2.1 Referential Noun Phrases

advertisement
Three Perspectives on Swedish Indefinite Determiners
Joakim Nivre
Växjö University
School of Mathematics and Systems Engineering
SE 351 95 Växjö, Sweden
E-mail: nivre@msi.vxu.se
1 Introduction
In Swedish, as in many other languages, the unmarked determiner in singular indefinite noun
phrases is the indefinite article (IA), realized as en (common gender) or ett (neuter gender).
Semantically, noun phrases determined by the indefinite article are usually taken to express
some kind of existential quantification; pragmatically, they are often used to introduce new
discourse referents. Thus, a typical use of the sentence in (1a) entails the pseudo-logical form
in (1b) and introduces a new referent for the car that has been bought:1
(1)
a.
b.
Hon har köpt en bil.
she has bought IA car
‘She has bought a car.’
x[car(x)  she has bought(x)]
In this kind of use, the indefinite noun phrase is often said to have specific (but indefinite)
reference, meaning that the speaker assumes that there is a particular, presumably unique,
referent (in this case, a particular car) that supports the truth of the sentence.
However, the indefinite article is not the only determiner that may occur in indefinite noun
phrases in Swedish. Another possibility is the determiner någon (neuter något, plural några),
which I will provisionally call the ‘existential determiner’ (ED). Now, it seems to be the
standard view that the indefinite article en and the existential determiner någon are both used
to express existential quantification, and that they differ mainly in their distribution. Thus, the
existential determiner would typically be used in negative polarity contexts, such as the
sentence (2a) with the interpretation indicated in (2b):
(2)
a.
b.
Hon har inte köpt någon bil.
she has not bought ED car
‘She hasn’t bought a/any car.’
x[car(x)  she has bought(x)]
1
For all examples in languages other than English, a word-by-word gloss (without punctuation or capitalization)
is given on the immediately following line, while a more idiomatic translation (enclosed in single quotes) is
given on the next line after that.
1
In this case, the indefinite noun phrase någon bil does not refer to any specific car, and is
therefore said to have non-specific (indefinite) reference.
Although the examples given so far may give the impression that the indefinite article and
the existential determiner are in complementary distribution, they can also occur in the same
contexts. And when they do, they seem to give rise to fairly subtle differences in meaning:
(3)
Hon har köpt någon bil.
she has bought ED car
‘She has bought some car.’
(4)
Hon har inte köpt en bil.
She has not bought IA car.’
‘She has not bought a car.’
In the appropriate contexts, both (3) and (4) are perfectly natural sentences, but (3) does not
express the same meaning as (1a), and (4) is not equivalent to (2a). For example, although
both (1a) and (3) can be taken to entail (1b), the use of (3) would also indicate that the
specific details of the car (e.g., what make it is) are unknown to the speaker, or at least
considered less important. Similarly, whereas both (2a) and (4) can be used to deny the
existence of a newly bought car, the use of (4) would normally trigger some additional
implication, e.g., that she bought something else (‘she hasn’t bought a car but a motorcycle’).
It is interesting to note that Swedish differs in this respect from languages as closely related
as Danish and Norwegian. The cognates of någon (nogen in Danish, noen in Norwegian) are
not used at all in sentences like (3), and the difference between sentences corresponding to
(2a) and (4) do not exhibit the same kind of contrast as the Swedish examples. In this article,
we will use the contrastive Scandinavian perspective as a key to analyzing the system of
indefinite determiners in Swedish, and especially what appears to be a wider use of the
existential determiner någon.
In delimiting the subject area, we will make two important restrictions. First, we will limit
our attention to the two determiners discussed so far, i.e., the existential determiner någon and
the indefinite article en. Secondly, we will only discuss their use with singular nouns, in
particular singular count nouns. The reason for the second restriction, besides making the
topic manageable, is that the extended use of the existential determiner in Swedish seems to
be exclusively confined to singular count nouns. Therefore, plurals and mass terms will be
discussed only to the extent that they can contribute further to our understanding of the way in
which indefinite determiners are used with singular count nouns in Swedish.
The aim of the article can be seen as twofold. The most important aim is obviously to
present an adequate analysis of Swedish indefinite determiners. At the same time, however,
we want to make a methodological point in favor of contrastive and corpus-based analysis as
underestimated tools in semantic studies. The contrastive perspective will mainly be adopted
in section 2, where we will analyze the use of the Swedish existential determiner någon in the
light of contrastive data from Danish and Norwegian.2 The corpus-based perspective will be
dominant in section 3, where we present the results of a corpus study of indefinite determiners
both in written and spoken Swedish, providing quantitative data on different uses in different
contexts.
Finally, we will try to put the pieces together and present a more theoretical analysis of the
meaning and use of indefinite determiners. As before, the discussion will primarily concern
2
Other examples of the use of contrastive analysis in semantic studies can be found in reports from the project
Comparative Semantics for Nordic Languages (NordSem), of which the study presented here is a part. Reports
are available at http://www.ling.gu.se/research/projects/nordsem/.
2
the Swedish situation, but we will also try to draw out some general implications for semantic
and pragmatic theory. The theoretical discussion is found mainly in section 4, while section 5
summarizes the main points of the article.
2 A Contrastive Scandinavian Perspective
When confronted with a possible ambiguity in a particular language, Kripke (1977) suggests
two methodological considerations that can be used to test whether the linguistic expression is
in fact ambiguous or not:
First, then, we can consult our linguistic intuitions, independently of any empirical investigation. Would
we be surprised to find languages that used two separate words for the two alleged senses of a given
word? If so, then, to that extent our linguistic intuitions are really intuitions of a unitary concept, rather
than of a word that expresses two distinct and unrelated senses. Second, we can ask empirically whether
languages are in fact found that contain distinct words expressing the allegedly distinct senses. If no such
language is found, once again this is evidence that a unitary account of the word or phrase in question
should be sought. (Kripke 1977:261)
In the original context, these tests were intended as tools to reduce the tendency to postulate
ambiguities merely as a way to save one’s favorite theory, what Kripke calls ‘the lazy man’s
approach in philosophy’. However, we believe that they can also be used positively as a way
of adding plausibility to an analysis that posits distinct uses (or senses) for a particular
linguistic expression or construction. Thus, to the extent that we would not be surprised to
find languages that use two separate words or constructions for the different uses, we would
be strengthened in our belief that they are distinct uses. And if languages are in fact found
that contain distinct words or constructions for the different uses, then this would provide
further support for the proposed analysis.
In this section, we will apply this method to the analysis of Swedish indefinite determiners,
in particular the use of the existential determiner någon. By comparing its use to that of its
cognates in Danish and Norwegian, we will try to show that it has three relatively distinct
uses, only one of which is found in the other languages. Of course, we do not want to claim
that these uses are unrelated — not in the way that the senses of homonyms are unrelated. On
the contrary, we will eventually try to give an account of the way in which they are related
semantically. In other words, while we will show that there are in fact languages (Danish and
Norwegian) that use separate means of expression for the different uses of någon in Swedish,
we would not be surprised to find languages where these uses did not correspond to distinct
expressions.
The fact that Swedish, Danish and Norwegian are historically related languages adds a
further dimension to the contrastive perspective. The existential determiners någon (Swedish),
nogen (Danish) and noen (Norwegian) clearly have a common origin in Old Norse, eventually
going back to the unattested *ne wait ek hwariR, meaning ‘I don’t know which’. Thus, the
present situation, where the Swedish expressions have a wider use than its Danish and
Norwegian cognates can only have arisen through a development where, either the Swedish
expression has acquired new uses, or the Danish and Norwegian expressions have become
more restricted in their use. Our impression, based on the contemporary situation, is that both
developments have probably taken place in parallel. And although a full investigation of the
diachronic evidence is outside the scope of this article, we will offer some speculations about
the historical development as part of the theoretical discussion in section 4.
The rest of this section is structured as follows. We begin, in section 2.1, by discussing the
use of indefinite determiners in referential noun phrases, first in negative polarity contexts,
which appear to be the prototypical contexts for the existential determiner in Swedish as well
3
as in Danish and Norwegian, and subsequently in other contexts. Section 2.2 deals with the
use of indefinite determiners in quantitative noun phrases expressing, for example, temporal
duration. In section 2.3, finally, we will draw some tentative conclusions concerning the
distribution of the existential determiner någon and the indefinite article en in Swedish, which
will serve as a point of departure for the empirical investigation in section 3 and the
theoretical discussion in section 4.
2.1 Referential Noun Phrases
Let us begin this section with a little note on terminology. We will use the term referential
noun phrase in a broad sense about any (indefinite) noun phrase that can be said to have a
referent, real or imagined, actual or possible. The reference can be specific as in (1a), repeated
as (5) below, or non-specific, as in the de dicto reading of (6):
(5)
Hon har köpt en bil.
she has bought IA car
‘She has bought a car.’
(6)
Hon vill köpa en bil.
she wants-to buy IA car
‘She wants to buy a car.’
The distinction between specific and non-specific uses, as defined here, only depends on
whether existential generalization is a valid inference and has nothing to do with whether the
speaker ‘has a particular referent in mind’ or not. The question of whether the speaker can
identify the referent or not will also be important for our analysis, but we will not use the term
‘specificity’ in this case.3 Our notions of specific and non-specific indefinite reference are
essentially the ones found in Lyons (1977). They are also the ones used by Haspelmath (1997)
in his typological study on indefinite pronouns and by Teleman et al (1999) in the analysis of
Swedish noun phrases. In addition to specific and non-specific reference, Teleman et al
(1999) also distinguish a weakly referential use of indefinite noun phrases, e.g. when used
predicatively:
(7)
Han är en god vän.
he is IA good friend
‘He is a good friend.’
We want to classify these uses as referential to distinguish them from the quantitative use of
indefinite noun phrases to be discussed in section 2.2 and exemplified in (8):
(8)
Hon har varit borta i en vecka.
she has been gone for IA week
‘She has been gone for a week.’
The noun phrase en vecka (a week) is not used to refer to a particular week (neither real or
imagined, actual or possible) but to a particular quantity of time (and the determiner en
functions primarily as a numeral, contrasting with två [two], tre [three], etc.). For example, it
3
The latter notion of specificity corresponds to what Dahl (1988) calls T-specificity, while our notion is the
same as Dahl’s P-specificity. For a discussion of different uses of the term ’specificity’ in the literatur, see
Jørgensen (1999).
4
would normally be rather odd to use en vecka as the antecedent of a subsequent anaphoric
pronoun.4 In cases like this, we will say that the noun phrase is used quantitatively and not
referentially. However, it is important to note that this is a classification of the use of a
particular noun phrase, not of the noun phrase per se. For example, the noun phrase en vecka
(a week) could very well be used referentially in another context, as illustrated in (9):
(9)
Jag minns särskilt en vecka förra sommaren.
I remember especially IA week last summer
‘I remember especially a/one week last summer’
The distinction between referential and quantitative noun phrases made in this article is not to
be confused with the distinction between referential and quantificational uses of indefinite
noun phrases found, for example, in Fodor and Sag (1982). All or most of the examples
discussed in connection with the latter distinction would qualify as referential in our sense.
More precisely, their referential use would normally correspond to our specific reference,
while their quantificational use would correspond to our specific or non-specific reference
depending on whether the existential quantifier has maximally wide scope or not.
In discussing the use of indefinite determiners in referential noun phrases in Swedish, it is
important to take into account whether the noun phrase occurs in a negative polarity context
or not, since this has a major influence on the choice of determiner. We will therefore begin
by considering the use of indefinite determiners in negative polarity contexts in section 2.1.1,
and discuss their use in other contexts in section 2.1.2.
2.1.1 Negative Polarity Contexts
It is well known that many languages contain expressions that typically occur in a specific
class of contexts, most prominently under the scope of negation. These expressions are
usually called negative polarity items (NPI) and their licensing contexts negative polarity
contexts (NPC). The following examples, taken from Krifka (1991), show some typical NPIs
in English:
(10)
a.
b.
He hasn’t seen any students.
*He has seen any students.
(11)
a.
b.
He hasn’t ever been to Yemen.
*He has ever been to Yemen.
(12)
a.
b.
She didn’t lift a finger to help him.
*She lifted a finger to help him.
(13)
a.
b.
It’s not worth a red cent.
*It’s worth a red cent.
Typical NPCs, besides negation, are interrogative sentences, the antecedents of conditional
sentences, generic sentences, comparative constructions and the scope of quantificational
noun phrases headed by certain determiners (such as no and few in English). However, not all
4
In a putative counterexample such as the following, we have the feeling that the anaphoric interpretation of the
pronoun den as referring to the week she has been gone is possible only through some kind of coercion: Hon har
varit borta i en vecka. Den har varit mycket händelserik. (She has been gone for a week. It has been very
eventful.)
5
NPIs can occur in all NPCs, and some NPIs apparently occur only under negation (Krifka
1991). Edmondson (1981) has proposed a scale Negative > Interrogative > Conditional >
Comparative, where a given NPI is more likely to occur in contexts towards the left of the
scale. Others have suggested that we need to distinguish between several different kinds of
NPC and corresponding kinds of NPI (see, e.g., Jackson 1995, Zwarts 1995). We will return
to the analysis of NPIs and NPCs in section 4, but for the moment we will define the class of
NPCs rather narrowly to include only sentences with negation and polar interrogatives
(although most of what will be said about Swedish indefinites probably holds true also of
conditionals and comparatives).
As indicated in the introduction, NPCs are usually considered the unmarked context for the
existential determiner någon in Swedish. This is the view expressed, for example, by the
authors of the Swedish Academy Grammar (Teleman et al 1999:411) when they assert that,
although någon is sometimes used to denote an approximate quantity or number, as
exemplified in (14), it is more typically used with a very weak sense of quantity, in particular
in negative polarity contexts,5 as illustrated in (15–17).
(14)
Han brukade komma hit någon gång varje år på den tiden.
he used-to come here ED time every year at that time
‘He used to come here some time every year in those days.’
(15)
Han har inte gjort någon kaninbur.
he has not made ED rabbit-cage
‘He hasn’t made a/any rabbit cage.’
(16)
Fick de något kaffe?
got they ED coffee
‘Did they get any coffee?’
(17)
Får du några pengar på torsdag?
get you ED money on Thursday
‘Will you get (any) money on Thursday’
In (14) the noun phrase någon gång varje år expresses the meaning ‘about once every year’,
which contrasts with en gång varje år (once every year) in being less exact. This use of någon
will be dealt with in section 2.3. In (15–17), by contrast, there is no sense of approximate
quantity, and the meaning expressed appears to be ‘pure’ existential quantification (with the
existence being denied in 15 and questioned in 16–17). In fact, the existential determiner
någon seems to be the unmarked choice of indefinite determiner for all nouns in NPCs,
regardless of whether it can occur with these nouns in other contexts. Thus, the sentences in
(15–17) can reasonably be seen as the unmarked negatives/interrogatives corresponding to the
declarative sentences in (18–20):
(18)
Han har gjort en kaninbur.
he has made IA rabbit-cage
‘He has made a rabbit cage.’
5
Teleman et al (1999) do not use the term ‘negative polarity context’ as such but instead speak of ‘nonaffirmative sentences’. However, since the latter notion is said to include interrogative, conditional, and negative
sentences, it seems that the term is more or less synonymous with our use of ‘negative polarity context’.
6
(19)
De fick kaffe.
they got coffee
‘They got (some) coffee.’
(20)
Du får pengar på torsdag.
you get money on Thursday
‘You will get money on Thursday.’
In (18) we have a singular count noun, construed with the indefinite article en. In (19) and
(20), we have a mass noun and a plural noun, respectively, both of which are construed with
no indefinite determiner at all. However, in all three cases, the existential determiner någon is
normally used in the corresponding NPCs.
The unmarked status of någon in NPCs is reflected also in the fact that, in cases where the
indefinite article is used instead, this invariably seems to give rise to a special and more
restricted interpretation (as opposed to mere existential quantification). For example, consider
example (21), which is the same as (15) except that the existential determiner någon has been
replaced by the indefinite article en:
(21)
Han har inte gjort en kaninbur.
he has not made IA rabbit-cage
‘He hasn’t made a rabbit cage.’
Although (21) is a perfectly natural sentence in Swedish, it would not normally be used to
simply assert that no rabbit cage has been produced. Rather, it would be used in contexts
where there would be a more specific reason for denying the truth of (18). Two such uses are
indicated in (22):
(22)
a.
b.
Han har inte gjort en kaninbur, han har gjort en fågelbur.
he has not made IA rabbit-cage, he has made IA bird-cage
‘He hasn’t made a rabbit cage, he has made a bird cage.’
Han har inte gjort en kaninbur, han har gjort många.
he has not made IA rabbit-cage, he has made many
‘He hasn’t made one rabbit cage, he has made many.’
In (22a), the negation applies primarily to the head noun. What is being denied is not the
existence of a produced cage, but the classification of this product as a rabbit cage. In (22b), it
is instead the cardinality of the set of produced cages that is at stake, i.e. the number of cages
is not one, but more than one. This interpretation is possible because the indefinite article in
Swedish is identical to the numeral meaning ‘one’, but it is more natural if the determiner is
stressed. The common denominator of these two interpretations is that the use of the
indefinite article seems to give rise to some kind of existential presupposition, which is never
the case when the existential determiner is used under the scope of negation. A similar
phenomenon is found in other types of examples where the indefinite article is used instead of
the existential determiner despite the context being, at least superficially, of the negative
polarity type:
(23)
Har du redan köpt en julgran?
have you already bought IA Christmas-tree
‘Have you already bought a Christmas tree?’
7
(24)
Kan du gå ner och köpa en tändsticksask åt mig?
can you go down and buy IA matchbox for me
‘Can you go down and buy me a matchbox?’
Teleman et al (1999) explain the use of en in these examples by referring to the speaker’s
assumption that there exists a referent for the indefinite noun phrases and that the proposition
expressed is or will be true. Thus, (23) is an example of a question presupposing a positive
answer. (The speaker assumes that a christmas tree will be bought; the question is when.) In
(24), we have an example of an interrogative sentence used to make a request. In this case, it
may be argued that we are not dealing with a proper NPC, since we are not dealing with a
genuine question. However, the crucial point is that the speaker expresses a propositional
content that he thinks (or hopes) will be true at some point in the near future, which means
that there will also be a referent for the noun phrase en tändsticksask (a matchbox). As a
tentative generalization, we may therefore conclude that the indefinite article en is used in
NPCs primarily when there is some kind of existence presupposition attached to the noun
phrase in question, whereas the existential determiner is the unmarked choice when no such
presupposition exists.
Let us now compare the Swedish data with the situation in the other two languages of
Mainland Scandinavian. In Norwegian, both the existential determiner noen and the indefinite
article en occurs regularly in negative polarity contexts, first exemplified in (25–26):
(25)
a.
b.
(26)
a.
b.
Hun møtte ikke noen student.
she met not ED student
‘She didn’t meet a/any student.’
Hun møtte ikke en student.
she met not IA student
‘She didn’t meet a student.’
Møtte hun noen student?
met she ED student
‘Did she meet a/any student?’
Møtte hun en student?
met she IA student
‘Did she meet a student?’
However, it seems that the existential determiner is used less widely than in Swedish and the
indefinite article is often preferred also in negative polarity contexts. Thus, the b-sentences in
(25–26), rather than the a-sentences, would appear to be the unmarked alternatives.6 With
regard to the interrogatives in (26) it may be noted that whereas (26b) is neutral with respect
to the expected polarity of the answer, (26a) seems to carry the implication that a negative
answer is expected, or at least that a positive answer is improbable. 7 This can be compared to
the situation in Swedish, where the existential determiner någon is normally preferred in
neutral questions, while the indefinite article en is often used when a positive answer is
expected (cf. example 23 above).
In Danish the distribution of the existential determiner nogen in negative polarity contexts
is even more restricted and seems to be confined exclusively to sentences with negation:
6
Corresponding to (25a) and (26a), there is also an alternative using the negative indefinite determiner ingen
(no) to replace the combination of negation and indefinite determiner: hun møtte ingen student (she met no
student). This alternative is available also in Swedish but has a fairly restricted distribution there.
7
It should be pointed out that not all speakers of Norwegian find (26a) fully acceptable. Apparently,
acceptability is improved if the pragmatic particle da (then) is added: Møtte hun noen student, da?
8
(27)
a.
b.
(28)
a.
b.
Hun mødte ikke nogen student.
she met not ED student
‘She didn’t meet a/any student.’
Hun mødte ikke en student.
she met not IA student
‘She didn’t meet a student.’
*Mødte hun nogen student?
met she ED student
Mødte hun en student?
met she IA student
‘Did she meet a student?’
In contrast to the corresponding Norwegian example (26a), the interrogative in (28a) is simply
ungrammatical and the indefinite article en is the only possibility. And even in sentences with
negation, exemplified in (27), the indefinite article en is the unmarked alternative. Thus, while
(27b) is a simple denial of the proposition that she met a student, (27a) is an emphatic version
(even though the determiner is unstressed) which can be paraphrased as ‘she met no student
whatsoever’. Again, this contrasts with the Swedish situation where någon is the unmarked
alternative with negation, whereas the indefinite article en triggers a special interpretation.
Summing up the discussion so far, we may conclude that the existential determiner (as well
as the indefinite article) occurs in negative polarity contexts in all three languages examined.
But there are also differences in the extent to which the existential determiner may be used as
an alternative to the indefinite article. Swedish clearly has the most extended use, making the
existential determiner the unmarked choice in NPCs, while Danish has the most restricted use,
allowing the existential determiner only in sentences with negation and even then with a
marked interpretation.8
2.1.2 Other Contexts
Let us now turn to the use of indefinite determiners in referential noun phrases occurring in
contexts other than NPCs. In most cases, the indefinite article en is the preferred alternative,
regardless of whether the noun phrase has specific or non-specific reference:
(29)
Jag talade med en medicinsk expert.
I spoke with IA medical expert
‘I spoke to a medical expert.’
(30)
De vill att jag ska gå en kurs.
they want that I shall go IA course
‘They want me to take a course.’
(31)
Alla borde läsa en bok då och då.
Everyone should read IA book then and then
‘Everyone should read a book from time to time.’
8
It should be remembered that these generalizations only concern the use of the existential determiner with
singular count nouns (which is the use where it contrasts with the indefinite article). With mass nouns and
plurals, the existential determiner is used more freely in NPCs in Danish as well as in Norwegian.
9
In (29) the noun phrase en medicinsk expert (a medical expert) refers to a specific person
whom the speaker has spoken to. Although the speaker does not identify the person as such,
we would normally expect him to be able to do so if asked about it. In (30) the noun phrase en
kurs (a course) occurs within the scope of a modal and can be interpreted either as specific
(‘there is a particular course they want me to take’) or non-specific (‘any course I choose’). In
(31) we have a similar kind of ambiguity involving the noun phrase en bok, although the most
likely interpretation is the non-specific one that results from giving the indefinite noun phrase
narrow scope with respect to the quantified noun phrase alla (everyone). However, if we
extend the sentence slightly, we see that a specific (wide scope) reading is also possible:
(32)
Alla borde läsa en bok då och då, nämligen Bibeln.
Everyone should read IA book then and then, namely Bible-DEF
‘Everyone should read a book from time to time, namely the Bible.’
Although the indefinite article en is the unmarked choice in these contexts, the existential
determiner någon can often be used in its place. However, this invariably seems to give rise to
a special interpretation. For example, Teleman et al (1999) observe that någon can be used
about a specific referent in cases where the speaker is not able or does not find it relevant to
identify the referent. Consider example (33):
(33)
Jag talade med någon medicinsk expert.
I spoke with ED medical expert
‘I spoke to some medical expert.’
By using the existential determiner någon, the speaker signals that he lacks information about
the referent of the noun phrase (or considers this information less important). It may be the
case, for example, that he only spoke to the expert on the phone and didn’t get the name, so
all he knows is that the person was a medical expert. In other cases, the lack of information
may concern not the identity of the referent but its subtype (within the type denoted by the
head nominal):
(34)
Hon åt någon smörgås till lunch.
she ate ED sandwich for lunch
‘She had some (kind of) sandwich for lunch.’
The most natural interpretation of (34) is one where the speaker expresses uncertainty about
the kind of sandwich she had for lunch, a content which can be expressed more explicitly with
one of the lexicalized phrases någon sorts (some sort of) and något slags (some kind of):
(35)
Hon åt någon sorts smörgås till lunch.
she ate ED sort-GEN sandwich for lunch
‘She had some kind of sandwich for lunch.’
Yet another variation on the ‘lack of information’ theme can be found in cases where there is
uncertainty about whether the head nominal is applicable to the referent at all:9
9
Interestingly, this interpretation is often possible also with the lexicalized phrases någon sorts and något slags
(some kind of), which means that what is referred to as någon sorts X (some kind of X) may not always literally
be a kind of X.
10
(36)
Jag fick prata med någon assistent eller (någon) sekreterare.
I got-to talk with ED assistant or (ED) secretary
‘They let me talk to some assistant or (some) secretary.’
Although it is possible to interpret (36) as saying that the person spoken to was either an
assistent or a secretary but the speaker does not know which, it is probably more commonly
used to express uncertainty about the person’s correct title (assistant, secretary or something
else). Again, a more explicit way of indicating this kind of uncertainty is available through the
addition of eller något (or something):
(37)
Jag fick prata med någon assistent eller (någon) sekreterare eller något.
I got-to talk with ED assistant or (ED) secretary or something
‘They let me talk to some assistant or (some) secretary or something.’
In all the cases considered so far, the use of the existential determiner någon instead of the
indefinite article en is most naturally interpreted as signaling lack of information on the part
of the speaker. In other cases, the information may not be lacking but is rather considered less
important:
(38)
Jag läste i någon tidning att besinpriset skulle gå upp.
I read in ED newspaper that gas-price-DEF should go up
‘I read in some newspaper that the gas price should go up.’
In uttering (38), the speaker may very well know in which newspaper he got the information
but wants to signal that the important piece of information is the prediction about the gas
price, not which newspaper it was found in. This ‘lack of importance’ use is often found in
derogatory remarks, especially when applied to human beings:
(39)
a.
b.
Hon skulle gå ut med någon överklasstönt.
she should go out with ED upper-class-jerk
‘She was going out with some upper class jerk.’
Jag blev påkörd av någon jävla idiot på cykel.
I was run-over by ED damn fool on bicycle
‘I was run over by some damn fool on a bicycle.’
Summing up the discussion so far, we may say that the indefinite article en is the unmarked
choice of determiner in indefinite noun phrases with singular specific reference, and that the
use of the existential determiner någon indicates that the speaker either lacks more specific
information about the referent (its identity, its kind, etc.) or considers such information less
important. In this respect, någon seems to be very similar to the English determiner some
(pronounced [sm], stressable) which, when used with a singular count noun, implies that the
speaker cannot identify the referent (Warfel 1972:43–44, cited in Haspelmath 1997):10
10
The parallel goes even further in that, in both Swedish and English, it is only the determiner that has this
restriction. English pronouns somebody, something, etc. are not limited to unknown referents, nor are pronominal
uses of Swedish någon and related pronouns such as någonting (something).
11
(40)
a.
b.
c.
John is looking for some book on reserve
(*and I know which one).
Hortense is watching for some sailor who’s due in port today.
(*He is a friend of mine.)
Ralph is worried because he lost some letter he was supposed to mail
(*but I have it right here).
Another way of bringing out the difference in interpretation between en and någon is to see
what happens when we add a phrase beginning in nämligen (namely) which gives more
information about the referent. In (41) we have modified the earlier examples (29) and (33) in
this way:
(41)
a.
b.
Jag talade med en medicinsk expert, nämligen doktor Frisk
I spoke with IA medical expert, namely doctor Frisk
‘I spoke to a medical expert, namely Doctor Frisk’
??
Jag talade med någon medicinsk expert, nämligen doktor Frisk
I spoke with ED medical expert, namely doctor Frisk
‘I spoke to some medical expert, namely Doctor Frisk.’
(41a) is a perfectly natural sentence, where the speaker uses the indefinite noun phrase en
medicinsk expert to introduce a referent and goes on to identify the referent also by name. By
contrast, (41b) gives a very odd impression apparently because the addition of more
information is incompatible with the ‘lack of information’ interpretation of the noun phrase
någon medicinsk expert.11
The contrast between en and någon in referential noun phrases is not limited to the case of
specific reference. A very similar contrast can be observed in non-specific referential uses and
even in weakly referential uses such as predicative noun phrases:
(42)
a.
b.
(43)
a.
b.
De vill att jag ska gå en kurs.
they want that I shall go IA course
‘They want me to take a course.’
De vill att jag ska gå någon kurs.
they want that I shall go ED course
‘They want me to take some course.’
Alla borde läsa en bok då och då.
Everyone should read IA book then and then
‘Everyone should read a book from time to time.’
Alla borde läsa någon bok då och då.
Everyone should read ED book then and then
‘Everyone should read some book from time to time.’
11
The ‘namely test’ goes back to Geach (1962), who speaks of specific NPs as ‘namely-riders’, which means
that they can meaningfully be followed by a phrase such as namely Bill. Note, however, that this characterization
is valid only if ‘specific’ is taken in the narrow sense of ‘the speaker having a specific individual in mind’ and
not in the wider sense in which the term is used in this article, since noun phrases determined by någon would
otherwise constitute a counterexample.
12
(44)
a.
b.
Hans fru är en berömd läkare.
his wife is IA famous physician
‘His wife is a famous physician.’
Hans fru är någon berömd läkare.
his wife is ED famous physician
‘His wife is some famous physician.’
As pointed out earlier, (42a) is ambiguous between a specific and a non-specific interpretation
for the noun phrase en kurs. The same kind of ambiguity is present in (42b) but the use of the
existential determiner någon adds a further component to each interpretation. If någon kurs is
given a specific interpretation then the effect of using the existential determiner någon is the
same as in the examples discussed earlier, that is, the speaker indicates that he cannot or will
not say which particular course it is. If, on the other hand, we interpret någon kurs as nonspecific, then the implication is that the persons referred to by they do not know or do not care
which course it is. This may seem surprising at first but has a straightforward explanation if
we assume that the non-specific interpretation is a de dicto interpretation, that is, the noun
phrase någon kurs is not the speaker’s description of the referent but a description attributed
to the referent of they. Thus, the common element in both interpretations is that någon signals
lack of information; in the specific (de re) case this lack of information is attributed to the
speaker; in the non-specific (de dicto) case it is attributed to whomever is the source of the
noun phrase description.
In (43) we have another kind of ambiguity due to the universally quantified noun phrase
alla (everyone). In this case, the non-specific interpretation of en bok/någon bok is simply a
narrow scope reading and the noun phrase description cannot be attributed to anyone else than
the speaker. The use of någon in this case seems to indicate that the speaker does not put any
constraints on which books people should read (any book will do), but the contrast with the
indefinite article is less salient than in the examples discussed earlier, presumably because the
non-specific reading in itself implies that the speaker has very little information about the
possible referents, regardless of which determiner is used.
Finally, example (44) shows that the contrast under discussion may be found also in
predicative position. The effect of using the existential determiner någon in (44b) is virtually
the same as in all the other cases discussed in this section. On the one hand, it may signal that
the speaker lacks information about the referent. He may not know her name, or what kind of
physician she is, or why she is famous. On the other hand, it may indicate that the speaker
considers such information unimportant, which may also be a way of implying that the person
herself is less important.
Let us now turn back to the contrastive analysis and examine the situation in Danish and
Norwegian. The most striking fact about these languages, when compared to Swedish, is that
the cognates of the existential determiner någon, Danish nogen, Norwegian noen, are not used
at all outside negative polarity contexts. Thus, regardless of whether the noun phrase is meant
to have specific or non-specific reference, we do not get an acceptable sentence by replacing
the indefinite article with the existential determiner. Instead, the Swedish examples involving
the existential determiner någon can only be rendered using the complex determiner en eller
anden (one or another) in Danish and the corresponding en eller annen in Norwegian. This is
illustrated in (45–46) for Danish and (47–48) for Norwegian, with (45) and (47) exemplifying
the case of specific reference, while the examples in (46) and (48) are ambiguous between a
specific and a non-specific interpretation (‘a particular student’ vs. ‘any student’).
13
(45)
a.
b.
c.
(46)
a.
b.
c.
(47)
a.
b.
c.
(48)
a.
b.
c.
Hun mødte en student.
she met IA student
‘She met a student.’
*Hun mødte nogen student.
she met ED student
Hun mødte en eller anden student.
she met one-or-another student
‘She met some student (or other).’
Hun ledte efter en student.
she searched for IA student
‘She was looking for a student.’
*Hun ledte efter nogen student.
she searched for ED student
Hun ledte efter en eller anden student.
she searched for one-or-another student
‘She was looking for some student (or other).’
Hun møtte en student.
she met IA student
‘She met a student.’
*Hun møtte noen student.
she met ED student
Hun møtte en eller annen student.
she met one-or-another student
‘She met some student (or other).’
Hun lette etter en student.
she searched for IA student
‘She was looking for a student.’
*Hun lette etter noen student.
she searched for ED student
Hun lette etter en eller annen student.
she searched for one-or-another student
‘She was looking for some student (or other).’
The lexicalized combination en eller annan (one or another) exists also in Swedish and can be
used as an alternative to någon, especially when the noun phrase has specific reference, as
pointed out by Teleman et al (1999). By and large, the complex determiner in Danish and
Norwegian supports the same interpretations as Swedish någon in referential noun phrases
outside NPCs but seems to have a less widespread use. This is probably due to the fact that it
is a longer, more cumbersome expression and therefore felt as stronger than Swedish någon.12
Thus, the ordinary indefinite article is preferred also in many cases where the speaker lacks
information about the referent and in principle could use the more marked expression.
Simplifying somewhat, we may say that Swedish någon in referential noun phrases
corresponds to its cognates Danish nogen and Norwegian noen when used as an NPI, and to
Danish en eller anden and Norwegian en eller annen in other uses. This pattern emerges very
clearly from a comparison of certain sentences with negation where it can be ambiguous
12
The same can probably be said about Swedish en eller annan as an alternative to någon.
14
whether the indefinite noun phrase is inside the scope of the negation or not. Consider first the
Swedish sentences in (49):
(49)
a.
b.
Hon träffade inte en student.
she met not IA student
‘She didn’t meet a student.’
Hon träffade inte någon student.
she met not ED student
‘She didn’t meet some/any student.’
For both sentences, the most obvious interpretation is one where the indefinite noun phrase in
object position is within the scope of the negation (‘it was not the case that she met a
student’). The preference for this interpretation is especially strong for (49b) because någon is
the unmarked determiner in NPCs. However, both sentences also allow a reading where the
indefinite takes wide scope (‘there is a student that she didn’t meet’), a reading which is
plausible, for example, when the question under discussion is whom she didn’t meet. If we
translate these sentences into Norwegian we get the following:13
(50)
a.
b.
c.
Hon møtte ikke en student.
she met not IA student
‘She didn’t meet a student.’
Hon møtte ikke noen student.
she met not ED student
‘She didn’t meet any student.’
Hon møtte ikke en eller annen student.
she met not one-or-another student
‘She didn’t meet some student.’
Whereas (50a) is ambiguous in the same way as (49a), (50b) can only correspond to the
narrow scope reading of (49b), while (50c) unambiguously expresses the wide scope reading
of (49b). This suggests that Swedish någon has two distinct uses in referential noun phrases.
On the one hand, it is used as a negative polarity determiner, corresponding to Norwegian
noen (Danish nogen); on the other hand, it is used as an ordinary indefinite determiner
indicating lack of information on the part of the speaker, corresponding to Norwegian en eller
annen (Danish en eller anden).
2.2 Quantitative Noun Phrases
In singular indefinite noun phrases used to denote a quantity, the indefinite article en retains
its original meaning ‘one’, while the existential determiner någon expresses an approximation
paraphrased by Teleman et al (1999) as ‘about one, hardly more than two’. This use is
especially frequent with nouns of temporal duration such as timme (hour), dag (day), vecka
(week), etc., and nouns of occasion such as gång (time), but it can also be found, for example,
with nouns of distance such as kilometer (kilometer). Here are a few illustrative examples:
13
We disregard for the moment the fact that the most natural translation of (48b) would in many cases be (49a)
because of the stronger tendency to use the indefinite article in NPCs in Norwegian.
15
(51)
a.
b.
(52)
a.
b.
(53)
a.
b.
Lisa var här för en vecka sedan.
Lisa was here for IA week since
‘Lisa was here a week ago.’
Lisa var här för någon vecka sedan.
Lisa was here for ED week since
‘Lisa was here about a week ago.’
Vi fortsatte en kilometer genom skogen.
we continued IA kilometer through forest-DEF
‘We continued for a kilometer through the forest.’
Vi fortsatte någon kilometer genom skogen.
we continued ED kilometer through forest-DEF
‘We continued for a kilometer or two through the forest.’
Han brukade hälsa på en gång i månaden.
he used-to visit PRT IA time per month
‘He used to come visit once a month.’
Han brukade hälsa på någon gång i månaden.
he used-to visit PRT ED time per month
‘He used to come visit about once a month.’
In Danish and Norwegian, only the indefinite article can be used in this way. The existential
determiners nogen (Danish) and noen (Norwegian) are completely out, and the ‘approximate
numeral’ meaning of Swedish någon can only be expressed using a paraphrase. Below is an
example from Danish:
(54)
a.
b.
c.
Det tager en uge.
it takes IA week
‘It takes a week.’
*Det tager nogen uge.
it takes ED week
Det tager en uges tid.
it takes IA week-GEN time
‘It takes about a week.’
2.3 Conclusion
When viewed from a contrastive Scandinavian perspective, the Swedish existential determiner
någon appears to have three fairly distinct uses, where it contrasts with the indefinite article
en in slightly different ways:
1. In NPCs någon is the unmarked determiner in noun phrases with indefinite reference. The
closest equivalent in Danish and Norwegian is nogen and noen, respectively, although the
latter expressions have a more restricted distribution.
2. In other contexts någon is a marked alternative to the indefinite article in noun phrases
with indefinite reference, usually indicating that the speaker is unable or unwilling to give
further information about the referent (or the class of possible referents). In Danish and
16
Norwegian a similar effect can be achieved only by using a complex determiner (Danish
en eller anden, Norwegian en eller annen).
3. In quantitative noun phrases någon can perhaps be described as an ‘approximate numeral’,
meaning ‘about one’ and contrasting with the indefinite article en in its numeral sense of
‘(exactly) one’. In Danish and Norwegian there is no single determiner that expresses this
sense.
Given the situation in Danish and Norwegian and the common historical origin of någon,
nogen and noen, it may seem natural to assume that the second and third uses are more recent
Swedish innovations. On the other hand, the etymology given earlier suggests that the second
use may in fact be the oldest one. And if the Swedish uses are innovations, they must be fairly
old innovations, since all three uses are attested at least as far back as the 18th century. The
following examples are taken from Svenska Akademiens Ordbok (Dictionary of the Swedish
Academy), the major historical dictionary of the Swedish language from 1521 and onwards:
(55)
Haar du watt i nåån Kyrckia här?
have you been in ED church here
‘Have you been to a church here?’
(56)
Jag har läst på något ställe i Cicero.
I have read on ED place in Cicero
‘I have read somewhere in Cicero’
(57)
Det händer någon gång at en förträffligt god Åker, fast illa häfdad, ger tämligt god
skörd.
it happens ED time that a superbly good field, although badly used, gives rather good
crop
‘It happens once in a while that a superbly good field, although badly used, gives a
rather good crop.’
The dates given for example (55), (56) and (57) in Svenska Akademiens Ordbok are 1620,
1788 and 1792, respectively. Whatever the historical truth of the matter is, we have to
conclude that the three different uses of någon in Swedish are clearly distinguishable, albeit
semantically related, since two of them are lacking in the historically related Danish nogen
and Norwegian noen. Therefore, we would also prefer an account of någon that explained
how the uses are related without making them mutually dependent on each other. We will try
to present such an account in section 4, but first we will try to get a better empirical basis by
examining the distribution of different uses in both spoken and written Swedish.
3 A Corpus Perspective
In the previous section, we examined the use of Swedish indefinite determiners, in particular
the existential determiner någon, contrasting the Swedish situation with that found in Danish
and Norwegian, the two most closely related Scandinavian languages. The outcome was that
there seemed to be three more or less distinct uses of någon with singular count nouns in
Swedish, only one of which was found in Danish and Norwegian. In this section, we will try
to put some more flesh on the bones by presenting the results of a corpus study of singular
indefinite determiners in Swedish. Besides giving us authentic examples of language use, this
17
approach will provide us with quantitative data on the relative importance of different uses for
the existential determiner as well as the indefinite article.
3.1 Data Collection and Analysis
The data for this study come from two tagged corpora of present-day Swedish:
1.
The PAROLE corpus of written Swedish, consisting of 25 million words taken from the
following text types: books, newspapers, periodicals and miscellaneous. (This corpus is
available at: http://spraakbanken.gu.se.)
2.
The Gothenburg corpus of transcribed spoken Swedish (GSLC), consisting of 1.2 million
words taken from a wide variety of social activities such as interviews, discussions,
formal meetings, court proceedings, auctions, travel agency dialogues, dinner
conversations, and roles plays (cf. Allwood 1999; see also http://www.ling.gu.se.)
The original plan was to obtain two samples from each corpus:
1.
An unbiased sample of singular indefinite determiners in order to estimate the relative
frequency of different determiners, especially en and någon.
2.
A balanced sample of 500 tokens, consisting of 250 occurrences of en and 250
occurrences of någon, that would be analyzed and coded manually.
However, because of various circumstances beyond our control, we were not able to apply the
same sampling methods to both corpora. Instead we have proceeded as follows:
1.
The unbiased sample from the PAROLE corpus consists of 1000 tokens tagged as
singular indefinite determiners (including also determiners other than en and någon),
while the corresponding sample for the GSLC corpus consists of all occurrences of någon
taggged as determiner and all occurrences of en regardless of annotation (about 25,000
tokens in total). In this way we can still compare the relative shares of the different
determiners, in spite of the two samples being of unequal size.
2.
The balanced sample in both cases consists of 250 tokens of en and 250 tokens of någon.
However, because of tagging errors in the corpus, some false instances (mostly
pronominal uses of någon) were included in these samples, which means that the samples
finally being analyzed contains somewhat fewer tokens (see Table 1). Still, the samples
are balanced enough to allow us to make all the comparisons that are of interest for the
present study.
The number of tokens finally analyzed in each of the samples is given in Table 1. From the
balanced sample have been excluded not only the false instances mentioned above, but also
occurrences that could not be unambiguously analyzed and coded, usually because of
insufficient contextual information.
18
Table 1
Number of tokens in samples analyzed
IA (en/ett)
ED (någon/något)
Total
Unbiased Sample
PAROLE GSLC
855
21454
77
3230
932
24684
Balanced Sample
PAROLE GSLC
236
226
205
190
441
416
Since we are not primarily interested in comparing spoken and written language, we will
mostly treat the balanced samples as constituting a single sample and comment on the relation
between speech and writing only in cases where there are interesting differences.
For the unbiased samples no further analysis has been performed, but for the balanced
samples each occurrence has been analyzed and coded with respect to its meaning and certain
features of the context. For the semantic/pragmatic classification, the following four
categories have been used (cf. section 2.1):




Specific referential
Non-specific referential
Weakly referential
Quantitative
For the contextual classification, the following parameters have been analyzed:



Negation, i.e. whether the indefinite occurs within the (potential) scope of negation.
Clause type, i.e. whether the indefinite occurs in an interrogative, conditional, etc. clause.
Modality, i.e. whether the indefinite occurs within the (potential) scope of a modal
operator.
Further concepts can be defined in terms of the basic coding categories. For example, a
negative polarity context (NPC) can be defined as a context where the indefinite occurs within
the scope of negation or in an interrogative or a conditional clause.
3.2 Overall Frequency Distributions
If we start by considering the relative frequency of indefinite articles and existential
determiners in the unbiased samples, we see that the indefinite article accounts for
approximately 92% of all the tokens in the PAROLE corpus and about 87% in the GSLC
corpus. With a 95% confidence interval, these estimates have a margin of error of about 2%
for the PAROLE sample (932 tokens) and 0.5% for the GSLC sample (24,684 tokens). In
other words, there seem to be a small difference to the effect that the existential determiner is
relatively more frequent in spoken language than in written language, at least for the types of
activities and texts contained in the two corpora used. More importantly, however, the results
confirm the assumption that the indefinite article is by far the most frequent determiner used
in singular indefinite noun phrases in Swedish.14
14
It is true that we have only compared two determiners, the indefinite article and the existential determiner.
However, the PAROLE sample of 1000 tokens also contained other determiners all of which were less frequent
than both the determiners studied in this article.
19
Turning now to the balanced sample and the relative share of different uses, Table 2
presents frequencies for the indefinite article and the existential determiner divided across the
three main uses discussed in section 2, i.e., negative polarity uses, referential uses in other
contexts, and quantitative uses. The negative polarity category here includes all non-specific
referential uses in interrogative and conditional clauses and under the scope of negation. This
means that the very few cases of specific referential uses in NPCs (3 occurrences altogether)
have been included in the category of other referential uses.
Table 2
Indefinite determiners in different uses/contexts
IA (en/ett)
ED (någon/något)
Total
Referential
Quantitative
Neg pol
Other
47
351
63
177
197
21
224
544
84
Total
462
395
857
Table 2 shows, first of all, that referential uses are much more frequent than quantitative uses,
for both determiners involved but especially for the existential determiner någon/något, where
quantitative uses only account for about 5% of all occurrences. Table 2 also shows that the
indefinite article is underrepresented in NPCs, with only 49 occurrences against an expected
frequency of 123,5, while the existential determiner is overrepresented (180 vs. 105,5). These
differences are statistically significant beyond the .001 level (according to a2 test with 2
degrees of freedom) and seem to confirm the view that there is a preference for the existential
determiner in NPCs. At the same time, it must be remembered that the indefinite article,
because of its greater overall frequency, still probably has a greater absolute frequency in
NPCs than the existential determiner. Moreover, the results do not directly support the view
that negative polarity uses are predominant for the existential determiner, since they account
for less than 50% of all the occurrences in the sample.
We will now turn to a closer examination of each category, where we will consider not only
quantitative data but also look at particular examples from the corpora. We will begin by
treating the referential uses in section 3.3, and move on to the quantitative uses in section 3.4.
3.3 Referential Noun Phrases
As in section 2, we will divide the discussion of referential noun phrases into two sections,
one dealing with negative polarity contexts (section 3.3.1) and one dealing with other contexts
(section 3.3.2).
3.3.1 Negative Polarity Contexts
Table 3 shows the distribution of the indefinite article and the existential determiner across
three types of negative polarity contexts: sentences with negation, interrogatives and
conditionals.15
15
The category “Negation” includes examples that qualify as NPCs only because of the presence of negation.
This means that negation may occur also in the examples categorized here as “Interrogative” (13 cases) and
“Conditional” (6 cases).
20
Table 3
Indefinite determiners in different negative polarity contexts
IA (en/ett)
ED (någon/något)
Total
Negation Interrogative Conditional
17
14
16
93
68
16
110
82
32
Total
47
177
224
While the relatively few occurrences of the indefinite article are evenly distributed across the
three contexts, the existential determiner is found above all with negation and in interrogative
sentences. This seems to suggest that the association of the existential determiner någon with
NPCs applies primarily to negative and interrogative contexts, and less clearly to conditional
contexts.16 The differences are again significant beyond the .001 level (2, df = 2).
Let us now consider some examples from the two corpora. In these and following
examples, we use ‘P’ to indicate that the example comes from the PAROLE corpus (written
Swedish) and ‘G’ for the GSLC corpus (spoken Swedish). Below we have some typical
negative polarity uses of the existential determiner någon, negative in (58–63) and
interrogative in (64–69):17
(58)
På allt sätt försökte jag få honom att förstå att jag inte hade något manus. (P)
in every way tried I get him to understand that I not had ED script
‘In every way I tried to get him to understand that I didn’t have a script.’
(59)
Inte ville jag ha något barn med Mark? (P)
not wanted I have ED child with Mark
‘I surely didn’t want to have a child with Mark?’
(60)
Men tro inte att jag fick någon egen cykel som belöning, inte ens en begagnad. (P)
but think not that I got ED own bike as reward, not even IA used
‘But don’t think that I got a bike of my own as a reward, not even a used one.’
(61)
… ja kan ju inte boka nån bil åt dej eller så … (G)
I can PRT not book ED car for you or so
‘You know, I can’t book a car for you or anything.’
(62)
… ja äh de låter kanske väldit fånit men ja har nästan aldri vart rädd för nån människa (G)
yes eh it sounds perhaps very silly but I have almost never been afraid of ED human-being
‘Well, eh … this perhaps sounds very silly, but I have almost never been afraid of a/any human being.’
(63)
… men de fick ja aldri nåt svar på (G)
but that got I never ED answer to
‘… but I never got an answer to that’
16
This may in turn be taken as support for Edmondson’s proposed hierarchy of NPCs (cf. section 2.1.1).
The examples from GSLC retains the original transcription in modified orthography without punctuation or
capitalization (cf. Nivre 1999). Note, for example, that the forms of the existential determiner are usually written
nån (for någon) and nåt (for något) in accordance with the most common pronunciation. The symbol ‘…’
indicates that part of the utterance has been omitted. In the English translations, punctuation and capitalization
has been introduced to facilitate understanding.
17
21
(64)
”Jag ska fråga bland mina kolleger om de vet något bra ställe”, sa sabotörens hustru.
(P)
I will ask among my colleagues if they know ED good place said saboteur-DEF-GEN
wife
‘ “I will ask among my colleagues if they know a good place”, said the saboteur’s
wife.’
(65)
Viveca frågade om de hade hört musik eller sett någon intressant teater. (P)
Viveca asked if they had heard music or seen ED interesting theater
‘Viveca asked if they had heard music or seen some interesting play.’
(66)
vill du ha nån påse å ta de i (G)
want you have ED bag to take it in
‘Would you like a bag to take it in?’
(67)
behöver vi nån påse vi har ju min väska (G)
need we a (paper-)bag we have PRT my (sports-)bag
‘Do we need a (paper) bag? We have my (sports) bag, you know.’
(68)
går de nåt senare flyg (G)
goes it ED later flight
‘is there a later flight’
(69)
har du haft nån naturupplevelse så där som du kommer ihåg speciellt … (G)
have you had ED nature-experience so there that you remember PRT especially
‘have you had an experience of nature that you remember in particular …’
Examples with the existential determiner in conditional clauses are less frequent in both
corpora:
(70)
var går gränsen då … om det nu finns nån gräns … (G)
where goes the limit then if there now is ED limit
‘Where is the limit then? … If there really is a limit.’
(71)
… de vore underbart om de kunde bli nån förbättring där … (G)
it were wonderful if it could become ED improvement there
‘It would be wonderful if there could be an improvement there.’
In both (70) and (71), the noun phrase introduced by the existential determiner not only has
non-specific reference but seems to apply to any object of the required sort, i.e., “any limit”,
and “any improvement” whatsoever. This can be contrasted with example (72) illustrating the
use of the indefinite article in a conditional clause, where the indefinite noun phrase gets a
more restricted interpretation:
(72)
Vill du ha en tablett så finns det där i hörnskåpet. (P)
want you have IA pill then is there there in corner-cupboard-DEF
‘If you want a pill, there are some there in the corner cupboard.’
It seems clear that the reference in (72) is not to any pill, but to a pill of a specific kind (say, a
pain killer) known to both speaker and hearer. Moreover, the existence of at least one pill of
the required sort is asserted in the second half of the sentence, whereas in the earlier examples
involving the existential determiner existence is not presupposed. In fact, in (70) the existence
22
of a limit is explicitly questioned. These examples thus reinforce the impression that, other
things being equal, the indefinite article is preferred when the existence of a potential referent
is presupposed or given in the context, while the existential determiner is preferred when such
existence is questioned.
The tendency is even clearer when we move to examples of the indefinite article in negative
and interrogative contexts:
(73)
borde man inte ha en diskussion om den psykolingvistiska metod som … (G)
should one not have IA discussion about that psycholinguistic method which
‘Shouldn’t we have a discussion about that psycholinguistic method which …’
(74)
… vill du ha en påse (G)
want you have IA bag
‘Would you like a bag?’
(75)
Jag kan inte ta hand om en unge. (P)
I can not take care of a kid
‘I can’t take care of a kid.’
Example (73), despite being both interrogative and negative in form, is contentwise a positive
suggestion that there should be a discussion. In a similar fashion, example (74), uttered by a
salesperson to a customer, is an offer for a bag. (It is even possible that the salesperson is
already handing the customer a specific bag.) Example (75), finally, is an example from a
novel where the character whose thoughts are described is pregnant and is contemplating an
abortion. Thus, even though the noun phrase en unge cannot be said to have specific
reference, a specific referent is in some sense potentially available in the context.
Other examples of the indefinite article in negative and interrogative contexts have a rather
special character. One such kind of example is where the noun phrase is used predicatively:
(76)
… de här e i å för sej inte en syllogism men … (G)
this here is in and for itself not a syllogism but
‘This is by the way not a syllogism but …’
Another special kind of example involves the preposition utan (without), which contains an
implicit negation but which nevertheless seems to favor the indefinite article:
(77)
Borta utan ett ord. (P)
gone without IA word
‘Gone without a word.’
(78)
Det är svårt att leva ensam tror jag, utan en man. (P)
it is hard to live alone think I without IA man
‘It is hard to live alone, I think, without a man.’
Example (77) almost has the flavor of an idiomatic expression, and it would be rather odd to
use the existential determiner there, despite the fact that there is no existential presupposition
whatsoever.
23
3.3.2 Other Contexts
Table 4 shows the distribution of the indefinite article and the existential determiner across
three different kinds of referential uses outside negative polarity contexts.
Table 4
Indefinite determiners in different referential uses
IA (en/ett)
ED (någon/något)
Total
Specific Non-specific
reference
reference
177
94
53
135
230
229
Weakly
referential
80
9
89
Total
351
197
548
The distributional pattern is similar to the one found in NPCs in that the indefinite article is
more evenly distributed across different uses than the existential determiner. This is probably
related to the fact that the indefinite article is more grammaticalized and therefore, on the
whole, less sensitive to contextual restrictions. However, it is also clear that the indefinite
article is preferred for specific reference, which is expected given that the conditions for using
the existential determiner in noun phrases with specific reference are fairly special. For the
existential determiner, on the other hand, non-specific reference seems to be the predominant
use. The differences are statistically significant at the .001 level (2, df = 2).
Turning now to examples, we begin by examining the existential determiner in noun
phrases with specific reference. Most of the examples in this category are cases where the
speaker uses the existential determiner to indicate that he is not able to identify the referent
more exactly or that he lacks information in some other way, as discussed in section 2.1.2:
(79)
de kommer ja ihåg för ja va junior och ni spela i nåt mästerskap på ånäsfältet (G)
that remember I PRT for I was junior and you-PL played in ED championship on
Ånäsfältet
‘That I remember because I was a junior and you played some championship on
Ånäsfältet.’
(80)
där lär han ha nåt stort residens (G)
there is-supposed he have ED big residence
‘There he is supposed to have some big residence.’
(81)
… en indonesisk folksång, som alldeles uppenbart var en tretakts fyllevisa inplanterad
av någon svensk sjöman … (P)
IA Indonesian folk-song, which quite obviously was a three-beat drinking-song
implanted by ED Swedish sailor
‘… an Indonesian folk song, which was quite obviously a three-beat drinking song
implanted by some Swedish sailor …’
In some cases, the speaker may even be uncertain whether the descriptive content of the noun
phrase applies to the referent at all (cf. section 2.1.1):
24
(82)
dom hade golfklubben i nåt slott eller så de såg ju nästan ut som nåt sånt här slottaktit
(G)
they had golf-club-DEF in ED palace or so it looked PRT almost PRT like ED such here
palace-like
‘They had the golf club … in some palace or something. It looked almost like
something palace-like.’
(83)
Några veckor före jul, eller någon dag i slutet av november, kom Lillemor till
terapitimmen … (P)
ED-PL weeks before Christmas or ED day in end-DEF of November came Lillemor to
therapy-hour-DEF
‘Some weeks before Christmas, or some day at the end of November, Lillemor came
to the therapy hour …’
By contrast, when the indefinite article is used in noun phrases with specific reference, there
is no sense of lacking information on the part of the speaker, as illustrated by the following
three examples all involving the noun phrase en dam (a lady):
(84)
… de va en dam ja träffade där borta …
it was IA lady I met there away
‘It was a lady I met over there.’
(85)
… å där stog ett par damer å då va de en dam som kom ikapp mej så sa hon ska du me
(G)
and there stood a couple-of ladies and then was there IA lady that caught up-with me
then said she will you with
‘And there were a couple of ladies and then there was a lady who caught up with me.
Then she said: “Do you want to come?”.
Note that in both these examples, it is perfectly possible that the speaker is unable to identify
the referent more exactly or considers her identity to be less important. However, whereas
substituting the existential determiner would have highlighted this fact, the use of the
indefinite article is neutral in this respect.
In noun phrases with non-specific reference, the contrast between the existential determiner
and the indefinite article is generally much weaker than in the specific case. As an example,
consider the following two examples from the GSLC corpus:
(86)
du kommer inte på några sammanträden eller sånt när vi ska ha nån eh sammankomst
(G)
you come not at ED meeting or such when we shall have ED eh gathering
‘You don’t come to any meetings or such things when we have some eh … gathering.’
(87)
… man ringer upp nån ska ha ett möte ont om tid va flera månader framåt då (G)
one calls up someone shall have IA meeting short of time PRT several months ahead
then
‘You call someone up, going to have a meeting, short of time, you know, several
months ahead, you know.’
These examples are similar in that they both discuss, in generic terms, the situation before a
meeting. In (86) the meeting is referred to by the noun phrase nån eh sammankomst, while in
(87) the noun phrase ett möte is used. It is possible that the use of the existential determiner in
25
(86) is due to the speaker being uncertain about whether sammankomst is the right noun to use
(the hesitation marker eh tends to strengthen this impression), but apart from that the semantic
difference is very slight.
Another example where the existential determiner is used to introduce a noun phrase with
non-specific reference, and where the indefinite article could have been used without much of
a difference is found in (88):
(88)
Det kan ju vara någon gratulant, så hon ropar på sin man. Men det är en arbetskamrat
som har fel på sin värmepanna.
it can PRT be ED person-congratulating so she calls on her husband but it is IA work
mate who has wrong on his heater
‘It could be someone wanting to congratulate, so she calls her husband. But it is a
work mate having a problem with his heater.’
Note, however, the switch to the indefinite article in the following sentence when referring to
the specific person who called (en arbetskamrat).
Quite a few of the examples involving the existential determiner can be seen as more or less
fixed phrases. A case in point is någon gång (some time), as exemplified in (89):
(89)
… jag hoppas bara du hinner lyssna på henne någon gång — du är ju läkare i alla fall.
(P)
I hope only you have-the-time listen to her ED time you are PRT doctor in any case
‘… I only hope you have the time to listen to her some time — you are after all a
doctor, you know.’
However, the indefinite article is also a possible in most cases:
(90)
Men när det här livet en gång tar slut, kan jag inte tänka mig någon annan tillvaro. (P)
but when this here life one time takes finish, can I not think myself ED other existence
‘But when this life is over some day, I can’t imagine any other existence.’
By contrast, the phrase på något sätt (in ED way) seems to be approaching the status of a fully
grammaticalized adverb meaning ‘somehow’. The semi-grammaticalized use is salient in (92)
below, whereas (91) can still be interpreted as more referential.
(91)
Jag vill använda den på något sätt i den bok som ska skrivas när fältarbetet är klart. (P)
I want use it in ED way in the book which shall be-written when field-work-DEF is
finished
‘I want to use it in some way/somehow in the book that is to be written when the field
work is finished.’
(92)
a de e väl både skämt å allvar på nåt sätt (G)
yeah that is I-suppose both joke and serious in ED way
‘Yeah, I suppose it is both a joke and serious somehow.’
3.4 Quantitative Noun Phrases
Table 5 shows the sample frequencies of the indefinite article and the existential determiner in
quantitative noun phrases.
26
Table 5
Indefinite determiners in quantitative noun phrases
IA (en/ett)
ED (någon/något)
Total
63
21
84
As noted in section 2.2, this use of the existential determiner is especially frequent with nouns
of temporal duration, as illustrated in (93–94):
(93)
a.
Jag hade ett svagt minne av att far kommenterat det där huset redan för något år
sedan … (P)
I had IA weak memory of that father commented that there house-DEF already
for
year since
‘I had a vague memory that father had commented on that house already about a
year ago.’
… för nåt år sen va ja ute i ett torp (G)
… for ED year since was I out in IA cottage
‘About a year ago I was out in a cottage.’
ED
b.
(94)
a.
b.
(G)
… över Kolahalvön ligger ett kallcentrum som fördröjer den definitiva våren i
Prästmon ytterligare någon vecka. (P)
over Kola-peninsula-DEF lies IA cold-center which delays the definitive spring in
Prästmon additionally ED week
‘There is cold weather over the Kola peninsula which delays the definitive spring
in Prästmon at least another week.’
ja såg till exempel i teve för nån vecka sedan erik bergstens program från japan
I saw for example on television for ED week since Erik Bergsten-GEN program
from
Japan
‘I saw for example on television about a week ago … Erik Bergsten’s program
from Japan.’
The examples in (93), with the head noun år (year), and in (94), with the head noun vecka
(week), can be contrasted with the examples in (95) and (96), involving the same head nouns
but with the indefinite article instead of the existential determiner:
(95)
Det hade fungerat så i ett år nu, … (P)
it had worked so for IA year now
‘It had worked like that for a/one year now, …’
(96)
En vecka senare flyttade Anita tillbaka till sitt inackorderingsrum. (P)
IA week later moved Anita back to her rented-room
‘A/one week later Anita moved back to her rented room.’
The difference between (93–94) and (95–96) is very clear. The existential determiner in this
kind of use expresses the meaning ‘about one, one or two’, while the indefinite article has its
original numeral sense ‘one’ (cf. section 2.2).
27
3.5 Conclusion
The corpus investigation presented in this section has substantiated the claims put forward in
section 2 in two different ways. From a quantitative point of view, the data clearly support the
claim that the existential determiner has a greater tendency to occur in NPCs, while the
indefinite article is associated more with specific referential uses. (The statistical significance
of these differences is beyond doubt.) From a qualitative point of view, we have been able to
find examples from actual use in spoken and written Swedish of practically all the different
uses of indefinite determiners distinguished in section 2, thus establishing a firm empirical
foundation for the analysis of these uses. What remains is to give a theoretical account of
these uses and the way in which they are related. It is to this task that we now turn.
4 A Theoretical Perspective
So far, the outlook in this article has been mainly empirical. In section 2, we used a
contrastive Scandinavian perspective to distinguish three main uses of the existential
determiner någon in Swedish, each of which contrasts with the corresponding use of the
indefinite article en in a different way. In section 3, we added a quantitative dimension to the
description of the Swedish data and supplemented the made-up examples from section 2 with
examples taken from two corpora of modern Swedish, representing both spoken and written
language. In this section, we will adopt a more theoretical perspective and try to provide a
semantic analysis of singular indefinite determiners with special reference to the Swedish
situation. In doing so, we will be particularly concerned with the way in which different uses
of the same determiner can be related to each other, and whether they can be derived from the
same semantic analysis, in combination with different contextual constraints and general
pragmatic principles.
The analysis will be couched in a formal semantic framework in the tradition of Lewis
(1970) and Montague (1973), using a categorial grammar for syntactic derivations and a
higher-order typed lambda calculus for semantic representations. This is mostly a choice of
convenience, since we can assume most readers to be familiar with this type of framework. In
reality, we will make very few specific assumptions and commitments concerning the
underlying framework, and there is nothing in the analysis presented below that prevents it
from being expressed in another formal framework. We will also ignore all problems having
to do with intensionality, and work with a purely extensional type system, where noun phrases
are of the type <<e,t,>,t> and hence denote sets of sets individuals.
Besides the formal semantic tradition, the theoretical discussion will draw on two different
sources. The first is the typological work on indefinites by Haspelmath (1997), which will
enable us to map out the functions of indefinite determiners in Swedish against the
background of cross-linguistically recurrent types of functions. The second source is work on
the historical development of indefinite determiners across languages (Heine 1997,
Haspelmath 1997), which will provide us with a key to the basic contrast between the
indefinite article and the existential determiner in Swedish.
The rest of this section is structured as follows. In section 4.1, we present the functions of
indefinite noun phrases distinguished by Haspelmath (1997) and relate them to the Swedish
situation. In section 4.2, we review relevant work on the historical development of indefinite
determiners. In section 4.3, we try to put the various pieces together and present the outline of
a formal semantic analysis of singular indefinite determiners in Swedish.
28
4.1 Referential Functions of Indefinite Noun Phrases
Haspelmath (1997), in a broad typological study of indefinite pronouns primarily involving
40 languages, distinguishes nine main functions of indefinite noun phrases,18 functions that
are based on important semantic and pragmatic distinctions that seem to recur crosslinguistically. These functions are (Haspelmath 1997:2–3):19
1.
Specific, known to the speaker
‘Somebody called while you were out: guess who!’
2.
Specific, unknown to the speaker
‘I heard something, but I couldn’t tell what kind of sound it was.’
3.
Non-specific, irrealis
‘Please try something else.’
4.
Polar question
‘Did anybody tell you anything about it?’
5.
Conditional protasis
‘If you see anything, tell me immediately.’
6.
Indirect negation
‘I don’t think that anybody knows the answer.’
7.
Direct negation
‘Nobody knows the answer.’
8.
Standard of comparison
‘Peter can run faster than anybody.’
9.
Free choice
‘Anybody can solve this simple problem.’
These functions can be grouped hierarchically, with a primary distinction between specific
reference (1–2) and non-specific reference (3–9). Within the non-specific group, functions 4–
8 can further be characterized as negative polarity functions (cf. Haspelmath 1997:52). (Note,
however, that these functions only account for the referential uses of indefinite noun phrases.
The quantitative uses discussed in section 2.2 and 3.4 fall outside the classification altogether.
We will leave the quantitative uses aside for the moment and return to them later in section
4.3.)
More important than the hierarchical grouping is the fact that the nine functions form an
implicational map in the following way:
18
Although Haspelmath’s study deals primarily with indefinite pronouns, this category is taken to include
indefinite determiners belonging to the same paradigm as indefinite pronouns (such as Swedish någon). The nine
functions can therefore be seen as functions not only of indefinite pronouns but of indefinite noun phrases more
generally.
19
The order in which the functions are listed here differs from the order in Haspelmath (1997:2–3) but reflects
the numbering used in the implicational map of Haspelmath (1997:4). In addition, some of the examples have
been changed so that all indefinites are noun phrases (Haspelmath’s study also includes pronominal adverbs like
somewhere and anywhere).
29
(4)
question
(1)
specific
known
(2)
specific
unknown
(6)
indirect
question
(7)
direct
negation
(3)
irrealis
non-specific
(5)
(8)
conditional comparative
(9)
free
choice
Figure 1
The implicational map for indefinite pronoun functions (Haspelmath 1997:4)
The implicational map is to be interpreted as making predictions about what combinations of
functions are possible for a single indefinite pronoun or determiner. More precisely, if we
consider the map as an undirected graph, with nodes labeled by functions and edges marked
by lines between nodes, then the prediction is that the set of functions associated with a
particular pronoun or determiner always label a connected subgraph, i.e. a set of nodes C such
that any pair of nodes in C are connected by a path only involving nodes in C. For example,
according to Haspelmath (1997:4), the English any-series covers functions 4–9, and the
Russian nibud’-series covers functions 3–5, both of which correspond to connected subgraphs
of the graph in Figure 1. More importantly, all of the languages in Haspelmath’s 40-language
have indefinite pronouns that conform to the predictions of the implicational map.
Furthermore, Haspelmath assumes that adjacency on the map can only be explained by
similarity of function or meaning and that the map may therefore be considered as a geometric
representation of its semantic or cognitive domain.
The predictions of the implicational map seems to hold good for Swedish singular indefinite
determiners as well.20 Thus, the indefinite article en covers functions 1–7, while the
existential determiner någon covers functions 2–8. This is illustrated in Figure 2.
20
Swedish is in fact one of the languages in the 40-language sample, but the analysis only takes into account
pronominal uses of någon.
30
(4)
question
IA (en)
(1)
specific
known
(2)
specific
unknown
(6)
indirect
question
(7)
direct
negation
(3)
irrealis
non-specific
ED (någon)
(5)
(8)
conditional comparative
(9)
free
choice
Figure 2
Functions of singular indefinite determiners in Swedish
The functions ‘standard of comparison’ and ‘free choice’ have not been discussed in earlier
sections, but it is fairly clear that neither of the two determiners can have the free choice
function and that only the existential determiner can be used in the comparative construction,
as evidenced by the following examples:
(97)
a.
b.
c.
(98)
a.
b.
Vilken idiot som helst kan lösa det här problemet.
which fool SOM HELST can solve this here problem
‘Any fool can solve this problem.’
En idiot kan lösa det här problemet.
IA fool can solve this here problem
‘A fool can solve this problem.’
Någon idiot kan lösa det här problemet.
ED fool can solve this here problem
‘Some fool can solve this problem.’
Peter kan springa fortare än någon klasskamrat.
Peter can run faster than ED class-mate
‘Peter can run faster than any of his class mates/some class mate.’
Peter kan springa fortare än en klasskamrat.
Peter can run faster than IA class-mate
‘Peter can run faster than a class mate.’
As illustrated in (97a), the free choice function is typically expressed in Swedish by an
expression of the Wh+som helst-series (cf. Sæbø 1999). It is true that (97b) has a generic
interpretation which comes close in meaning to (97a), but this is not a proper free choice
reading. By contrast, (97c) can only mean that there is at least one fool, whose identity is not
known to the speaker, who can solve the problem. Turning to the examples in (98), we note
first that (98b) can only mean that Peter can run faster than one of his class mates, whereas
(98a) is ambiguous between a negative polarity reading (‘Peter can run faster than any of his
class mates’) and a specific reading (‘Peter can run faster than one of his class mates’).
One thing that is striking about Figure 2 is the large amount of overlap between the two
determiners. However, if we look a little closer at the way they are distributed within each
main function, we find a pattern where the indefinite article is preferred for functions to the
left in the diagram, while the existential determiner becomes more common the further we go
to the right. Thus, the existential determiner is the marked alternative in noun phrases with
31
specific reference, even if the referent is unknown to the speaker. Conversely, the indefinite
article is less preferred in negative and interrogative contexts (possibly also in conditional
contexts) and occurs especially when the speaker expects a positive answer or makes some
sort of existential presupposition (cf. sections 2.2.1 and 3.3.1). Apparently, this is a pattern
that is very stable cross-linguistically, as Haspelmath (1997:52) observes: ‘In a number of
unrelated languages, the pragmatic contrast between positive and negative or neutral
expectations of the speaker is relevant for the choice of the indefinite pronoun within the
conditional and (polar) question functions. Invariably the indefinite series that is chosen for
positive expectations is the one that has more functions to the left of the question/conditional
functions on the map, and the series that is chosen for neutral or negative expectations is the
one that has more functions to the right on the map.’ In the case of our Swedish indefinite
determiners, we have seen that this pattern is also supported by a statistical corpus analysis.
Although the framework of Haspelmath (1997) permits us to place the Swedish data into a
wider typological context, and to see that it matches certain cross-linguistically valid patterns,
we still have to explain why these functions pattern in the way they do, and how the different
functions are related. Which are the basic uses of the indefinite article and the existential
determiner? And how can derivative uses be explained? In order to provide at least a partial
answer to these questions, we will now make a brief digression into the grammaticalization of
indefinite determiners (section 4.2). In section 4.3, we will then return to the analysis of
Swedish indefinite determiners.
4.2 The Origin of Indefinite Determiners
In section 2.3, we briefly considered some historical evidence bearing on the uses of the
existential determiner någon in Swedish. In this section, we are going to pursue the historical
perspective a little further by considering the way indefinite determiners typically develop in
natural languages. We will see that both the indefinite article and the existential determiner in
Swedish instantiate typologically well-known patterns and that these patterns provide a key to
their contrasting uses in modern Swedish.
Indefinite articles are atypical among grammatical categories in that, in all the languages
that have them, they invariably seem to be derived from the same source, viz., the numeral
meaning ‘one’ (Heine 1997). This is especially clear in the case of Swedish, where the two
items are still homonymous. It is not even clear that the two uses can always be clearly
distinguished, and it may in fact be more correct to regard it as one polysemous expression
(cf. Teleman et al 1999).
Heine (1997) presents a five-stage model for the evolution of the indefinite article from the
numeral, a model that can be interpreted both as a synchronic implicational scale 21 and as a
model of diachronic evolution. He also points out that the five-stage model only captures the
more salient steps in the evolution, which in reality proceeds gradually in a large number of
small, contextually defined extensions, but for the purposes of the present article this model is
more than adequate. The five stages are the following (Heine 1997: 72–73):
(99)
Stage I:
Stage II:
Stage III:
Stage IV:
Stage V:
The numeral
The presentative marker
The specific marker
The nonspecific marker
The generalized article
21
This means that if the indefinite article in some language have the properties associated with stage i it also has
the properties associated with all stages j such that j < i.
32
In stage I, the numeral is only used as a numeral and not with nouns that have indefinite
reference, whether specific or non-specific. Stage II is reached when the numeral/article is
used to introduce a new discourse referent, known to the speaker but presumed to be unknown
to the hearer, and this referent is taken up as definite in subsequent discourse. In stage III, the
article is still confined to noun phrases with specific reference, where the referent is known to
the speaker, but may be used even if the referent is not expected to be taken up in subsequent
discourse. In stage IV, the article can be used with referents that are unknown to both speaker
and hearer, typically with non-specific reference, but it is still restricted to singular count
nouns. Finally, in stage V, the use of the article may be extended to plurals and mass nouns.
By and large, modern Swedish seems to be in stage IV. The indefinite article is largely
confined to singular count nouns,22 but it may be used with all kinds of reference, specific and
non-specific, and the referent may be known or unknown to the speaker. In the present
context, however, we are more interested in its historical development. If we assume that the
Swedish indefinite article has developed according to Heine’s five-stage model, then two
things are worth noting. First of all, in stage I, noun phrases with the numeral/article would
normally be used quantitatively to assert that the cardinality of a particular set is one, and
would therefore presuppose the existence of a specific individual being the sole member of
this set. In fact, this is probably the very reason why indefinite articles tend to develop out of
the numeral ‘one’, being confined initially to noun phrases with specific reference, as pointed
out by Givón (1981). In any case, we may conclude that the original use of the numeral/article
expresses something stronger than mere existential quantification.
The second thing to note about the development of the indefinite article is that its earliest
referential function is that of ‘specific, known to the speaker’, or function 1 in Haspelmath’s
classification (cf. section 4.1). This means that the use of the indefinite article, even in this
stage, presupposes something more than the mere existence of a referent, namely some kind
of knowledge or information on the part of the speaker. These two elements in the early uses
of the indefinite article, presupposition of existence and information on the part of the
speaker, seems to provide the basis for a semantic contrast with the existential determiner that
is reflected even in present-day uses.
Let us now turn to the development of the existential determiner någon. As noted in section
2.3, the accepted etymology traces this expression back to a phrase meaning ‘I don’t know
which’. This makes it a prototypical instance of the ‘dunno’ type in Haspelmath’s (1997)
survey of source constructions for indefiniteness markers, a type that is well attested
especially in European languages. The original function of this construction is clearly that of
‘specific, unknown to the speaker’, or function 2 in Haspelmath’s classification (cf. section
4.1). Thus, the use of this expression would explicitly deny that which was presupposed by
the indefinite article in its early development, namely knowledge of the referent on the part of
the speaker. In fact, the use of this expression would seem to express little more than ordinary
existential quantification, pure and simple.
4.3 Singular Indefinite Determiners in Swedish
In the previous section, we tried to establish, using indirect evidence from language typology
and grammaticalization studies, the original uses of the indefinite article en and the existential
determiner någon. If this reconstruction is correct, then the development of these determiners
started from a situation where en primarily had a quantitative use, but one that nevertheless
presupposed the existence of a specific referent, and which was first extended to specific
22
It is true that the indefinite article has a plural form, ena, but this form has a very restricted distribution. It is
mainly used as a so-called emotive article (cf. Teleman et al 1999).
33
referential uses where the referent was known to the speaker. By contrast, någon was used
about specific referents unknown to the speaker and thus presupposed nothing more than the
existence of the referent. For the original situation, we might propose the following semantic
representations for en and någon:
(100) a.
b.
enNUM
P.Q.!x[P(x)  Q(x)]
(101) a.
b.
någon
P.Q.x[P(x)  Q(x)]
Assuming that P and Q are variables of type <e,t> ranging over sets of individuals, the logical
expressions in (100b) and (101b) are both of the type <<e,t>,<e,t>,t>, which means that they
denote functions from sets of individuals to generalized quantifiers. In other words, both en
and någon are analyzed as determiners in generalized quantifier theory, even though the
analysis is not expressed directly in terms of relations between sets but in the more traditional
way using standard quantifiers and lambda abstraction (cf. Barwise and Cooper 1981,
Westerståhl 1985, Partee et al 1993).
Now, it is quite clear that the semantic representation for the numeral/indefinite article en is
no longer valid in all its uses, since as a result of grammaticalization this determiner does not
retain its numeral sense in all contexts anymore. For example, there seems to be nothing
contradictory about the exchange in (102):
(102) A:
B:
Har hon köpt en bil?
has she bought IA car
‘Has she bought a car?’
Ja, hon har till och med köpt två.
yes, she has to and with bought two
‘Yes, she has even bought two.’
On the other hand, it is still possible to pick up on the numeral meaning of en, as illustrated by
the fact that the exchange in (103) is also okay:
(103) A:
B:
Har hon köpt en bil?
has she bought IA car
‘Has she bought a car?’
Nej, hon har köpt två.
no, she has bought two
‘No, she has bought two.’
This can be compared with the situation for the existential determiner någon, which is quite
impossible to coerce into the sense ‘exactly one’:
(104) A:
B:
Har hon köpt någon bil?
has she bought ED car
‘Has she bought a car?’
Ja, hon har till och med köpt två.
yes, she has to and with bought two
‘Yes, she has even bought two.’
34
(105) A:
B:
Har hon köpt någon bil?
has she bought ED car
‘Has she bought a car?’
??
Nej, hon har köpt två.
no, she has bought two
‘No, she has bought two.’
Provisionally, we will therefore treat en in modern Swedish as polysemous, using the
semantic representation in (100b) to capture the numeral sense exemplified in (103). The
numeral sense has to be distinguished from the article sense, which is present in (102). The
question that now confronts is how to analyze the latter.
Let us first note that it will not be sufficient simply to remove the uniqueness condition
from (100b), since this would make the article sense of en equivalent to the sense of the
existential determiner någon, as represented in (101b). In this way, we would fail to capture
the fact that the indefinite article in its basic use seems to presuppose existence and
acquaintance, whereas the existential determiner is only used to assert existence (and to
indicate lack of acquaintance). Using the indefinite article (with specific reference) is a bit
like using existential quantification in intuitionistic logic, in the sense that such an assertion
has to be supported by a specific instance of the quantified statement. By contrast, the
existential determiner behaves more like existential quantification in classical logic, where we
can prove an existentially quantified without proving a specific instance.
The way we propose to capture this distinction within a classical framework is to assume
that the indefinite article (in its numeral sense) does not introduce a quantifier at all, but rather
a special kind of singular term. Broadly speaking, there are two main traditions within formal
semantics as regards the treatment of indefinite noun phrases. The older tradition, represented
by Lewis (1970), Montague (1973), Barwise and Cooper (1981), among others, treats all noun
phrases, including indefinites, as denoting generalized quantifiers. The more recent tradition,
which is mainly represented by discourse-oriented theories such as those of Webber (1978),
Kamp (1981), Heim (1982), Barwise and Perry (1983), instead treats indefinites as singular
terms, typically introducing new discourse referents. The present proposal follows the older
tradition by treating noun phrases determined by the indefinite article as denoting generalized
quantifiers (sets of sets of individuals), but follows the more recent tradition in that their
semantic representation will not contain a standard (existential) quantifier but a special kind
of singular term. More precisely, we propose the following semantic representation for the
article sense of en:
(106) a.
b.
enART
P.Q.[P(arb)  Q(arb)]
The individual term arb is a special kind of term that stands for an arbitrary object (Fine
1985, Steedman 2000). An arbitrary object is an object with which properties can be
associated but whose extensional identity in terms of actual objects is unspecified. In this
respect, arbitrary objects resemble Skolem terms, and the process of skolemization will
eventually play a part in the present account also. The idea behind this analysis is that the use
of the indefinite article, at least in its original specific use, introduces an individual whose
identity is unspecified relative to current universe of discourse (or common ground).
However, there is nothing that prevents the speaker from privately identifying this individual
with a specific known individual — ‘having a specific individual in mind’ as it were —
something that will typically be the case in assertions but not in questions. But this eventual
link to the speaker’s information state is something that is not captured in the semantic
representation per se.
35
The most important difference between the semantic representations in (100b) and (101b),
on the one hand, and the representation in (106b), on the other, is that the latter involves a
specific individual, albeit an arbitrary one. In this way, we propose to account for the
specificity and existence presupposition associated with the prototypical use of the indefinite
article. However, this means that the indefinite article will fail to satisfy one of the basic
conditions on determiners in generalized quantifier theory, viz., the condition of Quantity
(see, e.g., Partee et al 1993). This condition requires that only the number of elements in the
relevant sets of individuals, i.e., in the intepretations of whatever predicates are substituted for
P and Q in (106b), is relevant for the interpretation. We do not find this particularly worrying,
since what is ruled out by Quantity is precisely the kind of specificity in interpretation that
seems to be inherent in the original use of the indefinite article. Furthermore, the quantifiers
induced by the determiner interpretation in (106b) satisfies all the other conditions commonly
associated with natural language determiners, viz., Conservativity, Constancy and Variation
(Partee et al 1993).
Let us now consider some examples in order to see how the proposed analysis may account
for the data examined in sections 2 and 3. In discussing these examples, we will simplify both
the syntactic and semantic analysis in many respects that are not immediately relevant to our
present concerns. However, we still need to make certain minimal assumptions about both
syntax and semantics. When analyzing a sentence like (1a), repeated below as (107), we will
assume the semantic representations in (108–109) for the constituents hon (she) and har köpt
(has bought), where R is a variable of the generalized quantifier type <<e,t>,t>:
(107) Hon har köpt en bil.
she has bought IA car
‘She has bought a car.’
(108) a.
b.
hon
P.[P(s)]
(109) a.
b.
har köpt
R.y.[R(z.has-bought(y,z))]
The pronoun hon is simply translated into (a type-lifted version of) an individual variable,
while the verb complex har köpt (has bought) is taken to denote a relation between
individuals and generalized quantifiers (sets of sets of individuals), which is defined in terms
of a more basic relation between individuals. Except for being purely extensional, this
analysis follows the tradition from Montague (1973).
In deriving sentences and their semantic representations, we will assume some kind of
extended categorial grammar that will allow us to derive sentences like (107) in two different
ways in order to capture potential scope alternations. In addition to the orthodox analysis in
terms of functional application, represented in (110), we also want to allow a quantifier in
object position to take scope over the entire sentence. One way of doing this is through a
combination of forward composition (FC) and type raising (TR), as indicated in (111). 23
23
The literature on extended, or flexible, categorial grammar is by now very rich. The interested reader is
referred to Buszkowski, Marciszewski and van Benthem (1988) and Oehrle, Bach and Wheeler (1988) for a
representative sample of different but related approaches. For a more basic introduction, see Wood (1993).
36
(110) Hon
<<e,t>,t>
har köpt
<<<e,t>,t>,<e,t>>
<e,t>
t
(111) Hon
har köpt
<<e,t>,t>
<<<e,t>,t>,<e,t>>FC
<<<e,t>,t>,t>
t
en bil
<<e,t>,t> FA
FA
en bil
<<e,t>,t>
<<<<e,t>,t>,t>,t>
TR
FA
In what follows, we will not be concerned with the precise details of the syntactic derivations,
nor with the corresponding derivations of semantic representations, which will be left as the
proverbial exercise to the reader.
Given the semantic representations in (106), (108) and (109), the sentence in (107) will be
assigned the following semantic representation:
(112) a.
b.
Hon har köpt en bil.
she has bought IA car
‘She has bought a car.’
[car(arb)  has-bought(s,arb)]
This can be compared with the corresponding sentence involving the existential determiner
which, given the semantic representation in (101), comes out as follows:
(113) a.
b.
Hon har köpt någon bil.
she has bought ED car
‘She has bought some car.’
x[car(x)  has-bought(s,x)]
Whereas (112) refers to a specific, though as yet unidentified individual, which has the
property of being a car and having been bought by the referent of hon, (113) merely asserts
that there exists at least one individual having this property. Thus, (112) can be said both to
presuppose the existence of a referent and to be more specific than (113), since the former
entails the latter but not vice versa.
One thing that is not directly captured in this account is the fact that the use of the
existential determiner indicates that the speaker does not know which individual supports the
truth of the existential quantification. Originally, before the expression was grammaticalized,
this was part of the explicit content asserted (‘I don’t know which’). At present, it is unclear
what status we should ascribe to this content. One option is to treat it as a kind of implicature,
arising in virtue of the contrast between the existential determiner and the indefinite article. In
other words, if the speaker had knowledge of the individual, he would normally use the
indefinite article to make an assertion about this particular individual (though without
identifying him). By using instead the weakest possible expression, a noun phrase introduced
by the existential determiner, the speaker invites the inference that he lacks such knowledge
(or considers it less important). We are not sure whether this is the correct analysis in the end,
but we will leave the matter for now and stick with the semantic representation already
proposed for the existential determiner.
We will now consider some other uses of the indefinite article and the existential
determiner in order to see whether the resulting semantic representations conform to our
expectations. We will focus on what we have so far called referential uses and save the
37
quantitative use for special consideration later. Let us begin with some examples involving
negation:
(114) a.
b.
c.
(115) a.
b.
c.
Hon har inte köpt en bil.
she has not bought IA car
‘She hasn’t bought a car.’
[car(arb)  has-bought(s,arb)]
[car(arb)  has-bought(s,arb)]
Hon har inte köpt någon bil.
she has not bought ED car
‘She hasn’t bought a/any car.’
x[car(x)  has-bought(s,x)]
x[car(x)  has-bought(s,x)]
Let us first note that, given the possibility of derivations like that in (111), we will be able to
get the representations in (114c) and (115c), which capture the specific (or ‘wide scope’)
readings of (114a) and (115a), respectively (cf. section 2.1). However, it is more interesting to
compare the readings resulting from the standard derivation using only functional application,
which are given in (114b) and (115b). As we can see, (115b) simply denies the existence of a
car being bought by the referent of hon, which squares well with the intuition that (115a) is
the unmarked and neutral way of denying either (112a) or (113a). By contrast, (114b) asserts
of a specific but unidentified individual that it fails to satisfy the conjunctive condition of
being a car and being bought by the referent of hon. Again, this conforms to our intuitions
insofar as we noted in section 2.1.1 that the indefinite article is appropriate in negative
contexts especially when the existence of a referent is somehow presupposed. Note also that
there are three different ways in which an individual can fail to satisfy a conjunctive
condition, each of which corresponds to a natural continuation of (114a) (cf. also section
2.1.1):
(116) a.
b.
c.
Hon har inte köpt en bil, hon har köpt en motorcykel.
she has not bought IA car she has bought IA motorcycle
‘She hasn’t bought a car, she has bought a motorcycle.’
Hon har inte köpt en bil, hon har hyrt den.
she has not bought IA car she has rented it
‘She hasn’t bought a car, she has rented it.’
Hon har inte köpt en bil, hon har hyrt en motorcykel.
she has not bought IA car she has rented IA motorcycle
‘She hasn’t bought a car, she has rented a motorcycle.’
car(arb)
has-bought(s,arb)
car(arb)
has-bought(s,arb)
However, we must not forget that en also has its numeral sense, which means that we have
two additional semantic representations for (114a):
(117) a.
b.
c.
Hon har inte köpt en bil.
she has not bought IA car
‘She hasn’t bought a car.’
!x[car(x)  has-bought(s,x)]
!x[car(x)  has-bought(s,x)]
In this case, the specific reading (‘there is exactly one car that she hasn’t bought’), represented
in (117c) seems hard to get in most contexts, but there is no reason to rule it out in principle.
38
Turning to the non-specific reading in (117b), we note that there may be two different reasons
for making such an assertion, as illustrated in (118) (cf. also section 2.1.1):
(118) a.
b.
Hon har inte köpt en bil, hon har köpt två.
she has not bought IA car she has bought two
‘She hasn’t bought a/one car, she has bought two.’
Hon har inte köpt en bil, inte en (enda).
she has not bought IA car not IA (single)
‘She hasn’t bought a/one car, not (a single) one.’
>1
<1
For an account that associates the indefinite article with existence presuppositions, an
example like (118b) may appear problematic, since it does not seem to presuppose the
existence of a referent any more than (115a). However, we believe that this example involves
the numeral sense of en, rather than its article sense. Some support for this claim can be
derived from the fact that it can appear without a head noun in an emphatic continuation like
inte en (not one).
Next, we will consider non-specific uses resulting from the interaction with quantified noun
phrases such as alla studenter (all students). For the existential determiner, we simply get the
two expected readings, one specific (wide scope) and one non-specific (narrow scope):
(119) a.
b.
c.
Alla studenter har läst någon bok.
all students have read ED book
‘All (the) students have read some book.’
y[student(y)  x[book(x)  has-read(y,x)]]
x[book(x)  y[student(y)  has-read(y,x)]]
For the indefinite article, the situation is more complicated. As far as the derivation of
semantic representations is concerned, we only get one analysis:
(120) a.
b.
Alla studenter har läst en bok.
all students have read IA book
‘All (the) students have read a book.’
y[student(y)  [book(arb)  has-read(y,arb)]]
However, if we assume that the term arb is really a Skolem term, then skolemization may be
applied either before or after the term has come under the scope of the universal quantifier,
which will result in two distinct interpretations (cf. Steedman 2000):
(121) a.
b.
y[student(y)  [book(f)  has-read(f)]]
y[student(y)  [book(f(y))  has-read(y,f(y))]]
In (121a), the Skolem term consists of a zero-place functor, which means that skolemization
has taken place outside the scope of any universal quantifier. This yields a specific
interpretation of the indefinite noun phrase, i.e., there is a specific, unidentified book, which
all students have read. In (121b), the Skolem term consists of a one-place functor applied to
the variable y bound by the universal quantifier, which is the usual result of skolemization
within the scope of a universal quantifier. This means that the interpretation of the arbitrary
object term is dependent upon the value of the variable y, which in turn means that it may
refer to different books for different students, thus giving us a non-specific interpretation of
the indefinite noun phrase.
39
The analysis of other non-specific uses, such as those resulting from the interaction with
modal operators and intensional predicates (cf. section 2.1.2), requires theoretical concepts
that have not been discussed and would carry us to far in the present context. Suffice it to say
that, to the extent that these contexts can be analyzed in terms of scope-bearing elements,
there is no apparent reason why the analysis sketched above could not be extended to these
contexts as well. Needless to say, this is something that will have to be proved in the future.
Before we close the survey of examples, it may be worth noting that the analysis proposed
for the indefinite article and the existential determiner also seems to work for their predicative
uses. Given that the copula verb är is analyzed as in (122), we seem to get reasonable
semantic representations for the sentences in (123) and (124).
(122) a.
b.
är
R.y.R(z.y = z)
(123) a.
b.
Hon är en berömd läkare.
[famous-doctor(arb)  s = arb]
(124) a.
b.
Hon är någon berömd läkare.
x[famous-doctor(x)  s = x]
Let us now return to the issue of negative polarity. We started out in section 2 by claiming
that the existential determiner is predominantly used in NPCs, where it is the unmarked
determiner. These claims were supported by the statistical analysis in section 3, which showed
a very clear association between the existential determiner and NPCs. Moreover, we saw in
section 2 that in Danish and Norwegian the existential determiner is used exclusively in
NPCs. The question then arises whether this development can be explained on semantic
grounds. Is there anything about the meaning of the existential determiner, as analyzed here,
that makes it particularly well suited for use in NPCs? In order to answer this question, we
must first make a brief digression and discuss some previous work on NPIs.
The literature on negative polarity items is extensive and we will not attempt to summarize
it here. Broadly speaking, there are two main traditions in trying to explain the distribution of
NPIs, a syntactic tradition starting with Klima (1964) and Jackendoff (1968), and a semantic
tradition going back to Horn (1972) and Fauconnier (1975a, 1975b, 1978, 1980). In what
follows we shall be concerned only with the semantic tradition, and more precisely with the
idea that NPIs denote end-points on semantic or pragmatic scales, or extreme elements among
a set of alternatives. This idea has been developed by Fauconnier (1975a, 1975b, 1978, 1980),
Ladusaw (1979, 1983), Heim (1987), and Krifka (1991, 1995). The presentation below is
primarily based on Krifka (1991, 1995).
The basic idea in the semantic approach to NPIs is that they denote low points on semantic
or pragmatic scales. Determining exactly what constitutes a scale in the relevant sense and
what counts as the low point on such a scale can be quite problematic, but in many cases the
notion of a scale can be reduced to an ordering with respect to specificity or entailment with
the low point being the least specific or weakest item. For example, consider the following
quantity scale (from Fauconnier 1980):
(125) <a drop, …, 1 pint, …, 1 quart, …, 1 gallon, …, 10 gallons, …>
On this scale, the expression a drop is taken to denote the minimum quantity. This means that
simple, positive assertions with a drop will be entailed by all corresponding assertions where
a drop is replaced by an expression higher on the scale. Thus, (126a) is entailed by (126b) and
(126c), and (126b) is in turn entailed by (126c):
40
(126) a.
b.
c.
He drank a drop.
He drank a pint.
He drank a quart.
However, in certain contexts, called downward-entailing or scale-reversing contexts, these
entailment relations are reversed, so that the assertions containing the weakest element in fact
are the strongest assertions. This is exemplified in (127), where the scale reversal is due to
negation, and where (127a) entails (127b) and (127c), and where (127b) entails (127c):
(127) a.
b.
c.
He didn’t drink a drop.
He didn’t drink a pint.
He didn’t drink a quart.
Given these observations, it seems that we should be able to explain the distribution of NPIs
as resulting from the interaction of their content, denoting low points on a scale, or weakest
possible alternatives, with general pragmatic principles such as Grice’s maxim of quantity.
The use of NPIs should be restricted to downward-entailing contexts, simply because these
are the only contexts where they can be used to make reasonably informative statements.
In reality, the situation is complicated by a number of factors. For example, there are NPCs
that are not downward-entailing, such as the protasis of conditionals (Heim 1987). On the
other hand, there are so-called strong NPIs (e.g., lift a finger, bat an eyelash, etc.) that require
contexts that are not only downward-entailing but also anti-additive24 (Krifka 1995). Another
problem for this approach is that it is not clear that all relevant scales can be reduced to
semantic notions like entailment or subset relations. Furthermore, it seems that many scales
are restricted to a particular set of alternative expressions, the precise delimitation of which is
subject to language-specific conventions rather than universal semantic principles.
A treatment of these problems is clearly beyond the scope of this article. In what follows,
we will simply assume that NPIs can somehow be analyzed in terms of semantic-pragmatic
scales, where NPIs denote the weakest element among a set of alternative expressions
forming such a scale. One way of formalizing this intuition is in terms of polarity lattices
along the lines of Krifka (1991), assuming for every NPI A (with denotation A’) a structure
LA, defined as follows:
(128) LA = < A’, LA, ≤A> is an NP lattice iff
a.
if A’ is of type , then LA is of type <,t>, i.e., a set of entities of type ;
b.
≤A is a quasi-order (preorder) on LA, i.e., ≤A is reflexive and transitive;
c.
A’  LA and A’  LA, i.e., LA contains at least one element other than A’;
d.
A’ is the unique Y such that for every X  LA, Y ≤A X (least element).
Returning now to Swedish indefinite determiners, it is not hard to see that we have a potential
NP lattice with the existential determiner being the weakest expression and denoting the lower
point on the scale. Using the lattice-theoretic framework of Krifka (1991), we may formalize
the situation by constructing the following NP lattice:
A truth-functional connective C is anti-additive iff [(p  q)Cr  (pCr  qCr)] and [pC(q  r)  (pCq  pCr)]
are logically valid (Zwarts 1995).
24
41
(129) Lnågon = < någon’, L någon, ≤någon> where
L någon = någon’, enNUM’, enART’
någon’ = PQx[P(x)  Q(x)]
enNUM’ = PQ!x[P(x)  Q(x)]
enART’ = PQ[P(arb)  Q(arb)]
≤någon = 
In this way, we can explain why någon is preferred in NPCs, even though its original use is
one of specific reference. Being the weakest expression among a set of alternatives, as shown
in (129), it qualifies as denoting the low point on a semantic-pragmatic scale and is therefore
at least a candidate for NPI-hood.
So much for Swedish, but where does this leave the other Scandinavian languages treated in
this article? The most likely explanation for the present situation would seem to be that the
existential determiner started out as a determiner for specific referents unknown to the
speaker. Given its basic contrast with the numeral/indefinite article en, it was gradually
extended to non-specific uses and specifically to negative polarity uses. All this seems to have
been common to Danish, Norwegian and Swedish, although I have no specific historical
evidence to back up this claim. Now, given Haspelmath’s contiguity condition (that all uses of
a single expression must form a connected subgraph of the implicational map), it seems that
uses can only be lost either from the left or from the right, so to speak. In Swedish, both the
‘specific unknown’ function and the ‘direct negation’ function have remained, and therefore
all the functions in between. But in Danish and Norwegian apparently functions have been
dropped from the left on the map, up to a point where modern Norwegian only retains
functions 4, 6 and 7, and Danish possibly only 7. At the moment of writing, we have not
historical data from Danish and Norwegian to support this hypothesis, but it would be very
interesting to see if such evidence is available.
One may speculate about the reasons for this differential development. One clue probably
lies in the parallel grammaticalization of the indefinite article, although it is very difficult to
say which is the cause and which is the effect. But it clearly seems that the grammaticalization
of the indefinite article has progressed further in Danish and Norwegian than in Swedish, as
evidenced for example by the fact that the indefinite article is the unmarked choice in NPCs
and does not carry any existence presupposition in the way that the Swedish indefinite article
often does in these contexts. This hypothesis is indirectly supported by the fact that the
differential development of någon is confined solely to its combination with singular count
nouns, which is the only context where it contrasts with the indefinite article. With plurals and
mass nouns (and in pronominal uses), Danish nogen and Norwegian noen both seem to be
alive and well and used to the same extent as Swedish någon.
Finally, we have to say something about the quantitative uses of en and någon in Swedish.
We will not present a formal account of these uses, since this would require an analysis of
adjuncts that go beyond the scope of the present article, but we want to make a few remarks
on their analysis. For en this is quite straightforward, since its sense in quantitative uses is
simply the original numeral sense. Thus, in example (51a), repeated below as (130a), en
contrasts with två (two), tre (three), etc.
(130) a.
b.
Lisa var här för en vecka sedan.
Lisa was here for IA week since
‘Lisa was here a week ago.’
Lisa var här för någon vecka sedan.
Lisa was here for ED week since
‘Lisa was here about a week ago.’
42
The question then arises why the article sense of en is excluded in this kind of construction.
On the present account there is a straightforward explanation if we assume that the temporal
adjunct involves quantification over periods of time. Since the article sense of en involves
reference to a specific individual, this sense is simply not relevant in this context (but cf.
example 9).
Concerning the use of någon in (130b), we may note that if what is going on is
quantification over periods of time, then the literal sense would be that Lisa was her at least
one week ago. Apparently, there has been a slight shift in meaning here from ‘at least one’ to
‘about one’. Alternatively, we can think of it as a natural semantic extension from ‘I don’t
known which’ (referential) to ‘I don’t know how many’ (quantitative). In any case, the
quantitative use seems to be derived from the ‘specific unknown’ function. In fact, indirect
evidence for this claim comes also from another Scandinavian language, viz., Icelandic.
In other respects, Icelandic is similar to Danish and Norwegian in that the cognate of någon,
nokkur, is mainly restricted to NPCs.25 However, there is another determiner, einhver, which
seems to cover roughly the same ground as Danish/Norwegian en eller anden/annen, which
means that it can be used about a specific referent unknown to the speaker. Apparently, this
determiner can also be used with the sense of ‘about one’, although this use is still considered
substandard from a normative point of view. This use is illustrated in (131), which could be
heard, for example, in a report from an ongoing football game:
(131) fia› er einhver mínúta eftir.
there is one-or-another minute left
‘There is about one minute left.’
We also find parallels in English and French, as exemplified in (132) and (133):
(132) We met some twenty years ago.
(133) Nous nous sommes rencontrés il y a quelque vingt ans.
we us are met there are some twenty years
‘We met some twenty years ago.’
However, in both these cases, the expression corresponding to the existential determiner
(some in English, quelque in French) is combined with an ordinary numeral, and it seems to
be rare that a single determiner expresses the approximate numeral sense ‘about one’, as in
Swedish and Icelandic.
5 Conclusion
In this article, we have examined the meaning and use of singular indefinite determiners in
Swedish from three different perspectives. We started, in section 2, by reviewing the Swedish
data in the light of contrasting data from the closely related languages Danish and Norwegian,
distinguishing three main uses of the existential determiner någon, in contrast with the
indefinite article/numeral en:
1.
Negative polarity uses (corresponding to Danish nogen, Norwegian noen)
25
The situation in Icelandic is more complex in that determiners can be both pre- and postposed, with different
interpretations. Moreover, the indefinite article is less grammaticalized in Icelandic, which means that many
indefinite noun phrases lack a determiner.
43
2.
3.
Other referential uses (corresponding to Danish en eller anden, Norwegian en eller
annen)
Quantitative uses (no corresponding determiner in Danish or Norwegian)
In section 3, we proceeded to back up our claims with empirical evidence from two corpora of
modern Swedish, one consisting of written language and the other consisting of transcribed
spoken language. Statistical analysis confirmed the association of the existential determiner
with NPCs as well as the preference for the indefinite article in noun phrases with specific
reference. Furthermore, we found that all the major uses discussed in section 2 were
represented in the corpus data.
In section 4, we tried to give a theoretical analysis of the indefinite article and the
existential determiner in Swedish. We started by exploiting typological studies of indefinites
as well as historical evidence on the development of indefinite determiners in order to get a
handle on the basic contrasts involved in the modern Swedish situation. We then proceeded to
give a formal semantic analysis of Swedish indefinite determiners that we hope can account
for most of the data discussed in sections 2 and 3. Although the analysis is still incomplete in
important respects, we hope to have gained a better insight into the meaning of indefinite
determiners and the mechanisms that influence their use. In addition, we hope to have shown
that the particular combination of methods and perspectives adopted in the present study is a
fruitful one that can help further our understanding of the semantics and pragmatics of natural
languages.
Acknowledgements
The work presented in this article has mainly been carried out within the project Comparative
Semantics for Nordic Languages (NordSem), sponsored by NOS-H. I want to thank my
colleagues in the project for many fruitful discussions on this and other topics: Jens Allwood,
Robin Cooper, Elisabet Engdahl, Jan Tore Lønning, Kjell Johan Sæbø, Finn Sørensen,
Ingebjørg Tonne, and Carl Vikner. Special thanks to Finnur Fridriksson, Torbjørn Nordgård,
Kjell Johan Sæbø, Hanne Erdmann Thomsen, Ingebjørg Tonne, Carl Vikner, and Bjarne
Ørsnes for providing data from Danish, Icelandic and Norwegian, to Christian Waldmann for
help with coding the corpus material, and to Lauri Carlson and Östen Dahl for useful
comments on an earlier report from this study.
References
Allwood, J. (1999) The Swedish Spoken Language Corpus at Göteborg University. In Fonetik
99: Proceedings from the Twelfth Sqeidsh Phonetics Conference. Gothenburg Papers in
Theoretical Linguistics. (Available at: http://www.ling.gu.se/fonetik99/manus/31.ps.)
Barwise, J. and Cooper, R. (1981) Generalized Quantifiers and Natural Language. Linguistics
and Philosophy 4, 159–219.
Barwise, J. and Perry, J. (1983) Situations and Attitudes. Cambridge, MA: MIT Press.
Buszkowski, W., Marciszewski, W. and van Benthem, J. (1988) Categorial Grammar.
Amsterdam: John Benjamins.
Dahl, Ö. (1988) The Role of Deduction Rules in Semantics. Journal of Semantics 1988, 1–18.
Edmondson, J. A. (1981) Affectivity and Gradient Scope. Chicago Linguistic Society 17, 38–
44.
44
Fauconnier, G. (1975a) Polarity and the Scale Principle. Chicago Linguistic Society 11, 188–
199.
Fauconnier, G. (1975b) Pragmatic Scales and Logical Structure. Linguistic Inquiry 4, 353–
375.
Fauconnier, G. (1978) Implication Reversal in Natural Language. In Guenthner, F. And
Schmidt, S. J. (eds) Formal Semantics and Pragmatics for Natural Languages. Dordrecht:
Reidel, 289–301.
Fauconnier, G. (1980) Pragmatic Entailment and Questions. In Searle, J. R., Kiefer, F. and
Bierwisch, M. (eds) Speech act theory and pragmatics. Dordrecht: Reidel, 57–69.
Fine, K. (1985) Reasoning with Arbitrary Objects. Oxford: Oxford University Press.
Fodor, J. D. & Sag, I. A. (1982) Referential and Quantificational Indefinites. Linguistics and
Philosophy 5, 355–398.
Geach, P. (1962) Reference and Generality. An Examination of Some Medieval and Modern
Theories. Ithaka: Cornell University Press.
Givón, T. (1981) On the development of the numeral ‘one’ as an indefinite marker. Folia
Linguistica Historica 2, 35–53.
Haspelmath, M. (1997) Indefinite Pronouns. Oxford: Clarendon Press.
Heim, I. (1982) The Semantics of Definite and Indefinite Noun Phrases in English. Ph.D.
Thesis. University of Massachusetts at Amherst.
Heim, I. (1987) A Note on Negative Polarity and Downward Entailingness. NELS 14, 98–107.
Heine, B. (1997) Cognitive Foundations of Grammar. New York: Oxford University Press.
Jackson, E. (1995) Weak and Strong Negative Polarity Items: Licensing and Intervention.
Linguistic Analysis 25, 181–256.
Horn, L. (1972) On the Semantic Properties of Logical Operators in English. Ph.D.
Dissertation. University of California at Los Angeles.
Jørgensen, S. W. (1999) Specificity and Danish Event Nominalisations. NORDSEM Report
No. 6. (Available at: http://www.ling.gu.se/research/projects/nordsem.)
Kamp, H. (1981) A Theory of Truth and Semantic Representation. In Groenendijk, J. A.,
Janssen, J. M. V. And Stokhof, M. B. J (eds) Formal Methods in the Study of Language.
Mathematical Centre Tract 135, 277–322. Amsterdam.
Krifka, M. (1991) Some Remarks on Polarity Items. In Zaefferer, D. (ed) Semantic Universals
and Universal Semantics. Foris Publications.
Krifka, M. (1995) The Semantics and Pragmatics of Polarity Items. Linguistic Analysis 25,
209–257.
Kripke, S. (1977) Speaker’s Reference and Semantic Reference. In Martinich, A. P. (ed)
(1985) The philosophy of Language. Oxford University Press, pp. 249–268. [Originally
published in French, P. A., Uehling, Jr, T. E., and Wettstein, H. K. (eds) Contermporary
Perspectives in the Philosophy of Language. University of Minnesota Press.]
Ladusaw, W. (1979) Polarity Sensitivity as Inherent Scope Relations. Doctoral Dissertation.
University of Texas at Austin.
Ladusaw, W. (1983) Logical Form and Conditions on Grammaticality. Linguistics and
Philosophy 6, 373–392.
Lewis, D. (1970) General Semantics. Synthese 22, 18–67.
Lyons, J. (1977) Semantics. Volume 1. Cambridge: Cambridge University Press.
Montague, R. (1973) The Proper Treatment of Quantification in Ordinary English. In
Hintikka, J., Moravcsik, J. M. E. And Suppes, P. (eds) Approaches to Natural Language:
Proceedings of the 1970 Stanford Workshop on Grammar and Semantics, 221–242.
Dordrecht: Reidel.
Nivre, J. (1999) Modifierad standardortografi (MSO6). Göteborg University: Department of
Linguistics. (Available at: http://www.ling.gu.se/SLSA/Postscripts/MSO6.ps.)
45
Oehrle, R. T., Bach, E. T. and Wheeler, D. (1988) Categorial Grammars and Natural
Language Structures. Dordrecht: Reidel.
Partee, B. H., ter Meulen, A. and Wall, R. E. (1993) Mathematical Methods in Linguistics.
Dordrecht: Kluwer.
Steedman, M. (2000) The Syntactic Process. Cambridge, MA: MIT Press.
Svenska Akademiens Ordbok [Dictionary of the Swedish Academy] (1898–) Stockholm:
Svenska Akademien.
Sæbø, K. J. (1999) Free Choice Items in Scandinavian. NORDSEM Report No. 4. (Available
at: http://www.ling.gu.se/research/projects/nordsem.)
Teleman, U., Hellberg, S. and Andersson, E. (1999) Svenska Akademiens Grammatik
[Grammar of the Swedish Academy]. Stockholm: Svenska Akademien.
Warfel, S. L. (1972) Some, reference, and description. In Battle, J. H. and Schweitzer, J. (eds)
Mid-America Linguistics Conference Papers, 41–49. Stillwater: Oklahoma State
University.
Webber, B. L. (1978) A Formal Approach to Discourse Anaphora. Ph.D. Thesis, Harvard
University.
Westerståhl, D. (1985) Logical Constants in Quantifier Languages. Linguistics and
Philosophy 8, 387–413.
Wood, M. McGee (1993) Categorial Grammar. London: Routledge.
Zwarts, F. (1995) Nonveridical Contexts. Linguistic Analysis 25, 286–312.
46
Download