be doubtful due to
the meaning of the verb ‘to cost’ which does not readily convey any idea of separation:
(11)

Kuinka
how

paljon pala
tätä hyvää kakkua /
much
bit:NOM this:PRTV good:PRTV cake:PRTV

/

litra
tuoretta
maitoa-si
maksaa?
litre:NOMfresh:PRTV milk:PRTV-2SG.POSS cost:3SG
‘How much does a bit of this cake /a litre of your fresh milk cost?’
These restrictions follow from the general meaning of the elative case and its relatively low degree of
grammaticalization in PCs, which will be discussed in section 3.2.
The division of labour between the two constructions is schematically represented below.

<ADD FIG. 2 SOMEWHERE HERE>

An interesting property of PPCs and PCs in Finnish is the ease with which their constituents are split
off and permuted, as in ex. (12). Contrasted with the otherwise rigid order of constituents in Finnish
NPs, this surprising looseness suggests that parts of PPCs and PCs have originated as separate single
NPs, each having its own role in the clause, rather than together constituting one NP. In the next section
we will elaborate on this idea.
(12)

a. Heillä
oli
kokonainen
pullo
They:ADESS be:PRET.3SGwhole:NOM bottle:NOM
ranskalaista viiniä.
French:PRTVwine:PRTV
b. Heillä oli ranskalaista viiniä kokonainen pullo.
c. Ranskalaista viiniä heillä oli kokonainen pullo.
d. Kokonainen pullo heillä oli ranskalaista viiniä.
‘They had a whole bottle of French wine.’ (Seppänen 1983:165)

3.2. Diachronic developments and synchronic results: MOTION FROM -> (P)PC
Let us now discuss the facts presented above from the broader synchronic and diachronic prospective,
starting with the general properties of the elative and the partitive cases.
The elative case in Finnish is primarily used in constructions referring to ‘MOTION FROM INSIDE’. As
such, it is a member of the INNER LOCAL case series which, in Modern Finnish, is opposed to the OUTER
LOCAL case series. Each series involves three cases which express STATIC POSITION, MOTION TO and
MOTION FROM (see fig.3 below). The usage of the elative case has been extended from purely
local/directional/separative to other contexts, including PCs.
The partitive case is one of the four central grammatical cases in the Finnic languages. One of its
main functions is to mark partial objects (and subjects) to verbs, finite and non-finite, where partiality is

9
determined by an interplay of at least three conditions: aspect, scope of negation and quantity.
Historically, however, the partitive is, together with essive and lative, a member of the most archaic local
case series of Finno-Ugric and was originally used to refer to SEPARATION AND MOTION FROM (see fig. 3
below). In modern Finnish, the local usages of these cases are restricted to certain postpositions and
adverbs, e.g. partitive: ulko-a ‘from outside’, kauka-a ‘from far away'; essive: ulko-na ‘outside’, kaukana ‘far away'; lative or translative3: ulo-s ‘out’, kauka-ksi ‘(towards) further away’. Otherwise this case
series has lost its original local function and is mainly used in more abstract senses, e.g. in nominal
predication in the case of the essive and translative (hän on oppetaja-na ‘She is a teacher’, mehu tuli
vahva-ksi ‘The juice became strong’), cf. Hakulinen 1957:63-65, Denison 1957:21-22, see also Stassen
this volume.
<ADD FIG. 3 SOMEWHERE HERE>
The partitive case has become particularly grammaticalized in Finnic languages whereas its Mordvin
counterpart (called ablative: -do, -”e) can still be used in a few ablative-like functions (Larsson 1983:
121). An important factor in the development of the Finnic partitive case into a grammatical case has
probably been its identification and association with the genitive case of the neighbouring IndoEuropean, especially Baltic languages, with which it shared some functions (Larsson ibid.: 139 - 147,
Koptjevskaja-Tamm & Wälchli this volume).
Thus, the two markers of PCs and PPCs in Finnish have actually very similar sources – some variant
of a separative case. As we will see in Section 6.2., separative constructions constitute the most popular
source for PCs across languages, whereas PPCs based on separative constructions occur less frequently.
This suggests that separative markers enter the semantic domain of nominal quantification via PCs, to
start with, and then in certain cases may gradually get extended into PPCs. The data, both Finnic and
cross-linguistic, suggest the following path of development / grammaticalization for such constructions:
1. Starting point
• Clauses referring to concrete separation of a part from the whole: predicates of separation
(such as ‘take away’, ‘cut’, ‘remove’ etc.) combine with two dependents, one referring to a part
and the other referring to the whole from which it comes.


‘Hecut[asli]fromhk’



2. Grammaticalization of PCs


Extension of “part“-expressions from concrete parts to a larger class of Measure expressions.



Reanalysis of the original construction – the “part“ and the “whole“ are being reanalyzed as
making up one constituent instead of being two different dependents to the same predicate; the
resulting constituent can, however, still retain a certain degree of “looseness“ and allow
splitting.

 ‘Hecut[asliofhk]’


Extension of predicates from those referring to concrete separation to others.
With these changes accomplished, the former separative construction has been
grammaticalized into a well-behaved PC, which in fact has not yet happened to the elative
case in Finnish.
3. Grammaticalization of PPCs



PCs can further develop into PPCs. The “bridge“ between the two is accomplished by cases
like ex. (9b), which are ambiguous between the ‘set’ and the ‘kind’ interpretations. Similarly,

10
“a glass of this wine“, as in “I would like a glass of this wine“, easily receives different
interpretations in different situations: uttered at a little party, it is most probably a PC – “this
wine“ refers to a definite quantity, e.g. to the wine in a bottle on the table; uttered at a
restaurant, it is rather a PPC – “this wine“ stands for “this kind of wine“, as opposed to the
others listed in a menu.
This grammaticalization path is represented in fig. 4.
<ADD FIG. 4 SOMEWHERE HERE>
The history of PCs and PPCs in Finnish also show two other processes which are often found in
grammaticalization:


at some stage during the grammaticalization process, the original separative meaning can gradually
get lost.



recycling: a new marker with the separative meaning can start expanding to partitive-like uses

Each of the stages in the development of Finnic PCs and PPCs finds parallels in a number of other
languages and the whole scenario, which accounts for the development of separative case-marker into the
marker of PCs and PPCs, looks like a plausible grammaticalization process, entirely motivated by
language-internal factors. However, it is highly probable that external factors, i.e. contacts with the IndoEuropean, primarily Baltic languages, have played a major role in this development. Thiswill be
discussed in Section 7.
3.3. Nominal vs. numeral quantifiers: Finnish and Estonian
Interestingly, constructions with nominal quantifiers in Finnish find close parallels in constructions
with numerals, including the word par i ‘couple’ with a similar distinction between indefinite
quantification (‘kind’) and selecting a subset from a larger definite set, cf. ex. (13) below:
(13)

a. Partitive complements
kaksi saksalaista
poikaa
two German:PRTV.SG boy:PRTV.SG
‘two German boys’ (Seppänen 1983:167)
b. Elative complements:
kaksi
hänen
veljistään
two:NOM (s)he:GEN brother:ELAT.PL:3POSS
‘two of his/her brothers' (ibid.:162)

Now, ex. (13b) is not surprising: it often happens that numerals and nominal quantifiers build their
PCs in the same way, and this is also true for Finnish PCs with elative-marked complements. Much more
striking is the fact that numerals in “normal“, non-PCs combine with partitive marked complements in a
way reminiscent of PPCs4 (cf., however, section 4.3. on the Slavic situation). However, there are also
important differences between the two types of constructions (cf. Seppänen 1983):


complements to numerals are always marked as singular5, whereas the number of the Substance in
nominal PPCs varies in accordance with its countability etc: thus, ‘German boys’ after the numeral
‘two’ in (13a) are in the singular, whereas ‘potatoes’ in (9a) and ‘boys’ in (11) are in the plural.

However, this difference exists only in the unmarked situation, i.e., when numerals immediately
precede their complements and both, thus, occur in their usual places. Numeral constructions can be
moved around and split off, just like PPCs (cf. ex. (12b-d)), but in such situations the number marking of
their complements has to be changed from the singular to the plural. Thus, ex. (14a) shows the normal,
unmarked numeral construction; the word ‘child’ follows the numeral ‘three’ and is in the singular

11
(lasta), while the plural is ungrammatical. In ex. (14b-d) the two words have moved around and now only
the plural form, lapsia, is allowed.
(14) Finnish (Seppänen 1983:165 - 169)
a. Heillä
on
kolme
lasta /
They:ADESS be:PRS.3 three:NOM child:PRTV.SG /
*lapsia.
*child:PRTV.PL
b. Heillä on *lasta / lapsia kolme.
c. *Lasta / Lapsia heillä on kolme.
d. Kolme heillä on *lasta / lapsia.
‘They have three children.’
In other words, combinations of numerals and their complements look much more similar to
combinations of nominal quantifiers and their complements when they do not occur in their most usual
places.
Another difference between constructions with numerals and those with nominal quantifiers has to do
with the partitive case-marking of the complement:


whereas the Substance in nominal PPCs is always marked with the partitive case, nominals in
numeral constructions appear in the partitive case only when the numeral itself is in the nominative or
accusative, otherwise the nominal and the numeral agree in case, cf. ex. (15) with (13a) :

(15)

Hän kirjoittaa usein kahdelle saksakaiselle
He:NOM writes often two:ALL.SG German:ALL.SG
pojalle
boy:ALL.SG
‘He often writes to two German boys.’ (Seppänen 1983:167)

In other words, the internal syntax of numeral constructions is sensitive to the syntactic function of
the whole NP and alternates between government (numerals in the nominative and accusative cases
govern nominal complements in the partitive case) and agreement6. The marking of the Substance in
nominal PPCs is not sensitive to the marking of the Measure. Now, since most numeral constructions
appear as subjects and objects of clauses, this distinction between constructions with numerals and
nominal quantifiers gets neutralized in the unmarked situation.
Thus, Finnish numeral constructions with the word order “numeral - complement“ constitute a
construction type of their own which is manifested in the following properties:


a highly idiosyncratic internal syntax, and



“fixed“ number marking of complements, which is not determined by any semantic considerations,
but follows unambiguously from their syntactic status.

Both these properties can readily be interpreted as characteristic of highly grammaticalized
constructions, i.e. constructions which have reached a very advanced stage of grammaticalization.
Permutations and splitting, as in (14), result in less grammaticalized combinations which fall prey to
semantic rather than purely grammatical considerations and betray their kinship with nominal PCs and
PPCs in the choice of plural marking on the nominal. In other words, all these observations support the
view that


markers of (P)PCs can spread further to numeral constructions, although the exact details of this
process remain unclear: either PPCs themselves give rise to numeral constructions, or both develop
from PCs. This is represented graphically in fig. 5. Again, as in Section 3.2, we do not take into
account the high probability of foreign influence on the development of numeral constructions (see
Section 7 for discussion).

12


<ADD FIG. 5 SOMEWHERE HERE>

Now, in Estonian, numerals behave just like their Finnish counterparts. However, part of what I said
about numeral quantifiers also holds for PPCs: the marking of the Substance alternates between the
partitive case and case agreement with the Measure in the way that is characteristic of numeral
constructions. Thus, in Estonian, nominal quantifiers share highly peculiar syntactic behaviour with
typical quantifiers – numerals.
(16)

Estonian (Erelt 1993: 144-145)
a. [Kott
kartuleid]
hakkas
[sack:NOM.SG potato:PRTV.PL]begin:PRET.3SG

otsa
end:GEN.SG

saama
get:mINF
‘The sack of potatoes is coming to an end.’
b. Kui palju sa
[koti
kartulite]
how much you:NOM [sack:GEN.SG potato:GEN.PL]
eest maksid?
for
pay:PRET.2SG
‘How much did you pay for the sack of potatoes?’
Thus, the Estonian facts show that nominal quantifiers can themselves acquire the morpho-syntactic
properties of numerals, i.e. that numeral constructions in one or another way can contribute to the
development of PPCs. The different stages in the development of PPCs in Estonian are seen in the
following diagram:




<ADD FIG. 6 SOMEWHERE HERE>

Table 1 summarizes the data on the internal syntax of PPCs in Finnish and Estonian as compared to
that of typical noun phrases (such as ‘Peter’s hat’ or ‘the roof of a house’) and that of numeral
constructions.
<ADD TABLE 1 SOMEWHERE HERE>
Thus, in Finnish, and even more so in Estonian, the syntax of nominal quantifiers distinguishes them
sharply from “normal“ nouns and can be interpreted as a symptom of their gradual alienation from the
class of nouns and association with typical quantifiers.
In the rest of this paper we will see more examples of similar processes.

4. Balto-Slavic circum-Baltic languages
The Baltic and most of the Slavic languages (apart from Bulgarian and Macedonian) have retained
archaic constructions with nominal quantifiers and use the genitive case to mark Substance nominals both
in PPCs and PCs, as illustrated in ex. (17):

13
(17)

a. Russian
stakan
sok-a /
glass:NOM
juice-GEN/
stakan
von to-go
sok-a
glass:NOM there that-GEN.SG.M juice-GEN
‘a glass of juice / a glass of that juice’
b. Latvian
glāze
tējas
/
glass:NOM tea:GEN
/
glāze
šīs
garšīgās
tējas
glass:NOM that:GEN good:GEN tea:GEN
‘a glass of tea / a glass of that good tea'

The genitive case in these uses is, thus, the Slavo-Baltic counterpart to the Finnic partitive and elative
cases. On the one hand, the connection between the Balto-Slavic genitive and the Finnic partitive does
not come as a surprise: both share a number of other functions. Thus, both Balto-Slavic genitives and
Finnic partitives mark partial objects and sometimes subjects (these uses are much more productive in
the Baltic languages than in Slavic) and also play a prominent role in the syntax of numeral
constructions. On the other hand, these numerous similarities between the Balto-Slavic genitives and
Finnic partitives betray language contacts as their plausible cause: some of the functions shared by the
two cases are typologically too infrequent to be explained by a coincident parallel development (see
Section 7). However, there is also an important difference between the two cases: the genitive, as
opposed to the partitive, is the normal case for attributive nominals. It should also be noted that
genitive-marked substance nominals have a broad distribution in other quantifying constructions, e.g.
after quantifiers such as ‘many’, ‘some’, ‘little’ etc.
Let us look at Slavic and Baltic separately (some more details can be found in Koptjevskaja-Tamm &
Wälchli this volume, Section 8.3.).
All Slavic languages have different morpho-syntactic rules for lower numerals (1-4) and higher
numerals (see Corbett 1978a, b, 1983 for an extensive discussion of the syntax of numeral constructions
in Slavic). The two extremes – the numeral ‘one’ and the highest numerals, like ‘thousand’ and ‘million’ –
show cross-linguistically well-attested patterns: ‘one’ behaves more or less like an adjective, agreeing in
gender and case with the accompanying number, whereas the highest numerals show fairly nominal
behaviour, governing the accompanying nominal in the genitive case. The rules for 2-4 spring from the
extension of nominal dual forms used after ‘two’ to other contexts and the later reanalysis of the dual; the
exact rules differ considerably across Slavic languages.
For our purpose, the behaviour of higher numerals (starting with 5) is particularly interesting in most
of the Slavic languages (with the exception of South-Slavic). Together with their Finnic counterparts they
share the typologically unique pattern of alternating between case-governing the accompanying nominal
and triggering its case agreement under exactly the same conditions (direct case vs. oblique case of the
whole NP). When governed, nominals in Slavic appear in the genitive case. However, their number
assignment is normally the genitive plural, in contradistinction to Finnic, where the singular is found
throughout7. Russian is alone in having genitive singular government for the numerals 2-4, but this is the
result of a reanalysis of dual and nominative plural forms. Thus, as the case was with Finnic, numeral
constructions in Slavic show an idiosyncratic behaviour typical of highly grammaticalized constructions.
Again it is reasonable to suggest the diachronic link between PPCs and numeral constructions: the
morpho-syntax of the latter is modelled, at least partly, on that of the former. Genitives normally follow
their heads, and thus (P)PCs look both like typical combinations of two nominals and like many numeral
constructions, cf. ex. (18) below.

14
(18)

Russian
a. kilogramm
jablok
kilogramm:NOM
apple:GEN.PL
‘a kilogram of apples’
b. pjat'
jablok
five:NOM apple:GEN.PL
‘five apples’
c. komnata
moix
dočerej
room:NOM my:GEN.PL daughter:GEN.PL
‘my daughters’ room’

Russian has a few masculine one-syllable mass nouns which may optionally use a form (sometimes
called “partitive“) different from their “normal“ genitive form in (P)PCs and as partial objects, e.g. vypit'
sok-u/-a ‘to drink juice-GEN/-PRTV’ and stakan sok-u/-a ‘a glass of juice-GEN/PRTV’ vs. cena sok-*u/a ‘the price of juice-*PRTV/-GEN’. These words behave, thus, somewhat similarly to their Finnic
counterparts in distinguishing between the adnominal genitive case and the case used to express partiality
/ indefiniteness. In Modern Russian, however, the partitive forms are used only sporadically (Paus 1994).
In Baltic, as in Slavic, higher and lower numerals follow different morpho-syntactic rules 8. Lower
numerals (‘1-9’) behave like adjectives in agreeing with quantified nominals in gender, case and number.
In Lithuanian, higher numerals (teens ‘11-19’ and tens, ‘10’, ‘20’ etc.) behave basically like nouns – they
inflect for case and always govern plural nominals in the genitive case. In Latvian, teens and tens are
themselves indeclinable, but the quantified nominal is either governed (in the genitive plural) or appears
in the case assigned to the function of the whole numeral construction. This latter option seems to be
gaining more and more ground in Latvian at the cost of government. Thus, on the whole Baltic numerals
govern nominals to a much lesser degree than in Slavic. However, numeral constructions share another
important characteristics with (P)PCs – the word order “quantifier - quantified“ (ex. (19a-b)). The word
order “ Head- Gen“ is otherwise highly marked among NPs, inwhich genitives normally precede their
heads (ex. (19c), cf. Christen this volume). Lithuanian allows the same order even with nominal
quantifiers, as a marked alternative. However, these word order peculiarities show that nominals as
quantifiers alienate themselves from nominals in other functions and associate themselves with
numerals and other quantifiers.
(19)Lithuanian (Ambrazas (ed). 1997: 587, 703)
a. pieno
stiklinė
milk:GEN glass:NOM
‘a glass for milk / a glass (full) of milk’
b. vienuolikavaikų
eleven
child:GEN.PL
‘eleven children’
c. stiklinė
pieno
glass:NOM
milk:GEN
‘a glass (full) of milk’
Thus, in Balto-Slavic we see the same tendencies for “cross-pollination“ between PPCs and numeral
constructions as in Finnic, even though we cannot always account for the actual historical processes
behind these observable similarities.
How did genitives come to be involved in (P)PCs? For Balto-Slavic we could hypothesize that these
uses of the genitive case were inherited from the Proto-Indo-European ablative case which, in these
languages, had merged with the older genitive case: according to this hypothesis, such constructions
could have followed the same grammaticalization path as Finnish and Estonian (P)PCs (see fig. 1). This,
however, cannot be the whole truth, since (P)PCs with genitive-marked Substance are attested at least in
some of the Indo-European languages in which the ablative differed from the genitive (even though the
ablative, on the whole, was weakly differentiated across Indo-European, cf. Brugmann & Delbrück 1909:
495):

15
(20)

a. Sanskrit
gōnam
ardhám
cow:GEN.PL half:ACC
‘half of the cows’
b. Latin
amphora
vini
amphora
wine:GEN
‘an amphora of wine’ (Brugmann & Delbrück 1909: 598)

Thus, separative / ‘from'-constructions are not the only source from which PCs and PPCs can develop.
We will return to this problem in Section 6.
In Latvian, nouns which denote parts may take dependents marked with the preposition no, ‘from,
off’ (cf. with what was said about most contexts for the elative-marked Substance nominals in Finnish in
Section 3.1):
(21)

Latvian
ikviens
varēja redzēt [laukumiņu no
[matroža
everyone could
see
small.patch:ACC of sailor:GEN
Jura Varapogas
krūtīm]]
un
uz
tām
Jura Varapoga:GEN chest:DAT.PL and on
they:F.DAT
[daļu
no
[tetovēta
pūķa
galvas
vai
part:ACC of
tattooed:GEN dragon:GEN head:GENor
astes]].
tail:GEN
‘...everyone could see the small patch of the sailor Jura Varapoga’s chest and on it a part of a
tattooed head or tail of a dragon’ (žIV: 246)

It seems that no sometimes marks the Substance even to other Measure nouns, e.g. viens piliens no mana
alus ‘one drop of my beer’, but such examples are considered substandard and are attributed to German
influence (see section 5.2 on von-PCs in German)9.
Figure 7 and table 2 summarize the discussion of (P)PCs for Russian and Latvian.
<ADD FIG. 7 SOMEWHERE HERE>
<ADD TABLE 2 SOMEWHERE HERE>

5. Germanic circum-Baltic languages
5.0. General
Nominal PCs and PPCs with genitive-marked Substance nominals are well attested in older Germanic
languages, which also used genitives in some other quantifying constructions, e.g. those with higher
numerals and words such as ‘many’. Modern Germanic languages have developed new types of PCs and
PPCs, which have replaced these archaic constructions, either completely, as in Continental Scandinavian
(Section 5.1.), or partially, as in German (Section 5.2.). Section 5.3. suggests possible motivations for this
development.

5.1. Scandinavian circum-Baltic languages
Constructions with genitive-marked Substance nominals, such as those discussed for Balto-Slavic,
also existed in the old Scandinavian languages:

16
(22)

Old Swedish
a. halfw-œn
span
korn-s
half-ACC.SG bushel barley-GEN
‘half a bushel of barley’
(Schwartz 1878: 125, from Upplandslagen Kk. 6:7, 1296)
b. attundœ löter
attung-z
eighth part:NOM.SG attung-GEN
‘an eighth part of an attung (division of the country)’
(Schwartz 1878:129, from Västgötalagen I.J.7:3, 1220-ies)

Such constructions are completely absent from Modern Swedish and Danish. Archaic PPCs with
genitivus generis, such as in ex. (22a) from the end of the 13th century, were replaced in Old Swedish and
Danish at an early stage by constructions involving juxtaposition of the Measure and the Substance or,
more rarely, introducing the Substance nominals with the preposition med/mäþ, ‘with’ in PPCs: ex. (23)
contains instances of all three construction types.
(23)

Old Swedish (Wessén 1970:111, from Östgötalagen Kr. 2: pr., end of the 13th century)
[þrea spän
hueti-s]
ok
[en þyn
rugh] ok
[three bushel wheat-GEN.SG] and [one barrel
rye]
and
[en þyn biug]...
ok
[fiura þyn-i
hästakorn]
ok
þär
[one barrel barley]
and [four barrel-PL horse.barley] and there
mäþ [tu
las
foþär] ok
[þry pund
mäþ smör]
with [two load forage] and [three pound with butter]
ok
[fiughur pund
mäþ fläsk]... ok
[et
halft pund
and [four
pound with pork]... and [one half pound
mäþ uax]
with wax]
‘Three bushels of wheat and one barrel of rye and one barrel of barley... and four barrels of horse
barley and therewith two loads of forage and three pounds of butter and four pounds of pork... and
one half pound of wax’

In PCs, the genitive sometimes alternated with the preposition av / af (which in its turn governed the
dative case on the accompanying nominal), as in ex. (24a) (cf. with (22b)). The details of the use of avcomplements are not quite clear from my sources, but it seems that in Old Swedish they could occur in
PPCs as well (Muriel Norde p.c. and 1997:214), cf. ex. (24b):
(24)

a. attundæ lot af attung-i
eighth part of attung-DAT.SG
‘an/the eighth part of an/the attung (division of the country)’
(Schwartz 1878:130, from Västgötalagen I.J.14:pr, 1220-ies)
b. en
alin af goþo
kirsko
one ell
of good:DAT.SG woollen.cloth:DAT.SG
‘one ell of good woollen cloth’
(Norde 1997: 214, ex. (14c) from Kopparbergsprivilegierna, 1347).

These usages are present in Modern Icelandic, in which the most frequent types of PCs and PPCs both
involve af-marked Measure nominals:

17
(25)

Icelandic (Delsing 1993:201)
a. eitt
kiló
a/one:N.NOM
kilo:NOM.SG
‘a/one kilo (of) butter’
b. eitt
kiló
a/one:N.NOM
kilo:NOM.SG
‘a/one kilo of this butter’

af smjöri
of butter:DAT.SG
af þessu
smjöri
of this:N.SG.DAT butter:DAT.SG

In older Scandinavian languages, the preposition av/af was originally used with ablative functions, to
refer to motion from / separation from, which gradually gave rise to other usages. In Modern
Scandinavian this concrete directional meaning hardly exists.
Modern Swedish and Danish make a sharp distinction between PCs and PPCs: PPCs are formed by
juxtaposing the Measure and the Substance, whereas, in PCs, the Substance is introduced by the
preposition av/af ‘of’10. A few Measure nominals referring to collections, such as grupp ‘group’,
mängder ‘lots’ can take av/af-complements even in PPCs, but the exact details differ for the two
languages.
(26)Swedish
a. ett
glas vin
a:N glass wine
‘a glass of wine’
b. ett
glas av det
god-a
vin-et
a:N glass of the:N.SG good-DEFwine-DEF.N.SG
‘a glass of the good wine’
There is also an alternative strategy for PPCs whereby the Measure noun is marked with the
preposition med ‘with’. Here it will suffice to mention that


not all Measure nouns allow this strategy. Thus, for Swedish, Delsing (1993:204) suggests that
Measure words fall into two categories:, genuine quantifiers (including conventionalized measures
and abstract quantity nouns) and pseudoquantifiers (e.g. container nominals, collections and forms),
of which only the latter can take med-phrases; cf. *ett antal /flertal med människor – ‘a
number/majority of people’, ??ett dussin/tjog med ägg – ‘a dozen eggs / a score of eggs’, ??en
liter/ett kilo med jordgubbar – ‘a liter/a kilo of strawberries’ vs. en grupp/hop med ungdomar – ‘a
group/crowd of youngsters’, en bukett/ett fång med blommor – ‘a bouquet/an armful of flowers’, en
låda/flaska med vin’ – ‘a case/bottle of wine'11.



while the two strategies are often in free variation, the Juxtapositional strategy is not allowed under
certain conditions. Thus, in Danish, nominal quantifiers with the suffixed definite article require medmarked Substance nominals, e.g. en spand (med) koldt vand ‘a pail (with) cold water’ vs. spanden
*(med) koldt vand ‘pail-the *(with) cold water’ (Heltoft 1996:23).
The discussion of (P)PCs in Swedish and Danish is summarized in figure 8 and table 3.
<ADD FIG. 8 SOMEWHERE HERE>
<ADD TABLE 3 SOMEWHERE HERE>

5.2. German
Compared to the languages presented so far, Modern German demonstrates a striking diversity of PC
and PPC types: archaic constructions with the genitive-marked Substance nominals and two newer types,
involving juxtaposition of the Measure and the Substance nominals, and marking the Substance nominal
with the preposition von.
In Modern German, archaic constructions – both PCs and PPCs – with genitive-marked Substance
nominals are used to a very limited degree and clearly belong to an elevated written style. Thus, in PPCs,

18
only Substance nominals with adjectival modifiers can normally be genitive-marked (cf. ex. (27)). This
option is particularly often chosen with Substance nominals in the plural (Hentschel 1993).
(27) German
eine
Flasche guten
Weins /
one:F.NOM bottle
good:M.SG.GEN
wine:GEN /
eine
Flasche des
besten
Weins
one:F.NOM bottle
the:M.GEN
best:M.SG.GEN wine:GEN
‘a bottle of good wine / a bottle of the best wine’
Single nominals with the genitive ending are sometimes possible, but have a clearly archaic and
poetic flavour, e.g. purpurn sind die dicken Tropfen Bluts ‘crimson are the thick drops (of) blood.GEN’
(Eschenbach 1993: 71).
The most frequent type of PPC involves juxtaposition of the Measure and the Substance, which is
reminiscent of the Swedish-Danish situation. However, there is a special problem here which neither
Swedish nor Danish have to solve: how do various syntactic positions which a whole PPC occupies
influence the morphosyntactic properties of its constituents? In other words, which of its constituents will
inflect for case? This problem obviously does not arise in Continental Scandinavian which has lost its
cases, apart from the somewhat problematic genitive. In general, PPCs occur very rarely in positions
others than that of subject and direct object, and are considered fairly marginal by native speakers of
German, who also differ in their intuitions about their acceptability. Here German has two main options.


First, both Measure and Substance can agree in case, as in ex. (28) below.

(28)

German:
a. eine
Flasche gut-er
Wein
one:F.NOM bottle
good-M.SG.NOM wine
b. trotz
ein-er
Flasche gut-en
Wein-s
in.spite.of
one-F.GEN
bottle
good-M.SG.GEN wine-GEN
c. mit
ein-er
Flasche gut-em
Wein(e)
with one-F.DAT
bottle
good-M.SG.DAT
wine(DAT)
‘(in spite of / with) one bottle of good wine’
(Eschenbach 1993: 71)

• Second, German may inflect the Measure part of PPCs which is seen in the form of the article in ex.
(29) below.
(29)

German
trotz/mit
ein-er
Flasche
in.spite.of/with one-F.GEN/DAT
bottle
‘in spite of / with one bottle of wine’
(Eschenbach 1993: 71)

Wein
wine

The choice between these options is partly dependent on the presence or absence of modifying
adjectives to the Substance – their absence favours exclusive inflection of the Measure.
In the most frequent type of PC in Modern German, the Substance nominal is marked with the
preposition von ‘of, from’, as in ex. (30a). Von - phrases occur sometimes in PPCs – when the Substance
is accompanied by an adjective (as in (30b)) and/or when the Measure is an abstract quantity noun or a
collection noun, e.g. eine Reihe von arabischen Staaten ‘a group of Arabic countries’, eine Unzahl von
Flüchtlingen ‘a number of refugees’ (Eschenbach 1993:74). Interestingly, the class of Measure words
allowing von-complements in German PPCs overlaps significantly with those allowing av / af
-complements in comparable Danish and Swedish constructions (cf. Section 5.1).

19
(30)

a. drei Liter von diesemWein
three litre of this:DAT.M wine
‘three litres of this wine’
???b. drei Liter von gutem Wein
three litre of good:DAT.M wine
‘three litres of good wine’

The discussion of German (P)PCs is summarized in figure 9 and table 4.

<ADD FIG. 9 SOMEWHERE HERE>
<ADD TABLE 4 SOMEWHERE HERE>
5.3. (P)PCs in the Germanic circum-Baltic languages: diachrony and synchrony
To summarize, the three modern Germanic languages under consideration have developed two main
types of new constructions as compared to the Balto-Slavic circum-Baltic languages, which came to
replace, to various extents, combinations with genitives:
-

analytic constructions, mainly restricted to PCs, and

-

juxtapositional PPCs.

Analytic constructions come from two sources:



the markers av/af and von, involved in the major type of PPCs and, to a certain degree in PCs have
originated in separative / ablative contexts;
the marker med, involved in the minor type of PPCs and PCs in Swedish and Danish, have originated
in comitative / associative contexts.

The former grammaticalization source is by now well known: Germanic (P)PCs with av/af/von and
Finnish (P)PCs with the partitive/elative are in a way “local“ variations on the same theme, i.e. different
language-specific versions of the same grammaticalization path (see section 3.2.).
The semantic reasons for the use of comitative markers in (P)PCs are also fairly clear in the case
of container nouns. A glass of water is in fact a glass with water in it, and a box of apples is a box with
apples in it; the difference, for English, consists in how much content there is in the container and/or
whether the quantificational aspect is interesting for the speaker. Thus, when the glass or the box is half
empty, the corresponding of-expressions would be definitely out of place. On the other hand, withconstructions do not necessarily imply that there is some space left in the container: if I say Could you
please help me to carry this box with apples in it? or Katti put a glass with water in it on the table and
started painting, both the box and the glass may very well be full, but the exact quantity is not so relevant
for me (it might become so, when Katti starts washing her brush in the glass and the water runs over the
edge of the glass). So the development probably follows the following path:
containers containing some / indefinite quantity -------->
---> containers measuring a certain quantity ----------> other measures
Now, what about the origin of juxtapositional PPCs?
On first sight, the emergence of juxtapositional PPCs appears to be well motivated by the decay and
collapse of the older case systems in the Germanic languages. I will argue, however, that this explanation
(at least on its own) is not feasible. I suggest that the association of PPCs with other quantifying,
primarily numeral constructions has led to no other explicit markers having arisen in PPCs.
Let us look at how the existence of juxtapositional PPCs relates to the collapse of the older case
systems in Germanic. The Mainland Scandinavian languages have witnessed a very peculiar
development from an old morphological genitive case to a phrase marking “genitive“ marker -s. Also, in
the older languages, the position of the genitive differed for different uses – from being free to being
rigidly preposed or postposed to the head, whereas s-marked genitive phrases in the modern languages

20
always precede their heads. Substance nominals always followed the Measure in the older languages and
do so in the modern languages too, which puts them outside of the domain in which s-genitive markers
operate. A straightforward and somewhat simplistic explanation for this development would, thus,
suggest that, while nominals in other functions could replace their old morphological genitive markers
with -s, this option was not available for Substance nominals; the old genitive case has simply been lost,
and as a result PPCs have to manage without any overt marker.
In German, the genitive case manifests itself as a morphological marker – but only for a subset of
nouns (mostly masculine and neuter nouns in the singular). However, case distinctions are effectively
expressed by means of inflecting articles, demonstratives and adjectives which normally accompany
German nouns, cf. der Mann ‘the:M.NOM man:NOM’ vs. des Mannes ‘the:M.GEN man:GEN’ and die
Frau ‘the:F.NOM woman’ vs. der Frau ‘the:F.GEN/DAT woman’. In fact, in Modern German, only
proper names can regularly appear as single genitive-marked nouns, as ein Glas Peters “a glass
Peter:GEN“ (‘Peter’s glass’), whereas all other genitive noun phrases involve determiners and/or
adjectives. Genitive marking of substance nominals in PPCs and PCs seems to fit in into this general
system: single (bare) Substance nominals are more or less avoided and the best examples are constituted
by Substance nominals in the plural accompanied by adjectives. However, even these examples belong to
an elevated and written style (cf. Section 5.2. for discussion).
The exposé above seems, thus, to show that the loss of genitive marking in PPCs (and PCs) in
Scandinavian and, to a lesser degree, in German follows from the more general decay of the
morphological genitive and the subsequent reorganization of the “genitive domain“ in these languages.
However, there are at least two additional considerations that are not covered by this explanation, namely
why no other marker has immediately replaced the older genitive marker in PPCs, and why juxtaposition
in PPCs extended even to those contexts where genitive markers are normally effectively used.
First, the allusion to the breakdown in the case systems provides only a partial explanation for the
emergence of juxtapositional PPCs in Swedish, Danish and German. Thus, it explains why Substance
nominals can never (in the case of Swedish and Danish) or need not (in the case of German) attach the
genitive marker. What is does not explain, however, is why they do not take any other explicit
marker. That is, other NPs with “old“ adnominal genitives, including PCs, have been replaced with NPs
involving prepositional phrases or compounding, e.g. the following example of the older genitivus
materiae, den Schlafrock echt ostindischen Stoffs ‘the gown of genuine east-Indian cloth’ (Behaghel 1923:
520) will correspond to den Schlafrock aus echten ostindischen Stoff in Modern German. Juxtaposition of
two nominals, one of which was previously marked with the genitive, is thus unique for PPCs – a fact
which has to be explained.
Second, the general decay in the nominal case systems cannot explain why words which still inflect
for case, such as adjectives in German, would still appear in the non-marked form in PPCs. Thus, in
German, the frequent coalescence of genitive and nominative forms led to zero-marked PPCs, starting in
the 15th century, that is, in effect, to juxtaposition. At first juxtaposition applied to single Measure nouns
(eine Tonne Holz ‘a ton of wood’), whereas a combination of Measure nouns with adjectives appear in
the genitive (eine Tonne gespaltenen Holz ‘a ton of chopped:GEN wood’). Later, however, the pattern
spread further to combinations of nouns and adjectives resulting in examples like eine Tonne gespaltenes
Holz ‘a ton of chopped:NOM wood’ (Behaghel 1923: 532). Thus, the older construction with the
nominal attribute, the genitive case of which coincided with the nominative case, came to be reanalyzed
as juxtaposition of two nominals, that is as a construction type of its own, specifically employed for PPCs
and different from normal combinations of a nominal head and a nominal dependent.
The question thus arises why the internal structure of PPCs differs from that of other combinations of
two nominals. My suggestion is that the reason for this lies in their semantically intermediate nature:
nominal quantifiers, although (originally) nouns, are used in functions, which are atypical for nouns in
general (cf. section 2.2). And, consequently, it is not surprising that nominal quantifiers alienate
themselves from typical nouns.
A clear manifestation of this tendency is the loss of inflectional distinctions for some of the most
usual nominal quantifiers (number in Swedish and Danish; number and case in German): e.g. in Swedish
två liter/*litrar mjölk – ‘two litre/*litre.PL milk’, tre kilo/*kilon smör – ‘three kilo/*kilo:PL butter’, fyra
meter/*metrar tyg – ‘four metre/*metre.PL cloth’ (cf. Delsing 1993:204) and in German drei Glas
/Gläser Bier – ‘three glass/glass.PL beer’ and nach drei Glas/Gläsern warmem Bier – ‘after three glass
/glass:DAT.PL warm:DAT. beer’ (cf. Plank 1981: 142 - 148).
The fact that nominal quantifiers take juxtaposed complements can be interpreted as a loss of typical
nominal syntactic properties and, thus, as another manifestation of the same tendency.

21
While disassociating themselves from typical nominal-nominal combinations, PPCs at the same time
come closer to other quantificational expressions, e.g. constructions with numerals. In a way, numeral
constructions constitute a natural focus of attraction for PPCs, as was demonstrated for Estonian (cf. ex.
(16)) and for Baltic (cf. ex. (19)). In other words,
• juxtapositional PPCs in Modern Germanic owe their existence partly to the influence from
juxtapositional numeral constructions.
An interesting peculiarity of PPCs in Danish and Swedish is their prosody12: the nominal quantifier
loses its stress, whereas numerals do not, e.g. en skål ´ris ‘a bowl of rice’, en gruppe ´børn ‘a group of
children’ (Da), ´två tunnor ´sill ‘two barrels of herring’ (Sw). The same de-stressing pattern is, in fact,
employed in both languages for a number of various functions, among others, in verb-incorporating
structures like læse ´tegneserier ‘to read cartoons’ (Da) and åka ´tåg ‘go by train’ (Sw.) (cf. Anward &
Linell 1976, Herslund 1995). A detailed account of exactly how prosody works in PPCs requires much
more research13. Here it will suffice to say that this particular pattern with a unitary stress to the right of
the de-stressed unit is generally used when the two formal units correspond to one conceptual unit (e.g.
when a verb and its object denote a special kind of activity) and/or when one (or even both) of the units is
not used in its literal meaning. Combinations of Measure and Substance nominals seem to share both
these properties, though to different degrees for different combinations: the meaning of Measure
nominals in PPCs is often different from that in other contexts, and, even more importantly, the whole
combination refers to a new countable entity (cf. section 2.3.).

6. (P)PCs in the European languages
6.1. The major (P)PC types and their distribution across the European languages
Let us now look at the structure of PCs and PPCs in a broader, pan-European typological prospective.
My database on PPCs covers most of the European languages, whereas the information on PCs is,
unfortunately, much more limited.
PCs across the European languages show the same “dislike“ for juxtaposition as the circum-Baltic
languages, i.e.:


PCs in the European languages tend to be formed with an overt marker associated with the Substance,
where overt markers are either inflectional (case endings) or analytical (prepositions).

PPCs across the European languages show the same two major types as in the circum-Baltic
languages:


PPCs without overt markers (juxtapositional)



PPCs with an overt marker associated with the Substance, where overt markers are either inflectional
(case endings) or analytical (prepositions);
This means that the following logically possible construction types are not attested among PPCs:

• *constructions with an overt marker associated with the Measure;
• *constructions with overt markers associated both with the Measure and with the Substance;
• *constructions with a “linker“ – an overt marker between the Measure and the Substance
Table 5 and map 1 show the occurrence of these types in the European language families and their
geographic distribution.
<INSERT TABLE 5 SOMEWHERE HERE>
<INSERT MAP 1 SOMEWHERE HERE>

22
As is clear from the table and the map, the juxtapositional PPC type represents the unmarked option
– it occurs in all the European language families, especially in two clear areas – the southern and southeastern parts of Europe, where different families meet, and in the Germanic (but only marginally in its
geographically most western members). In a few of these languages the juxtapositional type is clearly
new and came to replace the more archaic genitive construction – in addition to Germanic, also in
Bulgarian, Macedonian and Greek – whereas in most other languages this type has obviously existed for
a long time.
PPCs with prepositions have a very limited14 distribution and occur only among the most western
European Indo-European languages, Romance, Celtic and, to a limited degree, Germanic. In all these
cases the construction is of relatively recent origin.
Finally, PPCs using case inflection to mark Substance occur in three language families, those with
the genitive case in a few Indo-European and North-East Caucasian (Daghestanian) languages, and those
with the partitive case in several Finno-Ugric languages. The Daghestanian genitive-marking PPCs
constitute a clear island among the otherwise juxtapositional PPCs in this south-eastern European corner
with some languages alternating between the two construction types and at least one (Lezgian)
exclusively resorting to juxtaposition.
Within Indo-European, PPCs with genitives represent an archaic construction, which has a compact
distribution among Balto-Slavic (apart from the southern caseless languages Bulgarian and Macedonian)
and also occurs in an island-like fashion in Celtic (Irish and Scottish Gaelic) and in Germanic (German
and Icelandic, in both cases competing with more frequent and stylistically more neutral innovative
constructions). Among Finno-Ugrian languages, such PPCs occur mainly among the Finnic languages
and in Eastern Sami (e.g. in the Kildin dialect on the Kola peninsula, ex. (31)), i.e. among those
languages which have had the closest contacts with the most conservative “genitive-markers“ among the
Indo-European languages. The other languages (including other varieties of Sami) resort to the
juxtapositional type, as illustrated by ex. (32) below:

23
(31)

(32)

Eastern Sami (Itkonen 1973: 298)
munn
tonne
andDaù
[verc
jāvvε]
I:NOM you(SG):DAT give:1SG [sack:ACC flour:PRTV]
‘I am giving you a sack of flour’
a. Northern Sami (Inari dialect; Itkonen 1973: 303)
stuorra joavkku aalmug
ja kaalguh
big
number people:PRTV.SGand woman:PRTV.PL
‘a big number of people and women’
b. Udmurt (Pirrko Suihkonen p.c.)
kyk s'umyk
ts'aj
two cup:NOM tea:NOM
‘two cups of tea’
b. Meadow-Mari (Zorina et al. 1990: 93)
kok kilo
šere
olma
two kilogram:NOM sweet
apple:NOM
‘two kilograms of apples’
c. Erzya-Mordvin (Paasonen MW 2033))
kanst'
t'en'in'
[stopka vinin'e]
hand:PST.3PL
I:DAT [glass
wine:NOM]
‘they handed me a glass of wine’
d. Hungarian
három liter piros lé
three
litre red juice:NOM
‘three litres of red juice’

6.2. The origin of (P)PCs in the European languages
As for the origin of the overt markers in PCs and PPCs, the following generalization holds:

 overt markers in PCs and PPCs in the European languages very often originate as markers of 
‘direction FROM’ / ‘separation’ (Ablative and the like).

Section 3.2. suggested different steps on the grammaticalization path “SEPARATION FROM -> PCs“ for
Finnic languages, summarized in Fig. 1, which can possibly apply to the other languages as well. One of
the important suppositions was that the two main components of such constructions, the Measure and the
Substance, originate as two separate dependents to a verb and only gradually come to be related directly
to each other and build a single noun phrase via the process of syntactic reanalysis. We can thus talk
about a clausal source for (P)PCs. In addition to the numerous circum-Baltic examples discussed in the
previous sections, Armenian examples (6a) and (8) were just such a case, and the same point is made for
PCs by the Hungarian and Turkish examples, and for PPCs by the Welsh examples below:

24
(33)

(34)

Hungarian (Edith Moravcsik & Beata Megyesi p.c.)
a. PC
egy liter ab-ból
a
piros lé-ből
one litre that-ABL the red juice-ABL
‘a litre of that red juice’
b. Direction ‘FROM’
Péter Stockholm-ból érkeze-tt
tegnap.
Peter Stockholm-ABL come-PRET.3yesterday
‘Peter came from Stockholm yesterday. ’
Turkish (Kornfilt 1996: 114)
a. PC
Ahmet [pasta-dan iki
dilim] ye-di
Ahmet cake-ABLtwo slice
eat-PST
‘Ahmet ate two slices of the cake. ’
b. Direction ‘FROM’
Ahmet bakkal-dan
iki
şişe şarap çal-di
Ahmet grocer-ABL two bottle wine steal-PST

(35)

‘Ahmet stole two bottles of wine from the grocery store.’
Welsh
a. PPC
cwpanad o goffi
du
cup
of coffee black
‘a cup of black coffee’ (Susan Clack, Alan Thomas and Robert Borsley p.c.).
b. Direction ‘FROM’
Dwi'n
dod
o
Fangoryn
wreiddiol.
be.PRS.1SG-in come:VN from Bangor in
beginning
‘I come originally from Bangor.’ (King : 1993, p. 284)

In the same way as we have seen in the circum-Baltic languages (cf. the discussion of the partitive
case in section 3.2. and of the prepositions av/af and von in Sections 5.1. and 5.2), the derivational
relationship between PCs and “ablative“ constructions is often obscured by various other
grammaticalization processes which frequently apply to the latter. More specifically,



“FROM“ constructions serve as a popular grammaticalization source, e.g. for possessive NPs; and
the original “FROM“ meaning may gradually get bleached and even lost in the course of time.

As a result of these processes, PPCs with the prepositions de in most of the Romance15 languages, of
in English and, marginally, von in German look like possessive NPs, but in all these cases the similarities
are indirect and have to do with the ablative origin of both constructions.
It seems, however, that overt markers in PCs and PPCs do not necessarily originate as ablative-like
markers on the clause level. One example of a different grammaticalization source is the Scandinavian
‘with’-constructions (cf. ex. (22) and the discussion in section 5.1.). In this case, PPCs develop from noun
phrases with a clear noun-attribute structure: with tea is an attribute to cup in a cup with tea, which later
came to be re-interpreted as PPCs (‘a cup of tea’), and we can talk about phrasal sources for (P)PCs. In
such cases, the semantic changes leading to the development of PPCs do not presuppose the same degree
of syntactic reanalysis as was the case with formerly separational clauses which give rise to (P)PCs.
Also, as mentioned above, there is no historical evidence whatsoever for a separative / directional
origin of the genitive in a number of Indo-European languages where it has partitive and pseudo-partitive
uses. A similar problem is presented by genitive-marking PPCs in Daghestanian: to my knowledge, too
little is known about the origin of their genitive case (unfortunately, I lack information on PCs in
Daghestanian). In the absence of relevant data, we should not reject the possibility that PPCs and
“possessive“ structures can be related in other ways which do not necessarily have to do with directional /

25
separative meanings. – “Possessive“ structures, in a very broad sense of the word “possessive“, would
constitute another phrasal source for (P)PCs.
As is well known, possessive NPs across languages show a considerable degree of polysemy and are
frequently used to express PART-WHOLE and MATERIAL relations, just to name a couple. The PPC “a pile
of stones“, where “pile“ is a nominal quantifier, is semantically quite close to “a tower of stone“ – both
are made up of stones (and thus evoke the relation of MATERIAL), while the PC “a slice of the cake“ is
fairly close to “a corner of the room“ (where the relation of PART-WHOLE is evoked). Preliminary data
show that these “local“ similarities may result in different structures for the different nominal quantifiers
in one and the same language, but the details remain to be investigated.
The examples below show that possessive-like structures may sometimes be used as (P)PCs even in
those languages which normally resort to other types (cf. (36a) with (34a), and (36b) with (11)):
(36)

a. Turkish (Kornfilt 1996: 121)
Ahmet [şarab-in yari-sin-i]
iç-ti
Ahmet wine-GEN half-3SG.POSS-ACC drink-PST
‘Ahmet drank half of the wine’
b. Finnish (Leino 1993: 289)
tuotanno-n
valta-osa
production-GEN dominating-part
‘the bulk of production’

Fig. 10 summarizes the discussion of grammaticalization sources for (P)PCs involving overt markers
in the European languages.

<ADD FIG. 10 SOMEWHERE HERE>

In a still broader typological prospective, European PPCs with overt markers, in particular with
“genitive“/ possessive markers, appear as very rare, if not unique. Thus, even in the Sino-Tibetan
languages, notorious for their multifunctional possessive / attributive markers, which accompany almost
any type of dependents to a nominal, PPCs involve juxtaposition.
There are, as previously mentioned, at least two kinds of juxtapositional PPCs in the European
languages: those that have arisen in connection with the loss of former case inflections and those for
which no such development has been attested. I would suggest that the juxtapositional strategy on the
whole may be accounted for by the tendency to develop a unified treatment of nominal and other
quantifiers, in particular as cardinal numerals. At least for the European languages, nominals are normally
juxtaposed to numerals (Hurford forthc): numerals govern their nominal complements, as in Balto-Slavic
and Finnic languages, and especially the complicated alternation between government and agreement (in
Slavic and Finnic languages) is a unique areal feature (Sköld 1990, who, however, explains the Finnic
pattern by earlier contacts with Germanic languages).
Interestingly, the morpho-syntactic behaviour of Measure words in juxtapositional PPCs and the
mechanisms that shape it manifest a great deal of parallelism with numeral constructions in classifier
languages: classifiers start off as full-fledged nominals, but gradually lose characteristic nominal features
(cf. Bisang 1993, forthc.). However, it has been suggested that numeral classifier constructions normally
go through syntactic reanalysis and end up with the constituency structure [[Numeral Classifier] [N]]
which is not normally found in European languages (Croft 1996: 64). The problem is that usual
constituency tests work very poorly when applied to PPCs, especially to juxtapositional ones (David Gil
p.c.), and more comparative research is needed for generalizations on how the development of
juxtapositional PPCs bears on their headedness and constituency structure.

7. Conclusions: Circum-Baltic (P)PCs as an areal
phenomenon

26
Finally, to what degree are (P)PCs in the circum-Baltic languages interesting as an areal
phenomenon? The data on the PPC-types in the circum-Baltic languages is presented in fig. 11.
<ADD FIG. 11 SOMEWHERE HERE>
For the Indo-European languages, PPCs with case-marked Substance nominals are archaic
constructions which had a much wider distribution in the older languages, including the older stages of
Germanic (see Section 5). Thus, to judge from our data, a few centuries ago the circum-Baltic languages
were more similar in this respect, as shown in fig. 12:
<ADD FIG. 12 SOMEWHERE HERE>
However, cross-linguistically, PPCs with case-marked Substance nominals constitute a fairly unusual
option, both in a world-wide perspective and, to a lesser degree, among the European languages. Also
among the Finno-Ugric languages such constructions occur only in the circum-Baltic languages and in
Eastern Sami. Thus, the Finnic languages, in their PPCs, are much more similar to their Indo-European
neighbours than to their Finno-Ugric relatives,, most probably, as a result of contacts with the former.
The cross-linguistic data provided in this article confirm the earlier suggestions made by Kont 1963,
Larsson 1983 that some of the functions of the partitive case in Finnic were modelled on those of the
Baltic genitive. According to this hypothesis, the two cases had been similar enough for bilingual FinnicBaltic speakers to apply the same rules when speaking each of the languages. This hypothesis, thus, rests
on an assumption that the Finnic partitive and the Baltic genitive case shared some function(s), to start
with, and these functions were taken as a basis for identification of the two cases by bilingual speakers.
The question is which functions the Finnic partitive could have shared with the Baltic genitive
before the two were partly identified.
Typological considerations may be of great help in this matter. That is, if cross-linguistically highly
marked (unusual) phenomena occur in a number of genetically non-related neighbouring languages, the
obvious explanation for this “coincidence“ would be linguistic contacts, whereas development of crosslinguistically frequent phenomena in a language or several languages can normally be explained by
language-internal forces. Let us now have another look at the shared functions of the Finnic partitive
case, on the one hand, and those of the Baltic (and Slavic) genitive case and consider them from the point
of view of their cross-linguistic usualness (frequency), as well as their occurrence within the larger
families (Finno-Uralic and Indo-European respectively).

 Case­marking of Substance in PCs: the use of a separative case (or a separative marker in general) 
to mark Substance nominals is well attested cross­linguistically. In this respect Finnic languages 
seem to behave like their other relatives (cf. ex. (33a) from Hungarian). The origin of the genitive­
marking in Indo­European PCs is not quite clear – a separative source cannot be excluded, but can 
hardly be proven either. 
 Case­marking of Substance in PPCs: see the discussion above. 
 Case alternation for marking “total“ and “partial“ objects / subjects: although the details of this 
alternation differs considerably between Finnic, Baltic and Slavic, there is a considerable overlap 
here. Cross­linguistically the phenomena are extremely unusual (especially in case­marking of 
subjects), even though they have partial parallels in other languages. The distinction between total and
partial objects by means of case alternation (accusative vs. genitive) is attested in a number of older 
Indo­European languages; partial subjects are attested too, but to a very limited extent. Within Finno­
Ugric, outside Finnic, a certain parallel is provided by Eastern Sami (both for subjects and objects); in
Mordvin, indefinite objects to verbs of eating and drinking and a few others are marked by the 
ablative – genetically the same case as the Finnic partitive16. 
 Case­government of nouns by numerals: higher numerals govern their complements (determining 
either their case or adposition) in a number of languages, even though this is still a relatively 
infrequent phenomenon. It is also known from several older Indo­European languages, at least 
marginally. However, within the Balto­Slavic languages, including the Slavic languages outside the 
circum­Baltic region, these phenomena came to be much more prominent: even lower numerals case­

27
govern, in particular their complements, which is extremely unusual cross­linguistically. Nothing of 
this kind occurs anywhere in Uralic outside Finnic and Eastern Sami, which, in all probability, were 
influenced by the neighbouring Indo­European languages. It should also be noted that the rules 
governing alternation between government and agreement within numeral constructions are much 
more similar in Finnic and Slavic than in any of these languages and Baltic. Given the degree of 
complexity of these rules and their typological uniqueness a plausible hypothesis is that the Finnic 
system is in a certain sense “borrowed“. However, as far as we know, there were no historical 
preconditions for such an extensive influence on Finnic from Slavic, as opposed to that from Baltic. 
The only plausible conclusion is, thus, that at an earlier stage, the Baltic and Slavic rules for numeral 
constructions were much more similar, but were later simplified in Baltic.
 To summarize: of all the above­listed functions of the partitive (formerly separative) case in Finnic, 
case­marking of Substance nominals in PCs is the one which has almost certainly arisen due to the 
internal Finnic (Uralic) development. Since the Baltic genitive could also be used in PCs, this 
function could be taken as a basis for the identification of the two cases by bilingual speakers who 
would later extend the Finnic partitive case to the other functions of the Baltic genitive. However, 
even though I wish to make a case for the importance of Substance­marking in PCs as an original 
shared function of the Finnic partitive and the Baltic genitive case, I would not like to exclude the 
possibility of others.17 
Now, if the general conclusion is that (most of) the grammaticalized functions of the partitive case in
Finnic are modelled on those of the Baltic genitive case, what is the purpose of the grammaticalization
stories and diagrams (Fig. 4 - 6) presented in Sections 3.2. and 3.3.? It would have been sufficient to state
that the use of the partitive case was extended both to PPCs and numeral constructions under the Baltic
influence, without any further motivation. There is, however, an important methodological point here
related to the issue of internal reconstruction, areal linguistics and typology. More specifically, each of
the following grammaticalization steps suggested in Sections 3.2. and 3.3. represents a development
which finds numerous parallels in other languages:

 development of separative constructions into PCs 
 extension of PCs into PPCs / structural similarity between PCs and PPCs in a language
 modelling of numeral constructions on the basis of PPCs, or 
 development of partitive numeral constructions into non­partitive ones 
In other words, the developments sketched in 3.2. and 3.2. represent reasonable grammaticalization
paths which do not need to be externally motivated. In the eyes of traditional historical linguists this
could have been a sufficient plausible explanation for the Finnic situation: very often in historical
linguistics “the methodological inclination has been to consider the possibility of external causation only
when all efforts to find an internal motivation for some change have failed“ (Thomason & Kaufmann
1988:57). And even the fact that the Finnic partitive and the Baltic genitive case share numerous
similarities could have received internal explanation – each of the two cases followed more or less the
same grammaticalization path. In this situation, only broader cross-linguistic evidence provides necessary
arguments for the importance of external factors in the grammaticalization of the partitive case.
Although each of the steps on its grammaticalization path finds cross-linguistic parallels, their
cumulative effect is unique and is only shared by Baltic (and Slavic).
Interestingly, while the same Finno-Ugric languages use the partitive case to mark the Substance
in PPCs and partial objects, Swedish, Danish and German lost a similar distinction between whole
objects and partial objects, which were marked with the genitive case, just as they lost their PPCs with
genitive-marked Substance nominals. Thus, there exists a complicated relationship between the existence
in a language of an opposition between whole and partial objects by means of case alternation, on the one
hand, and the existence of PPCs with case-marked Substance nominals on the other. This relationship,
however, deserves a special study of its own (for some discussion see Koptjevskaja-Tamm & Wälchli this
volume).

References
Alho, Irja H. 1992. Distinguishing kind and set in Finnish. Studies in the Linguistic Sciences 22, 1: 1 16.
Ambrazas, Vytautas (ed.). 1997. Lithuanian Grammar. Vilnius: Baltos Lankos.

28
Anward, Jan & Per Linell. 1976. Om lexikaliserade fraser i svenskan. Nysvenska studier, årg. 55-56:
77 - 119.
Barwise, Jon & Robin Cooper. 1981. General Quantifiers in Natural Language. Linguistics and
Philosophy 4, 159-220.
Behaghel, Otto. 1923. Deutsche Syntax. Eine geschichtliche Darstellung. Band 1. Die Wortklassen
und Wortformen. Heidelberg: Carl Winter’s Universtitätsbuchhandlung.
Belletii, Adriana. 1988. The Case of Unaccusatives. Linguistic Inquiry 19, 1-34.
Bisang, Walter. 1993. Classifiers, quantifiers and class nouns in Hmong. Studies in Language 17:1, 1 51.
Bisang, Walter. forth. Classifiers in East and Southeast Asian languages. Counting and beyond. Ms.
Brugmann, Karl & Berthold Delbrück. 1909. Grundriß der vergleichenden Grammatik der
indogermanischen Sprachen. 2. Band: Morfologie, 2 Teil (2. neu bearbeitete Auflage.
Corbett, Greville. 1978a. Problems in the syntax of Slavonic Numerals. Slavonic and East European
Review, 56: 1-12.
Corbett, Greville. 1978b. Universals in the syntax of cardinal numerals. Lingua, 46: 355-68.
Corbett, Greville. 1983. Hierarchies, Targets and Controllers. Agreement Patterns in Slavic. London
& Canberra: Croom Helm.
Croft, William. 1991. Syntactic categories and grammatical relations. Chicago & London: The
University of Chicago Press.
Croft, William. 1993. What’s a Head? In Rooryck, Johan & Laurie Zaring (eds.), Phrase Structure and
the Lexicon. Studies in Natural Language and Linguistic Theory, 33. Dordrecht: Kluwer Academic, 35 75.
Croft, William. 1994. Semantic universals in classifier systems. Word 45.2, 145-171.
Delsing, Lars-Olof. 1993. The Internal Structure of Noun Phrases in the Scandinavian Languages. A
Comparative Study. Lund: University of Lund, Dept. of Scandinavian languages, dissertation.
Denison, Norman. 1957. The Partitive in Finnish. Annales Academiae Finnicae, ser. B, tom. 108.
Helsinki.
Erelt, Mati (ed.). 1993. Eesti keele grammatika. II. Süntaks. Tallinn: Eesti Teaduste Akadeemia Keele
ja Kirjanduse Instituut.
Eschenbach, Carola. 1993. Struktur- und Quantitätsbezug. Zähl- und Maßangaben in Wissens- und
Sprachverarbeitung. Hamburg: Universität Hamburg, dissertation.
Fairbanks, Gordon H. & Earl W. Stevick. 1975. Spoken East Armenian. New York: Spoken Language
Services, Inc.
Falk, Hjalmar & Alf Torp. 1900. Dansk-norskens syntax i historisk fremstilling. Kristiania: Forlagt af
H.Aschenhoug & Co. (W.Nygaard).
Greenberg, Joseph. 1989. The Internal and External Syntax of Numerical Expressions. Explaining
Language Specific Rules. In Kefer, Michael & Johan van der Auwera (eds.), Universals of Language.
Belgian Journal of Linguistics, 4: 105 - 118.
Hakulinen, Lauri. 1957. Handbuch der Finnischen Sprache. 1 Band. Wiesbaden: Otto Harrassowitz.
Heltoft, Lars. 1996. Det danske nominals udtrycks- og indholdssyntaks - et dependensanalytisk
forsøg. Ny forskning i grammatik, fællespublikation 3. Odense: Odense universitetsforlag, pp. 7 - 34.
Hentschel, Elke. 1993. Flexionsverfall im Deutschen? Die Kasusmarkierung bei partitiven GenetivAttributen. Zeitschrift für germanistische Linguistik, 21:320-333.
Herslund, Michael. 1995. The Object Relation and the Notion of Incorporation. In Schøslet, Lene &
Mary Talbot (eds.), Studies in Valency 1, 1-18. Odense: Odense University Press.
Herslund, Michael. 1997. Partitivity and inalienable possession. In Baron, Irène & Michael Herslund
(eds.), Possessive structures in Danish. Fagling-rapport nr. 3, 1-44, Copenhagen: Handelshøjskolen.

29
Hoeksema, Jacob (ed.). 1996. Studies on the Syntax and Semantics of Partitive and Related
Contructions. Berlin & New York: Mouton de Gruyter.
Hoop, Helen de. 1992. Case Configuration and Noun Phrase Interpretation. PhD dissertation,
University of Groningen.
Hopper, Paul J. & Sandra A. Thompson. 1984. The Discourse Basis for Lexical Categories in
Universal Grammar. Language, 56: 251 - 299.
Hurford, James. (forthc.). Numeral-noun interaction. In Plank, Frans (ed.), The Noun Phrase in the
Languages of Europe. Berlin: Mouton de Gruyter.
Itkonen, Erkki. 1973. Zur Geschichte des Partitives. Finnish-Ugrische Forschungen, 40. Helsinki.
King, Gareth. 1993. Modern Welsh. A comprehensive grammar. London: Routledge.
Kont, Karl. 1963. Käändsõnaline objekt läänemeresoome keeltes. Eesti NSV Teaduste Akadeemia
Keele ja Kirjanduse Instituudi uurimused, VI:190-199.
Kornfilt, Jaklin. 1996. Naked partitive phrases in Turkish. In Hoeksema, 107-143.
Larsson, Lars-Gunnar. 1983. Studien zum Partitivgebrauch in den ostseefinnischen Sprachen. Acta
Universitatis Upsaliensis, Studia Uralica et Altaica Upsaliensia, 15. Uppsala.
Leino, Pentti. 1993. Polysemia - kielen moniselitteisyys. Kieli 7. Helsinki: University of Helsinki,
Dept. of Finnish Language.
Mikkelsen, Kr. 1911. Dansk ordföjningslære. København: Lehmann & Stages forlag.
Nemvalts, Peep. 1996. Case Marking of Subject Phrases in Modern Standard Estonian. Acta
Universitatis Upsaliensis, Studia Uralica Upsaliensia 25. Uppsala.
Norde, Muriel. 1997. The History of the Genitive in Swedish: A Case Study in Degrammaticalization.
Doct. diss., University of Amsterdam.
Paasonen, Heikki 1996. Bd 4: Mordwinisches Wörterbuch under Mitarbeit von Hans-Hermann
Bartens bearbeitet und herausgegeben von Martti Kahla. Lexica Societatis Fenno-Ugricae
Paus, Charles. 1994.Social and pragmatic conditioning in the demise of the partitive case. Russian
Linguistics, 18: 249 - 266.
Penttilä, Aarni. 1957. Suomen kielioppi. Porvoo: Werner Söderström Oy.
Pitkänen, Antti J. 1979. Binominala genitiviska hypotagmer i yngre nysvenska. Doct. diss. Helsinki:
Svenska litteratursällskapet i Finland.
Plank, Frans. 1981. Morphologische (Ir)Regularitäten. Tübingen: Narr.
Press, J. Ian. 1992. The Structure of Quantifier Phrases in Russian. Irish Slavonic studies 13: 75 - 90.
Selkirk, E. 1977. Some remarks on noun phrase structure. In Culicover, P.W. & A. Akmaijan (eds.)
Formal Syntax. New York: Academic Press, 218 - 316.
Seppänen, Aimo. 1983. Finnish ‘kaksi poikaa’.Studia Linguistica 37, 2: 161 - 174.
Schwartz, Eugène. 1878. Om oblika kasus och prepositioner i fornsvenskan från tiden före år 1400.
Upsala Universitets Årsskrift. Filosofi, Språkvetenskap och Historiska Vetenskaper. II.
Sköld, Tryggve. 1990. The Finnish construction kolme poikaa and its background. Nyelvtudományi
Közlemények 91, 1-2: 203-211.
Smyth, Herbert Weir. 1956. Greek Grammar. Cambridge: Harvard University Press.
Thomason, Sarah Grey & Terrence Kaufman (1988) Language Contact, Creolization, and Genetic
Linguistics. Berkeley: University of California Press.
Wessén, Elias. 1970. Schwedische Sprachgeschichte. Band III: Grundriss einer historischen Syntax.
Berlin: Walter det Gruyter & Co.
Zorina, Zoja Georgievna, Galina Semenovna Krylova & Émma Semenovna Jakimova. 1990.
Marijskij jazyk dlja vsex: I. Joškar-Ola: Marijskoe knižnoe izdatel'stvo.

0I would like to express my gratitude to the many people who, in one or another way, have helped me in writing the paper.
First of all, to those who have provided me with the data on the various languages used in the paper:
Armenian - Natal'ja Kozinceva; Estonian - Diana Krull and Peep Nemvalds; Finnish - Irja Alho, Päivi Juvonen,
Johanna Laakso, Riitta Korhonen and Maria Vilkuna; Hungarian - Beata Megyesi and Edith Moravcsik; Irish – Dónall
P. ” Baoill; Latvian - Laimute Balode, Axel Holvoet and Baiba Kangere; Mari- Simon Christen; Mordvin Bernhard Wälchli; Old Swedish - John Swedenmark and Muriel Norde, Scottish Gaelic – Robert Mullally, Udmurt
- Pirkko Suihkonen; Welsh - Susan Clack, Alan Thomas and Robert Borsley.
The discussion in section 3.1 is, to a significant degree, based on generous assistance from Irja Alho, Riitta Korhonen and
Maria Vilkuna. At various stages in the preparation of this paper I have benefited a lot from discussions with Vytautas
Ambrazas, Jan Anward, Brita Bergman, Bill Croft, Östen Dahl, Kari Fraurud, Michael Herslund, Axel Holvoet, Baiba
Kangere, Edith Moravcsik, Muriel Norde, Tomas Stolz and Bernhard Wälchli. None of these kind people bears any
responsibility for the possible errors (Pierluigi Cuzzolin and Elisa Roma deserve a special thank for pointing out an error in
my analysis of Irish and Scottish Gaelic).
1To understand the connection between (organic) part of the whole and partitives consider the function of the word part in
the following examples, from merely indicating a part of the body-part to quantifying a subset separated from the whole set:
The upper part of his face is painted green - > A part of his face is painted green ->Part of my books have completely
disappeared from my shelves.
2Numerals, from the linguistic point of view, are a mystical category per se, but here I will refrain from any deep discussion
of their nature and part-of-speech properties.
3 The translative, a case marker of younger origin attested for Mordvin and Finnic, has assumed the place of the former
lative (-s, -k) in Finnic.
4Greenberg (1989: 111) quotes interesting anthropological observations on actual methods of counting which provide
evidence and an explanation for the close connection between between nominal quantifiers and, at least, higher numerals.
Thus, when Basque shephards count sheep, they put pebbles in heaps of ten. When there are ten such heaps, a pebble is put
aside to stand for one hundred. In this fashion, higher units become conceived of as objects which are themselves counted.
5The assignment of the singular number to nouns by numerals is a widely spread phenomenon (Hurford forthc.), also
among the Finno-Ugric languages, where it is considered to be an archaic feature (Sköld 1993).
6 With numerals in the partitive case, the distinction between government and agreement is of course blurred.
7The origin of number assignment in Finnic numeral constructions is not quite clear. According to one hypothesis it may be
accounted for by the earlier neutralization of number in the Finnic oblique cases. When the distinction between singular and
plural was later introduced into the paradign of oblique cases, it did not “hit“ highly grammaticalized numeral constructions
(similar resistance to younger morpho-syntactic rules is typical of old, highly grammaticalized constructions in general).
Another explanation motivates the singular number by considerations of economy (a numeral by itself signals that the noun
it combines with refers to plural objects), as well as by genetic factors: nouns in the singular appear in numeral
constructions in most of the other Uralic languages, even though the construction itself is different.
8 Note that the terms ‘higher’ and ‘lower’ are slightly misleading. Only the last digit in a numeral counts for morphosyntactic rules in Slavic and Baltic, thus, ‘322’ is a ‘lower’ numeral, whereas ‘20’ is ‘higher’ numeral. Teens are always
higher numerals.
9I am grateful to Baiba Kangere for drawing my attention to this construction in Latvian and providing me with relevant
examples.
10 In Northern Swedish local vernaculars, prepositions are sometimes found also in PPCs. Cf. Dalecarlian (Älvdalen) An
tsjyöpt tau tsijlo åv mjöli ‘He bought two kilos of flour’.
11Cf. this with ex. (22) from Old Swedish where mäþ occurs after pund - ‘pound’.
12I am grateful to Michael Herslund for drawing my attention to this fact.
13Pragmatic considerations and, in particular, a degree of lexicalization / novelty play a considerable role here. Thus, even
very lexicalized numeral expressions may be pronounced with the same unitary stress pattern, e.g. tio Guds bud ‘the ten
commandments’. On the other hand, less usual PPCs of the type en hink vin ‘a pail of wine’, though perfectly well-formed
in theory, are hardly ever used in reality, which makes it difficult to find a natural context for testing to what degree the
intonational pattern in PPCs is sensitive to their usualness / novelty.
14Of course, in terms of the number of speakers using this type and the geographic region covered by it, the distribution of
the prepositional PPC type is far from limited!
15In Rumanian, PPCs with the preposition de are opposed to possessive NPs in which the possessor attaches the genitive
case, cf. un pahar de vin ‘a:M glass of wine’ vs. casa fete-i - ‘house:DET girl-GEN.DET ’ (‘the house of the girl’). De is
however used as a marker of general attribution, as in o comoară de gospodină ‘a:F treasure of (a) housewife’.
16 A striking parallel to the partitive/ genitive case in the three above-listed uses of is, of course, provided by the French

partitive preposition / article de, which also started off as a separative marker and later, via PCs, came to be used both with
Substance nominals in PPCs and with partial objects and (existential) subjects.
17 Larsson (1983: 139 - 147), on the basis of Mordvin data, suggests that the partitive case was used for marking objects to
verbs of drinking and eating already at the Proto-Finnic stage, and that this could have been a shared function of the Finnic
partitive and the Baltic genitive case. However, the Mordvin ablative case is on the whole a “relic“ case in the sense that
most of its modern functions are obviously derived from the older more concrete ones which are by now lost. This relic
character of the Mordvin ablative case makes it a bad candidate for giving a fair picture of the Proto-FinnoVolgaic situation.

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close