The Syntax of Numerals in Bosnian (Leko)

The Syntax of Numerals
in Bosnian
Nedžad Leko
2009
L IN C O M E U R O P A
PVA
2009 . led by UNCOM GmbH 2009.
2085
UNCOM GmbH
Gmunderstr. 35
D-81379Muenchen
[email protected]
http://home.t-online.de/home/LINCOM.EUROPA
www.lincom-europa.com
webshop: lincom.eu
All rights reserved, including the rights of translation into any

foreign language. No part of this book may be reproduced in any
way without the permission of the publisher.
Printed in E.C.
Printed on chlorine-free paper
Die Deutsche Bibliothek - CIP Cataloguing-in-Publication-Data
A catalogue record for this publication is available from Die

Deutsche Bibliothek (http://www.ddb.de)
rw
f
I -■
/ V '
Bayerlsche
Staatsbiblio'hek
Miinchwn
The Syntax o f Numerals in Bosnian
Contents
1. What are numerals? 3

1.0. Inroduction 3
1.1. Some dictionary definitions of numerals 3
1.2. The categorial status of numerals 4
1.3. Cardinal numeral systems 5
1.4. Other numerals 6
1.5. Non-referential counting words and referential number words 7
1.6. Different kinds of referential number words 8
1.7. The grammatical properties of referential number words 9
1.8. Greenberg's Universals relating to numerals 12
2. Numerals in Bosnian 23
2.0. Introduction 23
2.1. Cardinal numerals 23
2.1.1. Jedan {I) 23
2.1.1.1. The structure of NP in Bosnian 25
2.1.1.2. The structural position of jedan 34
2.1.2. Dva (2) 36
2.1.3. Tri (3) 46
2.1.4. Četiri (4) 48
2.1.5. Pet (5) 50
2.1.6. Šest (6), sedam (7), osam (8), devet (9) 54
2.1.7. Jedanaest (11), dvanaest (12), trinaest (13),... 54
2.1.8. Deset (10), dvadeset (20), trideset (30),... 54
2.1.9. Sto (100) 55
2.1.10. Dvjesto (200), tristo (300), četiristo (400),... 58
2.1.11. Hiljada (1,000) 61
2.1.12. Pet hiljada (5,000),pedeset hiljada (50,000),... 62
2.1.13. Milion (1,000,000) 66
2.1.14. Pet miliona (5,000,000),pedeset miliona (50,000,000),... 68
2.2.15. Milijarda (1,000,000,000) 69
2.1.16. Summary 69
2.2. Ordinal numerals 72
2.2.1. P m (1st) 72
2.2.2. Drugi (2nd), etc. 74
2.3. Collective numerals 76
2.3.1. Dvoje (2) 76
2.3.2. Troje (3), etc. 78
2.3.3. Jedni, jedne, jedna (1) 79
2.3.4. Dvoji, dvoje, dvoja (2), etc. 79
2.4. Other numerals 80
2.4.1. Dvojica (2) 80
2.4.2. Trojica (3), etc. 81
2.4.3. Jedinica (1) 82
2.4.4. Dvica, dvojka (2), etc. 82
2.4.5. Approximative numerals 83
2.4.6. Fractions 84
2.4.7. Distributive numerals 85
1
The Syntax of Numerals in Bosnian
3. Concluding considerations 87
3.0. Introduction 87
3.1. The syntactic behaviour of numerals in Bosnian 87
3.2. Numerals that behave like Quantity Nouns 88
3.3. Numeral Quantifiers 91
3.4. Adjectival Numerals 96
References 102
2
1. What are numerals?
1.0. Introduction
This monograph deals with numerals with special emphasis on numerals in Bosnian
and their syntactic behaviour. At least eight types of numerals in Bosnian may be singled out,
and taking ‘five’ as an example, we have the forms pet, peti, petoro, petori, petorica, petica,
petina, po pet representing the various types of numerals. However, syntactically all these
numerals behave as three distinct categories, namely as quantifiers, nouns and adjectives.
Chapter 1 is devoted to numerals in general. Chapter 2 deals with numerals in Bosnian
describing them according to the traditional classification of numerals as cardinal, ordinal,
collective, etc. In Chapter 3, I will offer some theoretical considerations about the syntactic
behaviour of numerals in Bosnian. Theoretical considerations are based, of course, on real
language data, so it was inevitable to repeat in Chapter 3 some parts of the discussion from
Chapter 2, which is mainly concerned with the description of various types of numerals.
Those readers who are mainly interested in theoretical considerations may read first Chapter
3, which also gives some crucial descriptive facts that are extensively dealt with in Chapter 2.
1.1. Some dictionary definitions o f numerals
There is a lot of confusion about the numerals, and that is reflected in various standard
dictionaries, and specialized dictionaries of languages and linguistics.
For example, for Matthews (1997) a numeral is “one of a set of words or other
expressions indicating precise numbers: e.g. three or forty-nine; also the ordinal numerals:
third or forty-ninth” (p. 251). He specifies that an ordinal numeral is “one which indicates an
ordered position in a series, e.g. third’ (p. 259). And a cardinal numeral is “one which
indicates the number of individuals in a set, e.g. three, as opposed to an ordinal numeral, e.g.
third’ (p. 45).
Some linguists don’t include numerals at all in their discussion of parts of speech. So,
for example, there is no entry on numerals in Crystal (1997).
Frequently, an alternative term, number, is used for numerals. This is emphasized by
Hartmann and Stork (1972) who specify that a numeral is “a word denoting a number or
quantity. Two examples of numerals are cardinal numbers {one, two, three) answering the
question How many? and ordinal numbers {first, second, third) indicating the order in a
sequence. Alternative term: number” (p. 155).
For Richards, Platt, and Weber (1985) a numeral is “a word or phrase which is used to
name a number. In English, numerals may be cardinal numbers - one, two, three, or ordinal
numbers - first, second, third’ (p. 197). They also define an alternative term - number, and
specify that “numbers are used either as cardinal numbers (or cardinals) or ordinal numbers
(or ordinals). Cardinal numbers are used when counting; eg. 6 boys, 200 dollars, a million
years, and they may be used as nouns (eg. Count up to ten). Ordinal numbers are used when
we put things in a numerical order, eg. first, second, third’ (p. 197).
Trask (1997) gives very short and precise definitions of a cardinal numeral: “A
counting number like one, two, three” (p. 35), and an ordinal numeral: “An adjectival form of
a number: first, second, third’ (p. 158).
In the Oxford English Dictionary (Second edition, Vol. X) the following definitions of
a numeral are found: “ l.A word expressing a number; 2. A figure or character (or a group of
these) denoting a number” (p. 595).
3
On the other hand, the Concise Oxford English Dictionary (2006. Eleventh edition,
revised) contains somewhat different definitions of a numeral: “A figure, word, or group of
figures denoting a number. Origin ME: from Late Latin numeralis, from Latin numerus” (p.
981) and a number: “An arithmetical value, expressed by a word symbol, or figure,
representing a particular quantity” (p. 980).
Oxford Advanced Learner’s Dictionary o f Current English (2005. Seventh edition)
specifies that a numeral is “a sign or symbol that represent a number” (p. 1042), whereas a
number is “a word or symbol that represents an amount or a quantity” (p. 1041). On the other
hand, it is specified that a cardinal number is “a number used to show quantity rather than
order” (p. 222), whereas an ordinal number is “a number that refers to the position of
something in a series” (p. 1070).
Barnhart (1988) in his Dictionary o f Etymology says that the word ‘numeral’ in
English first appeared in “1530, in Palsgrave’s Lesclarcissement, word expressing a number;
later, figure standing for a number (1686) borrowed from Late Latin numeralis of or
belonging to a number, from Latin numerus number” (p. 714).
For Dixon (1994) cardinal and ordinal numbers are simply types of adjectives: “Types
of adjectives: quantifiers {many, few), cardinal {four), and ordinal {fourth) numbers” (p. 29).
Bussmann (1996) gives the most extensive definition of numerals: “Class of words
consisting primarily of adjectives {six months, double fault, threefold problem) as well as
substantives (a dozen eggs), indefinite pronouns {all, both, many, few), and adverbials (He
called twice already). Semantically they form a uniform group inasfar as they designate
numbers, quantities, and any other countable divisions. However, because their morphological
and syntactic behaviour varies in respect to declension, newer grammars classify them
differently, relegating them in part to pronouns, in part to adjectives. A basic division is made
between definite and indefinite {ten vs. several), where the definite numerals can be divided
into the following subgroups: a) cardinals - one, two, three, b) ordinals - first, second, third,
c) distributives - six each, d) iteratives - once, twice, thrice, e) multiples - eightfold, f)
collective numerals - a dozen, g) fractions - a tenth” (p. 334).
1.2. The categorial status o f numerals
Traditional grammarians usually consider numerals as a part of speech in the category

of substantives, together with adjectives, pronouns and nouns, as opposed to verbs and
adverbs. Most frequently numerals behave syntactically as adjectives in the sense that they
may modify nouns and agree with the noun in all relevant features. On the other hand, there
are numerals that clearly behave as nouns. It was noted (e.g. Štampe 1976, Corbett (1978))
that universally those numerals that denote lower numbers behave more like adjectives,
whereas higher numerals, complex numerals, and expressions designating bases (e.g. ‘five’,
‘ten’, ‘hundred’, etc.) behave more as nouns.
When the numerals behave syntactically as adjectives, they still have characteristics
that clearly differentiate them from proper adjectives. One important difference is the word
order, so that in languages where adjectives follow the head noun, numerals frequently
precede the noun. However, if both precede, or both follow the head noun, then the numeral is
the one that is more distant from the noun, e.g. six red cars vs. *red six cars. Further
difference between adjectival numerals and proper adjectives is their morphology. Very often,
adjectival numerals are inflected differently from proper adjectives. For example, they may
lack inflection for gender in languages where proper adjectives clearly distinguish gender, and
have separate gender inflections. Then in contrast to adjectives which may be predicative
adjectives, it is not the case with numerals (e.g. They are tall vs. *They are six). Proper
adjectives can be modified, whereas numerals cannot {very red cars vs. *very six cars).
Gradable adjectives can be marked to show comparative and superlative degree. That is not
4
the case with numerals. All these differences show that numerals are distinct from adjectives,
although they may have many adjective-like properties.
Those numerals that behave as nouns may have plural forms (e.g. hundreds). In
languages which have richer morphology, these numerals may have full nominal declension
with different forms for different cases (e.g. Bosnian: stotina-NOM, stotine-GEN, stotini-
DAT, stotinu-ACC, stotinom-INS/ LOC ‘hundred’). The nominal type of behaviour of such
numerals is reflected in the fact that they assign the genitive case to the following noun (e.g.
stotina vojnika-GEN PL ‘a hundred soldiers’) in the same way as nouns do (e.g. prijatelj
vojnika-GEN PL ‘aI the friend of soldiers’).
The discussion up to this point refers only to a specific type of numerals - cardinals.
This type of numerals is regarded as the basic, or unmarked type. In addition there are other
types of numerals. Ordinals are generally recognized as the most important and widespread
type. Various authors mention various other types of numerals, like adverbial cardinal (once,
twice ...), adverbial ordinal (the first time, the second time ...) and distributives (one each,
two each ...) (Greenberg 2000: 771). In addition to these, Bussmann (1996: 334) recognizes
also iteratives (once, twice, thrice), multiples (eightfold), collective numerals (a dozen) and
fractions (a tenth).
In contrast to cardinals which have the unmarked status, all other types of numerals
are marked in one way or another. Most frequently, this markedness is realized through
affixation to the unmarked cardinal numeral. This is the case, for example, with ordinal
numerals in English (e.g. fourth < four-th). The marked numerals may also be realized
through expressions of greater complexity, including the use of specialized nouns together
with cardinals, as with English adverbial cardinals (e.g.four times) or adverbial ordinals (e.g.
the fourth time). The unmarked status of cardinals is confirmed also by the fact that there are
languages which have only cardinal numerals.
1.3. Cardinal numeral systems
Cardinal numeral systems differ depending on what is taken as the numeral base of the
system, so that numeral systems may be quinary (5), decimal (10), duodecimal (12),
vigesimal (20), sexagesimal (60). The most widespread in world languages is the decimal
system. Mandarin Chinese is an example of a pure decimal system. Some decimal systems
have certain deviations, and therefore may be regarded as mixed systems. Such languages are
French and Welsh. For example, remnants of a vigesimal system in French are the numerals
denoting numbers from 80-99 (e.g. 80: quatre-vingt, literally ‘four-twenty’).
With complex cardinal numerals that are formed by addition usually the associative
‘and’ (or ‘with’) is used, or the superessive (Greenberg 2000: 777), an expression meaning
‘upon’ or something similar (e.g. Bosnian jedanaest < jedan na deset ‘one upon ten’). In
English, coordinators are optionally used in higher cardinal numerals (e.g. three hundred
(and) sixty).
There are also complex cardinal numerals that are formed as a result of multiplication.
Here, the base is usually treated syntactically as a noun which is multiplied (e.g. 30: trideset
‘three ten’). Multiplication with ‘one’ is possible only when the base is treated as a noun. So
in English one hundred or a hundred is possible, but not *one ten or *a ten. Exact parallelism
is found in Bosnian: jedna stotina ‘one hundred’ vs. *jedanjjedna/jedno deset ‘one ten’. In
addition to nominal stotina, Bosnian also has a quantifier sto for the number 100, but whereas
jedna stotina is possible, a quantifier form cannot be multiplied with ‘one’: *jedan/jedna/
jedno sto.
There are languages which use subtraction, division, and a process called overcounting
to form complex cardinal numerals. When such numerals are formed by subtraction, then they
usually include a separate morpheme with the meaning ‘from’, ‘without’, or ‘take away’. In
5
languages that use division to form complex numerals, division is usually by ‘two’ and,
rarely, by ‘four’. In this way complex numerals are formed usually in languages that have the
vigesimal numeral system. In such languages, 50 is most frequently formed by division and
expressed as ‘half of 100’, that is 100 divided by 2. Finally, an operation called overcounting
assumes overt expressions of this operation, so that the morpheme meaning ‘towards’ or
‘going-on’ is used. For example, in Ostyak, an Ugric language, 18 is expressed as ‘eight-
twenty’ and interpreted as ‘8 going-on 20’ (Greenberg 2000: 774).
It was already pointed out that numerals have many properties characteristic for
adjectives. They may resemble adjectives morphologically, and this is especially case with
lower numerals. In languages where adjectives agree with the head noun in features like
gender, number and case, numerals will agree also in the same features. However, some
features may be neutralized, and also numerals may have idiosyncratic morphology. For
example, numerals may agree in number with the head noun. So, the numeral 'one' may agree
with a singular noun, larger numerals may have morphogical plural characteristics and agre
with a noun in the plural. Those language which distinguish dual, in addition to singular and
plural, usually have a special, dual, morphology on the numeral 'two' which agrees with the
noun that is itself in the dual. Regarding number features of the noun preceded by the
numeral, not so rare is the case of the neutralization of number, so that the noun appears
always in the singular form when modified by a numeral (e.g. Hungarian and Turkish).
In contrast to lower numerals which may have adjectival characteristics, higher
numerals are morphologically often like substantives. This is especially the case with bases.
Often they do not agree with the noun, but rather assign the genitive case to it.
1.4. Other numerals
It was pointed out earlier that ordinals are often formed from cardinals by ading
affixes. Hower, lower numerals may be the exception to this tendency. The ordinal 'first' in
world languages is not derived from the cardinal 'one', but from some word meaning 'in front',
or 'foremost'. Although the ordinal 'second' is usually derived from the corresponding cardinal
numeral, often it may be derived from the word or root meaning 'following' (e.g. Latin
secundus from sequi 'to follow').
Languages also have special numerals that are used not for counting of nominal
objects, but rather for counting of verbal actions. They are called multiplicative numerals or
adverbial numerals. Languages may have both adverbial cardinals (e.g. once, twice..?) and
adverbial ordinals {the first time, the second time...). Adverbial cardinals in some languages
are formed by adding a distictive affix to the base of the cardinal numeral (e.g. Latin,
English), or by a particular word (like English time, French fois, Bosnian puta) combined with
a cardinal numeral. Finally, there are languages in which the adverbial expression is
incorporated into the cardinal numeral (e.g. German dreimal, Bosnian triput).
Adverbial cardinals and adverbial ordinals may modify only active predicates, never
stative ones. For example, in Bosnian (and English) the following expression is unacceptable:
*biti blijed dva puta 'to be pale twice', and the desired meaning can be expressed only using
an active predicate: poblijediti dva puta 'to become (turn) pale twice'. The distinction between
adverbial cardinals and adverbial ordinals may be illustrated with an interesting contrast in
Bosnian: dvaput/ dva puta vs. drugi put/ *drugi puta. Adverbial cardinals in Bosnian may be
formed either by incorporation {dvaput) or by combining puta with the cardinal numeral {dva
puta). On the other hand, with adverbial ordinals only one option exists: there is no
incorporation, and instead of the full form puta, only the reduced form put may be used.
The next category of numerals are so-colled collective numerals. They are frequently
used to modify the so-called pluralia tantum nouns (e.g. scissors, pants), or nouns with
different meanings in the singular and plural (e.g. Latin sg. llttera 'letter, character', pi. litterae
6
T. letters, characters; 2. written message'). Although there are no separate forms for collective
numerals in English, a cardinal numeral on its own cannot be used with pluralia tantum
nouns, but rather in combination with the word pair (e.g. two pair(s) o f scissors vs. *two
scissors). On the other hand, many languages have a set of collective numerals which are used
withpluralia.tantum nouns (e.g. Bosnian dvoje makaze 'two pair(s) of scissors' vs. *dva/ dvije
makaze). In cases when a plural form may have two different meanings, like in the case of
Latin litterae, then different numeral forms are used depending on the intended meaning. So,
the phrase including a cardinal numeral duae litterae is unambiguous and may mean only 'two
letters, characters (e.g. b and p). On the other hand, when the same plural form is used to
express a singular meaning, that is the meaning 'written message', then only a collective
numeral may be used with such a noun, e.g. binae litterae.
Collective numerals may be adjectival, or substantival, or even adverbial. When
modifying pluralia tantum nouns, they are adjectival (e.g. dvoje makaze-NOM). However,
when used with the so-called collective nouns in some languages (e.g. Bosnian djeca
'children', pilad 'chicken', dugmad ’buttons'), then the following noun is in the genitive (e.g.
dvoje djece-GEN vs. *dvoje djeca-NOM). An example of an adverbial form of collective
numerals are Bosnian forms udvoje, utroje, etc. These forms are used to indicate joint actions,
usually expressed in English with together.
Distributive numerals may be expressed in various ways. It may be achieved by using
a cardinal numeral with some adverb like 'each', as in English (e.g. one each, two each ...). In
some languages distributive meaning is expressed by reduplication of the cardinal numeral
(e.g. in Hebrew). There are also cases of using an adposition with a cardinal numeral. For
example, a preposition po is used which assigns dative to the cardinal numeral, whereas the
noun that follows is in genitive (e.g. Russian po pjati rublej 'five rubles each').
1.5. Non-referential counting words and referential number words
Wiese (2003) makes a distinction between non-referential counting words and

referential number words (expressions for numerical concepts). Non-referential counting
words would be, for example, English 'five', 'fifth', 'number five', or corresponding Bosnian
'pet', 'peti', 'petica'. Referential number words would be those occurring in expressions like
'five books', 'the fifth student', bus number five', or corresponding Bosnian expressions 'pet
knjiga’, 'peti student', 'autobus broj pet' or 'petica'. Counting words are non-referential in the
sense that they do not refer to anything, but are employed as tools for expressing numerical
quantities (e.g. 'five'), ranks (e.g. 'fifth'), and labels (e.g. 'number five'). Referential number
words are used to denote these quantities, ranks, and labels, as in expressions like 'five books',
'the fifth student', or 'bus number five1. The number words used in these English expressions
are identical to counting words in form. They differ in referentiality: number words are
referential, whereas counting words are not referential, but rather instrumental.
Number words refer to numerical concepts, but they are not used as numbers. On the
other hand, counting words are used as numbers, but they do not refer to numbers. The
expression 'five books' contains the cardinal number word which is referential: it refers to a
cardinality. On the other hand, the counting word 'five' used as a part of a numerical sequence
in counting is not referential, but instrumental. It is used as a tool in counting.
Languages often have the same form used both as a counting word and a referential
number word. For example, the same form 'five' in English is used in three ways:
1. as the counting word 'five' which occurs as one element in a counting sequence 'one, two,
three, four, five,...', and does not refer to anything;
2. as the cardinal number word 'five' which is used to refer to the cardinality of a set as in the
expression 'five books';
7
3. as the number word combined with the lexical item 'number' in expressions like 'bus
number five' with a nominal meaning.
In contrast to English with one form employed in three different functions, there are
languages which have different forms for different functions, so that counting words differ
from referential number words in form. This is especially the case with smaller numbers.
Wiese (2003) mentions Cantonese and Hebrew as examples of such languages. So, the form
yih is used in Cantonese as the counting word 'two', whereas the form leuhng is used as the
cardinal number 'two' which modifies a noun (e.g. leuhng gauh 'two cakes'). In Hebrew, the
counting word 'two' is shtdyim, but it can also be used as a nominalised number word. On the
other hand, the forms shney (with masculine nouns) and shtey (with feminine nouns) are used
as the cardinal number words which modify a noun, as in expressions like shney tapuchim
'two apples' or shtey bananot 'two bananas'.
Languages usually have one set of counting words, whereas they may have different
kinds of referential number words. In case a language has different cardinal and ordinal
number words, then the cardinal number words are often identical or similar in form to.
counting words. However, counting words do not appear in different inflectional forms,
whereas there are languages in which cardinal number words may have inflectional forms if
the noun phrase in which they occur takes a specific syntactic position. But in general, there is
similarity (often identity) between the counting words and cardinal number words. On the
other hand, the lexical items used to express ordinal or nominal number concepts are usually
derived items (e.g. 'tenth' in English is derived by suffixation: 'ten' + -th > 'tenth', and
corresponding Bosnian 'deset' + -i > 'deseti'), and sometime may include additional lexical
items (like 'number' in English to express a nominal number concept in 'bus number five').
When counting words and cardinal number words differ, it is mostly in the domain of
small numbers. So for example, the cardinal number word leuhng 'two' in Cantonese is not
derived from a corresponding counting word yih. In the same way, the ordinal number words
may also have suppletive forms in the domain of small numbers. For example, ordinal
numerals in Bosnian are derived from cardinals by -i suffixation, starting from 'five': pet 'five'
+ -i > peti 'fifth'. However, smaller ordinals have suppletive forms: jedan 'one' but prvi 'first',
dva 'two' but drugi 'second', tri 'three' but treći 'third', četiri 'four' but četvrti 'fourth'. In
English also ordinals 'first' and 'second' are not derived from corresponding counting words.
Wiese (2003) points out that it seems that there are no parallel examples of suppletive
forms within the nominal number words. Wiese (2003: 270) suggests that 'a reason why there
are no special suppletive forms for nominal number words might be that there seem to be no
particular (say, morphologically derived) nominal number words in the first place'. Instead of
a separate nominal number words, complex constructions are used which often involve
additional lexical items like 'number' in the English '(bus) number three'. However, Bosnian is
a language which actually can express this type of meaning by a separate nominal number
word. So, corresponding to the English example, a complex construction may be used, as in
(autobus) broj tri '(bus) number three', but nominal number words may be used to express the
same meaning: trica or trojka. So for example '(bus) number one/ two/ three...' would be
simply: jedinica, dvica/ dvojka, trica/ trojka, četvrtica/ četvorka, petica, šestica, etc.
1.6. Different kinds o f referential number words
Languages employ various kinds of number words to refer to various numerical

concepts. In English, three number constmctions may be distinguished: a construction with a
cardinal number word ('five buses'), a construction with an ordinal number word ('the fifth
bus'), and a 'number-construction ('bus number five'). However, in addition to these, there is
at least one additional number construction in Bosnian: a construction with a collective
number word (petoro djece 'five children').
8
There is one-to-one relation between cardinal and ordinal number words and
corresponding cardinal and ordinal concepts, so that constructions with cardinal number
words express concepts of numerical quantity, while constructions with ordinal number words
express concepts of numerical ranks. Number-constructions typically express concepts of
numerical label, as in 'bus number five'. However, such constructions may be used also to
refer to numerical ranks, and thus have ordinal, rather than nominal interpretations, as in the
following examples from Wiese (2003: 271):
(1) a. You are the number one in my life.

b. She is the number one tennis player in the world.
It should be pointed out that special nominal number words in Bosnian cannot be used
with ordinal interpretation, so that in constructions corresponding to (1) only structures with
'number' + cardinal numerals may be used. Bosnian translations of sentences in (1) are given
in (2):
(2) a. Ti si broj jedan u mom životu. vs. *Ti si jedinica u mom životu.
b. Ona je svjetska teniserka broj jedan. vs. *Ona je svjetska teniserka jedinica.
Consequently, we may conclude that there is no one-to-one relation between 'number-

constructions and concepts of numerical label in English. On the other hand, Bosnian nominal
number words do have one-to-one relation between number constructions and concepts of
numerical label. Therefore, we conclude that nominal number words in Bosnian are
specialised for nominal reference exclusively. On the other hand, English has ordinal number
words which are specialised for ordinal reference. There are languages, however, which lack
ordinal number words. For example, Cantonese has two types of 'number-constructions, one
being used with ordinal reference, and the other is used with nominal reference.
1.7. The grammatical properties o f referential number words
Referential number word constructions exhibit grammatical parallels to some non-

numerical expressions. Wiese (2003: 275) points out that cardinal number words behave
similarly to quantifier expressions like 'many' or 'few', as illustrated by the English examples
in (3) where the quantifier 'many' and the cardinal number word 'five' are combined with the
same noun:
(3) many apples/ five apples.
We find parallel examples in Bosnian:
(4) mnogo j abuka/ pet j abuka

many apples-GEN PL/ five apples-GEN PL
Notice that the norm jabuka 'apple' appears in the genitive plural form both after the
quantifier mnogo 'many' and pet 'five'. It means that these lexical items syntactically behave as
belonging to the same class - they are both quantifiers. Quantifiers in Bosnian assign genitive
case to their plural nominal complement. However, not all cardinal number words in Bosnian
behave as quantifiers.
Of course, the noun in (3) may be preceded by an adjective, so some authors suggest
that cardinal number words manifest an adjective-like behaviour (cf. Corbett (1978), Hurford
9
(1987)). In Bosnian the only cardinal number word that is completely adjectival is 'one'. It has
the adjectival declension and agrees with the noun in gender, number and case:
(5) a. jedna jabuka

one-FEM SG NOM apple-FEM SG NOM
b. jedan orah
one-MASC SG NOMnut-MASC SG NOM
c. jedno govedo
one-NEUT SG NOM head of cattle-NEUT SG NOM
But other cardinal number words (like pet 'five' in (4)) cannot be treated as adjectives,
since they don't agree with the noun in gender, number and case. The same form is used
regardless of the gender of the noun that follows:
(6) pet jabuka/ oraha/ goveda

five apples-FEM GEN PL nuts-MASC GEN PL heads of cattle-NEUT GEN PL
The same forms of nouns are found with quantifiers like mnogo 'many':
(7) mnogo jabuka/ oraha/ goveda

many apples-FEM GEN PL nuts-MASC GEN PL head of cattle-NEUT GEN PL
Therefore, we could conclude that cardinal number words like pet 'five' behave
syntactically in the same way as quantifiers. Wiese (2003) points out some parallels between
quantifiers and cardinal number words. For example, they occur in the same position,
preceding adjectives and nouns, as in (8):
(8) a. mnogo malih crvenih jabuka

many small-GEN PL red-GEN PL apples-GEN PL
b. pet malih crvenih jabuka

five small-GEN PL red-GEN PL apples-GEN PL
Both in English and Bosnian quantifiers and cardinal number words occur in front of
the adjectives. And this is, according to Greenberg (1966a), a general property of languages:
cardinal number words and quantifiers do not occur between the adjective and the noun.
Also, both quantifiers and cardinal number words allow so-called 'quantifier float',
appearing in positions removed from the noun phrase:
(9) Pisama sam dobio mnogo/ pet.

letters-GEN PL (I) received many/ five
'Letters I received many/ five'
But not all cardinal number words behave syntactically as quantifiers. There is a
tendency in world languages that higher cardinal number words mainfest a noun-like
behaviour, as in English 'millions of apples' (cf. Menninger (1958), Corbett (1978)). In
Bosnian the cardinal number words stotina ’hundred', hiljada 'thousand', tisuća 'thousand',
milion 'million', milijarda 'billion', etc. behave syntactically as nouns. They have full nominal
declensions. Parallel to this, there are also some expressions with quantificational force which
10
syntactically behave as nouns rather than quantifiers, as for example in English nominal
constructions like 'a huge amount of apples'. In Bosnian in addition to the quantifier mnogo
'many' there is a corresponding nominal expression mnoštvo. In contrast to mnogo which has
only one form, mnoštvo has the full nominal declension.
On the basis of examples of cardinal number words in Bosnian, like jedan/ jedna/
jedno 'one', pet 'five', stotina 'hundred', we can conclude that the grammatical behaviour of
cardinal number words is not idiosyncratic: some behave as adjectives (e.g .jedan), others are
quantifiers (e.g. pet), and still others have nominal characteristics (e.g. stotina).
On the other hand, ordinal number words usually manifest adjectival characteristics,
more specifically, characteristics of superlative adjectives. For example, they usually occur in
front of all other modifiers in definite noun phrases, that is in the same position as
superlatives:
(10) the fifth American president/ the oldest American president
In Bosnian, ordinal number words are fully adjectival - they have the adjectival
declension and agree with the head noun in gender, number and case:
(11) a. peti amerifiki predsjednik

fifth-MASC SGNOM American-MASC SG NOM president- MASC SG NOM
b. peta američka predsjednica

fifth-FEM SGNOM American-FEM SG NOM president-FEM SG NOM
c. peto američko dijete

fifth-NEUT SG NOM American-NEUT SG NOM child-NEUT SG NOM
The connection between ordinal number words and superlatives is documented in

Proto-Indo-European in the sense that the same suffix was used for deriving superlative forms
from the adjectival stem, and for deriving some ordinal number words from the numeral stem.
The remnants of this system are still visible in some Indo-European languages. For example,
the same suffix is used in German for both superlative forms and ordinal number words (e.g.
schon-st 'prettiest', drei/3ig-st 'thirtieth'). Notice also similarity in Bosnian between treći 'third'
< tret-ji and the comparative form žuči 'yellower' < žut-ji.
Wiese (2003) singles out the third class of number words, the so-called 'number'-
constmctions, pointing out that these constructions behave as proper names:
(12) ship number five/ ship Titanic
This example shows that the 'number'-item and the number word ('number' + 'five')
together behave like a proper name, so that 'number five' may be replaced by a proper name
like 'Titanic'. The construction in (12) has several possible variants: 'the number five ship', 'the
number five', 'ship five', 'the five'. But there is no separate number word that would be used
instead of these constructions in English. On the other hand, there are special number words
that are used in such constructions in Bosnian. For example, English 'tram number five' would
be simply petica 'five'. The corresponding 'number-construction is also used (tramvaj broj
pet), but a special number word with nominal function (e.g. petica) is preferred. Such number
words have full nominal declension, but like proper names, these number words don't take
complements.
In conclusion, number words syntactically do not behave as a unique class, but rather
share grammatical properties with other classes of words. So cardinals behave mainly as
11
quantifiers (although some cardinals manifest adjectival and nominal behaviour), ordinals
manifest adjectival behaviour, whereas 'number'-constructions behave as proper names.
Consequently, in generative grammar cardinal number words are usually analysed as
quantifiers, that is as heads of a functional projection QP above NP, whereas ordinal number
words are treated as adjectives, that is as heads of an adjective phrase AP which is integrated
into the NP (cf. Abney 1987, Giusti 1991).
1.8. Greenberg's Universals relating to numerals
Among typological studies of numerals, those by Greenberg were most influential. In

a number of his studies (cf. Greenberg 1966a, 1966b, 1978, 1989), various language
universals were proposed, including those that are directly, or indirectly, related to numerals.
In one of the first papers dealing with this topic, a paper presented at a conference in
1961 (Greenberg 1966a), he proposed 45 universals of grammar, paying special attention to
the morpheme and word order, based on the investigation of the thirty-language sample of the
world's languages (including European, African, Asian, Oceanian, and American Indian).
Tree of these universals are dealing directly, or indirectly, with numerals: Universals
18, 19, and 20 (Greenberg 1966a: 86-87). It is pointed out that the position of numerals (and
demonstratives) is related to that of descriptive adjectives in world's languages in the sense
that numerals usually precede descriptive adjectives. We can illustrate this with examples
from Bosnian, where the order in which a numeral follows the descriptive adjective is
ungrammatical, as in (13b):
(13) a. pet pametnih studenata

five clever-GEN PL students-GEN PL
'five tall students'
b. *pametnih pet studenata
In grammatical (13a), both the numeral and the descriptive adjective precede the noun. In
those languages, however, where the descriptive adjectives may appear after the noun,
nevertheless the numerals manifest a marked tendency to appear in front of the noun. For
example, in Bosnian, single adjectives do not appear in postnominal position, but adjective
phrases, or coordinated single adjectives, may have the position after the noun. In such a case,
the numeral cannot appear after the noun, but may only precede it, as in (14b) and (14e):
(14) a. pet pametnih i vrijednih studenata

five clever-GEN PL and diligent-GEN PL students-GEN PL
'five clever and diligent students'
b. pet studenata, pametnih i vrijednih
c. *studenata pet pametnih i vrij ednih
d. pet veoma pametnih studenata

five very clever-GEN PL students-GEN PL
'five very clever students'
e. pet studenata, veoma pametnih
f. * studenata pet veoma pametnih

12
Taking into account the facts illustrated with examples in (14), it is not surprising that in those
languages where the descriptive adjective precedes the noun, then the numerals (and
demonstratives) also always precede the noun. On the basis of this, the following universals
were formulated:
1. When the descriptive adjective precedes the noun, the demonstrative and the
numeral, with overwhelmingly more than chance frequency, do likewise (Greenberg's (1966a:
86) Universal 18).
2. When the general rule is that the descriptive adjective follows, there may be a
minority of adjectives which usually precede, but when the general rule is that descriptive
adjectives precede, there are no exceptions (Greenberg's (1966a: 87) Universal 19).
The examples in (14b) and (14e) are not counterexamples to Greenberg's Universal 19,
because a single descriptive adjective in Bosnian can never appear in postnominal positon, so
that the example in (15) is ungrammatical:
(15) *pet studenata pametnih
Compare (15) with (13a), the only acceptable order.

Greenberg (1966a: 87) also emphasises that the order of elements within the noun
phrase is usually fixed, in the sense that if all of the three types of noun modifiers precede the
noun, then the order is always the same: demonstrative, numeral, adjective, noun (e.g. those
five clever students). It is not surprising that such an order is a norm in so-called fixed word
order languages, like English, but we could expect deviations from this order in so-called free
word order languages, like Bosnian. The fact is that there are no deviations. Compare
examples in (16) from Bosnian:
(16) a. onih pet pametnih studenata

those-GEN PL five clever-GEN PL students-GEN PL
'those five clever students'
b. *pet onih pametnih studenata
However, the fact that we find the genitive plural (onih) in (16a) and not the expected
nominative plural form (oni) of the demonstrative shows that the demonstrative is first
merged with the NP pametni studenti 'clever students', as in oni pametni studenti 'those clever
students'. Then the numeral pet 'five' is merged to this complex and the numeral assigns the
genitive case to the following NP, so we get the form onih pametnih studenata. However, the
demonstrative cannot stay in this position, because it has to check additional features,
presumably [Ref] features (Referential features), and so it has to move to a higher projection,
above the projection that is hosting numerals. Consequently, we get the order, as in (16a).
On the other hand, in those languages where these modifiers (demonstratives,
numerals, and adjectives) appear after the noun, then we find the opposite order: noun,
adjective, numeral, demonstrative. However, there is a less frequent alternative in which the
noun is followed by modifiers in the same order as in the case of modifiers preceding the
noun, that is the order: noun, demonstrative, numeral, adjective. So, such an alternative
postnominal order appears, for example in Kikuyu, a Bantu language of East Africa, but such
an order is not so frequent. On the basis of these observations, the following universal was
formulated:
3. When any or all of the items (demonstrative, numeral, and descriptive adjective)
precede the noun, they are always found in that order. If they follow, the order is either the
same or its exact opposite (Greenberg's (1966a: 87) Universal 20).
13
These are three universals related to numerals that Greenberg proposed in his paper on
univerals of grammar in general. Subsequently, Greenberg proposed a number of
generalizations concerning numeral systems in world's languages in a separate study dealing
exclusively with numerals (Greenberg 1978). Here, I will single out and discuss those
generalizations that are most interesting from the point of view of our investigation.
Greenberg first studied cardinal numerals in the attributive constructions and proposed
54 generalisations. The three initial generalisations are related to the scope of cardinal
numeral systems:
4. Every language has a numeral system of finite scope (Greenberg's (1978: 253)
generalisation 1).
This means that every language has a defined set of numbers that can be expressed in
that language. For example, in American English that set ends with the number 1036 because
this number has the corresponding numeral expression, that is the lexical item decillion. Of
course, there are numbers larger than this, but there is no lexical item in American English
that would designate such a number. So, it is important to make the distinction between a
number and a numeral expression. For example, twelve is a numeral expresion for the number
'12' in English. But the same number may be expressed using a numeral expression dozen.
The same is true in Bosnian: both dvanaest and tuce may be used to designate '12'. On the
basis of such examples we can conclude that the set of numeral expressions is broader than
the set of cardinal numerals. However, the cardinal numeral system of a given language is
generated by the act of counting. The square root o f sixten, the square o f four, dozen, score,
etc., are also numeral expresions, but they are not included in the cardinal numeral system of
English because such expressions do not appear in the act of counting. The corresponding
numeral expressions in Bosnian (kvadratni korjen od šesnaest, četiri na kvadrat, tuce) also
are not a part of the cardinal numeral system of Bosnian. The numerals used in counting and
cardinal numerals used with nouns are usually identical in form. Few languages have
numerals used in counting different in form from the corresponding cardinal numerals.
Greenberg (1978: 253) defines 'the limit number L for each (numeral) system as the next
largest natural number after the largest expressible in the system. The reason for adding 1 is
that this will often give us a convenient round number.'
5. Every number n (0<n<L) can be expressed as part of the numeral system in any
language (Greenberg's (1978: 254) generalisation 2).
Here it should be pointed out that every language has the largest number expressible
(e.g. American English decillion), and this number may differ from language to language, and
all numbers lower than the largest number in that language (and larger than 0) can be
expressed. In other words, there is a continuity in expressing numbers.
6. Zero is never expressed as part of the numeral system (Greenberg's (1978: 255)
generalization 3).
This is obvious, since the numeral system is based on counting which begins with
'one'. It would be unusual to have a system in which a set without members is counted. Of
course, languages have expressions denoting 'zero', but these expressions cannot be regarded
as a part of the numeral system. For example, in Bosnian nula 'zero' is not a part of the
numeral sytem. It should be pointed out that nula is a noun of feminine gender with full
declension in singular and plural, and it can be counted (one zero / two zeros,...), as in (17):
(17) a. Taj broj ima pet nula.

that number has five zeros-GEN PL
'That number has five zeros',
b. Velika nula je napisana na tabli.
big zero is written-FEM 3P SG on blackboard
'A big zero is written on the blackboard.'
14
c. Velike nule su napisane na tabli.

big zeros are written-FEM 3P PL on blackboard
'The big zeros are written on the blackboard.'
The generalizations that follow take into account the arithmetical operations that are
employed in various numeral systems. Some languages have numeral systems which do not
employ arithmetical operations or use only addition.
7. In every numeral system some numbers receive simple lexical representation
(Greenberg's (1978: 255) generalization 4).
For example, in Bosnian the number '3' receives simple lexical representation: tri. On
the other hand, the cardinal numeral trideset tri expresses the number '33' as a function (a x b)
+ c in which the argument a has the value 3, b has the value 10, and c has the value 3: (3 x 10)
+ 3 = (tri x deset) + tri = (trideset) + tri = trideset tri. Every numeral system has at least one,
but frequently more numerals, which receive simple lexical representations. In Bosnian, the
following cardinal numerals have a simple lexical representation: jedan (1), dva/ dvije (2), tri
(3), četiri (4), pet (5), šest (6), sedam (7), osam (8), devet (9), deset (10), sto/ stotina/ stotinu
(100), hiljada/ hiljadu/ tisuća/ tisuću (1,000), milion (1,000,000), milijarda/ milijardu
(1,000,000,000), bilion (1,000,000,000,000), trilion (1,000,000,000,000,000). In Greenberg's
terminology, such numerals represent the atoms of the numeral system, and he distinguishes
simple atoms and bases. Simple atoms in Bosnian would be jedan, dva, dvije, tri, četiri, pet,
šest, sedam, osam and devet.
Typology of numeral systems distinguishes those that consist only of atoms, and those
that employ functions such as addition or multiplication. There are languages from South
America, Australia, New Guinea and South Africa with numeral systems consiting only of
atoms, but most languages employ at least addition in their numeral systems.
8. The largest value of L in systems with only simple lexical representation is 5 and
the smallest is 2 (Greenberg's (1978: 256) generalization 6).
Greenberg mentions Guana, an Arawakan language, with the system 1, 2, 3, 4, 'many'
as an example of the language with L = 5, and Botocudo, a Macro-Ge language in Brazil, with
the system 1, 'many' as the only example of the language with L = 2.
9. The smallest value for L in systems with arithmetical operations is 4 (Greenberg's
(1978: 256) generalization 7).
Greenberg mentions Port Essington, in Tasmania, with the system 1, 2, 2+1, 'many', as
an example of a language employing addition, and yet having a very limited numeral system.
On the other hand, Mullukmulluk in Australia and Aghu in New Guinea are languages with
numeral systems in which counting up to 30 is possible and addition is the only arithmetical
operation employed in their numeral systems.
Most languages have numeral systems which employ other arithmetical operations,
not only addition. The generalizations that follow are related to such numeral systems.
10. Of the four fundamental arithmetical operations - addition and its inverse,
subtraction, and multiplication and its inverse, division - the existence of either inverse
operation implies the existence of both direct operations (Greenberg's (1978: 257)
generalization 9).
This generalisation implies that inverse operations - subtraction and division - have
the marked status.
11. The existence of multiplication implies the existence of addition (Greenberg's
Greenberg points out that this is a near universal. He found just one example of a
numeral system with multiplication and without addition, and that is one subgroup of the
Yuman languages with the numeral systems in which L = 11 and the numerals may be
15
analyzed as 1, 2, 3, 4, 5, 3x2, 7, 4x2, 3x3, 10. On the other hand there are many numeral
systems with addition and without multiplication.
In addition to the four fundamental arithmetical operations that are used in the numeral
systems, some languages also employ an operation which is called the 'going-on' operation by
Greenberg (1978: 258) and 'overcounting' by Hurford (1975: 235). For example, in Ostyak, a
Finno-Ugric language, 18 is expressed as 8, 20, that is, '8 going-on 20'.
12. Subtraction is never expressed by the mere sequence of the subtrahend and
minuend (Greenberg's (1978: 258) generalization 11).
In the arithmetic operation of subtraction, the subtrahend is the number subtracted, the
minuend is the number from which subtraction takes place, and the remainder is the result. In
contrast to subtraction, simple juxtaposition is often found with numerals involving addition
or multiplication (e.g. in Bosnian 30 is trideset = tri 'three', deset 'ten').
13. Division is always expressed as multiplication by a fraction. Only units or
multiples of units are dividends, and the denominator of the fraction is always 2 or a power of
2 (Greenberg's (1978: 261) generalization 16).
Greenberg points out that almost all examples of division are examples of the number
50 expressed as 'A x 100, usually in a vigesimal system.
Most languages employ addition and multiplication in their numeral systems. In the
arithmetic operation of addition, a distinction is made between the augend, that which is
added to, and the addend, that which is added. The result is the sum, and any term of the sum
is called a summand.
In the arithmetic operation of multiplication, a distinction is made between the
multiplicand, that which is multiplied, and the multiplier, that which multiplies. The result is
the product, and the term 'factor' is used for both the multiplicand and the multiplier.
In the complex numerals we have a combination of several numbers to which some
arithmetic operations are applied. For example, in Bosnian the number 509 is expressed as the
complex numeral petsto devet, or alternatively, pet stotina (i) devet. This numeral is analyzed
as containing the arguments 5, 100, 9. The arithemtical function is (a x b) + c, where a = 5, b
= 100 and c = 9. The component numbers inside complex numerals may appear in different
phonological forms, as in Bosnian: četrnaest (14) vs. dvadeset četiri (24). The form četr- in
14 differs form četiri in 24. There are also examples of suppletion (e.g. in Chrau, an Austro-
Asiatic language, 2: var; 10: mat; 20: var jat, and not *var mat, so ja t is a suppletive alternant
for 10), and portmanteau expressions (e.g. in Russian, 40: sorok).
The arithmetic operations involved in the complex numeral expressions may be
represented by overt morphemes. Frequently, complex numerals contain the word or affix
meaning 'and' or 'with' which designates the operation of addition. Greenberg (1978: 264)
calls this a comitative link.
The operation of addition may be designated also by a word or affix meaning 'upon'.
Greenberg (1978: 265) calls this a superessive link. He gives an example from Old Church
Slavonic where 11 was jedinu na desete 'one on ten'. This origin of numerals 11-19 is still
visible in Bosnian. For example, the number 11 is expressed as jedanaest, derived from jedan
na deset 'one on ten'. This origin is not so obvius for native speakers of Bosnian, although it is
sometimes preserved even in pronunciation (especially among older generation) so that the
sound [n] in jedanaest is geminated in pronunciation of some native speakers: [jedannest]. A
rather rare way of expressing the operation of addition is by the possessive link, like in
Quechua 11 which is expressed as 'ten one-having', i.e. 'ten' which possesses a 'one'. Other
modes of expressing addition are those with complex numerals which involve the meaning 'to
be extra', 'to be added'. Sometimes complex numerals contain a form meaning 'to be left',
'remain', as in Anglo-Saxon twa-lif'two remain' for 12, the origin of Modem English twelve.
16
r
Finally, another important operation is serialization. For example, the numbers 21, 22,
etc. are expressed in English as twenty-one, twenty-two, etc., that is, as (2 x 10) + 1, (2 x 10) +
2, etc. Therefore, 20 is the augend by serialization, and 1, 2, etc. are addends.
The operation of serialization is a part of the following generalizations regarding
augends:
14. Every superessive or possessive augend is a serialized augend (Greenberg's (1978:
266) generalization 17).
For example, in Bosnian a superessive augend deset (10) was serialized. However, in
the development of the language; the original form used for serialization was reduced and
transformed: deset gave -est, so today we have the following cardinal numerals: jedanaest
(11), dvanaest (12), trinaest (13), četrnaest (14), petnaest (15), šesnaest (16), sedamnaest
(17), osamnaest (18), devetnaest (19).
15. A serialized augend is always larger than its addends (Greenberg's (1978: 266)
generalization 18).
In Bosnian, a serialized augend deset (10) is larger than the largest addend devet (9).
The following generalization introduces the notion of grouping, a process that
becomes operative when there are more than two addends.
16. Whenever there are three or more summands and at least one is a product,
parenthesization starts by separating the summand with the largest numerical value from the
rest. The same mle then applies to the remainder, if it consists of more than two summands,
and so on (Greenberg's (1978: 267) generalization 19).
For example, this generalization applies to the number 3333 in Bosnian with the
summands 3 x 1,000, 3 x 100, 3 x 10, and 3. Since 3,000 is the largest value here, we first
parenthesize (3 x 1,000) + [(3 x 100) + (3 x 10) + 3]. Then we parenthesize within the second
member: (3 x 100) + [(3 x 10) + 3]. It should be pointed out that this analysis is supported by
the fact that in Bosnian multiples of 10 and 100 are written as single words having one word
accent (trideset (30), tristo (300)), whereas multiples of 1,000 are written as two separate
words, with separate word accents (tri hiljade (3,000)).
In multiplication, we have a serialized multiplicand. According to Greenberg (1978:
269) a serialized multiplicand is a number whose successive multiplication by at least two
other numbers results in serialized products which may be expresed as simple lexemes or as a
product of the multiplicand and multiplier.
The operation of serialization is a part of the following generalizations regarding
multiplicands:
17. All adverbially or partitively expressed multiplicands are serialized multiplicands
For example, in Classical Greek the number 2,000 is expressed as dis-khilioi 'twice-
thousand'. So, instead of the cardinal numeral 'two', the numerical adverb dis 'twice' is used.
A serialized multiplicand is a base. Greenberg (1978: 270) concludes that a numeral system
which doesn't use multiplication and addition cannot have a base. Numeral systems which do
have a base, usually have more than one base (e.g. in Bosnian 10; 100; 1,000; 1,000,000). The
smallest base is called the fundamental base. Greenberg points out that there are only four
numbers which may be considered as fundamental bases in numeral systems. In order of
frequency, these are 10, 20,4, and 12.
The number 1 has its role in addition, but hardly so in multiplication. However, it is
sometimes expressed even in multiplication. Of course, multiplication of any number by 1 is
redundant, and therefore it is not usually expressed. Nevertheless, the following
generalization does hold:
18. Only a base is ever multiplied by 1 (Greenberg's (1978: 271) generalization 25).
Greenberg (1978: 272) points out that many languages may express the number 100
(and 1,000, etc.) in two ways, like in English 'one hundred' and 'a hundred'. Greenberg
17
concludes that the expressions like 'one hundred', 'one thousand', etc. are possible because
bases which are often called units are in effect being counted, hence one starts with 'one'.
However, the examples from Bosnian show that this conclusion is not correct. What is crucial
for the use of 1 as a multiplier is the morphological form and syntactic status of the base.
Only if the base behaves as a noun, then 1 may be used as a multiplier, as in Bosnian
examples in (18):
(18) deset (10) *jedan deset/ *jedna deset/ *jedno deset

sto (100) *jedan sto/ *jedna sto/ *jedno sto
stotina (100) jedna stotina
hiljada (1,000) jedna hiljada
tisuća (1,000) jedna tisuća
milion (1,000,000) jedan milion
milijarda (1,000.000.000) jedna milijarda
The numeral 'one' cannot be used as a multiplier with deset and sto because these forms
behave as quantifiers, not nouns. On the other hand, stotina, hiljada, tisuća, milion, milijarda
are nominal forms, and therefore 'one' may be used as a multiplier with these forms.
The generalizations that follow are concerned with the order of elements in numeral
expressions.
19. If in a language, in any sum the smaller addend precedes the larger, then the same
order holds for all smaller numbers expressed by addition (Greenberg's (1978: 273)
generalization 26).
For example, in Bosnian 19 is expressed as devetnaest (derived from devet na deset)
where 9 precedes 10, and all the smaller numbers, that is numbers from 11-19, are expressed
with the smaller number preceding 10.
20. If in a language, in any sum the larger addend precedes the smaller, then the same
order holds for all larger numbers expressed by addition (Greenberg's (1978: 273)
generalization 27).
For example, in Bosnian 21 is expressed as dvadeset jedan, where 20 precedes 1, and
all the larger numbers are expressed following this model. Greenberg's generalizations 26 and
27 assume that the constmction with the smaller preceding the larger is found in the smaller
numbers when a language has numerals with both orders. So, in Bosnian smaller numbers 11-
19 have the order with the smaller preceding the larger, whereas larger numbers, from 21
onwards, have the order with the larger preceding the smaller.
Greenberg (1978: 274) points out that languages with the order in which the larger
precedes the smaller are extremely common, whereas languages in which the smaller always
precede the larger are extremely rare (e.g. Malagasy), when the numerals are formed only by
addition.
21. If there are any numerals in which the expression of the multiplier follows that of
the multiplicand, the language is one in which the numeral follows the noun (Greenberg's
Greenberg (1978: 275) points out that the order with the multiplier preceding the
multiplicand is highly favored in world languages. This was noted also by Štampe (1976).
Bosnian is a language in which the multiplier precedes the multiplicand (e.g. 30: trideset
'three ten') and the numeral precedes the noun (e.g. tri knjige 'three books').
22. A link for addition is never initial in a numeral (Greenberg's (1978: 276)
generalization 30).
23. If a link for addition is final, the language is postpositional (Greenberg's (1978:
276) generalization 31).
18
24. If a link for addition occurs medially, it always goes with the following numeral in
a prepositional language and with the preceding numeral in a postpositional language
Addition in complex numerals is syntactically similar to nominal coordination. They
differ, however, because languages have overt markers of nominal cooordination, whereas
many languages form complex numerals without overt coordination, having only a simple
juxtaposition (e.g. in Bosnian 33: trideset tri 'thirty three'). In languages with overt
coordinators in complex numerals, the word for 'and' is usually a coordinator, but in many
languages it may be the same word as for 'with', so it means that it behaves as an adposition.
Also, in languages having the superessive link 'upon', it is an adposition (e.g. the Bosnian
superessive link na in numerals denoting numbers from 11-19 (e.g. sedamnaest '17') has a
prepositional origin).
The generalizations that follow are concerned with some characteristics of bases.
25. If 1 is expressed as a multiplier with a particular base, it is expressed with all
higher bases (Greenberg's (1978: 278) generalization 35).
As mentioned earlier, this generalization depends on the morphological form and
syntactic status of the base in languages like Bosnian: if the base behaves as a noun, 1 may be
used as a multiplier, but if the base behaves as a quantifier, 1 cannot be used (e.g. 10: *jedan
deset; 100: *jedan sto; 100: jedna stotina; 1,000: jedna hiljada/ tisuća; 10,000: *jedna deset
hiljada; 100,000: *jedna sto hiljada; 100,000: jedna stotina hiljada; 1,000,000: jedan milion;
10,000,000: *jedan deset miliona).
26. The only numeral expressions deleted are those for 1 and for bases of the system
In complex numerals involving the operation of subtraction, the numeral 'one' as a
subtrahend is often deleted. In complex numerals involving the operation of multiplication,
the numeral 'one' as a multiplier with bases is frequently omitted, and this omission may be
regarded as involving deletion (e.g. in Bosnian 100: jedna stotina, or stotina; 1,000,000: jedan
milion, or milion).
The generalizations that follow are concerned with some general organizing principles
of numeral systems.
27. In systems with more than one base, there is a base, the base for predictable
expression, above which in all numerals certain regularities hold. Such numerals, when
analyzed into their two principal constituents, will fall into two types, simple and complex. In
the simple numeral we have a product, or rarely a quotient. In the complex numeral the two
constituents are summands (Greenberg's (1978: 280) generalization 38).
For example, a simple numeral in Bosnian is šest stotina 'six hundred', whereas šest
stotina trideset tri 'six hundred and thirty-three' is a complex numeral. The principal
constituents of this complex numeral are šest stotina 'six hundred' and trideset tri 'thirty-three'.
In the simple constituent the factors or elements in a division are expressed in the
same way as when they occur in isolation.
For example, in Bosnian numeral šest stotina 'six hundred', the factors šest and stotina
are also found in that form in isolation. On the other hand, that is not the case with the
numeral šezdeset 'sixty': the factor šez does not appear in the form šest. Therefore, the base for
predictable expression in Bosnian is stotina 'hundred', not deset 'ten'.
Complex expressions fall into two parts, a product and a remainder.
For example, in Bosnian šest stotina trideset tri 'six hundred and thirty-three', šest
stotina 'six hundred' is a product and trideset tri 'thirty-three' is a remainder.
The remainder never has a larger value than the next lower base of the total
expression.
19
For example, in šest stotina trideset tri 'six hundred and thirty-three1, the value of the
remainder trideset tri 'thirty-three' is 33, and that is smaller than 100, which is the next lower
base of the total expression šest stotina trideset tri (633).
Finally, the remainder is expressed by the same mathematical function as when it
occurs in isolation.
For example, in šest stotina trideset tri 'six hundred and thirty-three', the remainder
trideset tri 'thirty-three' is expressed in the same way as in the numeral expression for 33 in
isolation. On the basis of these examples, we can conclude that 100 is the base for predictable
expression in Bosnian. Greenberg (1978: 281) claims that the base for predictable expression
is usually 100 in decimal systems.
28. The degree of morphological fusion varies inversely with the size of the numerical
value (Greenberg's (1978: 281 generalization 39).
In order words, more morphological irregularities are found with smaller numerals in
various numeral systems. In most languages the numerals denoting the larger value are more
marked, and morphological regularity is usually a property of marked expressions. So in
Bosnian the numerals expressing numbers from 11-19 are irregular in the sense that deset in
these numerals does not apper in the same form as in isolation. For example, instead of
jedannadeset, 'one-upon-ten' we have jedanaest (11). In this group (11-19) additional
irregularities are manifested by četrnaest (14) (instead of expected četirinaest), and šesnaest
(16) (instead o f šestnaest). In the decades, we find the following irregularities: četrdeset (40)
(instead of četirideset), pedeset (50) (instead ofpetdeset), šezdeset (60) (instead of šestdeset),
and devedeset (90) (instead of devetdeset). In the hundreds, we find irregularities only when
the form sto is used instead of stotina: dvjesto (200) (instead of dvijesto), četirsto (400)
(instead of četiristo), šesto (600) instead of (šeststo). There are no irregularities with stotina
(100), hiljada (1,000), milion (1,000,000), milijarda (1,000,000,000), which is in accordance
with Greenberg's generalization.
The generalizations that follow are concerned with the syntax of numeral expressions.
There are languages, like Turkish, in which nouns have singular and plural forms, but
nevertheles, they appear in singular when combined with numerals.
29. In languages in which the expression of plurality is facultative in the norm, the
singular may be used with numerals designating numbers > 1 (Greenberg's (1978: 282)
generalization 40).
Greenberg points out that in most instances the use of the singular is compulsory.
30. In languages with singular/ dual/ plural systems in the noun, if the plural is used in
any instances where a set of two objects is designated, the plural may be used with the
numeral for 2 (Greenberg's (1978: 282) generalization 41).
For example, in Homeric Greek the plural was often used in contexts where the dual
would be expected, and such contexts include phrases with 2.
31. If numeral expressions for the smallest addends take the plural of the noun when
they designate numbers > 1, then complex numerals with 'one' as an addend will take the
plural of the noun if'one' is not a separate word (Greenberg's (1978: 283) generalization 42).
For example, English numerals two, three, etc. require a plural noun, e.g. two/ three/...
students. The numeral for 10+1 is eleven, and it also requires a plural noun: eleven students.
This is also the case with jedanaest T T in Bosnian: jedanaest studenata 'eleven students-GEN
PL'. However, when 'one' is a separate word inside a complex numeral, then such a complex
numeral may or may not require the singular noun. For example, in English the plural is
required (e.g. twenty-one students), whereas in Bosnian only singular is possible (e.g.
dvadesetjedan student 'twenty-one student-SG').
32. Where there is rule-governed variation between the use of the singular and plural
with numerals, the use of the singular is favored with higher numbers, in measure
20
constructions, in indefinite constructions, and with nouns which are inanimate or impersonal
For example, in Modem Arabic dialects the plural form of the noun is used after
numerals 2 to 10 (or 3 to 10 in dialects with dual), whereas the singular form is used after
numerals larger than 10. In Amharic, either the singular or plural form is used with animate
nouns after numerals 2 to 99, and exclusively singular form with animate nouns following
numerals larger than 99. On the other hand, the singular or plural form is used with inanimate
norms following numerals 2 to 9, and only the singular form after numerals larger than 9. In
Tlappanec, a language of Nicaragua, personal nouns appear in the plural form when combined
with numerals, whereas impersonal nouns have the singular form in combinations with
numerals. In Akkadian, the singular form of nouns is used with measure expressions, and the
plural form with countable nouns. In Armenian and Ewe, the singular noun appears in
indefinite numeral constructions, whereas the plural form is used in definite constructions.
There is variation in the use of the singular or plural form of the noun with numerals in
Bosnian also, but it is not related to the higher or lower value of the number. It depends
exclusively on the syntactic status of the numeral. All numeral expressions that end in jedan,
jedna (FEM), jedno (NEUT) (1, 21,... 91, 101, 121, ...191, 1,001, etc.) maybe followed only
by a singular noun, as in (19a); all numeral expressions that end in dva, dvije (FEM), tri, and
četiri (2, 3, 4, 22, 23, 24,... 92, 93, 94, 102, 103, 104, etc.) require a special form of the noun
that follows, a form that looks like the genitive singular, but it is actually a remnant of the old
dual, which was lost in the development of the languages, except in contexts with dva, dvije,
tri, and četiri, as in (19b). Finally, the third form of the noun that is found in contexts with
numerals is the genitive plural form, as in (19c), and it appears in all contexts, except when
the noun follows numeral expressions with 1, 2, 3, and 4, as the only, or the final element of
the numeral expression (e.g., 21,22,23, 24).
(19) a. jedan/ dvadeset jedan/ sto jedan/ hiljadu jedan student

one twenty-one hundred and one thousand and one student-SG
b. dva/ tri/ četiri/ dvadeset dva/ sto dva studenta

two three four twenty-one hundred and one student-PAUC
c. pet/ šest/ sedam/ osam/ devet/ deset/ jedanaest/ ..devetnaest/ dvadeset studenata
five six seven eight nine ten. eleven nineteen twenty students-PL
So, in (19) the three forms of the noun appear: student, studenta, studenata. The singular form
student is found after the numeral jedan, and after any numeral expression ending in jedan.
The dual (or more properly, paucal) form studenta is found after the numerals dva, tri, and
četiri, and after numeral expressions ending in dva, tri, and četiri. Finally, the plural form
studenata is found after all other numeral expressions. Notice, that the plural form is required
also after the numerals jedanaest (11), dvanaest (12), trinaest (13), and četrnaest (14). It was
pointed our earlier that the origin of these numerals was the following: jedan na deset 'one-
upon-ten’, dva na deset, etc. In complex numeral expressions, the last item is the head of the
whole expression, determining the form of the noun that follows. In other words, jedanaest
behaves syntactically as deset, whereas dvadeset jedan behaves syntactically as jedan,
because jedan is the head of this complex numeral expression, and not dvadeset. Since deset,
and also jedanaest, behave as quantifiers, they take a plural noun as the complement and
assign genitive case to it. On the other hand, jedan, and also dvadeset jedan, behave as
adjectives and may modify only a singular noun. Numerals dva, dvije, tri, and četiri also
behave as quantifiers assigning the genitive case to the following noun which has neither
21
singular nor plural form, but rather a special paucal form, a remnant of the earlier dual
declension.
33. The order noun-numeral is favored in definite and approximative constmctions
There are two types of variation in the order of the noun and the numeral. First, there
are languages in which any numeral can either precede or follow the noun. Different orders
are related with different semantic or syntactic functions. Second, there are languages in
which some numerals precede and others follow the noun.
In Palestinian Arabic, numerals precede nouns, but when the order in which the
numeral follows the noun occurs, it is with the definite construction. In Russian, numerals
precede nouns, but the opposite order occurs with approximative constmctions (e.g., pjat'
karandašej ’five pencils1vs. karandašej pjat' ’about five pencils’).
In Bosnian, numerals always precede nouns, and approximative constructions do not
involve the opposite order of numerals and nouns. The preposition oko ’about’ is used to
express the concept of approximate number (e.g., oko pet olovaka ’about five pencils') in most
cases, but this concept may be also expressed by some other means. The form ending in -ak is
used only with numerals stotina, deset, and those ending in deset (dvadeset, trideset, etc.) to
express this concept: stotinjak 'about hundred', desetak 'about ten', dvadesetak 'about twenty',
etc. Finally, the approximative construction may involve using two numerals next to each
other: dvije-tri olovke 'two or three pencils', pet-šest olovaka 'five or six pencils'.
34. If a language has NG order in the possessive construction, it has QN order in the
partitive construction (Greenberg's (1978: 284) generalization 45).
NG order is the order with the genitive noun following the head noun in the possessive
construction (e.g., books o f the man). In such a case, the partitive constmction with the
numeral follows the same patem (e.g., five o f the books).
In Bosnian, the possessive construction is often expressed with a possessive adjective
preceding the head noun, but it may consist of the genitive noun phrase following the head
noun: knjige tog čovjeka 'books of that man'. If a possessor is indicated by a single noun, then
a structure with a possessive adjective is preferred: čovjekove knjige rather than knjige čovjeka
'books of the man'. And in the partitive constmction, the 'of phrase' follows the numeral: pet
od tih knjiga 'five of those books'
35. If a language has both partitive and adjectival QN constmctions, the smallest
number which employs the partitive is larger than the largest number which has the adjectival
constmction (Greenberg's (1978: 285) generalization 47).
This generalization points out to a preference for the partitive constmction with larger
numerals. Examples of such behavior are the following: In some Berber dialects, numerals
'one' and 'two' behave as adjectives, whereas other numerals establish a genitive-like
constmction with the following noun. In Rumanian, numerals from 20 are followed by de 'of,
horn' with the noun. Both partitive an adjectival constmctions are found in Welsh, but the
partitive constmction is preferred with higher numerals. Numerals from 1 to 9 behave as
adjectives in Lithuanian, whereas the genitive plural is used with larger numerals. In Russian
and some other Slavic languages, numerals larger than 'four' in the direct case contexts govern
the genitive plural of the following noun. This applies to Bosnian also, but in all contexts, not
only in the direct case contexts (e.g. pet knjiga-GEN PL 'five books'). However, it should be
pointed out that in Bosnian and Russian complex numerals have specific syntactic behaviour
determined by the smallest addend. For example, the complex numeral 21 behaves
syntactically in the same way as the numeral 1 (e.g. 1: jedan dan 'one day'; 5: pet dana-, 21:
dvadesetjedan dan vs. *dvadesetjedan dana).
22
r
2. Numerals in Bosnian
2.0. Introduction
Numerals in Bosnian have a very peculiar syntactic behaviour. Although semantically

clearly a single category, syntactically they behave as three distinct categories, namely as
quantifiers, nouns and adjectives. Traditionally, the distinction between numerals is made
taking into account semantics, and not their syntactic behaviour, so that three classes of
numerals are usually distinguished: cardinal, ordinal, and collective numerals:
(i) cardinal numerals: jedan (1), dva (2), tri (3), četiri (4),pet (5), šest (6), sedam (7), osam,...
(ii) ordinal numerals: prvi (1st), drugi (2nd), treći (3rd), četvrti (4th), peti (5th), šesti (6th),...
(iii) collective numerals: dvoje (2), troje (3), četvoro {A), petoro (5), šestoro (6), sedmoro (7),
I will follow this traditional division, and discuss first cardinals, then ordinals, and
then collective numerals. It should be pointed out, however, that this division can be
maintained only as far as semantics is concerned. Syntactically, quite another division of
numerals is necessary, and I will propose it in the conclusion of my discussion. But I will start
with what is traditionally called cardinal numerals.
2.1. Cardinal numerals
2.1.1. Jedan (1)
The numeral one is the only cardinal numeral that behaves as a tme adjective.1 It is a
syntactic modifier of a head noun in a noun phrase, and it agrees with the head noun in
number, gender and case, as any other adjective. It is the only cardinal numeral with separate
forms for three genders (masculine (jedan), feminine (jedna), and neuter (jedno)), two
numbers (singular (jedan, jedna, jedno) and plural (jedni, jedne, jedna)), and six cases
(nominative, genitive, dative, accusative, instrumental, and locative).2
To illustrate agreement of jedan with the head noun, I give the following examples
with nouns of masculine (1), feminine (2), and neuter gender (3):
1 Also all those cardinal numerals that end in one behave in the same way: dvadeset jedan
(21), trideset jedan (31),... etc.
2 Jedan has an adjectival declension, but interestingly, it declines as a definite form adjective,
and semantically behaves as an indefinite adjective. Namely, descriptive adjectives in Bosnian
can actually have two forms, labelled conventionally as indefinite and definite forms:
(i) crven stroj
red-INDEF machine
'a red machine'
(ii) crveni stroj
red-DEF machine
“the red machine1
This distinction will be discussed in more details in a separate section. These two types of
adjectives have different declensions. Jedan declines as crveni, but its meaning is that of an
indefinite adjective, so that jedan stroj could be translated in English as 'a/ one/ some
machine'.
23
a. Jedan muškarac je spavao.

one-MASC NOM man-MASC NOM AUX slept-3P SG MASC
b. Bojao se jednog muškarca.

(he) was afraid of one-MASC GEN man-MASC GEN
c. Pisao je jednom muškarcu.

(he) wrote (to) one-MASC DAT man-MASC DAT
d. Vidio je jednog muškarca.

(he) saw one-MASC ACC man-MASC ACC
e. Išao je sa jednim muškarcem.

(he) walked with one-MASC INS man-MASC INS
f. Pričao je 0 jednom muškarcu.

(he) talked about one-MASC LOC man-MASC LOC
a. Jedna žena je spavala.

one-FEM NOM woman-FEM NOM AUX slept-3P SG FEM
b. Bojao se jedne žene.

(he) was afraid of one-FEM GEN woman-FEM GEN
a Pisao je jednoj ženi.

(he) wrote (to) one-FEM DAT woman-FEM DAT
d. Vidio je jednu ženu.

(he) saw one-FEM ACC woman-FEM ACC
e. Išao je sa jednom ženom.

(he) walked with one-FEM INS woman-FEM INS
f. Pričao je 0 jednoj ženi.

(he) talked about one-FEM LOC woman-FEM LOC
a. Jedno govedo je spavalo.

one-NEUT NOM head of cattle-NEUT NOM AUX slept-3P SG NEUT
b. Bojao se jednog goveda.

(he) was afraid of one-NEUT GEN head of cattle-NEUT GEN
c. Pisaoje jednom govedu.

(he) wrote (to) one-NEUT DAT head of cattle-NEUT DAT
d. Vidio je jedno govedo.

(he) saw one-NEUT ACC head of cattle-NEUT ACC
e. Išaoje sa jednim govedom.

(he) walked with one-NEUT INS head of cattle-NEUT INS
24
f. Pričao je o jednom govedu.

(he) talked about one-NEUT LOC head of cattle-NEUT LOC
In all examples given in (l)-(3) numerals may be left out without any change of
morphological form of remaining lexical items in the sentence. This means that the numeral
one is an adjectival modifier which always receives gender, number, and case features from
the head noun by the feature spreading mechanism. The functioning of this mechanism is
elaborated in more detail elsewhere (Leko 1990). It is obvious from paradigms in (l)-(3) that
six cases may be distinguished in Bosnian: nominative, genitive, dative, accusative,
instrumental, and locative.3
Examples in (l)-(3) illustrate agreement in gender and case. What about agreement in
number? It is somewhat unusual that a numeral like one, denoting a single item, may have a
plural form, but it actually has. However, its use is restricted to modifying only pluralia
tantum nouns:
(4) a. Jedni svatovi su stigli.

one-MASC PL NOM wedding procession-MASC PL NOM AUX arrived-3P PL M
b. Jedne naočale su pukle

one-FEM PL NOM spectacles-FEM PL NOM AUX broke-3P PL FEM
c. Jedna kola su stigla.

one-NEUT PL NOM carriage-NEUT PL NOM AUX airived-3P PL NEUT
What is important for our discussion is that in all these examples one behaves
syntactically as an adjective, and it must agree with the noun it modifies in gender, number,
and case. Depending on gender and number of the head noun, one will have one of the
following six forms in subject nominative phrases:
( 5) ______________________ _______________ ____________ ___

MASC FEM NEUT
SG jedan jedna jedno
PL jedni jedne jedna
What is a position of one in the structure of NP? In order to answer this question I
have to elaborate on the possible structure of noun phrases in Bosnian.
2.1.1.1. The structure ofNP in Bosnian
In generative grammar it was originally assumed that a simple phrase like a book is a
Noun Phrase (NP) in which the noun (N) book is the head of the phrase, whereas the
determiner a is the Specifier of NP. With Abney (1987), the perspective changed. A
functional projection above NP was postulated, so that the determiner was no longer regarded
as the specifier of NP, but rather as the head of the phrase. Therefore, if the determiner (D) is
the head, then the whole structure must be the Determiner Phrase (DP) in which the D head
takes an NP as its complement. Consequently, a phrase like a book was no longer treated as
an NP, but rather as a DP. However, in a language like Bosnian, corresponding to a book, we
have simply knjiga, whereas in English *book without an article, or some other determiner, is
3 The seventh case - vocative - is not included in these paradigms, since it is rarely used, and
not relevant for our discussion.
25
ungrammatical. Therefore, we could say that noun phrases in English are indeed DPs, but in
Bosnian they are not DPs, but rather NPs.
Another possibility was to postulate a universal DP structure, in other words, noun
phrases are always DPs, even in languages without articles, like Bosnian, and the head of DPs
in such languages is simply a null element, or zero determiner. Many linguists adopted this
approach. However, it was noticed that languages with articles behave differently from
languages without articles with respect to certain syntactic phenomena. On the basis of this, it
was proposed, that the postulation of the universal structure of noun phrases as DPs cannot be
maintained, but instead we should distinguish between languages in which noun phrases are
DPs, and those in which noun phrases are simply NPs. For example, on the basis of detailed
argumentation, Bošković (2008) concludes that lexical items which are determiners in English
are actually not D heads in Bosnian, which argues in favour of the no-DP analysis of noun
phrases in Bosnian, and all other languages lacking articles. However, Bošković allows the
possibility that there is some functional structure in noun phrases of languages without
articles, but it cannot be DP.
I agree with Bošković that Bosnian doesn’t have any lexical items or morphological
elements that would qualify as D heads. Also I assume that all noun premodifiers are phrasal
and adjectival in character. All of them must agree with a noun in relevant agreement features,
that is phi-features and case. I agree with Bošković also that these premodifiers are base
generated inside an NP, and I assume that they are generated in multiple Spec positions of
NP. Bošković also allows the possibility of these elements being NP adjoined. However, I
differ from Bošković, because I don’t think that these adjectival elements stay inside an NP.
In highly inflected languages like Bosnian, it is necessary to postulate some kind of
agreement functional projection not only for clauses, but also for noun phrases.
The fact that there is definitely a default order of adjectival elements modifying a
noun, and frequently that default order is the only order possible, also suggests that there must
be a hierarchy of functional projections above an NP. In that hierarchy, the first one above an
NP, call it AgrP, would be a functional category responsible for checking relevant gender,
number and case features on agreeing adjectival elements. At the moment we are interested
only in the lower portion of a noun phrase, that is an NP proper. I assume that an NP is a
complement of a functional category Agr.
The head of NP is a nominal element: a noun proper, like pisac 'writer' in (6a), or a
quantity noun, like većina 'majority' in (6b). They both assign genitive case to their
complement:
(6) a. pisac knjiga

writer books-GEN PL
'a writer of books'
b. većina knjiga
majority books-GEN PL
'majority of books'
The head of NP may take a complement, which may be another NP (7a), AP (7b), PP (7c), or
a clause (7d), forming an N' projection:
(7) a. [n pisac [np zanimljivih knjiga]],

writer interesting-GEN books-GEN
'a/ the writer of interesting books'
26
b. [n' dječak [ap crven u lieu]]

boy 1red in face
'a/ the red-faced boy'
c. [n1stroj [pj>iz uvoza]]

machine from import
'an/ the imported machine'
d. [n1pisac [cp koji piše zanimljive knjige]]

writer who writes interesting-ACC books-ACC
'a/ the writer who writes interesting books'
The configuration of (7c) is represented in (8):
N PP
stroj iz uvoza
We will be more interested here in structures of premodification, that is structures consisting

of a noun head and its premodifier found in a Specifier (Spec) position. Typical noun
premodifiers are adjectives. I follow Jackendoff (1977) in postulating three levels of
adjectival structure inside an NP: appositive, restrictive, and functional.
It is easy to make distinction between restrictive and nonrestrictive complements of
nouns both in English and in Bosnian. For example, nonrestrictive relative clauses as
complements of nouns are distinguished from restrictive relative clauses by comma
intonation. However, it is difficult to establish a distinction between restrictive and appositive
adjectives in English, since there is no distinction in form - an adjective can have only one
form. Nevertheless, the fact that (9) may be interpreted in more than one way is a clear
indication that adjectives may have a restrictive and a nonrestrictive sense:
(9) Our excellent students like syntax.
The nonrestrictive interpretation of the adjective excellent in this example is that all
students, for example, in our department, are excellent and they like syntax. On the other
hand, the sentence can mean that only those students who are excellent like syntax, so that the
adjective excellent has the restrictive interpretation.
Although it appears that this distinction is only semantic, without parallel structural
differentation, the possibility of differentation by means of intonational pattern and stress is
evidence that the two different semantic interpretations correspond to a deeper structural
difference. The distinction between restrictive and appositive adjectives is easier to establish
in Bosnian, since a descriptive adjective actually can have two forms, labelled conventionally
as definite and indefinite forms. The distinction in form of descriptive adjectives is easily
perceived in singular, as in (10a,b), whereas plural distinctions depend on stress and vowel
length:
27
(10) a. mlad vojnik

young-INDEF soldier-MASC
b. mladi vojnik
young-DEF soldier-MASC
c. mladi vojnici
young-INDEF soldiers-MASC
d. mlddl vojnici
young-DEF soldiers-MASC
Comparing indefinite and definite singular forms of adjectives, it is clear that they
differ most in masculine gender, as in (10): the definite form has a long -i ending, whereas the
indefinite form ends in a consonant. Definite and indefinite forms of adjectives in feminine
and neuter gender differ only in the length of the final vowel, definite forms having long
endings: mlada (fem), mlado (neut).
I will show that a definite form of descriptive adjectives is a restrictive modifier,
whereas an indefinite adjective form in Bosnian is appositive in nature.
Both forms of adjectives, indefinite and definite, may be used as single modifiers of a
noun head, as in (10). However, they may be used in combination with other descriptive
adjectives, modifying the same head. In such a case, however, certain distribution restrictions
are percieved:
) a. nov, erven stroj

new-INDEF red-INDEF machine
b. novi crveni stroj

new-DEF red-DEF machine
c. nov, crveni stroj

new-INDEF red-DEF machine
d. erven, novi stroj

red-INDEF new-DEF machine
e. *novi erven stroj

new-DEF red-INDEF machine
f. *crveni nov stroj

red-DEF new-INDEF machine
Whenever adjectives of both forms appear in prehead positions modifying the same
noun, only combinations with indefinite form preceding definite form adjectives are allowed,
as in (llc,d). If the order is reversed, examples are unacceptable, as in (lle,f). I assume that
indefinite and definite adjectives operate on different levels, and since an indefinite adjective
is a modifier of a higher level, it must precede a definite form adjective. The fact that comma
intonation is obligatory in (1 lc,d) is an indication that indefinite adjectives are nonrestrictive.
Another piece of evidence that indefinite adjective forms are appositive is the fact that only
these forms may occur in postnominal positions. Definite forms never occur postnominally.
In Bosnian a single adjective as a part of NP regularly appears in prehead positions:
28
(12) a. dobri dječak

good-DEF boy
b. *dječak dobri
Left dislocation is possible, as in (13), but rarely used:
(13) Dječak dobri spava,

boy good-DEF sleeps.
This sentence is strongly marked and with a heavy stress on dječak.

However, conjoined adjectives and adjective phrases may occur either in prehead, or
posthead positions. When they are used after the head, certain distribution restrictions are
perceived:
(14) a. stroj, nov i erven

machine new-INDEF and red-INDEF
b. *stroj, novi i crveni

machine new-DEF and red-DEF
c. *stroj nov i erven
d. dječak, crven u licu

boy red-INDEF in face
'a red-faced boy1
e. *dječak crven u licu
Only examples (14a,d) are acceptable with indefinite forms of descriptive adjectives in a
postnominal position which must be interpreted as nonrestrictive, since the comma intonation
is obligatory. Examples (14c,e) lacking comma intonation are unacceptable. This shows that
indefinite adjectives are nonrestrictive.
Only indefinite form adjectives may be used in the predicate:
(15) a. Njegov kaput je nov.

his coat is new-INDEF
b. *Njegov kaput je novi.

his coat is new-DEF
However, there are descriptive adjectives which do not have indefinite forms, but only
definite forms ending in long vowels i, a, o. In such a case, the definite form adjective may be
used even in the predicate:
(16) a. Njegov kaput je mali.

his coat is small
b. *Njegov kaput je mal.
29
The form mal, as in (16b), simply does not exist.

Another piece of evidence that definite adjective forms are restrictive is the fact that
only these forms are used in complex lexical items like in (17):
slijepi miš
blind-DEF mouse
'bat'
crveni luk
red-DEF onion
'onion'
Of course, these lexical items can be modified by both definite and indefinite adjective forms,
whereas modification with a definite form would be impossible if an indefinite form is a part
of these lexical items, as in (18d):
odvratni slijepi miš

ugly-DEF blind-DEF mouse
'the ugly bat'
odvratan slijepi miš

ugly-INDEF blind-DEF mouse
'an ugly bat'
odvratan. slijep miš

ugly-INDEF blind-INDEF mouse
'an ugly blind mouse'
*odvratni slijep miš

ugly-DEF blind-INDEF mouse
Notice that (18c) is acceptable only with litteral interpretation. It no longer means 'an ugly
bat', but rather 'an ugly blind mouse'.
If these proposals regarding different levels of adjectives are correct, we would expect
that it would be possible to conjoin only adjectives of the same level. That is, we expect that it
will be possible to conjoin two (or more) indefinite adjective forms, or two (or more) definite
adjective forms, whereas conjoining of definite and indefinite forms should be unacceptable.
The examples of conjoined adjectives are given in (19):
(19) a. siromašan i bolestan dječak

poor-INDEF and sick-INDEF boy
b. siromašni i bolesni dječak

poor-DEF and sick-DEF boy
c. *siromasan i bolesni dječak

poor-INDEF and sick-DEF boy
d. *siromašni i bolestan dječak

poor-DEF and sick-INDEF boy
30
The examples (19c,d) are unacceptable because modifiers of different levels are conjoined.
Only modifiers of the same level can be conjoined, as shown in (19a,b).
There are other ways to prove that two types of descriptive adjectives, definite and
indefinite, indeed differ syntactically. One of them is to separate the adjective from its head
with parenthetical material. It appears that such material may be inserted only between an
indefinite adjective and its head, whereas separation of the definite adjective and the head
with the same material results in an unacceptable sequence:
(20) a. Uznemiren, ponekad bez razloga, pas laje.

upset-INDEF sometimes without reason dog barks
'An upset dog, sometimes without reason, barks.'
b. *Uznemireni, ponekad bez razloga, pas laje.

upset-DEF sometimes without reason dog barks
Another difference between the two types of descriptive adjectives is their potential
mobility, so that indefinite adjectives are quite easily preposed, whereas definite adjectives do
not allow such an option:
(21) a. Njegov pametan i precizan odgovor je iznenadio sve studente,

his clever-INDEF and precise-INDEF answer surprised all students
b. Pametan i precizan, njegov odgovor je iznenadio sve studente,

clever-INDEF and precise-INDEF his answer surprised all students
c. Njegov pametni i precizni odgovor je iznenadio sve studente,

his clever-DEF and precise-DEF answer surprised all students
d. *Pametni i precizni, njegov odgovor je iznenadio sve studente.

clever-DEF and precise-DEF his answer surprised all students
Indefinite adjectives can be actually moved in any direction, so that the posthead
position immediately after the noun is quite usual:
(21) e. Njegov odgovor, pametan i precizan, je iznenadio sve studente,

his answer clever-INDEF and precise-INDEF surprised all students
f. *Njegov odgovor, pametni i precizni, je iznenadio sve studente.

his answer clever-DEF and precise-DEF surprised all students
As (2If) shows, definite adjectives cannot be postposed. However, it appears that a single
indefinite adjective is less easily moved:
(21) g. ?Tačan, njegov odgovor je iznenadio sve studente.

correct-INDEF his answer surprised all students
h. *Njegov odgovor, tačan je iznenadio sve studente,

his answer correct-INDEF surprised all students
i. Njegov tačan odgovor je iznenadio sve studente,

his correct-INDEF answer surprised all students
31
This implies that coordination is sufficiently "heavy" to allow postposing.

There are adjectives in a complex structure of modification that always immediately
precede the head, being the most tightly bound to the head as such. These adjectives are
extensively studied by Levi (1978). She established that in English there is a class of
adjectives which are regularly used only in prehead positions, and cannot be used in the
predicate to modify the same noun. "Nonpredicating adjectives" are found also in Bosnian:
(22) a. mašinski tehničar

mechanical technician
b. Tehničar koji je mašinski

technician who is mechanical
c. mladi mašinski tehničar

young-DEF mechanical technician
d. ?mašinski mladi tehničar

mechanical young-DEF technician
A parallel example to (22a), with the same adjective mašinski 'mechanical', but used in
the relative clause, in (22b), is unacceptable. "Nonpredicating adjectives" usually do not allow
insertion of other adjectives between them and the head. Therefore, most speakers reject any
other order except the one in (22c), whereas some would allow (22d) as a stylistic variant.
"Nonpredicating adjectives" are similar to definite adjectives in the sense that they are also
restrictive, limiting the set of referents expressed by the head. However, in contrast to definite
form adjectives, they establish a functional relation with the head. To illustrate the difference
between a functional and nonfunctional relation, I will analyze modifiers of the noun stroj
'machine' in (23):
(23) električni šivaći stroj

electric sewing machine
The modifier električni 'electric' is not used to say anything about the function of the machine.
It describes a type of the machine - it is electric, rather than mechanical - that is, a physical
property of the machine. On the other hand, the modifier šivaći 'sewing' tells exactly what is
the function of the machine, that is, to sew.
On the basis of these facts we may conclude that three distinct levels of adjectival
structure should be postulated in Bosnian: appositive, restrictive, and functional. To illustrate
the difference between the three levels of adjectival structure, I will analyze adjectives
modifying the noun stroj 'machine' in (24):
(24) nov, crveni šivaći stroj

new-INDEF red-DEF sewing machine
This complex NP describes a new machine, which is red, and whose function is to
sew. Any other sequence of modifiers would be unacceptable. The lower level is a level of
functional arguments. It consists of adjectives strictly subcategorized by the head. They are
most tightly bound to the head. In (24) we find the adjective šivaći in this position. The higher
level is a level of restrictive adjectives. They limit the set of referents expressed by the head.
We find the definite form of the descriptive adjective crveni in this position in (24). Finally,
32
the highest level is a level of nonrestrictive, that is, appositive adjectives, and we find the
indefinite form of the descriptive adjective nov in this position in our example (24).
If the noun stroj is modified by two definite form adjectives, their order is flexible,
and native speakers accept both (25a) and (25b):
(25) a. sjajni crveni šivaći stroj

shiny-DEF red-DEF sewing machine
b. crveni sjajni šivaći stroj
However, in case of examples given in (26) with both functional and definite
adjectives present, a reversal of order would be regarded as a stylistic variant, if accepted at
all:
(26) a. crveni šivaći stroj

red-DEF sewing machine
b. ?šivaći crveni stroj
These facts may suggest that a single functional head (Def) has two Spec positions, a
higher one, for indefinite adjectives (or adjective phrases), and a lower one, for definite
adjectives (or adjective phrases). This is in line with a proposal by Chomsky (1994)
formulated in his bare phrase structure theory which allows multiple Specs in principle, as in
structure (27):
Speci X'
Spec2 X'
Head Complement
Alternatively, it may be proposed that definite and indefinite forms of descriptive

adjectives are projections of two separate heads, one specified positively for the feature
definiteness, and another negatively. Because of examples like (24) we are almost forced to
adopt an analysis with two separate Def heads, rather than a single one. This analysis is
unsatisfactory, however, primarily because it allows a single phrase to be specified both
positively and negatively with respect to the same feature, in this case the definiteness feature.
Therefore, I assume that there is only one Def head with a standard binary feature. In
(24) the feature is positively specified and the phrase has definite interpretation. We are
talking about 'the red machine', and the fact that it is new is an additional property, appositive
in nature. It is important to notice that (24) is acceptable only with comma intonation. This
may be an indication that adjunction is involved, rather than substitution or base generation. I
will pursue this proposal and suggest that indefinite adjectives in examples like (24) are
actually predicative adjectives which are adjoined to DefP without checking definiteness
features. Therefore, they are deprived of their indefiniteness force in such examples, and
consequently, there is no conflict of features: such phrases are interpreted as definite.
I assume that (24) has the configuration as represented in (28):
33
nov, crveni šivaći stroj
In highly inflected languages like Bosnian, it is necessary to postulate some kind of

agreement functional projection not only for clauses, but also for noun phrases. The adjectival
agreement must be checked in some functional projection. It is not obvious that this AgrP
projection should be a split projection, as proposed here. After all, one could claim that
definiteness, as well as gender, number and case are checked in the same projection.
However, these features differ. Gender, number and case are clearly agreement features in the
sense that adjectives must agree with the gender, number and case of the head noun. On the
other hand, definiteness is not an agreement feature in Bosnian. The definite adjective doesn't
agree in definiteness with the noun, since the noun is not inherently definite/ indefinite and
there is no article which would determine the definite/ indefinite status of the noun.
Consequently, another projection should be postulated in which only the definiteness of
descriptive adjectives would be checked. The fixed order of adjectives, with descriptive
adjectives preceding other types of adjectives, represents support for this proposal.
In conclusion, I assume that adjectival elements are base generated in Spec of NP.
However, they cannot remain in situ. They must move in order to check various features.
Nonpredicating adjectives must agree with the noun in gender, number and case. Therefore,
they move to Spec of AgrP to check these features establishing Spec-head agreement with
Agr head. Definite and indefinite form adjectives first check their agreement features in AgrP
and then move to Spec of DefP where they check their definiteness features.
2.1.1.2. The structural position of jedan
Now we can attempt to answer the question posed at the end of the section 2.1.1: what
is a position of one in the structure of NP? The question as it is formulated assumes that one is
a part of an NP. This assumption should not be controversial, since we have seen in our
examples (l)-(5) that one indeed behaves as any other adjectival lexical item in terms of
agreement with the head noun. Therefore, although it doesn't have all properties of an
adjective proper (e.g., it lacks comparative forms), I still treat it as a special type of adjective,
which may be called an adjectival numeral.
34
In the section 2.1.1.1,1 proposed three structural positions for adjectives in Bosnian.
We will now try to determine which of these positions is available to one, under the
assumption that it is an adjective.
In the section 2.1.1, a peculiar property of one was pointed out: it declines as a definite
form adjective, but semantically it behaves as an indefinite adjective. An example was given
that jedan stroj could be translated not only as 'one machine', but also as 'a machine', or even
'some machine'. Bosnian does not have articles. However, frequently used translation
equivalent of an indefinite article is jedan, in addition to neki 'some'. This implies that
indefmiteness is a feature of one, and that therefore it should be classed with indefinite
adjectives. In order to prove this, we should try to apply some of the tests used in a previous
section to determine syntactic behaviour of definite and indefinite adjectives. Is it possible to
combine one with other descriptive adjectives, and if so, what kind of combinations are
allowed?
One can combine with both definite and indefinite adjectives, but only if it precedes
these adjectives, as in (29a,b). If there is a reversal of order, it seems that only a combination
with an indefinite adjective preceding one is allowed, as in (29c):
(29) a. jedan visok čovjek

one tall-INDEF man
b. jedan visoki čovjek

one tall-DEF man
c. visok jedan čovjek

tall-INDEF one man
d. *visoki jedan čovjek

tall-DEF one man
One, coordinated with another indefinite adjective, may appear in the posthead
position, or in the predicate:
(30) a. život, jedan i dragocjen

life one and valuable-INDEF
b. *život, jedan i dragocjeni

life one and valuable-DEF
c. sin, jedan u roditelja

son one in parents
'the only son of his parents'
d. Život je jedan.
life is one
It was shown in the preceding section that only adjectives of the same level can be
conjoined. Therefore, jedan cannot combine with a definite form adjective in (30b). The
example (30d) shows that we find jedan also in the predicate, so that it cannot be treated as
belonging to the third class of nonpredicating adjectives. Also, examples (29)-(30) strongly
suggest that one behaves in many respects as an indefinite adjective.
35
Therefore, I assume that one is base generated in Spec of NP position. Then it moves
to Spec of AgrP in order to check gender, number and case features, like all other adjectives.
Finally, it moves to Spec of DefP where it checks its definiteness features.
2.1.2. Dva (2)
We are tempted to conclude that two does not differ syntactically from one because
parralel to a paradigm for one in (32a), we can propose a paradigm for two in (32b):
(32) MASC FEM NEUT

a. jedan jedna jedno
b. dva dvije dvoje
We find three different forms both in (32a) and in (32b), but the paradigm in (32b) is
artificial, being a combination of cardinal and collective numerals. The neuter singular form
dvoje is not a cardinal numeral, but rather a collective numeral, which selects exclusively
collective nouns (e.g. dvoje djece 'two children’), and will be discussed in more detail later.
So, in contrast to one, the cardinal numeral two has only two forms: the form dvije which
selects feminine nouns, whereas the form dva is used not only with masculine, but also with
neuter nouns. It appears that these two forms exhibit different syntactic behaviour. Let us first
consider examples with the form dvije. Parallel to examples in (2), we have sentences in (33):
(33) Dvije žene su spavale.

two-FEM NOM women-FEM NOM PL slept-3P PL FEM
b. Bojao se dviju žena

(he) was afraid of two-FEM GEN women-FEM GEN PL
c. Pisao je dvjema ženama.

(he) wrote (to) two-FEM DAT women-FEM DAT PL
36
d. Vidio je dvije žene.

(he) saw two-FEM ACC women-FEM ACC PL
e. Išao je sa dvjema ženama.

(he) walked with two-FEM INS women-FEM INS PL
f. Pričao je o, dvjema ženama.

(he) talked about two-FEM LOC women-FEM LOC PL
From (33) we could conclude that the feminine form of two is an adjective, because it
agrees with the head noun in all agreement features in the same way as the cardinal numeral
one does. However, whereas sentences with one can be expressed only as in (2), sentences
with two in (33) have alternative forms, so that parallel to (33e,f), equally acceptable, even
more frequent in colloquial speech, are the following sentences:
(34) a. Išaoje sa dvije žene.4

(he) walked with two women
b. Pričao je o dvije žene

(he) talked about two women
It should be pointed out that there is no difference in meaning between sentences in

(34) and corresponding examples in (33e,f). One would expect to find the same alternative
forms with (33b,c). Howevere, whereas some speakers find (35a) questionable, no native
speaker accepts (35b):
(35) a. ?Bojao se dvije žene.

(he) was afraid of two women
b. *Pisaoje dvije žene.

(he) wrote (to) two women
Yet, we find alternative forms with genitive and dative phrases, like (33b,c), in the
following examples:
(36) a. Bježao je od dviju žena.

(he) was running away from two-FEM GEN women-FEM GEN PL
b. Bježao je od dvije žene.
c. Pokazao je prema dvjema ženama.

(he) pointed towards two-FEM DAT women-FEM DAT PL
d. Pokazao j e prema dvij e žene.
Comparing (34), (35) and (36) we can conclude that alternative forms are possible
only inside prepositional phrases. Prepositions in Bosnian govern genitive, dative, accusative,
instrumental, and locative cases, so that we find alternative forms whenever a preposition
4 Whenever a numeral is dropped from PP, the noun must receive a corresponding case, here
instrumental: (i) *Išao je sa žene; (ii) Išaoje sa ženama..
37
which assigns one of these cases is present. The locative is the only case which is always
assigned by a preposition. Therefore, alternative forms are always possible in locative
phrases, as in (33f) and (34b). In contrast to locative, other cases are not exclusively governed
by a preposition. As a consequence, whenever a preposition is not present to assign case, the
alternative form is not allowed. Therefore, (35b) cannot replace a dative phrase given in
(33c).5 In addition to genitive and dative, instrumental doesn't have to be assigned by a
preposition. When the instrumental case is assigned by a verb, as in (37), then only one form
is possible. The alternative form, as in (37b), is ungrammatical:
(37) a. Kopao je dvjema lopatama.

(he) was digging (with) two-FEM INS spades-FEM INS PL
b. *Kopaoje dvije lopate.
We could conclude from this that the feminine form of two exhibits ambivalent
behaviour. It behaves as an adjective in (33), manifesting complete agreement with the head
noun in gender, number, and case. However, there is another lexical item two which behaves
as a quantifier, having only one form - dvije. This form can be used after prepositions, as in
(34a,b) and (36b,d). In all these examples where a preposition is followed by the phrase dvije
žene, the head of this phrase is the quantifier dvije, and žene is its complement. Quantifiers are
case-assigners in Bosnian and they typically assign genitive as their default, or structural case.
What is a case assigned by dvije to its noun complement? The feminine declension in
Bosnian has the same plural form in nominative and accusative - žene. Therefore, there is no
way to decide on case of žene on the basis of its form. However, following Chomsky (1981) I
assume that the nominative case is not lexically assigned, and it is not the case found on
complements to lexical items. Rather, nominative is typically a case of subjects, assigned by a
functional category INFL. Consequently, žene in our phrase dvije žene may be accusative, but
not nominative. Therefore, we could conclude that accusative is assigned to žene by the
quantifier dvije, which has case-assigning potential, but being undeclinable, itself cannot
receive a case. However, the whole phrase dvije žene with dvije as a head must be assigned a
case by the preposition, but its percolation to the head of the phrase is blocked, dvije being an
undeclinable lexical item.
We still have to explain why examples in (35b) and (37b) are unacceptable. We find
the phrases dvije žene and dvije lopate in these examples, which are identical in form to
accusative phrases:
(40) a. Vidio je dvije žene.

(he) saw two women-ACC
b. Kupio je dvije lopate.

(he) bought two spades-ACC
However, in the feminine declension, accusative always has a plural form different
from all other cases except nominative. Therefore, in all those unacceptable examples,
genitive, dative and instrumental phrases, respectively, are interpreted as accusative phrases,
since there is no formal clue to indicate that other cases, not accusative, are assigned. Such
formal clue is provided by prepositions sa 'with' and o 'about' in (34), and od 'from' and prema
5 The genitive phrase in (35a) is acceptable for most speakers although a preposition is not
present, and we will try to find explanation for this fact in later discussion.
38
'toward' in (36). These prepositions not only assign a corresponding case to the phrase headed
by dvije but also serve as a flag to indicate which case is assigned. Percolation of case from
the phrasal category to the head dvije is blocked, since it is an indeclinable lexical item.
I will briefly illustrate the mechanisms of case assignment and case percolation on
examples (33f) and (34b), repeated here as (41):
(41) a. Pričao je 0 dvjema ženama.

(he) talked about two-FEM LOC women-FEM LOC PL
b. Pričao je 0 dvije žene.

(he) talked about two women
Actually, I assume three independent processes by which an abstract Case is

ultimately realized as a morphological case. In this proposal, all lexical categories with Case
assigning potential may assign Case to their sister phrasal category. The configurations in (42)
represent prepositional phrases in (41a) and (41b), respectively, and illustrate Case-
assignment:
(42) a. b.
PP
P'
PP
P’
In (42a), corresponding to (41a), the only lexical category able to assign Case is a
circled category P and it assigns locative to its sister NP. Case-assignment is represented by a
single arrow. In (42b), corresponding to (41b), there are two lexical categories with Case-
assigning potential - the circled P and Q. They assign locative and accusative, respectively, to
their sisters. After they assign Case, [+N] lexical categories are free to receive case
themselves through the operation of another mechanism - Case-percolation. This mechanism
is responsible for the morphological realization of Case on the heads of NPs. Noun heads
eventually receive a case by percolation from the phrasal category which immediately
dominates them. Thus, Case-assignment is a process involving sisters, whereas Case-
percolation is a process involving mothers and daughters. Case-percolation is represented in
(43):
39
(43) a. b.
PP
PP
In (43a), corresponding to (41a), the phrasal category which is assigned Case is

circled, and its Case percolates down to the head of the phrase. Case-percolation is
represented by a double arrow. By the process of percolation Case is finally realized as a
concrete morphological case on the head of the phrasal category. The head noun žene thus
receives the locative case ženama. In (43b), corresponding to (41b), two phrasal categories are
assigned Case and both are circled. The head of QP is the quantifier dvije. Since it is an
indeclinable lexical item, the percolation of locative case from the phrasal category is
blocked. However, Case from the NP can percolate, because its head is a declinable noun, so
that the accusative case is morphologically realized on the noun žene.
Finally, I assume that the third mechanism - case-spreading - is responsible for case
realization on agreeing lexical categories. In other words, in the final step, the morphological
case of the head of the phrase will spread to all lexical categories that must agree with the
head in case features. This mechanism doesn't apply in (41 b), since there is no agreeing
categories. Its operation is represented in (44), corresponding to (41a):
(44)
PP
P NP
AdjP N'
Adj
N
o dvjema ženama
ttt_ Joe.
pl
fern
40
Feature spreading is represented by a triple arrow. In this process, all agreement

features, not only case, are transferred to the lexical item that agrees with the head noun in
case, number, and gender. So, the locative form dvjema is a result of feature spreading. We
can now make preliminary conclusions. It appears that there are two lexical items of the same
phonological shape - dvije. One behaves as an adjective modifying plural feminine nouns.
Another behaves as a quantifier. It requires a plural feminine noun as its complement
assigning accusative case to it.
Another form of two - dva - selects nouns of masculine and neuter gender. We will
first discuss examples with norms of masculine gender. Parallel to examples in (33), we find
the following sentences:
(45) a. Dva muškarca su spavala.

two man-MASC GEN/ACC SG slept-3P NEUT PL
b. Bojao se dvaju muškaraca.

(he) was afraid of two-GEN men-MASC GEN PL
c. Pisao je dvama muškarcima.

(he) wrote (to) two-DAT men-MASC DAT PL
d. Vidio je dva muškarca.

(he) saw two man-MASC GEN/ACC SG
e. ?Išao je sa dvama muškarcima.

(he) walked with two-INS men-MASC INS PL
f. ?Pričao je 0 dvama muškarcima.

(he) talked about two-LOC men-MASC LOC PL
Surprisingly, we do not find the expected nominative plural form muškarci in (45a),
and accusative plural form muškarce in (45d). Such structures are unacceptable:
(46) a. *Dva muškarci su spavali.

two men-NOM PL slept-3P MASC PL
b. *Vidioje dva muškarce.

(he) saw two men-ACC PL
The corresponding examples with a feminine noun in (33a,d) have nominative/

accusative form žene, as expected. We find the form muškarca in (45a,d) which may be either
genitive or accusative singular. I claim that this form is actually genitive, not accusative. Only
animate masculine nouns have identical genitive and accusative singular forms. Inanimate
nouns (e.g. bor 'pine') have identical nominative and accusative forms (bor-NOM/ ACC SG),
different from genitive (bora-GEN SG)
Therefore, I conclude that the form muškarca in (45a,d) is genitive, rather than
accusative. Notice also that there is no expected masculine form of the predicate in (45a), but
rather a neuter form su spavala. Obviously, the predicate does not agree with the masculine
form muškarca, but rather with dva. We could conclude from this that dva behaves as a
quantifier. Here we must assume that dva has default neuter features, like all quantifiers, but
in contrast to other quantifiers which may take complements of any gender, dva can combine
only with masculine and neuter norms, whereas another form - dvije - is used with feminine
41
nouns. Dva assigns genitive to its complement. What is peculiar about dva is its selection of
what appears to be a singular noun as a complement. However, it only appears to be a genitive
singular form, because when an adjective is added as a modifier of a noun in (45a), it must
have a form which is certainly not genitive singular, as obvious from ungrammatical (47b):
(47) a. Ona dva visoka muškarca su spavala,

those two tall men slept
b. *Ona dva visokog muškarca su spavala.

tall-GEN SG
The sentence with the genitive singular form visokog in (47b) is unacceptable, and
only (47a) with the adjective form visoka is possible. The descriptive adjectives have two
forms, indefinite and definite: visok and visoki, respectively. The form visoka is the genitive
singular of the indefinite form, whereas visokog is the genitive singular of the definite form.
However, the form visoka in (47a) cannot be the indefinite form of the adjective, because the
stress pattern is different, and also because the demonstrative is used in this phrase, and by
definition, the demonstrative cannot be followed by an indefinite adjective. Therefore, visoka
in (47a) cannot be the genitive singular of the indefinite adjective visok, but rather the genitive
dual form of the definite adjective visoki. This form visoka is actually a remnant of the old
dual. Bosnian used to have three numbers in nominal declension (singular, plural, and dual).
The dual number was lost as a separate category in the development of language. There are
few remnants of dual forms in contemporary language, and one of them may be the form of
the noun after the numeral two. So what appears to be a genitive singular form, should be
rather treated as a genitive dual.
I have claimed, discussing feminine forms of two, that there is an adjectival form dvije
with full adjectival declension. It appears that the masculine form dva does not have full
adjectival declension. Indeed, unacceptable examples in (46) prove that dva cannot be treated
as an adjectival lexical item in nominative and accusative contexts. Otherwise, sentences in
(46) would be acceptable, and dva would be the nominative and accusative agreement forms
of the adjectival two. And still, there are adjectival forms of dva in oblique cases, as obvious
from (45 b,c,e,f). It should be pointed out that forms in (45 e,f) are just marginally acceptable
for most native speakers, and corresponding sentences with undeclinable dva are preferred:
a. Išao je sa dva muškarca.

(he) walked with two man-GEN DU
b. Pričao je 0 dva muškarca.

(he) talked about two man-GEN DU
It is unusual that adjectival forms of dvije and dva apparently differ syntactically, the first
having a full declension, and the second only a partial one. I will try to show that both are
actually the same in having only a partial declension. This is not so obvious in the feminine
gender, because the noun following dvije in subject and object phrases has the form which
appears to be nominative/ accusative plural, and therefore, dvije is simply interpreted as a
nominative/ accusative agreement form of the adjectival two. I will propose, however, that
what appears to be the adjectival agreement form dvije in nominative and accusative is
actually a quantifier dvije which selects genitive dual form of feminine nouns as its
complement, in the same way as dva is a quantifier in corresponding examples of masculine
forms in nominative and accusative, which also selects genitive dual form of the following
noun.
42
Dva is possible in genitive phrases exactly because of the genitive case on the
complement noun. So in addition to (45b) with the adjectival form dvaju, the alternative form
with dva is allowed, as in (49a):
(49) a. Bojao se dva muškarca.

(he) was afraid of two man-GEN DU
b. Bojao se muškarca.
(he) was afraid of man-GEN SG
In both (49a) and (49b) we find the genitive form muškarca. In (49b) the case is
assigned by the verb bojati se, and in (49a) by dva. The important thing is that the genitive
case is present, and its presence actually licences the structure. There is no difference in form
between genitive singular and dual of masculine nouns. In feminine declension, however,
genitive dual is identical in form with nominative/ accusative plural, and this is probably a
reason that (35a) with a quantifier form dvije in a genitive context is less acceptable for some
speakers.
Whereas (45b) may be alternatively expressed as in (49a), no alternative form is
possible with dative phrases in (45c), so that (50) is unacceptable:
(50) *Pisaoje dva muškarca.

(he) wrote (to) two man-GEN DU
There is no formal signal in (50) to indicate that dative is assigned to a phrase dva
muškarca, and the sentence is therefore unacceptable.
Parallel to examples in (36), we find alternative forms inside genitive and dative
prepositional phrases:
a. Bježaoje od dva muškarca.

(he) was running ( away) from two man-GEN DU
b. Bježaoje od dvaju muškaraca.

(he) was running (away) from two-GEN men-GEN PL
c. Pokazao je prema dva muškarca.

(he) pointed towards two man-GEN DU
d. ?Pokazaoje prema dvama muškarcima.

(he) pointed towards two-DAT men-DAT PL
The example (51d) is questionable, and ( 51c) is a preferred alternative for expressing
this meaning. The same is true for instrumental and locative phrases in (45 e,f), which have
the same form as the phrase in (5Id) - dvama muškarcima. This shows that adjectival forms
of two are probably in retreat, and that this process first affects the form dvama. We can even
speculate that the quantifier form of two will stay as the only one in the further development
of language. The following examples may support this assumption:
(52) a. Kopao je sa drugovima.

(he) was digging with ffiends-MASC INS PL
43
b. *Kopao je sa pijucima.
(he) was digging with picks-MASC INS PL
c. Kopao je pijucima.
(he) was digging (with) picks-MASC INS PL
d. ?Kopao je dvama pijucima.

(he) was digging (with) two-INS picks-MASC INS PL
e. Kopao je sa dva pijuka.

(he) was digging with two picks-GEN DU
The instrumental case is assigned by the preposition sa 'with' only to animate nouns,
as in (52a), whereas verbs assign instrumental to inanimate nouns, as in (52c). The
preposition sa implies company, meaning 'together with'. Therefore, (52b) is unacceptable,
since it would mean that he was digging together with picks, implying that picks are animate,
and this is, of course, nonsense. Only (52c) is acceptable, implying means of performing the
action. Consequently, only (52d) should be acceptable, but it sounds so awkward that
probably no native speaker would use it in everyday speech. Instead, (52e) would be used,
although prescriptive grammarians disallow it. The fact that (52e), and not (52d), is regularly
used may be support for a speculation that the quantifier form of two will eventually totally
substitute the adjectival form. Another speculation is that it may happen sooner with the
masculine form, since parallel feminine adjectival forms are still used, so that we find both
(53a) and (53b):
(53) a. Kopao je dvjema lopatama.

(he) was digging (with) two-INS spades-FEM INS PL
b. Kopao je sa dvije lopate.

(he) was digging with two spade-FEM GEN DU
To conclude our discussion of two used with masculine nouns, we may say that there
are actually two lexical items. One behaves as a quantifier and it has default neuter features. It
selects a masculine dual noun as its complement, and assigns genitive case to it. Other
behaves as an adjective of incomplete declension modifying plural masculine nouns in
oblique cases.
The same form of two (dva) as in examples with masculine nouns is used with neuter
nouns. Parallel examples to those in (45) are given in (54):
(54) a. Dva goveda su spavala..

two head of cattle-NEUT GEN DU slept-3P NEUT PL
b. Bojao se dvaju goveda.

(he) was afraid of two-GEN head of cattle-NEUT GEN PL
c. ?Prišaoje dvama govedima.

(he) approached (to) two-DAT head o f cattle-NEUT DAT PL
d. Vidio j e dva goveda.

(he) saw two head of cattle-NEUT GEN DU
44
r
e. ?Išaoje sa dvama govedima.

(he) walked with two-INS head of cattle-NEUT INS PL
f. ?Pričao je 0 dvama govedima.

(he) talked about two-LOC head of cattle-NEUT LOC PL
From these examples, it is obvious that the quantifier form dva used in (45 a,d) with
complements of masculine gender does not differ from dva in (54 a,d) used with complements
of neuter gender. Again, dva is a head of the phrase and it assigns genitive to its complement.
Also we find the same plural form in the predicate. We can conclude from this that dva
behaves as a quantifier with default neuter features which assigns genitive to masculine or
neuter dual nouns.
We can conclude from examples in (54) that adjectival use of two is even more
restricted with neuter nouns, than with masculine nouns. Namely, in oblique cases only (54b)
is unanimously acceptable by native speakers. Therefore, alternative forms are more
frequently used, except in dative phrases:
(55) a. Bojao se dva goveda.6

(he) was afraid of two head of cattle-NEUT GEN DU
b. *Prišao je dva goveda.

(he) approached two head of cattle-NEUT GEN DU
c. Išaoje sa dva goveda.

(he) walked with two head of cattle-NEUT GEN DU
d. Pričao je o dva goveda.

(he) talked about two head of cattle-NEUT GEN DU
The example in (55b) is unacceptable for the same reason as (50).

The quantifier use of two is predominant in combinations with neuter nouns.
Adjectival forms of two are most frequently used with feminine nouns. It should be pointed
out that adjectival forms of two are the same for both masculine and neuter gender: dvaju
(genitive), and dvama (dative, instrumental, locative). This should not be surprising, however,
since it is a characteristic of regular adjectival declension to have the same forms in oblique
cases for masculine and neuter gender.
Our discussion of two showed that there are more than one lexical item two\
1. dvije, which behaves as a quantifier which requires a feminine plural form of the predicate;
2. dva, which behaves as a quantifier which requires a neuter plural form of the predicate;
3. adjectival two in oblique cases, which behaves as an adjective of incomplete declension.
Dvije assigns genitive to its complement - a dual feminine noun.
Dva assigns genitive to its complement - a dual noun of masculine or neuter gender.
The adjectival form of two agrees with its head noun in gender and case. Agreement in
case is manifested only in genitive, dative, instrumental, and locative, since adjectival two has
an incomplete declension. Genitive forms are: dviju for feminine, and dvaju for masculine and
neuter gender. In other cases we find dvjema for feminine, and dvama for masculine and
neuter gender.6
6 The form goveda in (55a) and (54b) appear the same in writing, but they are clearly
distinguished by stress and final vowel length.
45
2.1.3. Tri (3)
In contrast to two with two phonological shapes in nominative subject phrases - dvije
and dva - three has only one form in subject phrases - tri.
(56) a. Tri muškarca/ goveda su spavala.

three man-M GEN PAUC head of cattle-N GEN PAUC slept-3P PL NEUT
b. Tri žene su spavale.

three women-FEM GEN PAUC slept-3P PL FEM
If we compare these examples with those in (33a), (45a), and (54a), we will discover
parallels in syntactic behaviour of tri in the same environment in which dvije and dva are
found. Nouns following tri have the same form as those following dvije and dva. Therefore,
we conclude that the grammatical number of these nouns is not dual, but rather paucal, since
these noun forms are selected not only by two, but also by three. We could conclude from
examples in (56) that, in the same way as there are two lexical items two which behave as
quantifiers - dvije and dva - there are also two lexical items three, but of the same
phonological shape - tri. Tri\ behaves as a quantifier with feminine features and imposes
feminine plural agreement form of the predicate, whereas tri2 behaves as a quantifier with
default neuter features imposing the neuter plural form of the predicate.
Let us now consider object accusative phrases. We find, parallel to (33d), (45d), and
(54d), the following examples:
(57) Vidio je tri muškarca/ goveda/ žene.

(he) saw three man-M GEN PAUC head of cattle-N GEN PAUC women-F GEN PA
The same phrases are found if preceded by a preposition:
(58) Pričao je sa/ o tri muškarca/ goveda/ žene.

(he) talked with/ about three man-GEN PA head of cattle-GEN PA women-GEN PA
Since there is a complete parallelism in syntactic behaviour of quantifiers two and

three, we expect to find adjectival forms of three also. And indeed, we find the following
forms in genitive phrases:
(59) a. Bojao se triju muškaraca/ goveda/ žena.

(he) feared three-GEN men-GEN PL head of cattle-GEN PL women-GEN PL
And similarly with dative phrases:
(60) a. Pisao je trima ženama.

(he) wrote (to) three-DAT women-FEM DAT PL
b. ?Pisaoje trima muškarcima.

(he) wrote (to) three-DAT men-MASC DAT PL
c. ??Prišaoje trima govedima.

(he) approached (to) three-DAT head of cattle-NEUT DAT PL
46
The same phonological shape as in dative phrases {trima) is found, but less frequently,
in instrumental and locative phrases:
a. Pričao je sa/ 0 trima ženama.

(he) talked with about three women
b. ?Pričao je sa/ 0 trima muškarcima.

(he)talked with about three men
c. ??Pricao je sa/ 0 trima govedima.

(he) talked with about three head of cattle
Let us now consider the possibility of adjectival forms of three in subject and object
phrases. Only forms in (56) and (57) are possible, and not, for example, those in (62):
(62) a. *Tri muškarci su spavali.

three men-MASC NOM PL slept-3P MASC PL
b. *Vidio je tri muškarce.

(he) saw three men-MASC ACC PL
These examples are unacceptable because there is no adjectival form of three in

nominative and accusative. Let us now consider examples with feminine nouns:
a. Tri žene su spavale.

three women-FEM slept-3P FEM PL
b. Vidio je tri žene.

(he) saw three women-FEM
The feminine nouns like žena 'woman' have the same nominative and accusative plural
forms - žene. Therefore, it appears that tri agrees with these forms. However, tri in (63)
actually behaves as a quantifier which assigns genitive case to its paucal complement which
has the same form as nominative/ accusative plural.
We will next consider examples with neuter nouns:
(64) a. Tri goveda su spavala.

three head of cattle-NEUT slept-3P NEUT PL
b. Vidio je tri goveda.

(he) saw three head of cattle-NEUT
Neuter nouns like govedo have identical forms in genitive singular, nominative plural,
and accusative plural. Therefore, it is difficult to decide on the basis of examples with neuter
nouns like govedo whether tri behaves as an adjective or a quantifier in corresponding subject
and object phrases. There is no way to decide what is the case of goveda in (64). However,
there are neuter nouns with different forms in genitive singular and nominative/ accusative
plural. One of them is bure 'barrel'. Again, the phonological shape of two forms is the same,
47
but they differ in stress: biireta (genitive singular) vs. bureta (nominative/ accusative plural),
the falling stress in the first, and the rising stress in the second.
Therefore, only examples with the falling stress on bureta are allowed in subject and
object phrases, as in (65 a,b), which means that the form in question may be genitive singular,
but not nomimative/ accusative plural. Examples (65 c,d) are unacceptable because they have
the rising stress, which means that they are nominative/ accusative plural forms:
(65) a. Tri biireta su pala.

three barrels-NEUT GEN SG fell-3P NEUT PL
b. Vidio je tri biireta.

(he) saw three barrels-NEUT GEN SG
c. *Tri bureta su pala.

three barrels-NEUT NOM PL fell-3P NEUT PL
d. *Vidio je tri bureta.

(he) saw three barrels-NEUT ACC PL
These examples helped us to solve the puzzle of case on the noun goveda in (64). It
must be genitive. Therefore, we conclude that tri does not behave as an adjective in subject
and object phrases, but rather as a quantifier. Adjectival forms of three are found only in
genitive, dative, and less frequently, in instrumental and locative. Notice that adjectival forms
are the same for all three genders: triju in genitive, and trima in other oblique cases. So, in the
same way as with two, there are three lexical items three-.
1. tri\, which behaves as a quantifier which requires a feminine plural form of the predicate;
2. tri2 , which behaves as a quantifier which requires a neuter plural form of the predicate;
3. adjectival three in oblique cases, which behaves as an adjective of incomplete declension.
Trii assigns genitive to its complement - a paucal feminine noun.
Ih '2 assigns genitive to its complement - a paucal noun of masculine or neuter gender.
The adjectival form of three agrees with its head noun in case. Agreement is
manifested only in genitive, dative, instrumental, and locative, since adjectival three has an
incomplete declension. The genitive form is triju, whereas the form trima is found in other
oblique cases.
2.1.4. Četiri (4)
There is only one form offour in subject phrases - četiri. Our examples in (56), (57),
and (58) will not change syntactically if we replace tri with četiri. This means that four
behaves syntactically in the same way as three. There are two lexical items four of the same
phonological shape - četiri. We also find adjectival forms of four in the same examples as
those for three, but less frequently:
(66) ?Bojao se četiriju žena/ muškaraca/ goveda.

(he) was afraid of four-GEN women-GEN PL men-GEN PL head of cattle-GEN PL7
7 There are four so-called accents in Bosnian: two rising - short and long, and two falling -
short and long accents. Traditionally, the following diacritics are used to denote them, for
example on the vowel e: e - long rising, e - short rising, e - long falling, e - short falling.
48
Adjectival forms are even less frequently found with dative phrases:
(67) ??Prišaoje četirma ženama/ muškarcima/ govedima.

(he) approached four-DAT women-DAT PL men-DAT PL head of cattle-DAT PL
Most native speakers find adjectival forms totally unacceptable in instrumental and
locative phrases:
(68) ?*Pričaoje sa/ o četirma ženama/ muškarcima/ govedima.

(he) talked with about four-INS/LOC women-INS/LOC men-INS/LOC cattle-INS/LO
Not surprisingly, adjectival forms are unacceptable in subject and object phrases. We
would have the same acceptability judgements as in (62), (63), (64), and (65) if we replace tri
with četiri. Therefore, we conclude that there are also three lexical items four.
1. četirii, which behaves as a quantifier which requires a feminin plural form of the predicate;
2. četirii, which behaves as a quantifier which requires a neuter plural form of the predicate;
3. adjectival four in oblique cases, which behaves as an adjective of incomplete declension.
Četirii assigns genitive to its complement - a paucal feminine noun.
Četirii assigns genitive to its complement - a paucal noun of masculine or neuter
gender.
The adjectival form offour agrees with its head noun in case. Agreement is manifested
only in oblique cases, since adjectival four has an incomplete declension. However, dative,
instrumental and locative forms are extremely rare. The genitive form is četiriju, and the form
četirma is found in other oblique cases.
The syntactic behaviour of cardinal numerals higher than four is different from that of
lexical items two, three, andfour}
Therefore, before discussing five, I will make the following conclusions about two,
three, and four.
1. They manifest ambivalent syntactic behaviour:
a. as quantifiers, and
b. as adjectives of incomplete declension.
The forms corresponding to (a) are summarized in (69a), and those corresponding to
(b) in (69b):8
(69) a. ______________________ ____________

quantifiers dvfie dva
trii tri2
četirii četiri2
b.
adjectival forms Genitive dative, instrumental, locative
feminine Dviju dvjema
masculine, neuter Dvaju dvama
fem., masc., neut. triju trima
četiriju četirma
8 It should be pointed out that numerals dvadeset dva (22), dvadeset tri (23), dvadeset četiri
(24), and all higher numerals ending in dva, tri, and četiri behave syntactically in the same
way as two, three, andfour.
49
There are nine lexical items referring to two, three, and four. Six of them are
quantifiers, and three are adjectives. There are 12 different phonological shapes of these
lexical items. The quantifier three, as well as four, represent two lexical items of the same
phonological shape, one requires a feminine plural form of the predicate, and another requires
a neuter plural form of the predicate. There is a separate feminine adjectival declension of
two, whereas three and four have the same adjectival declension for all three genders.
2. Alternative forms of two, three, and four, a quantifier or/ and an adjective, are allowed only
when the quantified phrases are preceded by a preposition.
3. If there is no preposition in front of the quantified phrase, only adjectival forms are allowed
in oblique cases.
4. There are contexts, especially in dative phrases, when neither quantifier nor adjectival
forms of two, three, and four are possible. Another, descriptive way of expressing the desired
meaning must be found in such contexts.
5. Whenever alternative forms are allowed, quantifier forms of two, three, and four are
preferred.
6. Two, three, and four, which behave as quantifiers, are heads of phrases in which they
appear, assigning Case to their complements. They assign genitive Case to the following noun
which must be of paucal number. Two, three, and four as quantifiers are indeclinable, so that
they themselves cannot receive Case.
7. Two, three, and four, which behave as adjectives, have incomplete declensions. They agree
with the head noun which they modify only in oblique cases. Gender agreement is manifested
in NPs with two as a modifier, having separate feminine forms.
8. Adjectival use of two, three, and four is most restricted with neuter nouns. Adjectival forms
most frequently modify feminine nouns. Adjectival forms of four are least frequently used,
whereas adjectival use of two is most frequent. Genitive forms of two, three, and four are
quite often, and locative forms are extremely rare. On the basis of these facts I will propose
the following scale of frequency of adjectival forms:
(70) most frequent: feminine — > masculine —> neuter : leastfrequent

dva---- -— > tr i-----------> četiri
genitive---- > dative------> instrumental-------- > locative
This scale tells us that the most frequent adjectival form is the feminine genitive dviju. On the
other side of the scale we find neuter locative četirma as the least frequent adjectival form.
9. The preference for quantifier forms is an indication that these forms may eventually totally
substitute adjectival forms. Feminine adjectival forms may last longer than their masculine
and neuter counterparts. Also, adjectival forms of two are more stable than those of three and
four. We can speculate that the genitive form will survive longest. The last to surrender will
probably be the feminine genitive form dviju.
2.1.5. Pet (5)
There is only one form of the cardinal numeral five - pet. Let us illustrate its syntactic
behaviour first with examples including feminine nouns, parallel to those in (2) and (33):
(71) a. Pet žena je spavalo.

five women-GEN PL FEM slept-NEUT SG
b. Bojao se pet žena.

(he) was afraid of five women-GEN PL
50
c. *Pisao je pet žena.

(he) wrote (to) five women-GEN PL
d. Vidio je pet žena.

(he) saw five women-GEN PL
e. Išaoje sa pet žena.

(he) walked with five women-GEN PL
f. Pričao je 0 pet žena.

(he)talked about five women-GEN PL
In all sentences in (71), we find the same form of five - pet, and the same form of
women - the genitive plural form žena. Since there is no agreement in case, we conclude that
pet does not behave as an adjective, but rather as a quantifier which assigns genitive case to
its complement, and requires a neuter singular form of the predicate (je spavalo) in (71a). The
same form in all sentences - pet - proves that it is an indeclinable form. It also requires plural
noun in its complement. No alternative forms are possible in either of these examples, since
there are no declinable forms of pet, and therefore we conclude that, in contrast to two, three
and four, there is no adjectival five. We will see that the same is true for five appearing in
phrases with masculine and neuter nouns. There is only one form, a quantifier pet.
In my previous discussion it was pointed out that prepositions not only assign Case,
but also signal which Case was assigned. For example, although the same form pet žena is
found in (71 e,f), we know that these phrases are instrumental and locative, respectively,
because prepositions sa 'with' and o 'about' signal this. What happens in those examples, like
(71a,b,c,d), where the preposition is not present to signal which Case is assigned?
In (71c) the verb pisati 'to write' assigns dative to the phrase pet žena 'five women'.
The percolation of Case from the phrasal category to its head pet is blocked and the sentence
is unacceptable, because there is no formal signal to indicate that dative was assigned: neither
a preposition nor a dative ending are present. There is another way to express this meaning
with two, three and four.
(72) Pisao je dvjema/ trima/ (?)četirma ženama.

(he) wrote (to) two-DAT three-DAT four-DAT women-DAT PL
The alternative adjectival form is used, but the corresponding adjectival form petma
simply doesn't exist. Therefore, the dative meaning with five may be expressed only if a
preposition is present, as in (73):
(73) Pokazao je prema pet žena.

(he) pointed towards five women-GEN PL
In (71c) no preposition may be inserted, and consequently, the sentence is unacceptable. The
English sentence 'He wrote to five women' simply cannot be literally translated into Bosnian.
There are no prepositions in (71a,b,d) and yet, these sentences are acceptable. Let us
look first at (71b). The verb bojati se 'to be afraid of assigns genitive to the following phrase.
We find the genitive form žena 'women' which makes the sentence acceptable, although the
genitive case is assigned by pet, not by the verb. But the very presence of the formal genitive
ending on the noun licences the structure.
What about subject and object phrases in (71a,d)? Generally, subject phrases are
assigned nominative by the INFL element, whereas accusative is assigned to the object by the
51
verb. The fact that the sentences (71a,d) are acceptable is an indication that nominative and
accusative are actually realized. Pet is a quantifier with default neuter singular features, and it
is a characteristic of nominal neuter declension that nominative and accusative singular forms
are identical. Therefore, the phrases with the quantifier pet may appear in direct case contexts
and be interpreted as nominative, or accusative phrases. This is even more obvious when we
compare (71c) with the following example:
(74) Pisao je pet pjesama.

(he) wrote five poems-GEN PL FEM
'Ete was writing five poems.'
The verb pisati 'to write' may assign accusative, in addition to dative. In (74) it assigns
accusative to the following phrase pet pjesama 'five poems'. The sentence is acceptable
because the phrase pet pjesama is interpreted as the accusative phrase, and the form pet
allows the accusative interpretation. If the verb pisati assigns the dative case, then it cannot
percolate to the head of the phrase - pet - in (71c) and there is no other formal clue to indicate
that the dative is assigned. Therefore, (71c) is unacceptable.
We may suspect that it is only dative that cannot be morphologically realized on pet,
since all other examples in the paradigm (71) are acceptable. Let us check, therefore, the only
other case, in addition to those already discussed, that does not have to be assigned by a
preposition. It is instrumental, and we find the following examples parallel to those in (37):
(75) a. Kopao je sa pet lopata.

(he) was digging with five spades-GEN PL FEM
b. *Kopao je pet lopata.
The verb kopati 'to dig' in (75b) assigns instrumental to its complement phrase pet lopata 'five
spades'. The case cannot, however, percolate to the head of the phrase - pet, and since there is
no other formal signal that instrumental was assigned, the structure is unacceptable.
Therefore, in order to express the desired meaning, the sentence (75a) is used. The preposition
sa 'with' assigns instrumental case and it serves as a signal that this case is assigned to the
following phrase. Sentences like (75a) are widely used, although they are frowned upon by
prescriptive grammarians, since the preposition sa is claimed to mean only 'together with' and
should not be used with the meaning 'by means of, as in (75a).
Now we have to examine syntactic behaviour of five in phrases with masculine nouns.
Parallel to examples in (1) and (45), we find the following:
(76) a. Pet muškaraca je spavalo.

five men-GEN PL slept-NEUT SG
b. Bojao se pet muškaraca.

(he) was afraid of five men-GEN PL
c. *Pisaoje pet muškaraca.

(he) wrote (to) five men-GEN PL
d. Vidio je pet muškaraca.

(he) saw five men-GEN PL
52
e. Išao je sa pet muškaraca

(he) walked with five men-GEN PL
f. Pričao je 0 pet muškaraca.

(he) talked about five men-GEN PL
If we compare paradigms in (71) and (76), we notice close correspondences. The noun
muškarac 'man' is always in its genitive plural form - muškaraca. The form of the predicate in
(76a) is neuter singular - je spavalo. We conclude from this that pet in (76) doesn't differ from
pet in (71). No alternative forms are possible in (76), since there is no adjectival five. The
example (76c) is unacceptable for the same reason as (71c). But in contrast to (71c), the
deshed meaning here may be expressed in the following way:
(77) Pisao je petorici muškaraca.

(he) wrote (to) five-DAT men-GEN PL
The declinable form petorica is not a cardinal numeral, and will be discussed in a separate
section. It refers only to males, and there is no corresponding form for females.
In the same way as with feminine nouns, the meaning of instrumental case cannot be
expressed without a preposition. Therefore, parallel to (75), we find the examples in (78) with
a masculine noun. The discussion of examples in (52) applies here, too.
(78) a. Kopao je sa pet pijuka.

(he) was digging with five picks-GEN PL MASC
b. *Kopaoje pet pijuka.
Finally, let us consider examples with five in phrases containing neuter nouns:
(79) a. Pet goveda je spavalo.

five head of cattle-GEN PL slept-NEUT SG
b. Bojao se pet goveda.

(he) was afraid of five head of cattle-GEN PL
c. *Prišao je pet goveda.

(he) approached five head of cattle-GEN PL
d. Vidio je pet goveda.

(he) saw five head of cattle-GEN PL
e. Išao je sa pet goveda.

(he) walked with five head of cattle-GEN PL
f. Pričao je 0 pet goveda.

(he) talked about five head of cattle-GEN PL
Comparing (79) with (71) and (76), we discover that there is no difference in syntactic
behaviour of pet. We find again genitive plural noun as its complement - goveda, and the
neuter singular agreement form in the predicate - je spavalo. Therefore, the conclusion is that
53
there is a single lexical item five. It behaves as a quantifier which assigns genitive to its
complement of any gender.
In the same way as with feminine and masculine nouns, alternative adjectival forms
are not possible in (79), since adjectival five does not exist. Since there is no adjectival five,
and there is no preposition to assign and signal dative case, (79c) is unacceptable. There is no
form corresponding to petorica, as in (77), that would refer to neuter nouns, and therefore the
English sentence 'He approached five head of cattle' cannot be literally translated into
Bosnian.
A preposition is necessary to assign instrumental case to phrases with neuter nouns,
like in examples of phrases with feminine and masculine nouns. Parallel to (75) and (78), we
find only (80a), but not (80b), with neuter nouns like štilo 'pen-holder':
(80) a. Kopao je sa pet stila.

(he) was digging with five pen-holders-GEN PL NEUT
b. *Kopao je pet stila.
To summarize:
There is only a single lexical item five - pet, which behaves as a quantifier with default
neuter singular features. It requires a neuter singular form of the predicate. Pet assigns
genitive to its plural complement of any gender.
2.1.6. Šest (6), sedam (7), osam (8), devet (9)
Cardinal numerals six, seven, eight, nine have only one form each: šest, sedam, osam
and devet. Our examples in (71), (76), (79) will not change syntactically if we replace pet with
any of these numerals. This means that all these numerals behave syntactically like pet. They
all behave as quantifiers.9
2.1.7. Jedanaest (11), dvanaest (12), trinaest (13), četrnaest (14), petnaest (15), šesnaest
(16), sedamnaest (17), osamnaest (18), devetnaest (19)
All these numerals also behave as quantifiers, and any of them may replace pet in our
examples (71), (76), (79), which will not change syntactically.
2.1.8. Deset (10), dvadeset (20), trideset (30), četrdeset (40), pedeset (50), šezdeset (60),
sedamdeset (70), osamdeset (80), devedeset (90)
These cardinal numerals also belong to the same class as pef. they all behave as
quantifiers if inserted instead of pet in our examples in (71), (76) and (79). All higher cardinal
numerals ending in one of these behave in the same way (e.g. 110, 120..., 210, 220..., etc).
9 Numerals dvadeset pet (25), dvadeset šest (26), dvadeset sedam (27), dvadeset osam (28),
dvadeset devet (29), and all higher cardinal numerals ending in pet, šest, sedam, osam and
devet, behave syntactically in the same way as pet.
54
2.1.9. Sto (100)
The English sentence 'A hundred men were sleeping.' can be translated in three
different ways into Bosnian:10
(81) a. Sto muškaraca je spavalo.

hundred men-GEN PL slept-NEUT SG
b. Stotinu muškaraca je spavalo.

hundred men-GEN PL slept-NEUT SG
c. Stotina muškaraca je spavala.

hundred men-GEN PL slept-FEM SG
Neither of these three forms of hundred is adjectival, since we do not find a

nominative plural form muškarci with them, but rather a genitive plural form muškaraca. The
form stotina in (81 c) clearly behaves as a feminine noun, since we find a feminine agreement
form in the predicate - je spavala. On the other hand, we find a neuter singular form of the
predicate both in (81a) and (81b) - je spavalo. It means that sto in (81a) and stotinu in (81b)
behave as quantifiers. It seems that these two forms belong to the same class of quantifiers as
pet. If it is so we would expect to find the same forms in object accusative phrases. But we
would also expect to find the third form stotina. However, in contrast to subject phrases, only
two forms are acceptable in accusative phrases:
(82) a. Vidio je sto muškaraca.

(he) saw hundred men-GEN PL
b. Vidio je stotinu muškaraca.
c. *Vidio je stotina muškaraca.
In genitive phrases given in (83), three forms are possible:
a. Bojao se sto muškaraca.

(he) was afraid of hundred men-GEN PL
b. Bojao se stotinu muškaraca.
c. Bojao se stotine muškaraca.

(he) was afraid of hundred-GEN men-GEN PL
In (83a,b) we find the same forms sto and stotinu as in (81a,b) and (82a,b). However,
in (83c) we find a different form - stotine - from the one in (81c) - stotina. It means that the
feminine form of hundred is a fully declinable noun, and (82c) is unacceptable because the
form stotina is not accusative of the feminine noun stotina. We have to check syntactic
behaviour of hundred in other grammatical cases. If the feminine form of hundred is fully
10 I will continue to use examples with masculine nouns like muškarac 'man', since nothing
will change syntactically, if we use feminine or neuter nouns instead.
55
declinable, then we should not have problems to express dative phrases, which we
encountered in examples (71c), (76c), and (79c):
(84) a. *Pisaoje sto muškaraca.
b. *Pisaoje stotinu muškaraca.
c. Pisao je stotini muškaraca.

(he) wrote (to) hundred-DAT men-GEN PL
In (84c) the dative case is assigned by the verb pisati 'to write' to the NP stotina muškaraca
'hundred men'. Since the head of the phrase - stotina - is a fully declinable noun, dative may
percolate from the phrasal category to the head of the phrase and be morphologically realized
in the form stotini. Its percolation, however, is blocked when the head of the phrase is a form
sto in (84a) or stotinu in (84b), since these forms are not nouns and dative cannot be
morphologically realized on them. As a consequence, sentences in (84a,b) are unacceptable.
The locative case is the only one which is always assigned by a preposition. Since the
preposition is always present to signal that locative is assigned, we would expect that locative
meaning with hundred may be expressed in three ways, which is confirmed by (85):
(85) a. Pričao je o sto muškaraca.

(he) talked about hundred men-GEN PL
b. Pričao je o stotinu muškaraca.
c. Pričao je o stotini muškaraca.

(he) talked about hundred-LOC men-GEN PL
It is a characteristic of feminine declension in Bosnian that dative and locative cases have the
same form, so that we find the same form stotini both in (84c) and (85c). It is still another
proof that stotina indeed behaves as a fully declinable feminine noun.
Whenever instrumental case is assigned by a preposition, we find all three forms of
hundred, as in (86):
(86) a. Išaoje sa sto muškaraca.

(he) walked with hundred men-GEN PL
b. Išaoje sa stotinu muškaraca.
c. Išaoje sa stotinom muškaraca.

(he) walked with hundred-INS men-GEN PL
However, if a preposition is not present to signal that instrumental is assigned, the

forms with sto and stotinu should be disallowed, and only the instrumental form of stotina
should be acceptable. This is exactly what we find in (87):
(87) a. *Kopaoje sto pijuka.
b. *Kopao je stotinu pijuka.
56
c. Kopao je stotinom pijuka.

(he) was digging (with) hundred-INS picks
Now it is clear that the accusative form of stotina is not missing, as it looked like in
(82c). Rather, the accusative form of the nominal stotina is identical in form to the quantifier
stotinu. This is an explanation for the fact that we have only two acceptable sentences in (82)
instead of expected three. The form stotinu in (82b) may be interpreted either as the quantifier
or as the accusative form of the noun stotina.
The numeral stotina behaves as a fully declinable noun, and therefore we expect to
find also its plural forms. The full paradigm is given in (88):
(88) a. Stotine muškaraca su spavale.11

hundreds-NOM men-GEN PL slept-3PS FEM PL
b. Bojao se stotina muškaraca.

(he) was afraid of hundreds-GEN men-GEN PL
c. Pisao je stotinama muškaraca.

(he) wrote (to) hundreds-DAT men-GEN PL
d. Vidio je stotine muškaraca.

(he) saw hundreds-ACC men-GEN PL
e. Išaoje sa stotinama muškaraca.

(he) walked with hundreds-INS men-GEN PL
f. Pričao je 0 stotinama muškaraca.

(he) talked about hundreds-LOC men-GEN PL
The form of the predicate in (88a) is the feminine plural - su spavale - in agreement
with the feminine plural head - stotine - of the subject phrase. We find the same genitive
plural form of the complement noun - muškaraca - in all examples in (88). Finally, the case
endings of the plural forms of stotina are those characteristic of the feminine declension. This
proves that stotina behaves as a feminine noun with full singular and plural declensions. In
that respect stotina differs from all other cardinal numerals discussed so far. However, we
find, parallel to (88), the following plural paradigm:
(89) a. Desetine muškaraca su spavale.

tens-NOM men-GEN PL slept-3PS FEM PL
b. Bojao se desetina muškaraca.

(he) was afraid of tens-GEN men-GEN PL
c. Pisao je desetinama muškaraca.

(he) wrote (to) tens-DAT men-GEN PL
11 The masculine agreement form of the predicate in (88a) is also possible - su spavali. This is
a semantic agreement with what is felt to be a semantic head of the phrase, namely the
masculine noun muškaraca.
57
d. Vidio je desetine muškaraca.

(he) saw tens-ACC men-GEN PL
e. Išaoje sa desetinama muškaraca.

(he) walked with tens-INS men-GEN PL
f. Pričao je o desetinama muškaraca.

(he) talked about tens-LOC men-GEN PL
Obviously, the Bosnian numeral ten is not as simple as originally postulated in the
section 2.1.8. There I simply stated that it behaves as a quantifier with a single form - deset.
From (89) we conclude that there must exist another lexical item ten. Comparing (88) and
(89) we conclude that both stotine and destine behave as feminine plural nouns. Is there a
singular form corresponding to plural desetine? If there is, then it should be possible to
translate the English sentence 'Ten men were sleeping' in more than one way:
(90) a. Deset muškaraca je spavalo,

ten men-GEN PL slept-NEUT SG
b. *Desetinu muškaraca je spavalo.
c. %Desetina muškaraca je spavala.

ten-NOM men-GEN PL slept-FEM SG
The only way of expressing the desired meaning is as in (90a), using the cardinal
numeral deset. The quantifier form desetinu, corresponding to stotinu, simply does not exist.
Therefore, (90b) is unacceptable. But (90c) is grammatical. However, whereas there is no
difference in meaning between sto and stotina, so that (81a) and (81c) are exact parallels, the
form desetina acquired a different shade of meaning. The meaning of (90c) could be
paraphrased as: 'A group of ten men taken as an entity was sleeping.' Therefore, the only
literal translation of the English sentence 'Ten men were sleeping' is the one given in (90a).
The nominal form desetina also has a full feminine declension as stotina with the
same case endings. The forms desetine (gen), desetini (dat), desetinu (acc), desetinom (ins),
and desetini (loc) may be inserted in place of corresponding forms of stotina in examples
(82)-(86).
We conclude from this that there are two lexical items ten\ deset, which behaves as a
quantifier; and desetina, which behaves as a fully declinable feminine noun.
2.1.10. Dvjesto (200), tristo (300), četiristo (400), petsto (500), šesto (600), sedamsto (700),
osamsto (800), devetsto (900)
All these cardinal numerals behave syntactically in the same way. Since they end in
sto, we would expect to find, parallel to (81), the following sentences with two hundred, for
example:
(91) a. Dvjesto muškaraca je spavalo.

two hundred men-GEN PL slept-NEUT SG
b. *Dvjestotinu muškaraca je spavalo.
58
c. Dvije stotine muškaraca su spavale.

two hundred-GEN PAUC FEM men-GEN PL slept-FEM PL
Notice, first, that there is no form dvjestotim corresponding to stotinu in (81b). Therefore,
(91b) is unacceptable. However, there is a form corresponding to stotina. It is a form dvije
stotine in (91c). What is a head of the subject phrase in (91c)? The form of the predicate is the
feminine plural su spavale, and it may be in agreement with either stotine, as in (88a), or with
dvije, as in (33 a). It may be suggested that stotine is a nominative plural form and a head of
the subject phrase. This would mean that dvije is an adjectival nominative form. This
assumption must be rejected, however, since adjectival two, being of incomplete declension,
doesn’t have nominative and accusative forms, as pointed out in the section 2.1.2. Therefore,
we conclude that dvije in (91c) actually behaves as a quantifier and a head of the subject
phrase. The following example will support this conclusion:
(92) Pet stotina muškaraca je spavalo,

five hundred-GEN PL FEM men-GEN PL MASC slept-NEUT SG
In (92) we find a neuter singular form in the predicate - je spavalo, which may be only in
agreement with the default neuter features of pet, since other two lexical items in the subject
phrase are of feminine (stotina) and masculine gender (muškaraca). These examples help us
to confirm our earlier conclusions about the form stotina. It indeed behaves as a noun of
feminine gender, since dvije requires a feminine noun as its complement, and it also has a
plural declension, since pet selects a plural noun as its complement.
On the basis of ungrammatical (91b), we could conclude that cardinal numerals 200,
300, 400, 500, etc. do not have forms corresponding to stotinu. To prove that this is indeed the
case, I will give examples with five hundred in accusative contexts, as in (93):
(93) a. Vidio je petsto muškaraca.

(he) saw five hundred men-GEN PL
b. *Vidio je petstotinu muškaraca.
c. Vidio je pet stotina muškaraca.

(he) saw five hundred-GEN PL men-GEN PL
Clearly, higher numerals ending in hundred have only two forms. In the case of five hundred,
these forms are petsto and pet stotina. The first one behaves exactly as sto, that is, as a
quantifier. The other one is a complex lexical item, with the leftmost element as its head, so
that pet stotina behaves in the same way as pet, that is, as a quantifier.
In (93c) we find two genitive plural forms following pet. It is clear that genitive is
assigned to stotina by pet. The question is what assigns case to the noun muškaraca? In order
to answer this question, I will represent case assignment inside the VP by the following
configuration of the VP from our example (93c):
59
Verbs, nouns and quantifiers are lexical categories with Case-assigning potential. They
assign Case to their sister phrasal categories. The verb vidjeti ‘to see’ has Case-assigning
potential. It assigns accusative to its sister QP. We have seen that pet behaves as a quantifier,
and therefore it also has Case-assigning potential. However, it assigns genitive to its sister
NP. We have seen that stotina behaves as a noun and it also assigns genitive to its sister NP.
The noun muškarac also can assign Case, but there is no available NP to receive it, so that the
Case-assigning potential of the noun muškarac is not realized in this particular example.
Lexical categories with Case-assigning potential are circled in (94a) and Case assignment is
represented by a single arrow. The result of Case-assignment is represented by circled
categories in (94b). These categories carry Case: the highest one, accusative, and the other
two, genitive.
In the next step, Case is morphologically realized on the heads through the operation
of another mechanism - Case-percolation. This process is represented by double arrows in
(94b). The genitive of the higher NP percolates down and is morphologically realized as
stotina. The lower NP also carries genitive which percolates to the head of the phrase and gets
the morphological realization as muškaraca.
However, this analysis doesn’t take into account the fact that pet stotina is a single
lexical item, a complex one, but still a single item, on a par with petsto. Therefore, the
configuration of the VP in (93c) must take into account this fact, so that pet stotina must be
represented as a single constituent:
60
(95) b.
VP VP
V V
The lexical item pet stotina is felt as a compound, and therefore dominated by Q. This
is in agreement with a theoretical proposal that only lexical categories, in addition to INFL,
may assign Case. The node Q dominating pet stotina has Case-assigning potential and it
assigns genitive to its sister NP node. The theory assumes that Case may be assigned by
lexical categories only to their sister phrasal nodes. Therefore the node N dominating stotina
is not able to assign Case to NP dominating muškaraca, since these two nodes are not sisters.
On the other hand the node dominating pet may assign genitive to its sister NP node. Case
assignment is represented by single arrows in (95a) and lexical categories assigning Case are
circled.
Case percolation is represented by double arrows in (95b). The morphological
realization of genitive case on stotina and muškaraca is a result of case percolation from NP
nodes immediately dominating these lexical items. Therefore, we conclude that (95) is an
adequate representation of the VP from (93c) and the mechanisms involved in morphological
realization of Case.
2.1.11. Hiljada (1,000)
The meaning of the English sentence ‘A thousand men were sleeping’ may be
expressed in two ways in Bosnian:
(96) a. Hiljadu muškaraca je spavalo.

thousand men-GEN PL slept-NEUT SG
b. Hiljada muškaraca je spavala.

thousand-NOM men-GEN PL slept-FEM SG
There is no shorter form of thousand, which would be parallel to the short form sto. In
all other respects thousand behaves syntactically in the same way as hundred, so that the form
61
hiljadu is parallel to stotinu, whereas the form hiljada parallels the behaviour of stotina. The
first form - hiljadu - behaves as a quantifier, and the second form - hiljada - as a feminine
noun. As expected, only one form is possible in object accusative phrases:
(97) Vidio je hiljadu muškaraca.

(he) saw thousand-ACC men-GEN PL
The nominal form hiljada has the identical form in accusative as a corresponding
quantifier - hiljadu, and therefore only one sentence is possible with object phrases, as in (97).
Owing to the fact that hiljada is fully declinable, it is possible to express the dative meaning,
as in (98):
(98) a. Pisao je hiljadi muškaraca,

(he) wrote (to) thousand-DAT men-GEN PL
b. *Pisao je hiljadu muškaraca.
Of course, (98b) is unacceptable, because the percolation of the dative case is blocked,
since hiljadu behaves as a quantifier, and therefore is not able to receive the dative case.
If the form hiljada behaves in all respects as stotina, having the same case endings, we
would expect to find its plural forms. And indeed, we find the following sentence,
corresponding to (88a):
(99) Hiljade muškaraca su spavale.

thousands-NOM PL men-GEN PL slept-FEM PL
We conclude that the form hiljada also has the full feminine declension as stotina,
with the same case endings. The forms hiljada, hiljadama, hiljade, hiljadama, hiljadama may
be inserted in place of corresponding forms of stotina in (88b-f). Both stotina and hiljada
behave as feminine nouns with full singular and plural declensions.
2.1.12. Pet hiljada (5,000), pedeset hiljada (50,000), pet stotina/ petsto hiljada (500,000)
Corresponding to the sentence in (93c), we find the sentence with pet hiljada in (100):
(100) Vidio je pet hiljada muškaraca.

(he) saw five thousand-GEN PL men-GEN PL
The VP in (100) would have the same structure as the one in (95). What would be the
structure of the VP with both stotina and hiljada in it, as in (101)?
(101) Vidio je pet stotina hiljada muškaraca.

(he) saw five hundred-GEN PL thousand-GEN PL men-GEN PL
We find three genitive forms in a row following pet. What assigns Case to all these
forms? It may be proposed that the structure of the VP in (101) has the configuration as
represented in (102):
62
(102) a.
VP
V’
b.
VP
In (102a) the verb assigns accusative to its sister QP. The highest Q node assigns
genitive to its sister NP, which is then morphologically realized by percolation in the form
muškaraca. The node Q dominating pet assigns genitive to its sister NP. This case then
percolates to the head of the phrase and gets morphological shape stotina. Finally, the genitive
case is assigned to the lowest NP by the N dominating stotin-. This case is realized in the
63
form hiljada by percolation, as represented in (102b). There is an alternative way of

representing the structure of the VP in (101), as in (103):
(103) a.
VP
b.
VP
Configurations in (102) and (103) differ in constituent structure. In (102) stotina hiljada is a
constituent, and in (103) pet stotina is a constituent. Is there any reason to prefer one structure
64
over the other? It all depends on which structure better reflects psychological reality. Do
native speakers perceive the numeral 500,000 as 5X100,000 or as 500X1,000. The structure in
(102) would be appropriate for pet stotina hiljada perceived as 5X100,000. The table in (104)
will help us to solve this problem:
(104) a. 5 pet
b. 10 deset
c. 50 = 5X10 pedeset <= pet deset
d. 100 stotina *deset deset
e. 500 = 5X100 pet stotina *pedeset deset
f. 1,000 hiljada ?deset stotina
g- 5,000 = 5X1,000 pet hiljada ?pedeset stotina
h. 10,000 = 10X1,000 deset hiljada ?stotina stotina
i. 50,000 = [5X10]X1,000 pedeset hiljada *pet deset hiljada
j- 100,000= 100X1,000 stotina hiljada *deset deset hiljada
k. 500,000 = [5X100JX1,000 pet stotina hiljada *pedeset deset hiljada
1. 1,000,000 milion
From this table it is clear that there are lexical items indicating 5X10, 5X100, 5X1,000
but there is no lexical item indicating 5X10,000, since pet deset hiljada is unacceptable.
Instead we find pedeset hiljada, which corresponds to 50X1,000. Consequently, there should
not be a lexical item corresponding to 5X100,000, but rather one corresponding to 500X1,000
so that pet stotina hiljada should be analyzed as [pet stotina\ hiljada, and not as pet [stotina
hiljada\. This conclusion may be supported by the following facts:
(105) a. 100,000 = sto hiljada stotinu hiljada stotina hiljada

b. 500,000 = ?pet sto hiljada *pet stotinu hiljada pet stotina hiljada
c. 500 = petsto *petstotinu pet stotina
d. 500,000 = petsto hiljada *petstotinu hiljada pet stotina hiljada
If 500,000 is a combination of 5 and 100,000, then it should be possible to express this

number in three ways, since there are also three ways of expressing 100,000, as in (105a).
However, pet stotinu hiljada is unacceptable, and pet sto hiljada is unusual, with pet and sto
pronounced as separate words. In addition to pet stotina hiljada in (105b), the only alternative
way of expressing 500,000 is one given in (105d) - petsto hiljada, which means that 500,000
is a combination of 500 and 1,000. Therefore, the correct structure of (101) is the one given in
(103).
We can conclude that all higher numerals ending in hiljada are complex lexical items
in which a numeral preceding hiljada is the head. It is obvious from the form of the predicate
which agrees with the head of the subject phrase, and the head in such examples is always a
numeral preceding hiljada, as in (106), with the head in italics:
(106) a. Pet/ dvadeset/pedeset/ dvjesto/pet stotina/ petsto hiljada muškaraca je spavalo.

5 20 50 200 5 100-FEM/ 500 1000-FEMmen slept-NEUT
b. Dvije hiljade muškaraca su spavale.

two-FEM thousand men slept-FEM PL
c. Dvije stotine hiljada muškaraca su spavale.

two-FEM hundred thousand men slept-FEM PL
65
We find feminine agreement predicate forms with numerals 2,000 and 200,000 in
(106b,c). The head of these complex lexical items is the numeral dvije, which behaves as a
quantifier with default feminine features, and it determines syntactic behaviour of these
complex lexical items.
In contrast to this, we find neuter singular agreement forms of the predicate with
numerals 5,000; 20,000; 50,000; 200,000; 500,000 in (106a), since the heads of these
complex lexical items behave as quantifiers with default neuter features: pet, dvadeset,
pedeset, dvjesto, petsto, pet stotina.
2.1.13. Milion (1,000,000)
The English sentence ‘A million men were sleeping’ can be translated in two ways
into Bosnian, as in (107):
(107) a. Milion muškaraca je spavalo.

million men-GEN PL slept-3PS NEUT SG
b. Milion muškaraca je spavao.

million men-GEN PL slept-3PS MASC SG
In corresponding examples with hundred (81a,b,c) and thousand (96a,b), more than
one translation is also possible, but we also find more than one form of these numerals. In
case of million, however, a single form - milion - is found in both (107a) and (107b).
Sentences differ only in the form of the predicate. In (107a) it is a neuter singular - je spavalo,
and in (107b) a masculine singular form - je spavao. It appears that milion is a single lexical
item of both masculine and neuter gender. When one of them prevails, for example, the
masculine gender, we find a masculine agreement form in the predicate, as in (107b). This
sounds unusual, but examples of nouns with ambivalent gender are actually found in
Bosnian.12
Before accepting this proposal, we should check which forms are found in other
grammatical cases. The same form as in (107) is found in accusative object phrases:
(108) Vidio je milion muškaraca.

(he) saw million men-GEN PL
So it seems that milion is in the same class of numerals as pet, that is, it behaves as a
quantifier which assigns genitive to its plural complement. What is peculiar about it is that it
has an ambivalent gender (masculine and neuter). In that respect milion would be in a class of
its own, because no other numeral discussed so far exhibits this peculiar behaviour regarding
gender.
If milion behaves as a quantifier, we would expect that dative could not be expressed.
Yet, we find that sentences containing a dative phrase with milion are acceptable:
(109) Pisao je milionu muškaraca.

(he) wrote (to) million-DAT men-GEN PL
Now we have another form - milionu, and therefore we are forced to reject our original
assumption that milion behaves as a quantifier. It seems now that milion behaves as a fully
12 In Leko (1986) I discussed nouns like gazda ‘landlord’, which is a masculine noun in
singular, but in plural may behave also as a feminine noun.
66
declinable noun, having the same nominative and accusative forms, but with an additional
peculiarity of being both masculine and neuter. Before accepting this proposal and rejecting
the previous one, let us check what happens in other grammatical cases. We will start with the
locative, since that is the only case which is always assigned by a preposition:
(110) a. Pričao je o milion muškaraca,

(he) talked about million men-GEN PL
b. Pričao je o milionu muškaraca,

(he) taiked about million-LOC men-GEN PL
Suddenly, two forms are possible - milion and milionu. And the same happens in the
instrumental case:
( 111) a. Išao je sa milion muškaraca,

(he) walked with million men-GEN PL
b. Išao je sa milionom muškaraca.

(he) walked with million-INS men-GEN PL
To complete our paradigm of cases, we have to check what happens in genitive:
( 112) a. Bojao se milion muškaraca,

(he) was afraid of million men-GEN PL
b. Bojao se miliona muškaraca.

(he) was afraid of milliOn-GEN men-GEN PL
We notice exact parallels with corresponding examples including numerals hundred

and thousand. Therefore, we conclude that in the same way as there are two lexical items for
thousand - hiljadu and hiljada, there are also two lexical items for million, but of the same
phonological shape - milion. One of them - milioni - behaves as a quantifier, in the same class
of quantifiers as pet. We find this form in (107a), (110a), (111a) and (112a). The other one -
milioni - behaves as a declinable norm, and we find it in (107b), (109), (110b), (111b) and
(112b). The form milion found in (108) may be the quantifier form milioni with default neuter
features, or milion2 , the accusative singular form of a masculine noun. It is a characteristic of
masculine declension in Bosnian that nominative and accusative singular forms of inanimate
nouns are identical.
It is clear now that milioni, except for the difference in gender, behaves syntactically
very much like desetina (10), stotina (100), hiljada (1000). All these numerals have plural
forms, and we expect to find also plural forms of milioni. Its plural declension is given in
(113):
(113) a. Milioni muškaraca su spavali.

millions-NOM men-GEN PL slept-3PS MASC PL
b. Bojao se miliona muškaraca,

(he) was afraid of millions-GEN men-GEN PL
c. Pisao je milionima muškaraca.

(he) wrote (to) millions-DAT men-GEN PL
67
d. Vidio je milione muškaraca.

(he) saw millions-ACC men-GEN PL
e. Išao je sa milionima muškaraca,

(he) walked with millions-INS men-GEN PL
f. Pričao je o milionima muškaraca,

(he) talked about millions-LOC men-GEN PL
The form of the predicate in (113a) is an expected masculine plural form - su spavali.
It may be objected that this form is due to agreement with a masculine noun muškaraca. Let
us check therefore an example with a feminine noun in the complement of the numeral
milioni:
(114) Milioni žena su spavali.

millions-NOM women-GEN PL slept-3PS MASC PL
It is obvious now that the form of the predicate (su spavali) is in agreement with a
masculine plural head of the subject phrase - milioni. We fmd the same genitive plural form
of the noun complement (muškaraca) in all cases in (113), which also proves that milion is
the head of the phrase. It has case endings characteristic for masculine declension. So, there
may be no doubt that milioni indeed behaves as a masculine noun with full singular and plural
declensions. Among cardinal numerals discussed so far, this is the only one which behaves as
a masculine noun.
2.1.14. Pet miliona (5,000,000), pedeset miliona (50,000,000), pet stotina miliona (500M)
Let us consider now the following examples with higher numerals ending in million'.
(115) a. Pet/ dvadeset/pedeset/ dvjesto/pet stotina/ petsto miliona muškaraca je spavalo.

5 20 50 200 5 100-FEM/500 million-MASC men slept-NEUT
b. Dva miliona muškaraca su spavala.

two million-GEN men-GEN PL slept-3PS NEUT PL
c. Dvije stotine miliona muškaraca su spavale.

two hundred-GEN million-GEN men-GEN PL slept-3PS FEM PL
These are all complex lexical items, like those with hiljada in (106). The examples in
(106) and (115) are parallel, except for the difference in (b) examples, with feminine plural
and neuter plural predicate forms, in agreement with dvije and dva, respectively. Since milion
behaves as a masculine noun, it may be the complement of dva, not dvije. Therefore, we fmd
dva miliona in (115b) and dvije hiljade in (106b). Complex lexical items ending in milion do
not behave syntactically in the same way as a single lexical item milion. Rather, they behave
in the same way as the head of these lexical items, and it is not milion, but the first numeral in
the complex lexical item, which is written in italics in our examples in (115).
68
2.2.15. Milijarda (1,000,000,000)
The English sentence ‘A billion men were sleeping’ has double translation in Bosnian:
(116) a. Milijardu muškaraca je spavalo,

billion men-GEN PL slept-3PS NEUT SG
b. Milijarda muškaraca je spavala.

billion men-GEN PL slept-3PS FEM SG
If we compare these examples with those in (96), we will notice that there is no
difference between them, except for the fact that we have milijardu instead of hiljadu in
(116a) and milijarda instead of hiljada in (116b). It means that milijardu, like hiljadu,
behaves as a quantifier with default neuter singular features, and milijarda, like hiljada, as a
feminine noun. In examples corresponding to (97) and (98) we would have forms milijardu
and milijardi, respectively.
The feminine form milijarda has the same case endings as stotina and hiljada. We find
the form milijarde in genitive phrases, milijardom in instrumental phrases, and milijardi in
locative phrases. Not surprisingly, milijarda has a full feminine declension as stotina and
hiljada, so that we fmd plural forms with the same case endings as those for stotina and
hiljada, except for one difference. Compare the following example with those given in (93c)
and (100):
(117) Vidio je pet milijardi muškaraca,

(he) saw five billion-GEN PL men-GEN PL
Instead of the form milijarda, parallel to genitive plural forms stotina and hiljada, we
have the form ending in -i: milijardi. However, this is only a reflection of the regular pattern
that -a declension feminine nouns ending in consonantal groups have -i instead of -a in
genitive plural. And milijarda ends in a consonantal group -rd. The expected forms are found
in other grammatical cases of plural declension: milijarde (nominative and accusative) and
milijardama (dative, instrumental, locative).
2.1.16. Summary
Most cardinal numerals behave syntactically as nouns or quantifiers. There is only one
cardinal numeral which is fully adjectival - jedan ‘one’. One is always a modifier of the head
noun in a noun phrase, agreeing with the head in gender, number and case. There are also
adjectival forms of two, three, and four, but they are lacking nominative and accusative cases.
These forms are given in the table (69b).
There are only five cardinal numerals that behave in all respects as any other noun, in
the sense that they are fully declinable with singular and plural declensions. Majority of
cardinal numerals behave as quantifiers, having only one form, as represented in (118):
( 118) ______________________________________________________________________________
Singular & plural Paucal Singular
desetina, stotina,
all Cases hiljada, milion2 ,
milijarda
single dvije, trii, četiri], all the rest
form dva, tri2 , četiri2 _____ (Pet>.....)
69
There is only one masculine noun among all cardinal numerals - milion2 . Majority of
cardinal numerals behave as quantifiers with default neuter features, as represented in (119):
(119) _ _ _ ___________________ ________________________________________

Singular & plural Paucal Singular
MASC milion2
desetina, stotina,
FEM dvije, trii, četirii
hiljada, milijarda
NEUT dva, tri2 , četiri2 all the rest (pet....... )
Nouns generally assign genitive as their default Case, and cardinal numerals do not
differ from majority of norms in this respect. Cardinal numerals select plural complements,
except two, three, and four. Dvije, tri\, četirii require paucal feminine complements, and dva,
tri% ietirh require masculine or neuter complements, as represented in (120):
( 120) ______________________________________________________ __________________ __________________

MASC & MASC, NEUT
FEM
NEUT & FEM
dvije, trii, četiri] —>gen pauc N
dva, tri2 , četiri2 —> gen pauc N
all the rest (pet,..)—> gen pl N
From tables (118-120) it is clear that there are only eleven cardinal numerals that
differ in one way or another from the rest of numerals: these are neuter and feminine
varieties of two, three, and four, and desetina, stotina, hiljada, milioni, and milijarda. Of
course, we should add here adjectival forms of two, three, and four, and the only true
adjective - jedan. The main syntactic characteristics of these numerals are the following:
-1. jedan - behaves as an adjective which agrees with the head noun in gender, number and
case;
2. dviju, dvjema, dvaju, dvama, triju, trima, četiriju, četirma - adjectival case forms of 2, 3, 4;
3. dvije, trii, četirii - behave as quantifiers with default feminine feature which assign genitive
case to the feminine paucal complement;
4. dva, tri2, četiri2 - behave as quantifiers with default neuter features which assign genitive to
the masculine or neuter paucal complement;
5. milion2 - behaves as a fully declinable masculine noun which assigns genitive to a plural
complement of any gender;
6. desetina, stotina, hiljada, milijarda - behave as fully declinable feminine nouns which
assign genitive to a plural complement of any gender;
7. pet, ... and all the rest - behave as quantifiers with default neuter features which assign
genitive to a plural complement of any gender.
We can conclude from this that majority of cardinal numerals behave as quantifiers or
nouns, and may be classified in two groups:
a) cardinal numerals that behave as quantifiers having only one form;
b) cardinal numerals that behave as frilly declinable nouns.
In the (a) group, there are three quantifiers with default feminine features, whereas all
other quantifiers have default neuter features. All quantifiers with default neuter features are
also singular, except dva, tri2, četiri2 which are paucal. In the group (b), there is one numeral
of masculine gender - milion2, the rest are feminine. Cardinal numerals are listed in (121):
70
(121) 1 jedan, jedna, jedno

2 dva, dvije
3 tri
4 četiri
5 pet
6 šest
7 sedam
8 osam
9 devet
11 jedanaest
12 dvanaest
13 trinaest
14 četrnaest
15 petnaest
16 šesnaest
17 sedamnaest
18 osamnaest
19 devetnaest
10 deset
20 dvadeset
30 trideset
40 četrdeset
50 pedeset
60 šezdeset
70 sedamdeset
80 osamdeset
90 devedeset
100 sto, stotina, stotinu

200 dvjesto, dvjesta, dvije stotine
300 tristo, trista, tri stotine
400 četiristo, četiri stotine
500 petsto, pet stotina
600 šesto, šest stotina
700 sedamsto, sedam stotina
800 osamsto, osam stotina
900 devetsto, devet stotina
1000 hiljada, hiljadu, tisuća, tisuću

2000 dvije hiljade, dvije tisuće
3000 tri hiljade, tri tisuće
4000 četiri hiljade, četiri tisuće
5000 pet hiljada, pet tisuća
1.000. 000 milion

2.000. 000 dva miliona
3.000. 000 tri miliona
4.000. 000 četiri miliona
5.000. 000 pet miliona
71
1.000. 000.000 milijarda, milijardu

2.000. 000.000 dvije milijarde
3.000. 000.000 tri milijarde
4.000. 000.000 četiri milijarde
5.000. 000.000 pet milijardi
2.2. Ordinal numerals
2.2.1. Prvi (1st)
The ordinal numeral prvi 'first' behaves as a true adjective. It is a syntactic modifier of
a head noun in a noun phrase, and it agrees with the head noun in number, gender and case, as
any other adjective. The ordinal numeral prvi has separate forms for three genders (masculine
(prvi), feminine (prva), and neuter (prvo)), two numbers (singular (prvi, prva, prvo) and
plural (prvi, prve, prva)), and six cases (nominative, genitive, dative, accusative, instrumental,
and locative). Prvi has an adjectival declension, and it declines as a definite form adjective.
To illustrate agreement of prvi with the head noun, I give the following examples with
nouns of masculine (122), feminine (123), and neuter gender (124):
a. Prvi muškarac je spavao.

first-MASC NOM man-MASC NOM AUX slept-3P SG MASC
b. Bojao se prvog muškarca.

(he) was afraid of first-MASC GEN man-MASC GEN
c. Pisao je prvom muškarcu.

(he) wrote (to) first-MASC DAT man-MASC DAT
d. Vidio je prvog muškarca.

(he) saw first-MASC ACC man-MASC ACC
e. Išaoje sa prvim muškarcem.

(he) walked with first-MASC INS man-MASC INS
f. Pričao je o prvom muškarcu.

(he) talked about first-MASC LOC man-MASC LOC
a. Prva žena je spavala.

first-FEM NOM woman-FEM NOM AUX slept-3P SG FEM
b. Bojao se prve žene.

(he) was afraid of first-FEM GEN woman-FEM GEN
c. Pisao je prvoj ženi.

(he) wrote (to) first-FEM DAT woman-FEM DAT
d. Vidio je prvu ženu.

(he) saw first-FEM ACC woman-FEM ACC
72
e. Išao je sa prvom ženom.

(he) walked with first-FEM INS woman-FEM INS
f. Pričao je 0 prvoj ženi.

(he) talked about first-FEM LOC woman-FEM LOC
(124) a. Prvo govedo je spavalo.

first-NEUTNOM head of cattle-NEUT NOM AUX slept-3P SG NEUT
b. Bojao se prvog goveda.

(he) was afraid of first-NEUT GEN head of cattle-NEUT GEN
c. Pisao je prvom govedu.

(he) wrote (to) first-NEUT DAT head of cattle-NEUT DAT
d. Vidio je prvo govedo.

(he) saw first-NEUT ACC head of cattle-NEUT ACC
e. Išao je sa prvim govedom.

(he) walked with first-NEUT INS head of cattle-NEUT INS
f. Pričao je 0 prvom govedu.

(he) talked about first-NEUT LOC head of cattle-NEUT LOC
In all examples given in (122)-(124) the ordinal numeral may be left out without any
change of the morphological form of remaining lexical items in the sentence. This means that
the ordinal numeral first is an adjectival modifier which always receives gender, number, and
case features from the head noun by the feature spreading mechanism.
Examples in (122)-(124) illustrate agreement with the head noun in gender, case and
number (singular). What about agreement in plural? The following examples will illustrate
that type of number agreement:
(125) a. Prvi muškarci su spavali.

first-MASC NOM PL man-MASC NOM PL AUX slept-3P PL MASC
b. Bojao se prvih muškaraca.

(he) was afraid of first-MASC GEN PL man-MASC GEN PL
c. Pisaoje prvim muškarcima.

(he) wrote (to) first-MASC DAT PL man-MASC DAT PL
d. Vidio je prve muškarce.

(he) saw first-MASC ACC PL man-MASC ACC PL
e. Išao je sa prvim muškarcima.

(he) walked with first-MASC INS PL man-MASC INS PL
f. Pričao je 0 prvim muškarcima.

(he) talked about first-MASC LOC PL man-MASC LOC PL
73
(126) a. Prve žene su spavale.

first-FEM NOM PL woman-FEM NOM PL AUX slept-3P PL FEM
b. Bojao se prvih žena.

(he) was afraid of first-FEM GEN PL woman-FEM GEN PL
c. Pisao je prvim ženama.

(he) wrote (to) first-FEM DAT PL woman-FEM DAT PL
d. Vidio je prve žene.

(he) saw first-FEM ACC PL woman-FEM ACC PL
e. Išaoje sa prvim ženama.

(he) walked with first-FEM INS PL woman-FEM INS PL
f. Pričao je 0 prvim ženama.

(he) talked about first-FEM LOC PL woman-FEM LOC PL
(127) a. Prva goveda su spavala.

first-NEUT NOM PL head of cattle-NEUT NOM PL AUX slept-3P PL NEUT
b. Bojao se prvih goveda.

(he) was afraid of first-NEUT GEN PL head of cattle-NEUT GEN PL
c. Pisao je prvim govedima.

(he) wrote (to) first-NEUT DAT PL head of cattle-NEUT DAT PL
d. Vidio je prva goveda.

(he) saw first-NEUT ACC PL head of cattle-NEUT ACC PL
e. Išao je sa prvim govedima.

(he) walked with first-NEUT INS PL head of cattle-NEUT INS PL
f. Pričao je 0 prvim govedima.

(he) talked about first-NEUT LOC PL head of cattle-NEUT LOC PL
In all these examples first behaves syntactically as an adjective, and it must agree with
the noun it modifies in gender, number, and case. Depending on gender and number of the
head noun,./;rat will have one of the following six forms in subject nominative phrases:
(128) _ _ _ _________ _____________________

MASC FEM NEUT
SG prvi prva prvo
PL prvi prve prva
2.2.2. Drugi (2nd), etc.
All other ordinal numerals behave syntactically in the same way as first. It means that
ordinal numerals in Bosnian behave syntactically as adjectives. In (129) the ordinal numerals
are listed, following the pattern of cardinal numerals in the section 2.1.16:
74
(129) singular plural

1st prvi, prva, prvo prvi, prve, prva
2nd drugi, dmga, drugo drugi, dmge, druga
3rd treći, treća, treće treći, treće, treća
4th četvrti, četvrta, četvrto četvrti, četvrte, četvrta
5th peti, peta, peto peti, pete, peta
6th šesti, šesta, šesto šesti, šeste, šesta
7th sedmi, sedma, sedmo sedmi, sedme, sedma
8th osmi, osma, osmo osmi, osme, osma
9th deveti, deveta, deveto deveti, devete, deveta
llth jedanaesti, jedanaesta, jedanaesto jedanaesti, jedanaeste, jedanaesta

12th dvanaesti, dvanaesta, dvanaesto dvanaesti, dvanaeste, dvanaesta
13th trinaesti, trinaesta, trinaesto trinaesti, trinaeste, trinaesta
14th četrnaesti, četrnaesta, četrnaesto četrnaesti, četrnaeste, četrnaesta
15th petnaesti, petnaesta, petnaesto petnaesti, petnaeste, petnaesta
16th šesnaesti, šesnaesta, šesnaesto šesnaesti, šesnaeste, šesnaesta
17th sedamnaesti, sedamnaesta, sedamnaesto sedamnaesti, sedamnaeste, sedamnaesta
18th osamnaesti, osmanaesta, osamnaesto osamnaesti, osamnaeste, osamnaesta
19th devetnaesti, devetnaesta, devetnaesto devetnaesti, devetnaeste, devetnaesta
10th deseti, deseta, deseto deseti, desete, deseta

20th dvadeseti, dvdeseta, dvadesto dvadeseti, dvadesete, dvadeseta
30th trideseti, trideseta, trideseto trideseti, tridesete, trideseta
40th četrdeseti, četrdeseta, četrdeseto četrdeseti, četrdesete, četrdeseta
50th pedeseti, pedeseta, pedeseto pedeseti, pedesete, pedeseta
60th šezdeseti, šezdeseta, šezdeseto šezdeseti, šezdesete, šezdeseta
70th sedamdeseti, sedamdeseta, sedamdeseto sedamdeseti, sedamdesete, sedamdeseta
80th osamdeseti, osamdeseta, osamdeseto osamdeseti, osamdesete, osamdeseta
90th devedeseti, devedeseta, devedeseto devedeseti, devedesete, devedeseta
100th stoti, stota, stoto stoti, stote, stota

200th dvjestoti, dvjestota, dvjestoto dvjestoti, dvjestote, dvjestota
300th tristoti, tristota, tristoto tristoti, tristote, tristota
400th četiristoti, četiristota, četiristoto četiristoti, četiristote, četiristota
500th petstoti, petstota, petstoto petstoti, petstote, petstota
600th šestoti, šestota, šestoto šestoti, šestote, šestota
700th sedamstoti, sedamstota, sedamstoto sedamstoti, sedamstote, sedamstota
800th osamstoti, osamstota, osamstoto osamstoti, osamstote, osamstota
900th devetstoti, devetstota, devetstoto devetstoti, devetstote, devetstota
lOOOthhiljaditi, hiljadita, hiljadito hiljaditi, hiljadite, hiljadita

5000th pet hiljaditi, pet hiljadita, pet hiljadito pet hiljaditi, pet hiljadite, pet hiljadita
1.000.000th milioniti, milionita, milionito milioniti, milionite, milionita

5.000.000th pet milioniti, pet milionita, pet milioniti, pet milionite, pet milionita
1.000. 000.000th milijarditi, milijardita, milijardito milijarditi, milijardite, milijardita

5.000. 000.000th pet milijarditi, pet milijadita, pet milijarditi, pet milijardite,
75
In complex ordinal numerals only the last element has the form of the ordinal numeral,
whereas the first part occurs in the form of the cardinal numeral, as in 21st - dvadest prvi, and
not *dvadeseti prvi.
2.3. Collective numerals
2.3.1. Dvoje (2)
The so-called collective numeral dvoje (2) belongs to the class of numerals that behave
as quantifiers and require the so-called collective nouns, such as djeca ‘children’, pilad
‘chicken’, dugmad ‘buttons’, etc., as their complement.
The numeral dvoje assigns genitive to its complement and triggers neuter singular
agreement on the predicate, as in (130a), where we find the neuter singular form stiglo. It can
appear in nominative contexts, as in (130a), accusative, as in (130b), and genitive contexts, as
in (130c), as well as in the complement of prepositions, as in (130e, f, g), but not in dative
contexts, as in (130d), unless there is a preposition assigning dative, as in (130e), or any other
oblique case, like instrumental in (130f), or locative in (130g):
(130) a. Dvojedjece je s tig lo .

tw o ch ild ren -G E N A U X -3PS.SG arrived -N EU T .S G
b. Vidim dvoje d je c e .
(I) see two ch ild ren -G E N
c. Bojim se dvoje d jece.

(I) am afraid of two ch ild ren -G E N
\ d. *Pišem dvoje d je c e .
(I) write to two ch ild ren -G E N
e. Gledam prema dvoje d je c e .

(I) look towards two ch ild ren -G E N
f. Idem sa dvoje d jece

(I) walk with two ch ild ren -G E N
g. Pričamo dvoje d je c e .
(I) talk about two ch ild ren -G E N
On the basis of these examples we could conclude that dvoje is a quantifier which has
only one form, and it assigns genitive to its complement, and also triggers neuter singular
agreement of the predicate. So it behaves syntactically exactly as the cardinal numeral pet
‘five’ with the only difference that it requires a specific type of noun as its complement - a
so-called collective noun.
However, there is an important difference. Pet indeed has only one form, whereas
dvoje actually has different case forms: dvoga or dvojega in genitive, dvoma, dvome or
dvojemu in dative and locative, and dvoma in instrumental. We could suspect then that dvoje
is parallel to the cardinal numeral dva! dvije in being a quantifier in direct case contexts, and
an adjective in oblique case contexts. If so, we would expect that in oblique case contexts we
can find examples with oblique case forms of dvoje modifying a noun and agreeing in case
with it. Let us consider the examples in (131):
76
(131) a. Dvoje djece je spavalo.

two children-NEUT GEN PL slept-3P NEUT SG
b. *Bojao se dvoga djece.

(he) was afraid of two-GEN children-NEUT GEN PL
c. *Pisaoje dvoma djeci.

(he) wrote (to) two-DAT children-NEUT DAT PL
d. Vidio je dvoje djece.

(he) saw two children-NEUT GEN PL
e. *Išaoje sa dvoma djecom.

(he) walked with two-INS children-NEUT INS PL
f. *Pričaoje o dvoma djeci.

(he) talked about two-LOC children-NEUT LOC PL
Examples in (131b, c, e, f) show that it is not possible to have the forms of dvoje
agreeing in case with the folowing noun. On the basis of this we may conclude that oblique
forms of dvoje do not behave as adjectives.
If these forms do not behave as adjectives, then we could suspect that they behave as
nouns, having different case forms in oblique cases, in the same way as the cardinal numeral
stotina behaves as a noun with full nominal declension, always assigning genitive case to its
complement. If so, we would expect that in oblique case contexts we can find examples with
oblique case forms of dvoje followed by a genitive form of the collective noun. Let us
consider the examples in (132):
(132) a. Dvoje djece je spavalo.

two children-NEUT GEN PL slept-3P NEUT SG
b. *Bojao se dvoga djece.

(he) was afraid of two-GEN children-NEUT GEN PL
c. *Pisao je dvoma djece.

(he) wrote (to) two-DAT children-NEUT GEN PL
d. Vidio je dvoje djece.

(he) saw two children-NEUT GEN PL
e. *Išao je sa. dvoma djece.

(he) walked with two-INS children-NEUT GEN PL
f. *Pričao je 0 dvoma djece.

(he) talked about two-LOC children-NEUT (*EN PL
The fact that examples (132b, c, e, f) are unacceptable in the same way as (13 lb, c, e,
f) would mean that oblique forms of the collective numeral dvoje exhibit a very peculiar
syntactic behaviour which resembles neither adjectives nor nouns. So, what are actually the
contexts in which the oblique forms of dvoje occur? First, these forms are used very rarely
77
and second, the contexts are limited to expressions in which oblique forms of dvoje occur
without any complement, as in (133):
(133) a. Šta izabrati od toga dvoga?

what choose-INF from those-GEN SG two-GEN
b. Dajem poklone njima dvoma.

(I) give presents (to) them-DAT two-DAT
On the basis of these examples we conclude that the oblique forms of dvoje behave as
nouns, but nouns which cannot take any complement and must have some premodifier which
agrees in case with the corresponding form of dvoje.
2.3.2. Troje (3), etc.
All other collective numerals behave syntactically in the same way as dvoje. It means
that collective numerals in Bosnian behave syntactically as quantifiers when used in direct
case contexts. Collective numerals are listed in (134):
(134) 2 dvoje
3 troje
4 četvoro, četvero
5 petoro, petero,
6 šestoro, šestero
7 sedmoro, sedmero
8 osmoro, osmero
9 devetoro, devetero
11 jedanaestero, jedanaestero
12 dvanaestoro, dvanaestoro
13 trinaestoro, trinaestero
14 četrnaestom, četmaestero
15 petnaestoro, petnaestero
16 šesnaestoro, šesnaestero
17 sedamnaestoro, sedamnaestom
18 osmanaestoro, osamanestero
19 devetnaestom, devetnaestero
10 desetoro, desetero
20 dvadesetero, dvadesetero
30 tridesetom, tridesetom
40 četrdesetoro, četrdesetoro
50 pedesetom, pedesetero
60 šezdesetom, šezdesetom
70 sedamdesetom, sedamdesetom
80 osamdesetoro, osamdesetero
90 devedesetom, devedesetom
It should be pointed out that the largest collective numeral is 99 - devedet devetoro.
The forms *stotoro, *stotero, *stotinoro, *stotinero, *hiljadoro, *hiljadero, *milionoro,
*milionero, *milijardoro, *milijardero do not exist. In complex collective numerals, like 99,
78
only the last element has the form of the collective numeral, whereas the first element has the
form of the cardinal numeral.
In additon to oblique case forms of dvoje, there are also oblique case forms of
collective numerals troje and četvoro, which also behave syntactically as nouns that occur
without any complement. The oblique forms of troje are: troga or trojega in genitive, troma,
trome or trojemu in dative and locative, and troma in instrumental. The oblique forms of
četvoro/ četvero are: četvorgal četverga in genitive, četvormal četverma, četvoromel
četverome in dative and locative, and četvormal četverma in instrumental. However, these
oblique forms sound rather archaic, and they are rarely used, especially the oblique forms of
the collective numeral four.
2.3.3. Jedni, jedne, jedna (1)
Plural counterparts of cardinal numerals, like jedni, jedne, jedna 'one- M ASC/ FEM/
are treated in traditional grammars as plural forms of collective numerals. They behave
n e u t ’,
as adjectives and they modify only specific kinds of norms. These are nouns which are plural
in form, but singular in meaning, known as pluralia tantum nouns, like svatovi ‘wedding
procession’, naočale ‘spectacles, glasses’, kola ‘car, carriage’, etc., as in (135):
(135) a. Jedni svatovi su stigli.

one-N O M .M A SC w e d d in g p rocession -N O M .M A S C arrived -M .P L
b. Vidim jedne svatove.

(I) s e e one-A C C .M A SC w e d d in g p ro cession -A C C .M A S C
Nouns with similar characteristics are those denoting a pair of separate items which
are used together and regarded as a unified entity, like rukavice ‘(a pair of) gloves’, cipele ‘(a
pair of) shoes’, opanci ‘(a pair of) peasant shoes’, etc. All these nouns may occur in quantified
phrases, but only with what is traditionally described as plural forms of collective numerals.
These numerals have three forms for three genders: masculine, as in (136a), feminine, as in
(136b), and neuter, as in (136c):
(136) a. Jedni sv a to v i s u s tig li,

one-M A SC w e d d in g p r o c e s s io n s -M A S C arrived-M A SC .PL
b. Jedne naočale s u r a z b ije n e ,

one-F EM g la sses-F E M w e r e broken-FE M .PL
c. Jedna k o la s u s tig la .
one-N E U T carriages-N E U T arrived-N EU T .PL
Numerals modifying nouns in (136) are also completely adjectival, agreeing with a noun in
gender and case. Masculine forms are jedni-NOM, jedne-ACC, jednih-GEN, jednim-
DAT/INS/LOC. Feminine forms are: yWne-NOM/ACC.yWnzTi-GEN.y'ednzm-DAT/INS/LOC.
Neuter forms are: yW«a-NOM/ACC,yW«i/!-GEN,yWnz'm-DAT/INS/LOC.
2.3.4. Dvoji, dvoje, dvoja (2), etc.
All other plural forms of collective numerals behave syntactically in the same way as
jedni. It means that plural forms of collective numerals in Bosnian behave syntactically as
adjectives. The nominative forms of these numerals are listed in (137):
79
(137) 1 jedni, jedne, jedna

2 dvoji, dvoje, dvoja
3 troji, troje, troja
4 četvori, četvore, četvora
5 petori, petore, petora
6 šestori, šestore, šestora
7 sedmori, sedmore, sedmora
8 osmori, osmore, osmora
9 devetori, devetore, devetora
11 jedanaestori, jedanaestom, jedanaestom

12 dvanaestori, dvanaestom, dvanaestom
13 trinaestori, trinaestom, trinaestom
14 četmaestori, četrnaestom, četrnaestom
15 petnaestori, petnaestom, petnaestom
16 šesnaestori, šesnaestom, šesnaestom
17 sedamnaestori, sedamnaestom, sedamnaestom
18 osmanaestori, osamanestore, osamnaestom
19 devetnaestori, devetnaestom, devetnaestom
10 desetari, desetare, desetara

20 dvadesetori, dvadesetom, dvadesetom
30 tridesetori, tridesetom, tridesetom
40 četrdesetori, četrdesetom, četrdesetom
50 pedesetori, pedesetom, pedesetom
60 šezdesetori, šezdesetom, šezdesetom
70 sedamdesetori, sedamdesetom, sedamdesetom
80 osamdesetori, osamdesetom, osamdesetom
90 devedesetori, devedesetom, devedesetom
Here again the largest plural collective numeral is 99 - devedet devetori, devedest
devetore, devedeset devetora. In complex plural collective numerals, like 99, only the last
element has the form of the plural collective numeral, whereas the first element has the form
of the cardinal numeral.
2.4. Other numerals
2.4.1. Dvojica (2)
There are numerals ending in -ica: e.g. dvojica. These numerals require a human
masculine noun as their complement to which they assign genitive case. These numerals
behave like the so-called collective nouns such as djeca ‘children’. They have the nominal
feminine declension, but syntactically they behave as neuter plural norms, so that the predicate
shows up in a neuter plural form when the subject phrase contains one of these numerals, as in
(138a). They also may occur in any case context, including direct cases, nominative in (138a),
accusative in (138b), and oblique cases, like dative in (138c):
(138) a. Dvojica vojnika su stigla.

tw o-N O M so ld ie r s-G E N AUX -3PS.PL arrived-N EU T .PL
80
b. V id im d v o j ic u v o jn ik a .
(I ) s e e twO-ACC so ld ie r s-G E N
c. B o j im se d v o j ic e v o jn ik a .
( I ) a m a fr a id ( o f ) t w o - GEN so ld ie r s-G E N
d. P iše m d v o j ic i v o jn ik a .
(I ) w r ite (to ) tw o -D A T so ld ie r s-G E N
e. Idem sa d v o j ic o m v o jn ik a
(I ) w a lk w ith t w o - INS so ld ie r s-G E N
f. G o v o r im 0 d v o j ic i v o jn ik a
(I ) ta lk about t w o T LOC so ld ie r s-G E N
2.4.2. Trojica (3), etc.
All other numerals in -ica behave syntactically in the same way as dvojica. It means
that all numerals of this type behave as nouns. The nominative forms of these numerals are
listed in (139):
(139) 2 dvojica
3 trojica
4 četvorica, četverica
5 petorica, peterica
6 šestorica, šesterica
7 sedmorica, sedmerica
8 osmorica, osmerica
9 devetorica, deveterica
11 jedanaestorica, jedanaesterica
12 dvanaestorica, dvanaesterica
13 trinaestorica, trinaesterica
14 četmaestorica, četmaesterica
15 petnaestorica, petnaesterica
16 šesnaestorica, šesnaesterica
17 sedamnaestorica, sedamnaesterica
18 osamnaestorica, osamnaesterica
19 devetnaestorica, devetnaesterica
10 desetorica, deseterica
20 dvadesetorica, dvadeseterica
30 tridesetorica, trideseterica
40 četrdesetorica, četrdeseterica
50 pedesetorica, pedeseterica
60 šezdesetorica, šezdeseterica
70 sedamdesetorica, sedamdeseterica
80 osamdesetorica, osamdeseterica
90 devedestorica, devedeseterica
81
Here again the largest numeral in -ica is 99 - devedeset devetorica. In complex

numerals in -ica, like 99, only the last element has the form of the plural collective numeral,
whereas the first element has the form of the cardinal numeral.
2.4.3. Jedinica (1)
There is a type of numerals in Bosnian that are purely nominal. They are not used to
quantify a noun, but rather to denote a name of the numeral, as in English 'bus number one'. In
Bosnian it is possible, of course, to use an expresion corresponding to English and say
'autobus broj jedan', but this can be also expressed just with a special form of the numeral -
jedinica, as in the following examples:
(140) a. Autobus broj jedan je stigao.

bus-SG MASC number one arrived-3P SG MASC
b. Jedinica je stigla
one-NOM SG FEM arrived-3P SG FEM
This numeral behaves as a feminine noun with full singular and plural declensions.
The singular forms are: jedinica-NOM, jedinicu-ACC, jedinice-GW , jedinici-DAT/LOC,
jedinicom-INS. The plural forms are: jedinice-NOMIACC, jedinica-GEN, jedinicama-
DAT/INS/LOC
2.4.4. Dvica, dvojka (2), etc.
All other numerals of this type behave syntactically in the same way as jedinica. It
means that all numerals of this type behave as nouns. However, whereas jedinica has just one
form in nominative singular, other numerals have two or three forms in nominative singular.
The nominative forms of these numerals are listed in (141):
(141) 1jedinica
2 dvica, dvojka
3 trica, trojka
4 četvrtica, četvorka
5 petica, petorka
6 šestica, šestorka
7 sedmica, sedmorka
8 osmica, osmorka
9 devetica, devetka, devetorka
11 jedanaestica, jedanaestka, jedanaestorka

12 dvanaestica, dvanaestka, dvanestorka
13 trinaestica, trinaestka, trinaestorka
14 četmaestica, četmestka, četmaestorka
15 petnaestica, petnaestka, petnaestorka
16 šesnaestica, šesnaestka, šesnaestorka
17 sedamnaestica, sedamnaestka, sedamnaestorka
18 osamnaestica, osamnaestka, osmanaestorka
19 devetnaestica, devetnaestka, devetnaestorka
82
10 desetica, desetka, desetorka

20 dvadesetica, dvadestka, dvadesetorka
30 tridesetica, tridestka, tridesetorka
40 četrdesetica, četrdesetka, četrdesetorka
50 pedesetica, pedesetka, pedesetorka
60 šezdesetica, šezdesetka, šezdesetorka
70 sedamdesetica, sedamdesetka, sedamdesetorka
80 osmadesetica, osmadesetka, osamdesetorka
90 devedesetica, devedestka, devedesetorka
Notice that the English sentence 'Five presidents arrived' can be expressed in four
different ways in Bosnian:
(142) a. Pet predsjednika je stiglo.

five presidents-GEN PL MASC arrived-3P NEUT SG
b. Petorica predsjednika su stigla.

five presidents-GEN PL MASC arrived-3P NEUT PL
c. Petoro predsjednika je stiglo.

five presidents-GEN PL MASC arrived-3P NEUT SG
d. Petorka je stigla.
five arrived-3P FEM SG
e. *Petica je stigla.
five arrived-3P FEM SG
In (142a) the cardinal numeral pet is used. It behaves as a quantifier assigning genitive
case to its complement and it triggers neuter singular form of the predicate. In (142b) the
numeral form in -ica is used to denote that all presidents are male. This numeral behaves as a
noun which assigns genitive case to its complements and triggers neuter plural agreement
form of the predicate. In (142c) the collective numeral petoro is used, which indicates that the
presidents are not of the same sex. This numeral behaves as a quantifier which assigns
genitive case to its complement and triggers neuter singular agreement of the predicate.
Finally, the expression 'five presidents' may be replaced by a special nominal numeral
petorka, which indicates that the presidents form a group of five persons. This numeral
behaves as a feminine noun and it triggers feminine singular agrement form of the predicate.
Notice, that another form of the nominal numeral petica, as in (142e), cannot be used to
denote a group of five persons. However, (142e) is fully grammatical sentence with a possible
meaning 'The bus number five arrived'.
2.4.5. Approximative numerals
Some cardinal numerals have special forms ending in -ak which are used to express
the approximative value, e.g. desetak means 'not exactly ten, close to ten', and it may be either
less than ten, or more than ten, but close to ten. There are few numerals that may express the
approximative value in this way: most frequently numerals denoting tens may have these
forms, and stotina 'hundred', as in (143):
83
(143) 10 desetak
20 dvadesetak
30 tridesetak
40 četrdesetak
50 pedesetak
60 šezdesetak
70 sedamdesetak
80 osamdesetak
90 devedesetak
100 stotinjak
These numerals behave syntactically as nouns assigning genitive to its plural

complement and triggering neuter singular form of the predicate, as in (144):
(144) Desetak studenata je došlo.

10 students-GEN PL arrived-3P SG NEUT
The numerals from 'two' to 'nine' express the approximative value by putting two
cardinala numerals next to each other: dva-tri, tri-četiri, četiri-pet, pet-šest, šest-sedam,
sedam-osam, osam-devet.
2.4.6. Fractions
The numerals expressing fractions have the forms listed in (145):
(145) 1/2 polovina

1/3 trećina
1/4 četvrtina
1/5 petina
1/6 šestina
1/7 sedmina
1/8 osmina
1/9 devetina
1/11 jedanaestina
1/12 dvanaestina
1/13 trinaestina
1/14 četmaestina
1/15 petnaestina
1/16 šesnaestina
1/17 sedamnaestina
1/18 osamnaestina
1/19 devetnaestina
1/10 desetina
1/20 dvadesetina
1/30 tridesetina
1/40 četrdesetina
1/50 pedesetina
1/60 šezdesetina
84
1/70 sedamdesetina
1/80 osamdesetina
1/90 devedesetina
1/100 stotnina
Fractions behave syntacticaly as nouns of feminine gender with full singular and plural
declensions. They assign genitive case to the plural complement, as in (146):
(146) a. Polovina studenata je došla.

one half (of) students-GEN PL arrived-3P SG FEM
b. Vidio je polovinu studenata.

(he) saw one half-ACC SG (of) students-GEN PL
2.4.7. Distributive numerals
The expressions consiting of the so-called distributive preposition po and a cardinal

numeral express a distributive meaning which roughly corresponds to English 'each': po dva
'two each', po tri 'three each', etc. Bussmann (1996) gives the following definition of
distributives: ‘Subgroup of numerals formed by words or phrases such as apiece, each, per
person, where the units in question are distributed to some or all members of a group
individually: They will receive six books each’ (p. 136).
It is standardly assumed that po is a preposition in distributive expressions. However,
syntactic behaviour of distributive po-phrases in Bosnian shows that po is not a preposition in
such constructions, but rather a specific kind of a quantifier. It is typical for prepositions that
they assign various cases to their nominal complements, and we would expect that po also
assigns case to its complement. However, consider the following examples:
(147) a. Po jedan student ulazi svaki sat.

DIST one-NOMMASC SGstudent-NOM MASC SG enters each hour
b. Odriču se po jednog zahtjeva svakog mjeseca.

reject-3PL REFL DIST one-GEN M request-GEN M each month
‘They renounce one request each month.’
c. Nagradu dodjeljuju po jednom studentu svake godine.

award-ACC give-3PL DIST one-DAT student-DAT each year
‘They give the award to one student each year.’
d. Svaki student je pročitao po jednu knjigu,

each student AUX-3SG read DIST one-ACCMASC book-AC
e. Svaki knez vlada po jednim gradom.

each prince rules DIST one-INST MASC town-INST MASC
f. Svaki student je govorio o po jednom dogadjaju.

each student AUX-3SG spoke about DIST one-LOC M event-LOC M
These examples show that po doesn’t assign its own case, as already observed by
Dickey (1992). Rather, the appropriate case to the complement of po is always assigned by
85
another case assigner. locative in (147f) is assigned by a preposition o ‘about’, nominative in

(147a) by INFL, whereas genitive (147b), dative (147c), accusative (147d), and instrumental
(147e) axe assigned by respective verbs. Prepositions in Bosnian obligatorily assign case,
therefore we may conclude that po in (147) is not a preposition. It is an element with a
quantificational force, and therefore I include it in the category of quantifiers. It differs,
however, from other quantifiers in selectional properties, taking exclusively complements
containing a cardinal numeral. And it lacks case assigning potential. Neither can it receive a
case.
To support the claim that po in (147) is not a preposition, and that it does not assign
case, I offer the examples in (148) which should be compared with (147e):
(148) a. * Svaki knez vlada po dva grada.
b. * Svaki knez vlada po pet gradova.
c. Svaki knez vlada sa po dva grada.
d. Svaki knez vlada sa po pet gradova.
The ungrammatical structures in dative and instrumental contexts can be rescued by

inserting a preposition which has potential to assign dative and instrumental case,
respectively. The presence of the preposition which lacks that case assigning potential, like
po, cannot rescue the structure as obvious from ungrammatical (148a,b). I still claim,
however, that sa in (148c,d) doesn’t asssign instrumental to the quantified complement. It
only serves to licence the structure and actually performs the role of the instrumental case
ending since quantifiers lack morphological case endings to indicate that the instrumental was
assigned.
86
3. Concluding considerations
3.0. Introduction
In this chapter I would like to summarize the discussion from Chapter 2 concentrating
on some theoretical considerations regarding the syntactic behaviour of numerals in Bosnian.
Chapter 2 was mainly concerned with the description of various types of numerals and
Chapter 3 will mainly deal with theoretical considerations. However, theoretical
considerations are based on real language data, so it is inevitable to repeat parts of the earlier
discussion in this chapter also. Those readers who are mainly interested in theoretical
considerations, and want to skip Chapter 2, will find in this chapter crucial descriptive facts
that were extensively dealt with earlier.
3.1. The syntactic behaviour o f numerals in Bosnian
In Chapter 2 at least eight types of numerals in Bosnian were singled out, and taking
‘five’ as an example, we have forms pet, peti, petoro, petori, petorica, petica, petina, po pet
representing the various types of numerals. However, from the syntactic point of view, there
are no eight categories of numerals. Syntactically all these numerals behave as three distinct
categories, namely as quantifiers, nouns and adjectives.
Those numerals which behave as quantifiers, and those that behave as nouns are
syntactic heads assigning case to their complements, as in (la), where we find the quantifier
pet ‘five’ assigning genitive case to its complement vojnik ‘soldier’, and in (lb), where we
find the quantity norm stotina ‘hundred’ also assigning genitive case to its complement.
Notice the different agreement forms on the predicate, the default neuter singular form stiglo
in (la), and the feminine singular form stigla in (lb) in agreement with the quantity noun
stotina ‘hundred’. This shows that the semantic head of subject phrases in (la) and (lb),
namely vojnik, is not the syntactic head of these phrases, the syntactic heads being pet and
stotina, respectively. Those numerals that behave as adjectives are noun modifiers agreeing in
phi-features and case with the noun head, as in (lc), where we find the adjectival numeral peti
‘fifth’ agreeing in gender, number and case with the head noun vojnik ‘soldier’. Notice also
that the masculine singular form stigao in the predicate is in agreement with the head noun
vojnik ‘soldier’, indicating that vojnik is not only the semantic head, as in (la,b), but also the
syntactic head of the subject phrase in (lc):
(1) a. Pet v o jn ik a je s t ig lo .
five soldier-G E N .M A SC .PL A U X -3PS.SG arrived -N E U T .S G
b. Stotina v o jn ik a je s tig la .
hundred-NOM.FEM.SG soldier-G EN .M A SC .PL A U X -3PS.SG arrived-FE M .SG
c. Peti v o jn ik je s tig a o .
fifth-NOM.MASC.SG soldier-N O M .M A SC .SG aux - 3 p s .s g arrived-M A SC .SG
87
3.2. Numerals that behave like Quantity Nouns
I will first elaborate on the class of numerals as in (lb), that is numerals which behave
as nouns. They are fully declinable, and, like majority of other nouns13, they also assign
genitive case to their complement. Two groups of numerals are found in this category:
First, the numerals desetina ‘ten’, stotina ‘hundred’, hiljada ‘thousand’, milion
‘million’, and milijarda ‘billion’. All of them have the full declension as feminine nouns,
except milion, which declines as a masculine noun. These numerals take a genitive
complement and the predicate agrees with them in gender and number, as in (2a), where we
find the feminine singular form of the predicate stigla in agreement with the feminine singular
form stotina. This suggests that the syntactic head of the subject phrase in (2a) is the numeral
stotina, rather than the noun which follows stotina. Being fully declinable, these numerals
may occur not only in direct case contexts, like nominative in (2a), and accusative in (2b), but
also in oblique case contexts, like dative in (2c):
(2)a. Stotina vojnika/ vozila/ balerina je stigla.

hundred-NOM.FEM.SG soldiers-G EN / v ehicles-G E N / dancers-GEN arrived-FEM.SG
b. Vidim stotinu14 vojnika/ vozila/ balerina.

(I) se e hundred-ACC soldiers-G EN / v ehicles-G E N / dancers-GEN
13 Nouns are case-assigner in Bosnian and they typically assign genitive as their default, or
structural case, as in (i). However, some nouns may assign marked, or inherent cases, usually
instrumental and dative, as in (ii) and (iii):
(i) pisac knjige

writer book-GEN
‘a/ the writer of the book’
(ii) komandovanje batalj onima

commanding battalions-INS
‘commanding with battalions’
(iii) spomenik miru

monument peace-DAT
‘a/ the monument to peace’
14Notice that there is also a quantifier denoting ‘hundred’ (and also ‘thousand’ and ‘billion’),
in the form with -u ending: stotinu (and also hiljadu, tisuću, milijardu). This form is identical
to the accusative form of the noun stotina. However, the quantifier form ending in -u is used
only in direct case contexts, like other numerals that behave as quantifiers, for example pet
‘five’. The examples in (i) and (ii) illustrate that the quantifiers stotinu, hiljadu, and milijardu
indeed behave like the quantifier pet ‘five’ in all relevant syntactic respects:
(i) Pet/ stotinu/ hiljadu/ milijardu vojnika je stiglo.

fiv e / h u ndred/ th o u sa n d / b illio n soldier-GEN.PL arrived-NEUT.SG
(ii) Vidim pet/ stotinu/ hiljadu/ milijardu vojnika.

(I) s e e f iv e / hu ndred/ th o u sa n d / b illio n soldier-GEN.PL
In (i) we find neuter singular agreement on the verb, in contrast to (2a) with feminine singular
agreement, which clearly indicates that stotina behaves as a noun, and stotinu as a quantifier.
88
c. Pišem stotini vojnika/ balerina.

(I) w r ite (t o ) h u n d red -D A T s o ld ie r s-G E N / dan cers-G E N
Numerals that behave like quantity nouns can be modified by a demonstrative or by an

adjective that agrees with the numeral in phi-features and case, as in (4a), with nominative
forms ona 'that' and dobra 'good, at least' in agreement with nominative stotina 'hundred', and
as in (4b), with accusative forms onu and dobru in agreement with accusative stotinu:
(4) a. Ona/ dobra stotina vojnika je stigla.

th a t/ good-N O M .FE M h u n d red -N O M soldier-G E N .P L arrived-FE M .SG
‘T h a t/ a g o o d (a t le a s t ) o n e h u n d r e d m e n a r r iv e d .’
b. Vidim onu/ dobru stotinuvojnika.

(I) s e e that/ good-A C C .FE M h u n d red -A C C soldier-G E N .P L
These numerals do not allow movement of the demonstrative or the pronoun out of
their complement, so that (5b) is ungrammatical with the demonstrative ovih 'these' moved in
front of the numeral stotina 'hundred':
(5) a. Stotina ovih vojnika je stigla.

hu ndred th ese-G E N soldier-G E N .P L arrived-FE M .SG
b. *Ovih stotina15 vojnika je stigla.
The second group of numerals that behave as nouns are numerals ending in -ica:
dvojica, trojica, četvorica, etc., as in (7). These numerals require a human masculine noun as
their complement to which they assign genitive case. They behave like the so-called collective
nouns such as djeca ‘children’. These numerals, like collective nouns, have the nominal
feminine declension, but syntactically they behave as neuter plural norms, so that the predicate
shows up in a neuter plural form when the subject phrase contains one of these numerals, as in
(7a). They also may occur in any case context, including direct cases, nominative in (7a),
accusative in (7b), and oblique cases, like dative in (7c):
(7)a. Dvojica/ trojica/ četvorica/... vojnika su stigla.

tw o -N O M / th ree-N O M / four-N O M so ld ie r s-G E N aux - 3 p s .p l arrived-N E U T .P L
b. Vidim dvojicu/ trojicu/ četvoricu ... vojnika.

(I) s e e tw o -A C C / th ree-A C C / four-A C C so ld iers-G E N
15 If the numeral quantifier stotinu is used in the context of (5b), then the extraction of the
demonstrative is possible, as expected. So, although (5b) with the quantity noun stotina is
unacceptable, the corresponding examples with the quantifiers sto and stotinu would be
perfectly possible, as in (i):
(i) Ovih sto/ stotinu vojnika je stiglo.

th ese-G E N .P L hu ndred soldier-G E N .P L arrived -N E U T .SG
Notice in (i) a default neuter singular agreement in the predicate {stiglo), clearly indicating
that sto and stotinu are not quantity nouns, as stotina, but rather quantifiers. So, there is a
quantity noun stotina, with a full nominal declension; sto, a numeral quantifier comparable to
pet ‘five’; and also a quantifier form stotinu.
89
c. Pišem dvojici/ trojici/ četvorici ... vojnika.

(I) w r ite (t o ) tw o -D A T / th ree-D A T / fou r-D A T so ld iers-G E N
The complement of both kinds of numerals that behave as nouns can be a genitive
pronoun, like vas ‘you’ in (8).
(8) a. Stotina vas je stigla.

hundred you-G E N .P L aux - 3 p s .s g arrived-FE M .SG
'O n e h u n d r e d o f y o u a r riv ed .'
b. Dvojica vas su stigla.

two you-G E N .PL aux - 3 p s .p l arrived-N EU T .P L
'Two o f y o u a r riv ed .'
If the form of the predicate in (8) is kept constant and the order of the numeral and the
pronoun is reversed, then the resulting string is ungrammatical: the pronoun cannot precede
the numeral noun, as in (9):
(9) a. *Vas stotina je stigla.
b. *Vas dvojica su stigla.
However, if the form of the predicate is changed, then the sentence is acceptable, as illustrated
with the contrast in (10):
(1 0 ) a. * V a s d v o j ic a su s tig la .
you-G E N .PL tw o-N O M aux - 3 p s .p l arrived-N EU T .P L
; b. Vas dvojica ste stigli.

you-G EN /A C C t w o ( m a le s ) A UX-2PS.PL arrived-M A SC .PL
Notice the distinction in person in the predicate: 3rd person in (10a) vs. 2 nd person in (10b). So
in (10b) dvojica is no longer the head of the subject phrase, at least not the syntactic head.
Compare (8b) where the predicate has 3rd person plural neuter agreement, agreeing with the
head dvojica. Therefore, it means that (10b) is not a result of the movement of vas from the
complement of dvojica. I suspect that here we have an appositive structure: the head is vas,
and dvojica is in apposition to vas. The question is why we have the genitive/ accusative form
of the 2 nd person plural pronoun (vas), and not the nominative form (vi). Namely, the
nominative form vi appears in appositive structures with common nouns, as in (1 la):
(11) a. Vi s tu d e n ti s te s t ig li.
you-N O M .2PS .PL students-N O M .P L aux - 2 p s .p l arrived-M A SC .PL
b. *Vas s tu d e n ti s te s t ig li.
y o u - GEN/ACC students-N O M .P L AUX -2PS.PL arrived-M A SC .PL
We don’t want to say that a noun is a complement of a pronoun in (1 la), but it may be in
apposition to a pronoun. But the genitive/ accusative form vas is not possible in such contexts,
as ungrammatical (1 lb) illustrates.
However, vas, and not vi, may appear with all quantified expressions, but crucially,
always with the second person plural agreement form in the predicate, which tells us that the
90
predicate agrees with some head in the subject phrase which has second person plural
features. A quantifier, or a quantity noun for that matter, cannot have these features, and
consequently it is not the head of the subject phrase. Only a pronoun can have these features,
and therefore a pronoun must be the head of the subject phrase containing numerals like
dvojica, or the collective numeral dvoje, as in (12a), or the cardinal numeral dvije, as in (12c):
(12) a. Vas dvoje ste stigli.

you-G EN /A C C t w o AU X -2PS.PL arrived-M A SC .PL
‘You two (where ‘two’ maybe of mixed gender, or only masculine) arrived.’
b. *Vi dvoje ste stigli.
c. V as d v ij e s te s tig le .
you-G EN /A C C tw o-F E M AU X -2PS.PL arrived-FEM .PL
c. *Vi dvije ste stigle.
At this point, I don’t have a proposal for this mystery - why the genitive/accusative form vas,
and not the nominative vi, with quantified expressions.
In Giusti and Leko (2001), it was proposed that the categorial status of quantity
expressions that behave like the numerals discussed in this section is that of a Quantity Noun.
They have, in fact, their own functional projection where gender and case are realized. The
demonstrative that precedes them agrees with them, and it is therefore in Spec DP/KP.
Furthermore, they do not allow movement of their complement to a higher Spec. The
structure is given in (i):
(i)
DP/KP
ona stotina vojnika
3.3. Numeral Quantifiers
I will now elaborate on the class of numerals first illustrated in (la), that is numerals
that behave as quantifiers. They are indeclinable and take a genitive complement. Being
indeclinable, they may not appear in oblique case contexts.
Numerals which behave as quantifiers, contain three classes. First, cardinal numerals
from pet ‘five’ onwards, excluding compound numerals which end in ‘one’, ‘two’, ‘three’,
and ‘four’ (namely, numerals like ‘twenty-one, ‘twenty-two, etc.). So, cardinal numerals like
pet ‘five’ are found in direct case contexts, like accusative in (13a), but not in oblique case
contexts, like dative in (13b):
91
(13) a. Vidim pet/ šest/ sedam/ osam/ devet/ deset/.. .vojnika.

(I ) s e e f i v e / s ix / s e v e n / e ig h t / n in e / te n / sold ier-G E N .P L
b. *Pišem pet/ šest/ sedam/ osam/ devet/ deset/... vojnika.

T write to five/ six/ seven/ eight/ nine/ ten/ ... soldiers.’
These numerals cannot be modified by an agreeing adjective or demonstrative so that (14) is

ungrammatical. They allow a part of the complement noun phrase to move to SpecQP, as in
(15b) with a demonstrative/ adjective o v ih 'these'/ h ra b r ih brave' moved in front of the
numeral:
(14) *Ovo / h ra b ro pet v o jn ik a .

this-NO M .N EU T brave-N O M .N E U T f iv e soldier-G E N .P L
(15) a. Pet ovih/ h r a b r ih v o j n ik a j e s tig lo .

fiv e th ese-G E N b rave-G E N soldier-G E N .P L arrived -N E U T .SG
b. Ovih/ hrabrih pet vojnika je stiglo.

'These/ brave five soldiers arrived.'
Notice that in (15b) o v ih / h ra b r ih still modify the complement of the numeral. The
structure in (15b) derives from (15a) by movement of o v ih / h ra b r ih to the Spec of QP.16
Therefore, the interpretation of the two examples in (15) is basically the same.
Second, the so-called collective numerals, like d v o je , tro je , č e tv o r o , etc., also belong
to the class of numerals that behave as quantifiers. They require the so-called collective
nouns, such as d je c a ‘children’, p i l a d ‘chicken’, d u g m a d ‘buttons’, etc., as their complement.
These numerals also assign genitive to their complement and trigger neuter singular
agreement on the predicate, as in (16a), where we find the neuter singular form stig lo . They
can appear in nominative contexts, as in (16a), accusative, as in (16b), and genitive contexts,
as in (16c), as well as in the complement of prepositions, as in (16e), but not in dative
contexts, as in (16d), unless there is a preposition assigning dative, or any other oblique case,
as in (16e):
(16) a. Dvoje djece je stiglo.

tw o c h ild ren -G E N A U X -3PS.SG arrived -N EU T .S G
b. V id im d v o j e d je c e .
(I ) s e e t w o ch ild ren -G E N
c. B o j im s e d v o je d jece.
(I ) a m a fr a id o f tw o ch ild ren -G E N
d. * P iše m d v o je d jec e .
‘I w r ite t o t w o c h ild r e n .’
e. P o k a z u je m p rem a d v o je d je c e .
(I ) p o in t to w a r d s tw o ch ild ren -G E N
16 Corbett (1979) also argues that prequantifiers, as in (15b), are base-generated in a position
after p e t and then moved to a Spec position. This analysis is adopted also by Franks (1995).
Babby (1987), however, does not agree with this analysis for Russian.
92
Finally, the numerals that behave as quantifiers also include cardinal numerals dva
‘two’, tri ‘three’ and četiri ‘four’, but only when used in direct case contexts.1718Here we must
assume that dva has the default neuter features, like other quantifiers, but in contrast to other
quantifiers which may take complements of any gender, dva can combine only with
masculine and neuter nouns, as in (17a), whereas another form - dvije - is used with feminine
nouns, as in (17b):
(17) a. Dva vojnika/ vozila sustigla.

t w o soldiers-G E N .M A S C .PA U C / vehicles-G E N .N E U T .P A U C arrived-N E U T .PL
b. Dvije balerine su s tig le .

tw o dancers-G EN .FEM .PAU C AU X -3PS.PL arrived-FE M .PL
For tri and četiri we must assume that there are two homophonous forms of these
numerals: one which combines with masculine and neuter nouns, as in (18a), and one which
combines with feminine nouns, as in (18b):
(18) a. Tri/ č e tir i v o jn ik a / v o z ila s u s tig la .

th r e e / fo u r ■ soldiers-G EN .M A SC .PA U C / vehicles-G E N .N E U T .P A U C arrived-N E U T .P L
b. Tri/ č e tir i b a le r in e su s t i g l e . 18
th r e e / fo u r d an cers- n o m .f e m .p l a u x - 3 p s .p l arrived-FE M .PL
Masculine and neuter nouns that follow numerals dva, tri and četiri have a form
ending in -a, a genitive singular ending. However, it only appears to be a genitive singular
form, because when an adjective is added as a modifier of nouns in (17a) and (18a), it must
have the form which is certainly not genitive singular, as obvious from ungrammatical (19b):
(19) a. Ona dva/ tri/ četiri visoka vojnika/ vozila su stigla.

‘Those two/ three/ four tall soldiers/ vehicles arrived.’
b. * O n a d v a / tr i/ č e tir i v is o k o g v o j n ik a / v o z i l a s u s tig la .
tall-G EN .SG
17 The oblique forms of these numerals behave as adjectives.

181 assume that dva has the default neuter features, like all other quantifiers, and therefore we
find neuter forms on the verbs in (17a) and (18a). However, whereas all other quantifiers also
have default singular features, ‘two’, ‘three’, and ‘four’ are inherently paucal in grammatical
number, and the forms on the verb could be treated as paucal forms. But since these forms are
identical with plural neuter forms, we follow the generally accepted view (cf. Corbett (1983:
89-91)), that these should be treated as neuter plural forms. As far as (17b) and (18b), I claim
that the forms dvije, tri2, and četiri2have inherently feminine gender, and therefore we find a
feminine form of the verb. Crucially, I claim that numeral quantifiers are heads and the verbs
agree with them. They also assign case to their complements, but not grammatical number.
Number is not assigned, but numeral quantifiers ‘two’, ‘three’, ‘four’ require (select) paucal
nouns as their complements. In contrast to Franks (1995: 105) who treats paucal as a case,
here it is treated as a number. Also, in this system the numeral quantifiers ‘two’, ‘three’, and
‘four’ are not modifiers, but heads. In (17a) and (18a) they are nominative heads and verbs do
agree with them.
93
The sentence with the genitive singular form of the adjective v is o k o g in (19b) is unacceptable,
and only (19a) with the adjective form v is o k a is possible. The descriptive adjectives have two
forms, indefinite and definite: v is o k and v is o k i, respectively. The form v is o k a is the genitive
singular of the indefinite form, whereas v is o k o g is the genitive singular of the definite form.
However, the form v is o k a in (19a) cannot be the indefinite form of the adjective, because the
stress pattern is different, and also because the demonstrative is used, and by definition, the
demonstrative cannot be followed by an indefinite adjective. Therefore, v is o k a in (19a) cannot
be the genitive singular of the indefinite adjective v iso k , but rather the genitive dual/ paucal
form of the definite adjective v iso k i. Bosnian used to have three grammatical numbers -
singular, dual and plural. Dual was lost in the development of the language, but the remnants
survive in contexts with numerals d v a , tr i and č e tir i.
If v is o k a is genitive paucal, then v o jn ik a cannot be genitive singular, although the
form is identical to genitive singular, but must be genitive paucal, which happens to have the
same form as genitive singular. Therefore, we may conclude that the form of masculine and
neuter nouns after numerals d v a , tr i and č e tir i cannot be genitive singular, but rather genitive
dual/ paucal.
On the other hand, it seems that d v ije , and tr i and č e tir i which combine with feminine
norms, do not behave as quantifiers, but rather as adjectives, since the noun following them
has the nominative plural form, and the predicate is also feminine plural, as in (17b) and
(18b), where we find the form s tig le . This suggests that the syntactic head of the subject
phrase in (17b) and (18b) is the noun b a le r in e , rather than the numeral, in contrast to
examples in (17a) and (18a), where the numeral is the head of the subject phrase. But this
analysis must be rejected in view of examples like (20), where the phrase containing d v ije
appears after a preposition in the same context as d v a . The preposition s a assigns the
instrumental case, but the instrumental plural form cannot follow d v ije in the same way as it
cannot follow d v a , so that (20b) is unacceptable. Therefore, d v ije cannot be treated as an
adjective. And the same is true for the forms tr i and č e tir i which combine with feminine
nouns:
(2 0 ) a. P le še m sa d v ije b a le r in e / d v a v o jn ik a .
(I ) d a n c e w ith tw o dan cers-G E N .PA U C / t w o soldiers-G E N .P A U C
b. * P le še m sa d v ij e b a le r in a m a / d v a v o j n ic i m a
(I ) d a n c e w ith tw o d an cer-iN S T .P L / t w o sold iers-IN S T .P L
Therefore, we keep a unified treatment of d v a and d v ije , and claim that d v ije also takes
the genitive paucal complements which happen to be identical in form with the nominative/
accusative forms. In contrast to d v a , however, d v ije has feminine features, and therefore the
agreement form of the predicate is feminine, as in (17b).
Like Quantifiers proper, d v a , d v ije , tr i and č e tir i can appear in nominative contexts, as
in (17) and (18), accusative, as in (21a), and genitive, as in (21b), but not in dative, or other
oblique contexts, as illustrated with the ungrammatical (21c):
(21) a. Vidim dva vojnika

(I) s e e tw o soldier-GEN.PAUC
b. Bojim se dva vojnika.

(I) a m afraid of tw o soldier-GEN.PAUC
94
c. *Pišem dva vojnika.19

T write to two soldiers.’
It is unexpected that (21b) is acceptable, because d v a is a numeral quantifier which has only
direct cases, and therefore it is expected that it may appear only in nominative and accusative
contexts. In (21b) it appears in the genitive context. We think that (21b) is acceptable because
the morphological genitive is present in the structure, realized on the noun following d v a , and
its presence actually licenses the structure, although the genitive assigned by the verb b o ja ti s e
‘to be afraid (of)’ is not realized on d v a .
The phrase d v a v o jn ik a , although unacceptable in (21c) after the dative assigning verb
like p i s a t i ‘to write’, may appear after dative assigning prepositions, as in (22):
(22) Idem prema dva vojnika.

(I) w a lk to w a rd s tw o soldier-GEN.PAUC
The preposition overtly signals which case is assigned. For example, d v a v o jn ik a is possible
after the preposition p r e m a because that preposition not only assigns dative, but also signals
which case was assigned: p r e m a always assigns dative. On the other hand, the verb like p i s a ti
‘to write’ may assign not only dative, but also accusative, and therefore we don’t know what
the case of the phrase after the verb is, unless the morphological case ending indicates that.
Since the numeral quantifier d v a doesn’t have case endings for oblique cases, therefore it
cannot be used after this verb. Quantifiers (and d v a is a quantifier) may appear in oblique case
positions only if preceded by a preposition.
Numeral quantifiers allow (and actually prefer) movement of the modifier of the
following noun into the Spec of QP, as in (24b) where the demonstrative o n a 'those' is moved
into the SpecQP. But, differently from the other Quantifiers, they do not allow pronouns like
v a s 'you' in their complement, as in (25), possibly due to the absence of a paucal form of the
pronoun v i ‘you-PL’, taking into account that vi refers to plurality (notice that p e t v a s is
acceptable):
(24) a. ?Vidim dva ona vojnika.

(I) s e e tw o those-GEN.PAUC soldier-GEN.PAUC
b. Vidim ona dva muškarca.
(25) a. *Dva vas neka dodje sutra.20

tw o you-GEN let c o m e to m o rro w
b. *Vas dva neka dodje sutra.21
19 The ungrammatical example in (21c) becomes grammatical if instead of the quantifier form
dva, we use the adjectival form d v a m a , as in (i), or the nominal form d v o jic i, as in (ii):
(0 P iše m d vam a v o jn icim a .

(I) w rite (to) two-DAT soldier-D a t .m a s c .p l
(ii) P iše m d v o jici vo jn ik a .

(I) w rite (to ) two-DAT soldier-GEN.MASC.PL
20 Notice that p e t / š e s t / se d a m /... vas is acceptable, in contrast to *d v a / t r i / č e tir i va s.

95
1
In Giusti and Leko (2001), the structure given in (ii) is proposed for the quantity
expressions that behave like this class of numerals:
onih pet vojnika
3.4. Adjectival Numerals
Finally, numerals that behave as adjectives have a full adjectival declension. Three
subgroups of adjectival numerals maybe distinguished:
1. o r d in a l n u m e r a ls , s u c h a s p r v i, p r v a , p r v o 'first-M A SC / FEM/ NEUT'; d ru g i, d ru g a ,
d r u g o ‘s e c o n d ’ ; tre ći, tre ć a , tr e ć e ‘th ir d ’ ; č e tv r ti, č e tv r ta , č e tv r to ‘fo u r t h ’, e tc .
2. plural counterparts of cardinal numerals, such as je d n i, je d n e , j e d n a 'one- m a s c /
'two’; tro ji, tro je , tr o ja 'three'; č e tv o r i, č e tv o r e , č e tv o r a ‘four’,
FEM /NEUT'; d v o ji, d v o je , d v o ja
etc.2
122
3. adjectival forms of the cardinal numerals d v a ‘two’, tr i ‘three’ and č e tir i ‘four’,
which appear only in oblique cases: genitive: d viju -¥ E U , d v a ju - M ASC/ NEUT; triju , č e tir iju ;
dative/ instrumental/ locative: dvjem a-FEM , dvam a-M A SC f NEUT, trim a , č e tir m a .
21 In connection with ungrammatical (56b) the reader might jump to the conclusion that vas
dva is not possible at all. However, although dva vas doesn’t occur, vas dva actually does
occur, as the following examples from The Oslo Corpus o f Bosnian Text illustrate:
(i) Nisam je dopušio kad je došao njihov drugi vojnik i rekao: “Vas dva, za mnom!”
T hadn’t finished smoking it when another of their soldiers came and said: “You two,
follow me!”
(ii) “Posadite vas dva mene do njegove strane.”

‘You two, place me at his side.’
As (ii) shows, the agreement form of the imperative verb is the 2nd person plural (posadite),
just as with vas dvojica/ vas dvoje/ vas dvije above. So the only ungrammatical thing about
(25b) is the 3rd singular ending on the verb, and it would be grammatical with the imperative
verb in the 2nd person:
(iii) Vas dva dođite sutra.

you-ACC two come-3rd PERS PL tomorrow
Again, this proves the claim that dva is not the head of the subject phrase in (iii), but rather
stands in apposition to vas, which is the real head of the subject phrase.
22 These cardinals are treated in traditional grammars as plural forms of collective numerals
because they are morphologically plural and they require pluralia tantum nouns'as their head.
We treat these numerals as plural cardinals.
96
Ordinal numerals agree with the noun they modify in gender, number, and case, as in
(17), with masculine, feminine, and neuter nominative singular forms of ordinal numerals in
agreement with corresponding head nouns:
( 1 7 ) a. P r v i/ d r u g i/ tr e ć i/ ć e t v r t i / . . . v o j n ik j e s tig a o .
fir s t/ s e c o n d / th ir d / fourth-N O M .M A SC .SG sold ier-N O M .M A S C .S G arrived -M A S C .S G
b. Prva/ druga/ treća/ četvrta/ ... b a le r in a j e s tig la .

first/ s ec o n d / third/ fourth-NOM.FEM.SG dancer-N O M .FEM .SG arrived-FE M .SG
c. P rvo/ d r u g o /t r e ć e /č e tv r to /... v o z ilo j e s tig lo ,

fir s t/ s e c o n d / th ir d / fourth-N O M .N EU T.SG veh icle-N O M .N E U T .SG arrived-N E U T .SG
The declension of ordinal numerals is completely adjectival, so that their endings have
the same form as those of definite form adjectives, as in (18a) with nominative forms, (18b)
with accusative forms, and (18c) with dative forms:
(18) a. Prvi/ drugi/ t r e ć i/ č e t v r t i . .. h ra b ri v o j n ik j e s tig a o .

first/ second/ th ir d / fourth-N O M brave-N O M so ld ier-N O M a r riv ed
b . V id im p r v o g / d ru g o g / tr e će g / č etv rto g ... hrabrog v o jn ik a .

(I) s e e fir s t/ s e c o n d / th ir d / fou rth-A C C b rave-A C C so ld ier-A C C
c. Pišem p r v o m / d r u g o m / t r e ć e m / č e tv r to m ... hrabrom v o jn ik u ,

(I) write (to) fir s t/ secon d / th ir d / fou rth -D A T brave- dat so ld ier-D A T
Ordinal numerals are also adjective-like in the sense that they may occur with singular
n o u n s , a s in ( 1 9 a ) , a n d p lu r a l n o u n s , a s in (1 9 b ):
(1 9 ) a. Prva/ druga/ tr e ć a / č e tv r ta . .. b a le r in a j e s tig la .

fir s t/ secon d / th ir d / fourth-FEM .SG dan cer-FE M .SG arrived-FE M .SG
b. Prve/ d r u g e / tr e ć e / č e tv r te . .. b a le r in e s u s t ig le .
fir s t/ s e c o n d / th ir d / fourth-FEM .PL dancer-FE M .PL arrived-FE M .PL
Numerals that behave as nouns are N heads, whereas numerals that behave as
quantifiers are Q heads. They both take genitive DPs as their complements. What is the
syntactic position of adjectival numerals?
Following Cinque (1994), who argues that adjectival elements are generated in distinct
specifier positions, I assume that they are generated in Spec of NP. Adjectival numerals
always agree with the noun in gender, number and case. Therefore, they must move from their
base generated position in Spec NP to the extended projection of NP, presumably AgrP, in
order to check these features. The checking relationship is established by adjectival movement
to the specifier position of the Agr head, which doesn’t contain any lexical material, but rather
gender, number and case features.
Ordinal numerals may check all their agreement features in AgrP. There is no obvious
reason for their movement to a higher position, and yet I will show that they indeed occupy
some higher position. Ordering restrictions, illustrated in (20), show that this projection must
be higher than AgrP, because ordinal numerals, like other adjectival numerals, must precede
other adjectives, which also check their phi-features in AgrP:
f StaatsbibSiGihek
Bayerischej
i
97
"1
(20) a. Prvo/ drugo/ treće/ četvro...crveno vozilo je stiglo,

first/ second/ third/ fourth red vehicle arrived
b. *Crveno prvo/ drugo/ treće/ četvrto...vozilo je stiglo,

red first/ second/ third/ fourth vehicle arrived
The second group of adjectival numerals, i.e., the plural counterparts of cardinal
numerals, modify only specific kinds of nouns. These are nouns which are plural in form, but
singular in meaning, known as p lu r a lia ta n tu m nouns, like s v a to v i ‘wedding procession’,
n a o č a le ‘spectacles, glasses’, k o la ‘car, carriage’, etc., as in (21):
(21) a. Jedni/ dvoji/ troji/ četvori... svatovi sustigli.

one/ tw o / th ree/ four-NOM.MASC w e d d in g procession-NOM .M ASC arrived-M.PL
b. Vidim jedne/dvoje/ troje/ četvore... svatove.

(I) s e e o n e / t w o / th ree/ four-ACC.MASC w e d d in g procession-ACC.M ASC
Nouns with similar characteristics are those denoting a pair of separate items which
are used together and regarded as a unified entity, like ru k a v ic e ‘(a pair of) gloves’, c ip e le ‘(a
pair of) shoes’, o p a n c i ‘(a pair of) peasant shoes’, etc. All these nouns may,occur in quantified
phrases, but only with what is traditionally described as plural forms of collective nouns.
These numerals have three forms for three genders: masculine, as in (22a), feminine, as in
(22b), and neuter, as (22c):
(22) a. Jedni/ dvoji/ troji/ četvori... sv a to v i su stigli,

one/' two/ thee/ four-MASC w e d d in g processions-M A SC arrived-MASC.PL
b. Jedne/ dvoje/ troje/ četvore .. n a o č a le su razb ijene.

1 one/ t w o / three/ four-FEM glasses-FEM w e re broken-FEM.PL
c. Jedna/ dvoja/ troja/ četvora... k o la su stig la

one/ two/ three/ four-NEUT carriages-NEUT arrived-NEUT.PL
Numerals modifying norms in (22) are also completely adjectival, agreeing with a
noun in gender and case. Therefore, their agreement features would be checked in AgrP.
However, the movement to a higher projection would be required in order to check their
p lu r a lia ta n tu m feature.
The cardinal numeral ‘one’ should be included in this category, primarily because of
its morphosyntactic properties which differentiate it from all other cardinal numerals. It is the
only fully declinable cardinal numeral with three forms for three genders (je d a n - m a s c /
j e d n a - f e m / j e d n o - n e u t ) and completely adjectival, agreeing with a noun in gender and case.
It seems that all its agreement features may be checked in AgrP, and therefore, as in the case
of ordinal numerals, there is no obvious reason for a movement to a higher position. But this
cardinal numeral, like ordinals, must precede other adjectives, which would imply that it also
moves to a higher position.
Finally, the third group of adjectival numerals consists of adjectival forms of cardinal
numerals ‘two’, ‘three’, and ‘four’. The adjectival forms of these numerals are found only in
oblique cases, as illustrated with genitive in (32a), and dative in (32b):
(32) a. Bježimod dvaju/ triju/ četiriju vojnika.

(I) run fro m tw o / th ree/ four-GEN.MASC s o ld ie r s- g e n .m a s c
98
b. Pišem dvama/ trima/ četirma vojnicima.

(I) w rite (to ) tw o / th ree/ four-DAT.MASC soldiers-DAT.MASC
The adjectival ‘two’ has a separate form for feminine gender, whereas the same form
for all tree genders is found with ‘three’ and ‘four’. Genitive forms of these adjectival
numerals are: dviju, dvaju, triju, četiriju, whereas the forms dvjema, dvama, trima, četirma
are found in dative, instrumental and locative contexts. The endings of these forms,
specifically the genitive ending -ju, represent the remnants of the old dual declension, and
differ from endings in adjectival declension, which suggests that they cannot be checked in
AgrP, but that another projection must be postulated for their checking. Ordering restrictions
show that this projection must be higher than AgrP, and its head would be responsible for
checking specific dual endings on cardinal numerals ‘two’, ‘three’, and ‘four’.
The existence of a functional category which contains the number features of the Noun
Phrase was postulated for languages like Modem Hebrew (cf. Ritter 1991). It was argued that
the head of NumP is the locus of the number specification of NP. I assume that such a
projection is available in Bosnian. In my proposal, the specifier position of NumP is reserved
for adjectival numerals.
If we assume that the features of functional heads are binary, then we may propose
that the Num head would carry the feature [+ Dual], If the feature is positively specified, dual
like endings of cardinal numerals ‘two’, ‘three’, ‘four’ in oblique cases could be checked in
[Spec, NumP]. On the other hand, the [-Dual] feature would neither imply singular nor plural,
but rather a specific type of plural, that is pluralia tantum, and in such a case, numerals which
carry that feature could be checked in [Spec, NumP], So, only dual and pluralia tantum
features have to be checked in NumP, whereas singular and plural features would be checked
in AgrP.
Adjectival numerals cannot modify a pronoun, as shown in (33b) and (34b):23
(33) a. Oni su stigli.

they-MASC arrived-MASC PL
b. *Jedni oni su stigli.
(34) a. Bojao se njih.

(he) was afraid of them-GEN PL
b. *Bojao se dvaju ]
23 The fact that (34b) is unacceptable is not because of the archaic form dvaju. It is true that
adjectival forms of numerals ‘two’, ‘three’ and ‘four’ are dying out, and will probably
disappear sooner or later. They are rarely used in colloquial language, and remain mainly in
literary language. For comparison, these are figures from The Oslo Corpus o f Bosnian Texts:
numeral quantifiers: dva: 1026; dvije: 692; tri: 804; četiri: 387;
adjectival numerals: dvaju: 22; triju: 21; četiriju: 1
dvama: 2; dvjema: 20; trima: 1; četirma: 0
The example (34b) suggests that dvaju can’t modify a pronoun at all, but again, corpus data
reveal some instances of njih dvaju, as illustrated with examples (i)-(iii):
(i) Zajedničko svojstvo i karakteristika njih dvaju jeste jedna stvar:...

‘The common property and characteristic of two of them is one thing:...’
99
(he) was afraid of two-GEN them-GEN PL
The syntactic position of adjectival numerals is given in the structure in (iii), as

proposed in Giusti and Leko (2001) for the quantity expressions that behave as adjectives:
(iii)
DP/KP
Notice that in oblique cases such as genitive, as in (35a,b), and complements of

prepositions assigning oblique cases, as in (35c,d), the constructions allowed are two:
Quantifiers and Quantity Adjectives.24 We find Quantifiers dva in (35a) and dvije in (35c),
whereas Quantity Adjectives are found in (35b) (dvaju) and (35d) (dvjema):
(ii) Izmedju njih dvaju biće veo, a na uzvisinama će biti ljudi...

‘There will be a veil between two of them, and there will be people on the hills...’
(iii) Kojim bi se od njih dvaju, da nije toga bilo, naš narod danas...
‘Which one of two of them would our people today, if that didn’t happen...'
However, dvaju no longer modifies a pronoun, as it would in the unacceptable (34b). As in

our discussion in connection with examples vas dvojica! vas dvoje/ vas dvije/ vas dva, we
consider that dvaju in (i)-(iii) is in apposition to njih, and not a modifier of njih.
24 This is true also for a preposition assigning locative case:
(i) o tri knjige

about three book-GEN PAUC
(ii) o trima knjigama,

about three-LOC book-LOC PL
As for dative, we can give an example from The Oslo Corpus o f Bosnian Texts:
100
(35) a. Bojao se dva muškarca.

| (he) was afraid of two man-GEN PAUC
j b. Bojao se dvaju muškaraca.

j (he) was afraid of two-GEN men-GEN PL
i
i c. Razgovarao je sa dvije žene.
(he) talked with two women-GEN PAUC
1 d. Razgovarao je sa dvjema ženama.

(he) talked with two-INST women-INST PL
The three proposed categories of numerals are distinguished on the basis of a certain
number of diagnostics, and they are summarized in the following table:
Diagnostics in Bosnian N Q A
1. selection of a genitive complement + + -
2. default (neuter) agreement on the predicate - + -
3. agreement with the following noun phrase - +

4. co-occurrence with an agreeing demonstrative + - +
5. co-occurrence with a genitive demonstrative - + -
6. rich declension + - +
(iii) ka dvjema vrelim tačkama

towards two-DAT hot-DAT PL point-DAT PL
This can be also expressed as (iv):
(iv) ka dvije vrele tačke
101
References
Abney, Steven. 1987. The English Noun Phrase in Its Sentential Aspect. Ph.D. disseration,
MIT, Cambridge.
Alexiadou, Artemis, Liliane Haegeman, and Melita Stavrou. 2007. Noun Phrase in the
Generative Perspective. Berlin: Mouton de Gruyter.
Babby, Leonard. 1987. Case, Pre-Quantifiers, and Discontinuous Agreement in Russian.
Natural Language and Linguistic Theory 5: 91-138.
Barbiers, Sjef. 2007. Indefinite numerals one and many and the cause of ordinal suppletion.'
Lingua 117: 859-880.
Barnhart, Robert (ed.). 1988. The Barnhart Dictionary o f Etymology. The H.W. Wilson Co.
Barwise, Jon and Robin Cooper. 1981. Generalized Quantifiers and natural language.
Linguistics and Philosophy 4: 159-219.
Bengtson, John. 1987. Notes on Indo-European “10”, “100”, and “1000”. Diachronica IV:
1/2:^257-262.
Bošković, Željko. 2008. What will you have, DP orNP? In: Proceedings ofNELS 37.
Browne, Wayles. 1993. Serbo-Croat. In: The Slavonic Languages. Bernard Comrie and
Greville Corbett (eds.): 306-387. London: Routledge.
Bussmann, Hadumod. 1996. Routledge Dictionary o f Language and Linguistics, (translated
from German and edited by Gregory Trauth and Kerstin Kazzazi). London/ New
York: Routledge.
Cardinaletti, Anna and Giuliana Giusti. 2006. The Syntax of Quantified Phrases and
Quantitative Clitics. The Blackwell Companion to Syntax. Volume V. Martin Everaert
and Henk van Riemsdijk (eds.): 23-93. Oxford: Blackwell Publishing.
Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris.
—. 1994. Bare Phrase Structure. MIT Occasional Papers in Linguistics 5.
Cinque, Guglielmo. 1994. On the Evidence for Partial N-Movement in the Romance DP.
) Paths towards Universal Grammar. Studies in Honor o f Richard S. Kayne. Guglielmo
Cinque et al. (eds.): 85-110. Washington, D.C.: Georgetown University Press.
Comrie, Bernard. 1992. Balto-Slavonic. In: Indo-European Numerals. Jadranka
Gvozdanović (ed.): 717-833. Berlin: Mouton de Gruyter.
Concise Oxford English Dictionary, Eleventh edition, revised (2006). Catherine Soanes
and August Stevenson (eds.). Oxford: Oxford University Press.
Corbett, Greville. 1978. Universals in the syntax of cardinal numerals. Lingua 46: 355-368.
—. 1979. Adjective Movement. Nottingham Linguistic Circular 8: 1-10.
—. 1983. Hierarchies, Targets, and Controllers: Agreement Patterns in
Slavic. University Park: Pennsylvania University Press.
—. 1991. Gender. Cambridge: Cambridge University Press.
—. 1993. The head of Russian numeral expressions. In: Heads in
Grammatical Theory. Greville Corbett, Norman Fraser and Scott McGlashan (eds.):
11-35. Cambridge: Cambridge University Press.
—. 2001. Number. In: Language Typology and Language Universals. An
International Handbook. Vol. 1. Martin Haspelmath, Ekkehand Koenig, Wulf
Oesterreicher and Wolfgang Raible (eds.): 816-831. Berlin: Walter de Gruyter.
Corver, Norbert, Jenny Doetjes and Joost Zwarts. 2007. Linguistic perspectives on numeral
expressions: Introduction. Lingua 117: 751-757.
Crystal, David. 1997. A Dictionary o f Linguistics and Phonetics. 4th ed. Oxford: Blackwell.
Cruse, D. A. 1994. Number and Number Systems. In: The Encyclopedia o f Language and
Linguistics. R. E. Asher (ed.): 2857-2861. Oxgord: Pergamon Press.
Dickey, Stephen. 1992. Serbo-Croatian Distributive po. Paper presented at the Conference on
Balkan and South Slavic Linguistics, Literature, and Folklore, Chicago.
102
Dixon, R. M. W. 1994. Adjectives. In: The Encyclopedia o f Languages and Linguistics. Vol.
I. R.E. Asher (ed.). Oxford: Pergamon Press.
Flegg, Graham (ed.). 1989. Numbers Through the Ages. London: MacMillan in association
with The Open University.
Franks, Steven. 1994. Parametric Properties of Numeral Phrases in Slavic. Natural Language
and Linguistic Theory 12: 591-61A.
1995. Parameters o f Slavic Morphosyntax. Oxford: Oxford University Press.
Fraser, Norman, Greville Corbett and Scott McGlashan. 1993. Introduction. In: Heads in
Grammatical Theory. Greville Corbett, Norman Fraser and Scott McGlashan (eds.): 1-
10. Cambridge: Cambridge University Press.
Giusti, Giuliana. 1991. The Categorial Status of Quantified Nominals. Linguistisch Berichte
136: 438-454.
—. 1992. La sintassi dei determinanti. Padova: Unipress.
—. 1994. Enclitic Article and Double Definiteness. A Comparative Analysis of
Nominal Structure in Romance and Germanic. The Linguistic Review 11: 241-255.
—. 1995a. A unified structural representation of (Abstract) Case and Article. Evidence from
Germanic. In: Studies in Comparative Germanic Syntax. H. Heider, S. Olsen, and S.
Vikner (eds.): 77-93. Dordrecht: Kluwer.
—. 1995b. Heads and Modifiers among Determiners. Evidence from Rumanian. In: Advances
in Rumanian.G. Cinque and G. Giusti (eds.): 103-125.
—. 2002. The Functional Structure of Noun Phrases. A Bare Phrase Structure
Approach. Functional Structure in DP and IP. The Cartography o f Syntactic
Structures. Vol 1. Guglielmo Cinque (ed.): 54-90. Oxford: Oxford University Press.
—. 2006. Parallels in clausal and nominal periphery. Phases o f Interpretation. Mara Frascarelli
(ed.): 163-186. Berlin: Mouton de Gruyter.
Giusti, Giusti and Nedžad Leko. 1995. On the syntax of quantity expressions in Bosnian.
University o f Venice Working Papers in Linguistics, vol.5, no.2, 23-47.
—. 1996. Definite and indefinite quantity expressions in Bosnian. In: Determinatezza e
indeterminatezza nelle lingue Slave. Atti del Convegno svoltosi a Firenze, Ottobre
1995. R. Benacchio, F. Fici and L. Gebert (eds.): 147-162. Padova: Unipress.
—. 2001. The Categorial Status of Quantity Expressions. In: Current Issues in Formal Slavic
Linguistics ed. by Gerhild Zybatow et al. (eds.): 96-105. Frankfurt/Main: Peter Lang.
Greenberg, Joseph. 1966a. Some universals of grammar with particular reference to the order
of meaningful elements. In: Universals o f Language. Second edition. Josep Greenberg
(ed.): 73-113. Cambridge, MA: The MIT Press.
—. 1966b. Language Universals: with special reference to feature hierarchies. The Hague:
Mouton.
—. 1978. Generalizations About Numeral Systems. Universals o f Human Language. Volume 3
Word Structure. Joseph Greenberg (ed.): 249-295. Stanford: Stanford University Press
~. 1989. The internal and external syntax of numerical expressions. Explaining Language
Specific Rules. Belgian Journal o f Linguistics 4: 105-118.
—. 2000. Numeral. In: Morphology. An International Handbook on Inflection and Word-
Formation. Vol. 1. Geert Booij, Christian Lehmann, and Joachim Mugdan (eds.): 770-
783. Berlin: Walter de Gruyter.
Grimshaw, Jane. 1991. “Extended Projection”, ms., Brandeis University.
Gvozdanović, Jadranka. 1992. Remarks on numeral systems. In: Indo-European Numerals.
Jadranka Gvozdanović (ed.): 1-10. Berlin: Mouton de Gruyter.
—. 1999. Types of numeral changes. In: Numeral Types and Changes Worldwide. Jadranka
Gvozdanović (ed.): 95-111. Berlin: Mouton de Gruyter.
—. 1999a. Some remarks on numeral morphosyntax in Slavic. Numeral Types and Changes
Worldwide. Jadranka Gvozdanović (ed.): 187-196. Berlin: Mouton de Gruyter.
103
Hartmann, Reinhard and Stork, Francis. 1972. Dictionary o f Language and Linguistics.
London: Applied Science Publishers ltd.
Homby, A.S. 2005. Oxford Advanced Learner's Dictionary o f Current English. Seventh
edition. Oxford: Oxford University Press.
Hurford, James. 1975. The Linguistic Theory o f Numerals. Cambridge: Cambridge University
Press.
—. 1987. Language and Number. The Emergance o f a Cognitive System. Oxford: Basil
Blackwell.
—. 1991. Numerals. In: International Encyclopaedia o f Linguistics, Vol. 3. W. Bright (ed.):
131-132. Oxford: Oxford University Press.
—. 2007. A performed practice explains a linguistic universal: Counting gives the Packing
Strategy. Lingua 117: 773-783.
Jackendoff, Ray. 1977. X ’-Syntax: A Study o f Phrase Structure. Cambridge, MA: MIT Press.
Justus, Carol. 1988. Indo-European numerals and numeral systems. In: A Linguistic
Happening in Memory o f Ben Schwartz. Studies in Anatolian, Italic, and other Indo-
European Languages. Yoel L. Arbeitman (ed.): 521-541. Louvain-Ia-Neuve: Peeters.
Kobuchi-Philip, Mana. 2007. Floating numerals and floating quantifiers. Lingua 117: 814-831
Kordić, Snježana. 1997. Serbo-Croatian. Muenchen: Lincom Europa.
Leko, Nedžađ. 1986. Syntax o f Noun Headed Structures in Serbo-Croatian and
Corresponding Phrasal Structures in English. Ph. D. disseration, Indiana University.
—. 1990. Case of Noun Phrases in Serbo-Croatian and Case Theory. Folia Linguistica 23/1-2:
27-54.
—. 1995. Syntactic behaviour of numerals 1,2,3,4,5 in Bosnian and Slovene. Quaderni del
Dipartimento di Linguistica, Universita di Firenze 6: 213-228.
~. 1999. Functional Categories and the Structure of the DP in Bosnian. In: Topics in South
Slavic Syntax and Semantics. Mila Dimitrova-Vulchanova and Lars Hellan (eds.):
229-252. Amsterdam: John Benjamins Publishing Company.
Levi, Judith. 1978. The Syntax and Semantics o f Complex Nominals. New York: Academic
Press.
Loebel, Elisabeth. 1989. Q as a Functional Category. In: Syntactic Phrase Structure
Phenomena in Noun Phrases and Sentences. Christa Bhatt, Elisabeth Loebel and
Claudia Schmidt (eds.): 133-157. Amsterdam: John Benjamins Publishing Co.
Lujan Martinez, Eugenio. 1999. The Indo-European system of numerals from ‘1’ to ‘10’. In:
Numeral Types and Changes Worldwide. Jadranka Gvozdanović (ed.): 199-219.
Berlin: Mouton de Gruyter.
Manczak, Witold. 1985. Indo-European numerals and the sexagesimal system. Papers from
the 6thInternational Conference on Historical Linguistics. Jacek Fisiak (ed.): 347-352.
Poznan: John Benjamins Publishing company and Adam Mickiewicz University Press.
Matthews, P.H. 1997. The concise Oxford Dictionary o f Linguistics. Oxford: Oxford
University Press.
Mel’čuk, Igor. 1980. Animacy in Russian cardinal numerals and adjectives as an inflectional
category. Language 56: 796-811.
Meninger, Karl. 1969. Number Words and Number Symbols. A Cultural History o f Numbers.
Cambridge, MA: The MIT Press.
Neidle, Carol. 1988. The Role o f Case in Russian Syntax. Dordrecht: Kluwer.
Olsen, Susan. 1989. AGR(eement) in the German Noun Phrase. Syntactic Phrase Structure
Phenomena in Noun Phrases and Sentences. Christa Bhatt, Elisabeth Loebel and
Claudia Schmidt (eds.): 39-50. Amsterdam: John Benjamins Publishing Co.
Payne, John. 1993. The headedness of noun phrases: slaying the nominal hydra. Heads in
Grammatical Theory. Greville Corbett, Norman Fraser and Scott McGlashan (eds.):
114-139. Cambridge: Cambridge University Press.
104
Pesetsky, David. 1982. Paths and Categories. MIT dissertation. Distributed by MIT Working
Papers in Linguistics.
Peters, Stanley and Dag Westerstahl. 2006. Quantifiers in Language and Logic. Oxford:
Clarendon Press.
Richards, Jack, John Platt, and Heidi Weber. 1985. Longman Dictionary o f Applied
Linguistics. Harlow: Longman.
Ritter, Elizabeth. 1991. Two Functional Categories in Noun Phrase: Evidence from
Modem Hebrew. Syntax and Semantics 25. Susan Rothstein (ed.): 37-62. New York:
Academic Press.
Rutkowski, Pawel and Hanna Maliszewska. 2007. On Prepositional Phrases inside numeral
expressions in Polish. Lingua 117: 784-813.
Salzmann, Zdenek. 1950. A method for analyzing numerical systems. Word 6: 78-83.
Seiler, Hansjakob. 1990. A dimensional view on numeral systems. Studies in Typology and
Diachrony. Papers presented to Joseph Greenberg on his 75th birthday. William
Croft, Keith Denning and Suzanne Kemmer (eds.): 187-208. Amsterdam: John
Benjamins Publishing Company.
Shields, Kenneth. 1985. Speculations about the Indo-European cardinals 5-10. Diachronica
11:2: 189-200.
—. 1991. The Indo-European numeral “4”: a new etymology. In: Studia
Etymologica Indoeuropaea. L. Isebaert (ed.): 265-272. Leuven: Uitgeverij Peeters.
Shlonsky, Ur. 1991. Quantifiers as functional heads: A study of Quantifier Float in Hebrew.
Lingua 84: 159-180. *
Sportiche, Dominique. 1988. A Theory of Floating Quantifiers and Its Corollaries for
Constituent Structure. Linguistic Inquiry 19: 425-449.
Štampe, David. 1976. Cardinal Number Systems. Papers from the 12th regional meeting.
Chicago Linguistic Society. Salikoko Mufwene, Carol Walker and Sanford Steever
(eds.): 594-609. Chicago: Chicago Linguistic Society.
Suprun, Adam. 1958. Nekotorye obšije javljenija v istoričeskom razvitii iisliteljnyh v
slavjanskihjazykah. Frunze: Kirgizskij gosudarstvennyj universitet.
Szemerenyi, Oswald. 1960. Studies in the Indo-European System o f Numerals. Heidelberg:
Carl Winter, Universitatsverlag.
The Oxford English Dictionary. Second edition. Vol X. Oxford: Clarendon Press.
Trask, R.L. 1997. A Students’s Dictionary o f Language and Linguistics. London: Arnold.
Trenkić, Danijela. 2004. Definiteness in Serbian/Croatian/Bosnian and some implications for
the general structure of the nominal phrase. Lingua 114: 1401-1427.
Veselinova, Ljuba. 1997. Suppletion in the Derivation of Ordinal Numerals: A Case Study.
MIT Working Papers in Linguistics 31: 429-447.
Veselovska, Ludmila. 2001. Agreement patterns of Czech group nouns and quantifiers. In:
Semi-lexical Categories. The Function o f Content Words and the Content o f Function
Words. Norbert Corver and Henk van Riemsdijk (eds.): 273-322. Berlin: Mouton.
Villar, F. 1991. The numeral ‘two’ and its number marking. In: Perspectives on Indo-
European Language, Culture and Religion. Studies in Honor o f Edgar C. Polome.
Vol.l. Roger Pearson (ed.): 136-154. Washington: Journal of Indo-European Studies.
Wiese, Heike. 2003. Numbers, Language, and the Human Mind. Cambridge: Cambridge
University Press.
—. 2007. The co-evolution of number concepts and counting words. Lingua 117: 758-772.
Willim, Ewa. 2000. On the Grammar of Polish Nominals. In: Step by Step. Essays on
Minimalist Syntax in Honor o f Howard Lasnik. Roger Martin, David Michaels, and
Juan Uriagereka (eds.): 319-346. Cambridge: The MIT Press.
Worth, Dean Stoddard. 1959. Grammatical and lexical quantification in the syntax of the
Russian numeral. International Journal o f Slavic Linguistics and Poetics V II: 117-132
105
L IN C O M S t u d ie s in
S la v ic L in g u is t ic s
in this series
01 Laura A. Janda Back from the brink:
A study of how relic forms in languages
serve as source material for analogical
extension
02 Youri A. Poupynin Interaction between Aspect and Voice

in Russian
03 H. Schuster-Šewc, Grammar of the Upper Sorbian

Gary H. Toops, trans. Language
04 Maria D. Voeikova Russian Existential Sentences:

A Functional Approach
05 Edna Andrews The Semantics of Suffixation
06 Robert D. Greenberg The Balkan Slavic Appellative
08 William J. Sullivan Space and Time in Russian.

A Description of the Locus Prepositions
of Russian
10 SnježanaKordić Der Relativsatz im Serbokroatischen
13 Neil Bermel Register Variation and Language

Standards in Czech
15 Matthew Baerman The Evolution of Fixed Stress in

Slavic
18 Snj ežana Kordić Worter im Grenzbereich von Lexikon und

Grammatik im Serbokroatischen
21 K oH C T aH T H H B. Jlm jia H O B TEHE3HC CJIOBAIIKOrO

JIHTEPATYPHOrO R3LIKA
22 David S. Danaher The Semantics and Discourse
Function of Habitual-Iterative Verbs in
Contemporary Czech
23 Iwona Kraska-Szlenk The Phonology of Stress in Polish

24 Ursula Doleschal Genus als grammatische und
textlinguistische Kategorie
25 KOHCTaHTHH B. JlH ^aH O B H3MK BOCTOHHOCJIOBAIJKHX

nyBJIHKAPHH B CUIA
26 Ian Press A History of the Russian Language

and its Speakers
27 Stefan M. Pugh A New Historical Grammar of the East

Slavic Languages
Volume 1: Introduction and Phonology
28 Mojmlr Dočekal, Czech in Generative Grammar

Petr Karllk,
Jana Zmrzllkova (eds.)
29 Noel C. Brackney * The Origins of Slavonic
30 Marc L. Greenberg A Short Reference Grammar of Slovene
31 Ian Press Topics in the History of Russian
32 Janina Molczanow The Phonology of Glides in Russian
33 NedžadLeko The Syntax of Numerals in Bosnian
Bayerlsch© ^
Staatsbibliothek
M iln ch eriy

The Syntax of Numerals in Bosnian (Leko)

Uploaded by

Copyright:

Available Formats

The Syntax of Numerals in Bosnian (Leko)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Syntax of Numerals in Bosnian (Leko)

Uploaded by

Copyright:

Available Formats

The Syntax of Numerals

All rights reserved, including the rights of translation into any

Die Deutsche Bibliothek - CIP Cataloguing-in-Publication-Data

A catalogue record for this publication is available from Die

1. What are numerals? 3

1. What are numerals?

1.1. Some dictionary definitions o f numerals

1.2. The categorial status o f numerals

Traditional grammarians usually consider numerals as a part of speech in the category

1.3. Cardinal numeral systems

1.4. Other numerals

1.5. Non-referential counting words and referential number words

Wiese (2003) makes a distinction between non-referential counting words and

1.6. Different kinds o f referential number words

Languages employ various kinds of number words to refer to various numerical

(1) a. You are the number one in my life.

Consequently, we may conclude that there is no one-to-one relation between 'number-

1.7. The grammatical properties o f referential number words

Referential number word constructions exhibit grammatical parallels to some non-

(3) many apples/ five apples.

We find parallel examples in Bosnian:

(4) mnogo j abuka/ pet j abuka

(5) a. jedna jabuka

(6) pet jabuka/ oraha/ goveda

(7) mnogo jabuka/ oraha/ goveda

(8) a. mnogo malih crvenih jabuka

b. pet malih crvenih jabuka

(9) Pisama sam dobio mnogo/ pet.

(10) the fifth American president/ the oldest American president

(11) a. peti amerifiki predsjednik

b. peta američka predsjednica

c. peto američko dijete

The connection between ordinal number words and superlatives is documented in

(12) ship number five/ ship Titanic

1.8. Greenberg's Universals relating to numerals

Among typological studies of numerals, those by Greenberg were most influential. In

(13) a. pet pametnih studenata

b. *pametnih pet studenata

(14) a. pet pametnih i vrijednih studenata

b. pet studenata, pametnih i vrijednih

c. *studenata pet pametnih i vrij ednih

d. pet veoma pametnih studenata

e. pet studenata, veoma pametnih

f. * studenata pet veoma pametnih

(15) *pet studenata pametnih

Compare (15) with (13a), the only acceptable order.

(16) a. onih pet pametnih studenata

b. *pet onih pametnih studenata

(17) a. Taj broj ima pet nula.

c. Velike nule su napisane na tabli.

(18) deset (10) *jedan deset/ *jedna deset/ *jedno deset

(19) a. jedan/ dvadeset jedan/ sto jedan/ hiljadu jedan student

b. dva/ tri/ četiri/ dvadeset dva/ sto dva studenta

Numerals in Bosnian have a very peculiar syntactic behaviour. Although semantically

2.1. Cardinal numerals

2.1.1. Jedan (1)

a. Jedan muškarac je spavao.

b. Bojao se jednog muškarca.

c. Pisao je jednom muškarcu.

d. Vidio je jednog muškarca.

(18) deset (10) jedan deset/ jedna deset/ *jedno deset

( 5) ______________ _ ____ _