1. Introduction
In this paper, I look at the construction consisting of a pronoun followed by a noun within the same noun phrase constituent, typically with a human referent, as it is used in the Scandinavian languages, illustrated in (1) and (2) by Norwegian and Danish.
- (1)
a.
‘My god! Look at that bloke!’
b.
‘My god! Look at that bloke!’
c.
‘My god! Look at that bloke!’
- (2)
a.
‘That girl is totally gorgeous!’
b.
‘My god, that boy was gorgeous!’
c.
‘That bloke is not stupid.’
I will call this construction the han mannen construction, and refer to the pronoun in this construction as a pronoun demonstrative, in order to be as descriptive and agnostic as possible. This is the same as Johannessen's (Reference Johannessen, Solstad, Grønn and Haug2006) psychologically distal determiner, but is distinct from Delsing's (Reference Delsing1993:54) preproprial article. The han mannen construction is interesting for several reasons. Firstly, the noun in all varieties of Scandinavian (with the possible exception of a variety of Icelandic) must be definite, even in varieties like Danish which normally only have one definite element in a phrase. Secondly, across different varieties of Scandinavian, different forms of the pronoun are required, e.g. Norwegian requires the nominative form (1b) versus (2b), while Danish requires the objective case form (1c) versus (2c), regardless of the grammatical function of the entire noun phrase (Sunnhordlandsk Norwegian does not have a separate object form of third person pronouns, as indicated in the glosses). Lastly comes the question of how the pronoun demonstrative differs from other demonstratives, in particular, what is the difference between han mannen ‘this/that man’ and den mannen ‘this/that man’.
In this paper I address each of these points, and offer an account of the construction within a Lexical-Functional Grammar framework. My proposal is essentially lexical, and is based on the fact that the underlying semantics of the pronoun when it functions as a demonstrative involves a specific referent, while this is not necessarily the case with other definite demonstratives.
The organisation of this paper is as follows. I begin by examining the han mannen construction in a single variety of Norwegian, namely Sunnhordlandsk Norwegian (section 2). This is because the construction appears to have a very straightforward account in this dialect, based on morphological definiteness agreement between demonstratives and the noun. I then present some relevant aspects of Lexical-Functional Grammar in section 3, and show how the han mannen construction can be described within this framework. The account offered here is compatible with that of the Parallel Grammars project (Butt et al. Reference Butt, Dyvik, King, and Rohrer2002), in particular with the implementations for Danish (based on the account of extracted subject case-marking in Danish by Ørsnes Reference Ørsnes2002) and Norwegian (based on the XLE Web Interface LFG parser, http://decentius.aksis.uib.no/logon/xle.xml).
I then look at the han mannen construction in Oslo Norwegian (section 4), based on data from the TAUS corpus of spoken Oslo Norwegian and Johannessen's recent work (2006, 2007, 2008), among others. The picture in Oslo Norwegian is complicated by the enduring influence Danish has had on literary Norwegian. Thus, here we find both demonstratives with definite nouns (see e.g. Vangsnes Reference Vangsnes1999, Reference Vangsnes2001, and others), and demonstratives with non-definite nouns, Danish-style. This requires an examination of the differences between the pronoun demonstrative and other demonstratives in Norwegian, since the pronoun demonstrative never occurs with a non-definite noun (section 4.1).
In section 5 I outline this construction in the other Scandinavian languages, including showing examples where the construction is ungrammatical, as it is for example in Faroese. I summarise the main argument of the paper in section 6.
The data in this paper thus comes from a variety of sources, including tagged corpora, the Internet, and native speaker judgements on constructed examples.
2. HO SUNNHORDLANDSKE DAMO ‘THAT WEST NORWEGIAN WOMAN’
Consider the scenario given in (3).
(3)
a.
look at she woman.def you
‘My god! Look at that woman!’
b.
look at she woman.indef you
‘My god! Look at that woman!’
In Sunnhordlandsk Norwegian (SHLN), spoken in Western Norway, the utterance given in (3a) is perfectly acceptable, yet raises several interesting questions. Firstly, what word class does the pronoun belong to? Secondly, why is the noun in the definite form? That is, why is (3b) ungrammatical?
If we assume that the answer to the first question is that the pronoun is a demonstrative, then the definiteness agreement on the noun will occur in the same way as with all definite determiners in SHLN, and the presence of the definite noun is accounted for trivially.
In support of this we can note several facts. Firstly, both ho and den in (3a) and (4a) are more or less equivalently deictic (or ‘contextually indexed’) demonstratives, as indicated by the nearly identical free translations.
- (4)
a.
‘My god! Look at that woman over there!’
b.
Intended: ‘My god! Look at that woman over there!’
A key defining feature of demonstratives is that they ‘display a heightened sensitivity to speakers’ extralinguistic demonstrations’ (Büring, to appear:3), hence the inclusion of the pictures accompanying (3), illustrating the referent and a prototypical (or exaggerated) reaction which often accompanies the use of the han mannen construction. (The picture of the small girl is not meant to imply that this is only used by children, but it is certainly a part of spoken rather than written language.)
In addition, there is clear morphological evidence that the pronoun in (3a) is a definite determiner in SHLN. In this dialect, all definite determiners, including demonstratives, require definiteness agreement on the noun, as shown by the ungrammatical example (5a) and the grammatical version in (5b), and by the a and b sentences in (3) and (4) above. In addition, adjectives appear in the definite (weak) form with both normal and pronoun demonstratives, illustrated in (6).
(5)
(6)
Thus, the assumption that the pronoun in the han mannen construction is a demonstrative is not contradicted by available evidence. Cross-linguistically, demonstratives are generally definite (Lyons Reference Lyons1999), so SHLN is typical in this respect.
I would like to make clear that I am not claiming that the suffix is only a definite/demonstrative agreement marker, since it also appears on nouns without a prenominal determiner.
(7)
However, since the suffix must occur when there is a separate definite article or demonstrative, it also functions as morphological agreement.
There is other evidence that pronouns in Norwegian generally are demonstratives, as given by e.g. Faarlund, Lie & Vannebo (Reference Faarlund, Lie and Vannebo1997), Fretheim & Amfo (Reference Fretheim and Amfo2005) and Johannessen (Reference Johannessen, Johannessen and Hagen2008). Pronouns, like determiners, occur where other demonstratives occur: with adjectives (8a), with PPs (8b), with other adverbial phrases (8c), as well as alone (8d). In addition, each of these demonstratives can be used alone as a pronoun. This is further, distributional, evidence that the personal pronouns are actually demonstratives.
- (8)
a.
b.
c. den/han [adv der borte]
that/he there over
d. [Denna/Den/Han] var fin!
‘This/This/He is lovely!’
Thus, semantically, morphologically and distributionally, pronouns can be considered definite determiners in SHLN, and semantically it is clear that they are demonstratives since they may be used deictically to identify a referent in the immediate physical context.
3. THEORETICAL BACKGROUND
The account presented here is set within the Lexical Function Grammar (LFG) framework. This is a lexically-driven approach to grammar, which allows a strict separation of morphology, syntax and semantics. Here I give a brief introduction to the relevant parts of LFG.
3.1 Lexical-Functional Grammar
LFG utilises two distinct syntactic structures – c(onstituent)-structures, or phrase structure trees, and f(unctional)-structures, which are attribute–value matrices – and constraints may be placed separately on either of these structures. Importantly for our analysis, ‘individual c-structure elements, including words, may specify complex f-structures’, that is to say that a single word may play more than one role functionally or may satisfy more than one part of the corresponding f-structure (Nordlinger & Bresnan, to appear:4, my emphasis). A simple illustration of this is given in (9) below, from Nordlinger & Bresnan (to appear:3).
- (9)
a. Mary sees Sue, c-structure
b. Mary sees Sue, f-structure
Here, the verb sees comprises information about its subject and object, as well as the tense of the clause, corresponding to two different parts of the f-structure. This can also be seen in the lexical specification for sees, as given in (10) (from Nordlinger & Bresnan, to appear:4).
(10)
In a similar way, the word mannen contains information about definiteness, number, person and gender.
(11)
LFG assumes the Lexical Integrity Principle (Bresnan & Mchombo Reference Bresnan and Mchombo1995), which states that the morphemic composition of words is subject to rules independently from the f-structure and c-structure, given in (12). Combined, this means that c-structures apply to single and entire words, while f-structural information need not, as illustrated in the examples in (13) and (14). ‘Words are constructed in the lexicon, while c-structure and f-structure form the core of the syntactic component’ (Nordlinger & Bresnan to appear:5).
(12) Lexical integrity
Morphologically complete words are leaves of the c-structure tree and each leaf corresponds to one and only one c-structure node.
Lexical items belong to a word class, and carry feature specifications. For example, in (10), sees carries the specification of present tense, and the specification that its subject must be third person singular (3sg) (indicated in the last two rows in (10)). Notice that this is a different way of viewing the agreement morphology to one where the -s is assumed to appear in the c-structure due to the presence of a 3sg subject. Recall that c-structure operations apply to entire words. Thus, in the c-structure, the word sees will need to unify with a 3sg subject, and the 3sg subject will need this particular form of the verb in order to unify and produce a grammatical string. In the f-structure, the specifications which will be carried by the predicate see and the suffix -s will be separate.
Pronouns typically carry specifications for person, number and gender, while determiners carry specifications for the type of determiner they are (e.g. demonstrative). A definite noun like mannen carries feature specifications including humanness, gender, number and definiteness. These specifications are unified in the f-structure and c-structure.
The representation in (11) shows simplified lexical entries for mann and the definite suffix -en. The pred value of mann is ‘man’, and this semantic information and the syntactic features of third person, singular number and masculine gender will be sent to the parent node in a c-structure, as indicated by the ↑ in the lexical specifications. The -en will do likewise, but it will contribute the feature of definiteness rather than a predicate value. (It will be shown in the next section that the suffix is actually ‘specific’ rather than ‘definite’, but the principle is the same.)
These features can be seen in the c-structure in (13). The symbols (↑ = ↓) can be translated roughly as ‘my features also belong to my mother’, thus ensuring the lexical features percolate appropriately throughout the tree.
(13)
In the f-structure, the feature of definiteness is represented by both the demonstrative and the suffix, as shown in (14).
(14)
With regard to the nominal phrase structure, LFG assumes a DP structure when there is a D head present, but it is also possible for a determiner to not head its own phrase, and to be a specifier within the NP, as in (13). It will be suggested later that pronoun demonstratives in Danish and Swedish are Ds which take an NP object, as opposed to the Norwegian pronoun demonstratives which are specifiers within the NP.
3.2 Lexical features of han mannen in SHLN
My analysis of the han mannen construction in SHLN is simply that third person pronouns may be demonstratives, which therefore require morphological agreement when used with a noun. The lexical entries for han and ho therefore need to include, in addition to the usual person, number and gender features, the specification that they are demonstratives in the han mannen construction, as shown in (15). Also, ho and han must have human referents when they are demonstratives, while den need not and so is unspecified for humanness.
(15)
Essentially, if we assume that pronouns are categorially demonstratives, then the definiteness agreement on the noun will occur in the same way as with all definite determiners in SHLN, and the presence of the definite noun is accounted for trivially.
4. OSLO NORWEGIAN
While the account of the form of the noun in the han mannen construction given above is trivial, in Standard Bokmål/Oslo Norwegian, there is no morphological basis for calling pronouns demonstratives. In Oslo Norwegian, the definite form of the noun is optional with other demonstratives, but obligatory with the pronoun demonstrative. Thus, in Standard Bokmål, den mann is possible, as is den mannen, but *han mann is not. In addition, the form *ham mannen, with the objective form of the pronoun, is also ungrammatical.
This raises two questions:
(i) Is han really a demonstrative in this construction in Standard Bokmål?
(ii) Why is ham mannen not allowed?
The second question is undoubtedly related to the issue of default case marking generally, but I will not address this question further here. (Notice that this problem doesn't arise in SHLN, since the third person singular pronouns do not make any case distinctions.) The answer to the first question is ‘yes’, for several reasons.
Firstly, the primary use of the pronoun in this construction even in Oslo Norwegian is to ‘point out’ a referent, either in the immediate physical context or, more typically, ‘identifiable’ in the immediate discourse context, as illustrated in the constructed example in (16), an example from the NoTa corpus of spoken Oslo Norwegian in (17), which includes extensive contextualising in order to illustrate the discoursal usage, an online example in (18) and other examples below. This categorisation has been offered for Norwegian and other Scandinavian languages, e.g. the Norwegian reference grammar (Faarlund et al. Reference Faarlund, Lie and Vannebo1997), and others who have worked on this construction, e.g. Johannessen (Reference Johannessen, Solstad, Grønn and Haug2006) for Norwegian, and Josefsson (Reference Josefsson2006) for Swedish. Other demonstratives will of course also require the definite form of the noun in each of these contexts here.
- (16)
a.
‘My god! Look at that woman!’
b.
‘My god! Look at that woman!’
(17)
“Saw” # har du sett den? . . . # det var sånn at de ble satt inn på et rom ikke sant # to stykker # og så # . . . og så og så våkner det de to opp da og de kjenner ikke hverandre # og så våkner de opp med sånn lenker til foten, og de skjønner jo ingenting ikke sant så ser dem på hverandre og blir jo drit-redde # skjønner ingenting, og så # og så # viser det seg for å være # at han ene # han er lege # og han andre er ikke noe sånn spesielt #, mm # og så # men han legen han hadde noen som pasienter da # som hadde blitt # utsatt for han morderen # og morderen er sånn # han dreper ingen men han # vil at de skal drepe seg selv
(Søndre Nordstrand, Oslo; male, age 17 years; # represents a new intonational phrase)‘“Saw”, have you seen it? . . . There was like these two guys who get put in a room y'know, and like, and so they wake up, these guys, and they don't know each other, and so they wake up with like chains around their ankles, and they have no idea what's going on y'know, and like they look at each other and totally shit themselves, they've got no idea, and like, and so, it turns out, that the first guy, he's a doctor, and the other guy isn't anyone special, mm, and like, but this doctor he had this patient, who'd been attacked by this killer, and the killer's like, he doesn't kill anyone but he, wants people to kill themselves’
(18) [In a forum discussing mental drivers, and why one should always carry one's camera with one, just in case . . .]
Men toppen av kransekaka var vel hun dama som parkerte i VENSTRE FELT i TÅSENTUNNELEN . . .
‘But the icing on the cake is really that woman who parked in the LEFT LANE in the TÅSEN TUNNEL’
(http://www.bilforumet.no/medlemmers-egne-bilder/88786-bilder-hverdag-rare-ting-i-trafikken.html)
Secondly, while the demonstrative den is homophonous with the definite article den, Fretheim & Amfo (Reference Fretheim and Amfo2005:106) showed that den and han have the same prosody when used demonstratively, and that this is different to den used as a definite article. This is also different to the preproprial article, which is typically very unstressed, as opposed to the pronoun demonstrative, which may never be unstressed.
Thus, it is a fact that both han and den are demonstratives in Standard Bokmål as in SHLN. This leaves the question then of what the difference is between han and den, i.e. why is den mann allowed in Oslo Norwegian, but *han mann is not?
4.1 The specific difference between han and den in Norwegian
Johannessen (Reference Johannessen, Solstad, Grønn and Haug2006, Reference Johannessen and Arboe2007, Reference Johannessen, Johannessen and Hagen2008) develops a case for showing that the pronoun demonstrative has a psychological deictic function in Oslo Norwegian and other dialects, which contrasts with the (physical) spatial deictic function of the demontratives den, denne, etc.
However, there is more that can be said here, in particular, there is a crucial difference between the pronouns han/ho and den in that han/ho can only ever have specific reference (explained in more detail below), while den can, but needn't. A nice contrast illustrating the difference in specificity between han/ho and den is given by Lødrup (Reference Lødrup1982:55 Note 1) in a short note in a short article, where he mentions that:
Legg merke til at hun/han som bestemmere ikke er synonyme med den. En nominalfrase med den som bestemmer kan ha spesifikk eller ikke-spesifikk referanse, mens en med hun/han bare kan ha spesifikk referanse. . . . Han/hun er på denne måten som den der(re).
[Note that hun/han as determiners are not synonyms with den. A noun phrase with den as a determiner can have specific or non-specific reference, while one with hun/han can only have specific reference. [Examples translated below as (19a) and (19b).] Han/hun is like den der(re) in this way.]
- (19)
a.
‘I'd love to meet the girl (i.e. any girl) who can do that.’
b.
‘I'd love to see that girl (that you're talking about) who can do that.’
The two examples in (19) show the difference in the presupposition of existence of a particular referent, or the speaker having a particular referent in mind. Lødrup (Reference Lødrup1982) refers to this as specificity, as do e.g. Lyons (Reference Lyons1999:165), von Heusinger (Reference von Heusinger2002), Riley (Reference Riley2007:833), Strahan (Reference Strahan2007), Farkas (Reference Farkas2002:239) and Johannessen (Reference Johannessen, Johannessen and Hagen2008:9) among others.
4.1.1 Interlude: specificity and definiteness
Note that we are interested here in specificity as it is associated with definiteness, not indefiniteness, although the underlying assumption of a presupposition of existence (and possibly identifiability) is the same. As noted in Lyons (Reference Lyons1999:165, 168f.), both definites and indefinites may be specific or non-specific. Anderssen (Reference Anderssen2007:256) also makes this point, noting that in Norwegian a full analysis of definiteness requires recourse to specificity, although she is concerned with child acquisition of the definite article and not demonstratives. Her analysis of the features of pronouns, determiners and the definite suffix in the Tromsø dialect of Norwegian is very similar to the analysis which will be proposed here, namely that the -en suffix is primarily specific, although it may also be unique, if there is no prenominal definite article. (19a) above is a nice example of a non-specific definite noun phrase, as is (25a) below. (19b) with the pronoun demonstrative is also formally definite, but only has a specific interpretation.
Following Lyons (Reference Lyons1999:278), I assume that definiteness is a formal syntactic or morphological feature which is the realisation of underlying (semantic or pragmatic) identifiability and/or uniqueness. Identifiability itself can also be understood in discourse or cognitive terms, being related to accessibility (e.g. Ariel Reference Ariel and Brown2006), information structure (e.g. Polanyi, van den Berg & Ahn Reference Polanyi, den Berg and Ahn2003), and the ability of the speaker to estimate the processing effort required by the hearer to retrieve or interpret the reference of the noun phrase (Ariel Reference Ariel and Brown2006:15). Identifiability is generally understood to be hearer- or discourse-oriented (Vangsnes Reference Vangsnes1999, Anderssen Reference Anderssen2007).
Lyons’ (Reference Lyons1999:9) examples of hypothetical situations show the difference between uniqueness and identifiability clearly – if the competition is not yet over in (20a), and there is not yet a person accompanying the speaker in (20b), then the referents cannot be identifiable. However, they are unique, in that there will be a single winner of the competition and a single co-traveller.
- (20)
a. The winner of this competition will get a week in the Bahamas for two.
b. The person who comes with me will not regret it.
Uniqueness can also be thought of as the ability to individuate the referent within the discourse world (Ward & Birner Reference Ward and Birner1995); it needn't imply that the actual referent is identifiable.
With respect to demonstratives, Lyons (Reference Lyons1999:279) suggests that they are not themselves lexically specified as [+def], rather that their feature [+dem] implies that their referent is unique and thus identifiable. In those languages where identifiability is realised as definiteness, demonstratives will necessarily be interpreted as definite. This is clearly the case in Norwegian. This non-direct link between demonstratives and definiteness will be useful in understanding why in literary Oslo Norwegian the demonstrative den may occur with either a definite or an indefinite noun, and is related to the question of why a noun phrase with a demonstrative may have either a specific or non-specific interpretation.
A specific referent is one that the speaker has in mind. Specificity is related to Fodor & Sag's (Reference Fodor and Sag1982) referentiality. It is associated, and sometimes conflated, with identifiability (Farkas Reference Farkas, Nash and Tsoulas1994), wide-scope quantification (Fodor & Sag Reference Fodor and Sag1982), ‘having a particular referent in mind’ (Vangsnes Reference Vangsnes1999:44), the phrase a certain x (Vangsnes Reference Vangsnes1999), noteworthiness (Ionin Reference Ionin2006), individual reference (Julien Reference Julien2003:240), discourse anaphoricity (Enç Reference Enç1991) and referential anchoring within a discourse (von Heusinger Reference von Heusinger2002).
From this, we can say that there are semantic features of [specificity], [identifiability] and [uniqueness], all of which may underly morphosyntactic definiteness. These features may be lexically specified, and in fact, we see that [specificity] is the most important feature for our analysis. Crucially, a specification of any of these features will be realised morphosyntactically in Norwegian as definite; on the other hand, being assigned to the category of demonstratives will imply, but not necessarily entail, definiteness.
Anticipating the final account, this means that there will be unification restrictions or requirements on whether another specific or definite element is permitted/needed within the same noun phrase, or whether the noun phrase as a whole is definite.
4.1.2 Examples contrasting den and han/ho
There are lots of examples to be found on the web, for both Oslo Bokmål and SHLN, where the demonstrative den is present, and where the pronoun demonstrative would not be possible, due to the lack of a specific referent. In (21), the referent of the noun phrase den jento ‘the girl.def’ is not presupposed to exist in the real world, its reference is only the idea of a ‘girl who I trust most’. This noun phrase has only a de dicto reading, while ho jento ‘she girl.def’, with the pronominal demonstrative, has a de re reading, and therefore cannot be used here.
- (21)
a. Å Siren e den jento på denna planeten eg stole mest på!!
‘And Siren (a girl's name) is the girl on this planet I trust most in.’
b. *Å Siren e ho jento på denna planeten eg stole mest på!!
Notice that this account also predicts that ho jento will not be felicitous with a presentational reading, which is exactly the case with (22a), although a non-existential reading is possible, as in (22b) (thanks to Höskuldur þráinsson p.c. for discussing a similar issue with me). Presentational (there-existential) constructions have been shown to disallow specific referents rather than just definite ones (Ward & Birner Reference Ward and Birner1995).
- (22)
a.
and so was there yes she girl.def in car.def my
Intended: ‘And so there was a girl in my car.’
b.
and so was there yes she girl.def in car.def my
‘And so this girl (e.g. that we were just talking about) was like in my car!’
In a similar fashion, den mannen in (23) does not have a real-world referent, it is merely an ideal to which Ove strives to please Sara. Therefore, han mannen is not a possible option here.
- (23)
a. Forholdet er, frå Ove sitt synspunkt, ganske perfekt, og han forandrar seg ein god del for å bli meir lik den mannen han trur Sara vil ha.
‘The relationship is, from Ove's perspective, quite perfect, and he has changed a good deal to become more like the man he thinks Sara wants.’
b. *Forholdet er, frå Ove sitt synspunkt, ganske perfekt, og han forandrar seg ein god del for å bli meir lik han mannen han trur Sara vil ha.
Finally, notice that in SHLN, the use of den with a definite noun alone (assuming no special prosody such as that mentioned by Vangsnes (Reference Vangsnes1999:77ff.) in utterances such as ‘he had ˈDA ˈSMILET/that smile on his face’) is pragmatically ‘incomplete’, and requires some extra information such as den der ‘that there’, as shown in (24a–c) (the hash # in (24a) indicates pragmatic incompleteness). The use of the pronoun demonstrative does not require any extra information, although locational devices such as der borte ‘over there’ may also be used, as in (24e). Typically, both den and the pronoun demonstrative are accompanied by a gesture, and online and spoken corpus (e.g. TAUS) searches reveal that extra information is often included with both den and han/ho, in the form of a relative clause.
- (24)
a.
look at dem woman.def you
b. sjå på den der damo du
c.
look at dem woman.def over there you
d. sjå på ho damo du
e. sjå på ho damo der borte du
This again supports the idea that den alone does not select a specific referent, or at least not to the same extent as the pronoun determiner does. The question of whether den can be said to be specific at all will be addressed in section 4.2.
Notice also that the pronoun demonstrative is found in informal texts. Native speakers of Norwegian in general report that it is only felicitous in speech, and is definitely not allowed in (formal) writing. This is supported by the not infrequent appearance of the pronoun demonstrative in the TAUS corpus of spoken Oslo Norwegian, and by the range of discourse uses identified by Strahan (Reference Strahan2007).
4.2 Den mann vs. den mannen in Oslo Norwegian
We have now established that han mannen must have a specific referent, while den mannen does not seem to. In Oslo Norwegian there is also a contrast between den mannen with the demonstrative and definite suffix, and den mann with no suffix. To understand better the difference between these two forms, let us consider some further examples. Julien (Reference Julien2003:240, ex. (21)) gives a nice contrasting pair of sentences, cited here in (25) (Julien glosses den as ‘definite’).
- (25)
a. Den kvite mann-(en) har undertrykt andre kulturar.
‘The white man has oppressed other cultures.‘
b. Den kvite mann-*(en) åt ein is.
‘The white man ate an ice-cream.’
The definite suffix is optional in (25a), where the interpretation of den kvite mann(en) is most likely generic. In (25b), which allows only a specific referent, the suffix is obligatory. Julien (Reference Julien2003:240) states that the suffixed article appears to be related to ‘individual reference’, what is being termed here ‘specificity’.
Helge Dvyik (p.c./post to LFG mailing list) also makes this point. In (26a), the den mann form is non-specific, while the presence of the suffix in (26b) allows either a specific or non-specific reading, as the glosses show.Footnote 1 (Recall from section 2 that (26a) is ungrammatical in SHLN.)
- (26)
a. den mann som sier slikt . . .
‘the (kind of) man who (would) say such things’
b. den mannen som sier slikt . . .
‘the (kind of) man who (would) say such things’
‘the man who (is) say(ing) such things’
Notice that den in (26a, b) is a demonstrative, even though it is translated into English with the (compare this with han mannen som snakker no, which could be translated as ‘the man talking now’). Now, since the only overt difference between a and b here is in the presence or absence of the suffix, and the semantic difference is between a non-specific versus an optionally specific referent, the suffix itself must have some (potential) specificity associated with it. It is important to bear in mind that the suffix is not purely a specificity agreement feature, since it can be used when the NP is non-specific, as seen in e.g. Lødrup's example (19), and in (21) and (23).
4.2.1 The specificity of den in natural discourse
While specific den is possible, it is useful to examine discourse-based evidence from natural language corpora in this respect. Johannessen has done exactly this for den phrases in Oslo Norwegian. She has found that the use of den mannen type phrases in the TAUS corpus of spoken Oslo Norwegian is consistently non-specific: ‘Hver av nominalfrasene som er innledet med den, viser til en type person, ikke en bestemt person’ [Each of the noun phrases which are introduced with den refers to a type of person, not a particular person] (Johannessen Reference Johannessen, Johannessen and Hagen2008:8). This is interesting because while den need not be non-specific, in natural conversation this would appear to be the default usage.
In addition, searches in the NoTa spoken corpus of Oslo Norwegian for den mann, den dame, den gutt, den lege and den kvinne return zero matches. This indicates that this form is not usual in conversational Oslo Norwegian. These two findings also indicate that specific reference is not achieved through the den demonstrative in Norwegian.
4.2.2 The specificity of den online
Judging from the results of a search for this string on Google, the construction den mannen most often has a non-specific referent, as Johannessen found for spoken Oslo Norwegian. The instances where den mannen has a specific referent are often found in headlines, e.g. (27a–c). Some specific uses of den mannen were found in running text, but not many; cf. (27d–e) (even (27c) is questionable as to the referential status of den mannen).
- (27)
a. «Gi den mannen en xylofon!»
(http://pub.tv2.no/nettavisen/innenriks/ioslo/article1611221.ece)
‘Give the man a xylophone!’
b. «Hva går den mannen på?» Bjarne Betong-Hanssen imponerer meg.
(http://voxpopulinor.blogspot.com/2006/09/hva-gr-den-mannen-p.html)
‘“What is that man on?” Bjarne Betong-Hanssen impresses me.’
c. «Den mannen gjør meg syk» Trusler. Svertekampanjer. Frysebokstaktikk. Det er like mange metoder som mobbere.
(http://www.orapp.no/_den_manne/)
‘“That man makes me sick.” Threats. Smear campaigns. Freeze-tactics. There are as many methods as there are bullies.’
d. Djupedal må være den eneste statsråd i historiene som greier å fylle sin tabbekvote i løpet av 3 måneder! Jammen skal det bli spennende å se hva den mannen kan få vridd ut av seg i løpet av neste 6 mnd.
(http://www.vgb.no/129/perma/16122)
‘Djupedal must be the only minister in history who has managed to fill his quota of mistakes in the course of 3 months! But god it's going to be exciting to see what this man can wring out of himself in the course of the next 6 months.’
e. Dei som var om seg, fekk helst på den læraren frå Griffith University som skal komme til Volda og vere sensor på eksamen
(www.hivolda.no/jpv/reisebrev2003.htm)
‘Those who were there got to meet the teacher from Griffith University who's going to come to Volda and be a supervisor for our exam’
4.2.3 Genre of den mann
A comment on the genre or formality of the den mann construction is in order here. Online searches for den mann type expressions return hits from sources like the Bible (www.biblegateway.com/passage/?book_id=5&chapter=17&version=5), famous quotations (www.ordtak.no/index.php?emne=Menn) and poetry (dikt.org/Håvamål_01_I). In other words, this construction does not appear to be part of the conversational grammar of Oslo Norwegian. In support of this view, note that den mann type phrases are categorised as ‘dpDanish’ (Borthen, Brøseth & Fretheim Reference Borthen, Brøseth and Fretheim2007), contrasting with plain DPs, in the Norwegian NP-form project being conducted out of Trondheim.
4.2.4 The -en suffix
It is standard now to assume that double definiteness in Mainland Scandinavian languages is compositional, where both the prenominal article and the suffix contribute separate semantic aspects of definiteness. Under this assumption, the suffix is generally considered to contribute semantic specificity (e.g. Vangsnes Reference Vangsnes1999, Julien Reference Julien2005, Anderssen Reference Anderssen2007:254f.). This can be seen in examples like the following (where # represents pragmatic incongruity):
- (28)
a. Æ spiste ikke [den minste bit] av kaka. # Den spiste han Derek.
‘I didn't even eat a small slice of the cake. It was eaten by Derek.’
b.
‘I didn't eat the smallest slice of the cake. It was eaten by Derek.’
(28a) is infelicitous since there is no specific referent indicated by the noun phrase. This contrasts with (28b), which does have a specific referent, due to the presence of the suffix, and the noun phrase can thus be referred to with a pronoun.
A similar pattern occurs in Swedish. The examples in (29) and (30) from Delsing (Reference Delsing1993:128f.) illustrate this. (29a) is ungrammatical because the idea of ‘the most beautiful princess’ here does not mean a particular individual, rather it refers to the concept of the princess who is the most beautiful one (of that kind). (Compare this with (19)–(23) above for Norwegian.) In a similar way, if there is ‘no reason to doubt’ something, then it follows that the nonexistent reason cannot be a specific one, since it does not exist. Therefore the suffix indicating a specific referent cannot be used.
(29)
(30)
Anderssen (Reference Delsing2007:258) also notes that the suffix may spell out both specificity and uniqueness when it occurs without the prenominal article.
Therefore, the suffixed article in Norwegian (and Swedish) is better considered specific rather than merely definite, although other aspects of definiteness, such as uniqueness and identifiability may also be associated with it.
4.2.5 Summary of Oslo Norwegian data
To summarise so far:
(i) The pronoun demonstrative cannot be used if the referent is non-specific. It is fairly common in natural, casual conversation, but is never found in formal texts.
(ii) Den with a suffixed noun may be used when the referent is either specific or non-specific, although in natural conversation and online den is typically non-specific.
(iii) Den with a non-suffixed noun can only ever be non-specific. This form is also not found in natural, casual conversation, and appears only in more formal contexts.
(iv) The -en suffix is primarily (although not exclusively) a specificity suffix, rather than being a general definite suffix.
4.3 Lexical features of han mannen in Norwegian
My analysis of the han mannen construction in Oslo Norwegian is therefore parallel to the explanation given for SHNL, except that here we need to recognise that third person pronouns are specific demonstratives, which therefore require specific morphological agreement when used with a noun. The lexical entries for han and ho therefore look something like in (31), which includes the specification that they are demonstratives [(↑dem) = +] and that they are specific [(↑spec) = +].
(31)
Given the similar semantics between the han mannen construction in SHLN and Oslo Norwegian, the lexical features of the SHLN pronoun demonstratives will be the same as those given in (31). Therefore, my conclusion is that pronoun demonstratives are specific demonstratives, and the presence of the specific noun is accounted for trivially. The c-structure for the Norwegian han mannen is given in (32), modified from the version with den given in (13) earlier.
(32) c-structure of Norwegian han mannen
5. THE OTHER SCANDINAVIAN LANGUAGES
In the following sections, I very briefly present data from the other Scandinavian languages involving this construction. The question of whether the language has double definiteness or not is important, to determine whether the pronoun demonstrative fits the definiteness agreement patterns of the other demonstratives and definite articles. In addition, the case form of the pronoun is relevant. Where only one case form is used (as in Danish and Norwegian), it is noted whether this is the standard non-local-subject or non-coargument object form (i.e. the default form), or whether the case must change, as it does in Icelandic.
5.1 Danish
Here I look at the han mannen construction in Standard Danish and the dialect of Vestjysk Danish.
5.1.1 Standard Danish
The essential Danish data for this construction is identical to Norwegian, with two crucial differences. Firstly, the pronoun used is in the object form, not the subject form, as in (33a, b). Secondly, Danish does not have double definiteness, yet still uses the definite noun with the pronoun demonstrative, as in (33c).
(33)
Neither the suffix nor the demonstrative den can be described as being [spec = +], since they do not co-occur and the distinction found in Norwegian between den mann (must have a non-specific referent) and den mannen (may have a specific or non-specific referent) does not exist in Danish. However, like in Norwegian, the referent of ham manden phrases must be specific. Thus, just as in Norwegian, (34) is ungrammatical in Danish, and (33a) cannot refer to a type of man or to a general man.
(34) *Og Karen er hende damen på den her planet jeg stoler mest på!!
Intended: ‘And Karen is the woman on this planet I trust most in.’
Interestingly, it is difficult to create a minimal pair in Danish like that in (19) above for Norwegian. The ‘any girl’ reference is straightforward (35a), but the corresponding sentence with the specific hende pigen is highly dispreferred (35b). This may be due to the idiomatic nature of the phrase itself, that it always refers to some unknown and thus non-specific person. In (35c) there is clearly reference to a girl who is present in either the immediate physical or discourse context, and the use of hende pigen is felicitous.
- (35)
a.
‘I'd love to meet the girl (i.e. any girl) who can do that.’
b.
‘I'd love to see that girl (that you're talking about) who can do that.’
c.
‘I'd love to see that girl (that you're talking about) who can do that.’
Finally, Danish clearly has a definite noun phrase, rather than just a definite noun, in combination with the pronoun demonstrative, as shown in (36a, b).
- (36)
a. Det er ham den store mand med sækken på ryggen.
(http://www.123hjemmeside.dk/qiterlia/2630743)
‘It's that big guy with the bag on his back.’
b. Den lille kække Fiat Seicento er kommet på markedet herhjemme i en frisk specialmodel kaldet Brush, ikke Bush, som ham den store mand der ovre i USA.
(http://www.dba.dk/asp/sektion/artikler/detail.asp?ArtikelId=13025)
‘The little sprightly Fiat Seicento has come on the market here at home in a fresh special model called Brush, not Bush like the big man over there in the USA.’
In Danish, unlike in Norwegian, when an adjective is present in the ham manden construction, then the definiteness is prenominal rather than suffixed to the noun. This exactly parallels the situation for definiteness in ordinary definite noun phrases (as discussed by e.g. Hankamer & Mikkelsen Reference Hankamer and Mikkelsen2002). As shown in (37a), double definiteness is not permitted in Danish – either the noun is definite or the prenominal determiner is, but never both. When an adjective modifies the noun, then the only option is to use the prenominal determiner (compare b and c).
- (37)
a. *den mand-en ‘the man-def’
b. den store mand ‘the big man’
c. *store manden ‘big man-def’
Bjarne Ørsnes (p.c.) says that in Danish, this lack of double definiteness in noun phrases can be captured by using ‘instantiated symbols’, as was employed in the ParGram (Parallel Grammars) project. The lexical entry for den carries the specification [def = +_], where the underscore indicates that this specification cannot unify with anything else as the PRED value, i.e. it cannot co-occur with another [def = +] expression. This correctly rules out *den damen, as both den and damen have the specification [def = +_]. This captures the intuition that definiteness in Danish can only be supplied from one source. Notice that this is a constraint on the surface definiteness, not on the underlying uniqueness or specificity, neither of which it is necessary to appeal to in order to correctly describe Danish definiteness.
If we then assume that the Danish pronoun demonstrative is a head D that takes a definite NP as its object, we can account for this data. The lexical entry for ham is given in (38). Notice though that the definiteness of the NP object is motivated still by the specificity feature of the pronoun demonstrative, just as it is in Sunnhordlandsk and Oslo Norwegian.
(38)
The appearance in the c-structure of the definiteness in the noun phrases ham manden and ham den store mand is governed by the instantiated symbols as described above, while the f-structures are the same (except for the absence/presence of the adjective stor ‘big’). Notice that the definiteness in (39a) is realised in the pronoun demonstrative and the suffix, while in (39b) it is realised in the pronoun demonstrative, definite article and definite form of the adjective. The c-structure, as given in (39c), illustrates how the object of the pronoun demonstrative is definite, which is realised either with a definite noun manden, or with the prenominal article den when there is an adjective present.
- (39)
a. F-structure for ham manden
b. F-structure for ham den store mand
c. C-structure of Danish ham manden/ham den store mand
This is different from the Norwegian pronoun demonstrative, which occurs as a specifier within the NP, as shown in (32) above. This difference is necessary, since *han den store mannen is ungrammatical in Norwegian.
5.1.2 Vestjysk Danish
The han mannen construction occurs in Vestjysk Danish (VJD) according to recent investigations by the author. VJD famously does not have a suffixed article, yet the pronoun demonstrative can be analysed in the same way as Standard Danish, in that it requires a definite object. Since VJD does not have a definite suffix, it is the prenominal article that must appear with the pronoun demonstrative. As in Standard Danish, the pronoun demonstrative in VJD is ham, the object form.
- (40)
a.
b.
c. hin den bette
her the little.one
d. ham Johannes har ringet
him Johannes has called
The basis of the definiteness agreement in Norwegian and Danish is then that the pronoun demonstratives are specific, as well as being underlyingly definite. The Norwegian pronoun demonstratives follow the usual definiteness agreement rules for definite determiners in Norwegian and thus always appear with a definite noun, but never with another definite article. In Danish the pronoun demonstratives take a definite NP object, where the requirement for a definite object is motivated by the specificity of the pronoun demonstrative. This means that the pronoun demonstrative may occur with a prenominal definite article, either due to the presence of an adjective as in Standard Danish, or because there is no suffixed article as in Vestjysk Danish.
5.2 Swedish
Both Standard Swedish and ‘dialectal’, non-standard Swedish are looked at here. This construction is not found in Standard Swedish, but is acceptable in at least some colloquial/dialectal varieties (Josefsson Reference Josefsson, Riemsdijk and Hellan1994).
5.2.1 Standard Swedish
Standard Swedish does not allow the han mannen construction at all, and every Swede I have ever consulted about this construction (about a dozen) has rejected all of my initial attempts at contextualising it. (After 5 to 10 minutes of persistent questioning, most Swedes relent and concede that it is maybe possible or that it really is okay, but that it is probably slang. I conclude that it is highly dispreferred in Standard Swedish.) A search on Google for “hon kvinnan” ‘she woman.def’ reveals just three instances of this construction in the first eight pages of hits, e.g. (41a); the other hits typically are subject–verb inversion constructions, where hon and kvinnan do not belong to the same constituent, e.g. (41b), or appositional constructions, e.g. (41c). No relevant examples of hon kvinna ‘she woman.indef’ were found, showing that, to the extent that this construction is possible, the noun must be definite, as is the case in every Scandinavian language. Swedish is a double-definiteness language, although, like Norwegian, non-specific definites do not have the suffix (see (29) and (30) above).
(41)
5.2.2 Non-standard Swedish
Just why the han mannen construction is ungrammatical in Standard Swedish is unclear, since it is grammatical in some Swedish dialects. Where it does occur, the nominative form of the pronoun is used, whether the noun phrase is a subject (42a) or object (42b), and the noun must be definite.
(42)
Furthermore, the han mannen construction is possible with a doubly definite noun phrase, as shown in (43). This indicates that the non-standard Swedish pronoun demonstrative is syntactically more like the Danish than the Norwegian one, heading its own DP, and taking a definite NP object. As with the Norwegian and Danish examples above, there is no intonation break between the pronoun demonstrative and the rest of the noun phrase (Josefsson Reference Josefsson2006:1357).
(43)
Interestingly, Josefsson (Reference Josefsson2006) shows that the gender of the pronoun demonstrative (along with pronouns and predicative adjectives in general) agrees with the sex of the referent, rather than with the gender of the noun. As shown in (44), the neuter nouns biträdet ‘the clerk’ and statsrådet ‘the secretary of state’ may occur with either hon ‘she’ or han ‘he’; notice also the adjective agreement here.
- (44)
a.
‘She/the clerk was sick yesterday.’
b.
yesterday
This pattern has been identified in Norwegian and Danish pronoun demonstratives (Johannessen Reference Johannessen and Arboe2007), e.g. han spøkelset ‘he ghost.n.def’, especially used in discussion of Harry Potter ghosts, but it is received poorly by native speakers when I have attempted to confirm its acceptability.
This agreement with the underlying sex of the referent is interesting, as it supports the idea that it is the semantic features of the actual referent which are relevant to the syntax, including the specificity/existence of that referent, and their natural gender. It is possible that, if a plain demonstrative is used, then the predicative adjective may have to agree with the noun rather than the referent. That is, is Dette statsrådet er sjukt ‘this secretary.of.state.neut.sg is sick-neut.sg’ a possible sentence? (In Icelandic it is certainly the case that people are referred to with the ‘wrong’ pronoun – in a recent conversation, a teacher, who was female, was referred to consistently as hann ‘he’, because the gender of kennari ‘teacher’ is masculine, and it was the role of teacher rather than the person themselves who was relevant to the discussion. Thráinsson (Reference Thráinsson2005:517f.) mentions this mismatch between grammatical and natural gender as a source of variation among speakers of Icelandic, with natural gender generally preferred, especially when the referent is well-known.)
I am uncertain as to whether this is the same variety that permits the ‘normal’ examples in (44). If it is, then the double definiteness here suggests that non-standard Swedish pronoun demonstratives are like Danish, rather than Norwegian ones, in that the pronoun demonstrative takes a definite NP object, as shown in (45). It might also be necessary for the pronoun demonstrative in non-standard Swedish to carry a specification of natural sex rather than grammatical gender, in order to allow the examples in (44).
(45)
Thus, there are two patterns emerging for the pronoun demonstrative. In the first pattern, the pronoun demonstrative is a normal demonstrative which can only occur with a definite/specific noun, and never with a definite article, as in Norwegian. In the second, the pronoun demonstrative takes a definite NP object, which may be realised with just a suffixed noun (Danish, non-standard Swedish), just a prenominal definite article (Vestjysk Danish), or both a suffix and a prenominal article (non-standard Swedish). The presence of ‘double definiteness’ in Norwegian and Swedish results in different possible combinations with the pronoun demonstrative according to whether the pronoun demonstrative is part of the NP or the head of its own DP.
5.3 Icelandic
The han mannen construction does exist in Icelandic, although its use is very marginal. When it does occur, the definite form of the noun is required and the case of the pronoun is the appropriate case for the noun phrase as a whole, see (46a). The indefinite form is only allowed if there is an intonational break between the pronoun and the noun, see (46b). The hash sign indicates that most people reject this construction when it is presented to them, but some speakers (e.g. Ásgrímur Angantýsson and Jóhannes Gísli Jóhannsson, p.c.) believe that, given the right context, for example, including extra material such as þarna ‘there’, this construction is possible, see (47). The key point here is the fact that, with an indefinite noun, the pronoun cannot be considered part of the same constituent, rather it seems to be in apposition with the noun.
(46)
It appears to be more acceptable with additional material, as illustrated in (47). Notice that a is fine while b, with an indefinite noun, is completely ungrammatical.
(47)
The examples in (47) are reminiscent of the facts for the Norwegian demonstrative den, which often appears with extra material. I suspect it is related to the fact that the Icelandic definite suffix is not specific in the way it is in Norwegian, but this is an as yet untested hypothesis and requires further investigation.
The fact that the definite noun is required with the pronoun is noteworthy, since Icelandic is like Danish in that it generally only allows one definitely-marked element per noun phrase, with either demonstratives or definite articles (Thráinsson Reference Thráinsson2007:89). The exception to this is the demonstrative hinn/hin/hitt ‘the other’, which does require a definite noun.
(48)
However, having said that Icelandic does not have double definiteness in general, there are in fact two common constructions where double definiteness and specificity are involved, namely a gapping construction, and a preproprial use.
The gapping construction (Josefsson Reference Josefsson, Riemsdijk and Hellan1994, Thráinsson Reference Thráinsson2007:89) is far more common in Icelandic than the han mannen construction. It consists of a pronoun followed by a definite noun, (49a), or more commonly, a proper name, (49b). The ‘gapping’ label stems from the interpretation of this construction as the referent of the noun and at least one other referent, which would give a set of referents compatible with the person (and gender) of the pronoun, (49c, d). For this reason, (49e) is infelicitous, since the referents of þeir must be male/masculine.
- (49)
a.
‘We girls; (me and) the girls often go to the theatre.’
b.
‘Your and Jón's friend; Jón's (and your) friend often goes to the theatre.’
c.
‘(She/they (females)) and María often go to the theatre.’
d.
‘(He/they (includes at least one male)) and María often go to the theatre.’
e.
they. m María go.3 pl often in theatre
In addition to this gapping construction, pronouns in Icelandic are often used with proper names. This is very similar to the Norwegian and Swedish preproprial article (Delsing Reference Delsing1993), where an obligatory pronoun occurs before a name in some dialects. Recall that the preproprial article in Norwegian must be unstressed while the pronoun demonstrative must be stressed. In addition, the pronoun demonstrative occurs with a range of common nouns with human referents, while the preproprial article must occur before a name. The Icelandic examples in (50) below may be unstressed like the Norwegian prepropial article, but they are not obligatory.
Except for the starred b, the examples in (50) are completely acceptable. In a, the possessive pronoun þín clearly fills the role of making the noun phrase ‘your sister’ specific, much as the definite suffix does, or rather, more than the definite suffix alone does, since (50b) is highly dispreferred and probably completely ungrammatical. Notice also the dative case marking on the pronouns and head noun in (50a) and the accusative case of the pronoun and head noun in c, due to the quirky case marking required by the verbs finna ‘find/like’ and langa ‘want’. In addition, names (including kinship terms used in a name-like way) are common in this construction. I believe that this is because proper names of known individuals have specific reference, but there is not the space here to argue for this position.
- (50)
a.
‘That sister of yours likes black and white movies a lot.’
b.
c.
‘Your mum wants to go to to Norway.’
d.
‘Maria is a cool (female) person.’
Pronoun determiners need more investigation in Icelandic, to establish whether or not they function as demonstratives, and thus whether the han mannen construction exists in this language. Pronoun–noun combinations seem to have similarities with the han mannen construction in Norwegian and Danish, in that the referent must always be specific, and that it may be realised as either definite, as in (49a), or specific (e.g. by being possessed, as in (50a)). However, there are further restrictions on the semantics of the noun in this construction in Icelandic, in that not all common nouns may be used, but names of animates (including pets), kin terms and kinship-like terms such as ‘friend’ are nearly always felicitous.
5.4 Faroese
The han mannen construction does not exist in Faroese. As shown in (51a), the pronoun honum ‘him’ is not used in translations of Danish Se på ham manden ‘Look at that guy’, where a normal deictic demonstrative hasin/hasari is used instead, see (51b, c).
- (51)
a.
b.
‘Look at that man.’
c.
‘Look at that woman.’
As shown in (52a), a bare definite noun may be used deictically. A pronoun may not be used as a demonstrative with either a common noun, (52b), or a name, (52c). Topicalisation as in (52d) is the only possible combination of noun/name and pronoun in Faroese, but never with a pronoun demonstrative.
- (52)
a.
woman-def is interesting
‘The/that woman is interesting.’
b.
c.
d. Maria, hon er ein ahugaverd kona.
‘Maria, she's an interesting woman.’
Even the Icelandic pattern, where a pronoun is permissible in front of kin terms like ‘your sister’, is not allowed in Faroese, as shown in (53).
- (53)
a.
‘Your sister thinks that black and white films are interesting.’
b.
Finally, the Icelandic partitive pronoun demonstrative construction (the ‘gapping’ construction) is also not permitted in Faroese. (54a) is thus more like the English ‘we girls’ rather than the Icelandic við stelpurnar. Notice also that the indefinite form of gentur ‘girls’ is used, and not the definite form genturnar, see (54a, b).
- (54)
a.
‘We girls often go to the theatre.’
b.
c.
‘Jon's and your friend often goes to the theatre.’
The han mannen construction does not exist in Faroese. Pronouns in Faroese do show other typically Scandinavian traits, for example, they may be modified by relative clauses or PPs, but the reason behind the absence of this construction in Faroese must be left for further research.
6 SUMMARY AND CONCLUSIONS
Pronoun demonstratives in Norwegian, Danish and Icelandic, along with some Swedish dialects, are lexically specified as [specificity = +], and must always have a specific referent. There is syntactic variation within these languages as to whether the pronoun demonstrative appears as a specifier within the NP, as in Norwegian and Icelandic, or whether it heads its own DP and takes a definite NP object, as in Danish and non-standard Swedish.
Specificity, uniqueness and identifiability are assumed to all be underlying semantic features which are realised as morphosyntactic definiteness. Under the account given here, the definiteness of the noun (phrase) is motivated by a need to unify with the specificity feature of the pronoun demonstrative. Definiteness on the noun or in the noun phrase may be realised as the specific (definite) suffix, a definite article, a proper name, or having a possessor.
Although there are restrictions on what kinds of nouns may appear with the pronoun demonstrative, in those varieties that allow it at all, the noun must be human, anthropomorphised or a proper name. Each variety has further restrictions on the type of noun which may occur with the pronoun demonstrative, for example, in Icelandic, most examples with common nouns are infelicitous but all names including pets' names are allowed, whereas in Norwegian the referent must be human.
Standard Swedish and Faroese do not have this construction at all.
ACKNOWLEDGEMENTS
Financial support for the collection of this data comes from a NORMS Postdoctoral Fellowship and NORMS travel grants for 2008. I would like to heartily thank the following people for providing native speaker judements and opinions on the han mannen construction: Anne Vad and Karen Thagaard Hagedorn (Danish), Daniel Sävborg and Gustav Lindström (Stockholm Swedish), Elisabeth Skogseth (Sunnhordlandsk Norwegian), Ásgrímur Angantýsson, Heimir Freyr Viðarsson and Jóhannes Gísli Jónsson (Icelandic) and Yrsa Hauksdóttir (Faroese). In addition, I would like to warmly thank the following people for their helpful comments and feedback on the ideas presented here: Matthew Whelpton, Jóhannes Gísli Jónsson, Höskuldur þráinsson, Ásgrímur Angantýsson, Halldor Ármann Sigurðsson, Janne Bondi Johannessen, Bjarne Ørsnes and an anonymous reviewer. I am of course responsible for errors and misinterpretations that remain.
