1. Introduction
This squib deals with the word class of the Danish word rundt ‘around’, a word that has recently been argued to be a postposition in certain contexts (Nguyen Reference Nguyen2020a, Reference Nguyen2020b).Footnote 1 This is interesting given that it is generally accepted that Danish is a prepositional language. To the limited extent that postpositions (or postposition-like words) are mentioned at all in the research literature on Danish, it is recognized that only a few such words exist, purportedly with a limited distribution and occurring only in fixed expressions and the like (for an overview, see Nguyen Reference Nguyen, Christensen, Jørgensen and Wood2019, Reference Nguyen2020a). Nevertheless, in a number of recent publications, I have argued that not only are there words in Danish that can be used as postpositions, but also that they are far more numerous and have a wider distribution than has previously been acknowledged (Nguyen Reference Nguyen, Christensen, Jørgensen and Wood2019, Reference Nguyen2020a, Reference Nguyen2020b, Reference Nguyen, Andersen and Jensen2021a, Reference Nguyen, Goldshtein, Hansen and Hougaard2021b, Reference Nguyen2022b).
This squib focuses on one of these words, rundt ‘around’, as seen in the examples below, where rundt follows the relevant NP (rundt is italicized throughout):
Modern dictionaries uniformly classify rundt as an adverb, and to the extent that rundt is dealt with in grammatical descriptions, it is likewise usually classified as an adverb (see Mikkelsen Reference Mikkelsen1975: 101, 103, 467; Thomsen Reference Thomsen1998: 78).
Note, however, that rundt may also marginally precede an NP with which it forms a constituent. This construction is noted in Aa. Hansen (Reference Hansen1967 III: 308) and in the Ordbog over det danske Sprog [Dictionary of the Danish language] (ODS), which describes Danish in the period 1700-1950. According to Aa. Hansen and the ODS, in this case rundt is used as a preposition. A modern example is given below:
The construction ‘rundt + NP’ is marked, and many speakers do not accept it. Modern dictionaries do not mention the construction, and in Nguyen (Reference Nguyen2020a, Reference Nguyen2022a), it is judged as ungrammatical.Footnote 3 The construction is likely a relic from an earlier stage of Danish. Nevertheless, if Aa. Hansen and the ODS are correct that rundt is a preposition when preceding an NP, it is reasonable to consider whether it is a postposition when following an NP, given that the two constructions have similar semantics.
In principle, whether a word is a postposition or not does not hinge on whether it has a prepositional use. In Swedish and Norwegian, two neighbouring languages to Danish, the equivalents of rundt may also both precede and follow the NP. In the former, it is regarded as a preposition; in the latter, it is regarded as a preposition that follows its complement (Teleman et al. Reference Teleman, Hellberg and Andersson1999, Faarlund et al. Reference Faarlund, Lie and Vannebo1997) – in effect a postposition (or a postposition-like word). Thus, some scholars of Swedish and Norwegian seem to implicitly assume that a postposition must necessarily have a prepositional use; plausibly, this may be the case for some scholars of Danish as well. If this is true and if the ‘rundt + NP’ construction is largely unknown, it becomes clear why rundt in ‘NP + rundt’ is not regarded as a postposition in Danish. This reasoning is problematic: If it is true that a preposition does not necessarily have a postpositional use, it follows that a postposition does not necessarily have a prepositional use. (see also Nguyen Reference Nguyen, Christensen, Jørgensen and Wood2019, Reference Nguyen2020a, Reference Nguyen2020b, Reference Nguyen2022b).
This squib aims to provide arguments in favour of the view that rundt can be a postposition. In section 2, I discuss the functional, semantic, and syntactic properties of adpositions and why rundt may conceivably be an adposition.
2. The function, semantics, and syntax of adpositions
Both prepositions and postpositions belong to the adposition word class. Insofar as postpositions and postpositional phrases (PostPs) exist in Danish, they presumably have the same kind of functional, semantic, and syntactic properties as prepositions and prepositional phrases (PrePs) (Nguyen Reference Nguyen, Christensen, Jørgensen and Wood2019, Reference Nguyen2020a, Reference Nguyen2020b, Reference Nguyen, Andersen and Jensen2021a, Reference Nguyen, Goldshtein, Hansen and Hougaard2021b).
In terms of functional properties, we may conceptualize an adposition as a word that typically expresses a spatial relation: it introduces an NP as its complement and relates this NP to another element (see much more detailed analyses in e.g. Hagège Reference Hagège2010). For instance, the Danish preposition på ‘on’ in Vi bor på øen ‘We live on the island’ expresses a spatial relation by introducing the NP øen ‘the island’ as its complement and relating it to the verb bo ‘live’. An adposition may also relate its NP-complement to a noun, for instance the preposition til ‘to’ in Rejsen til øen tager to timer ‘The trip to the island takes two hours’. Parallel to this, rundt can be regarded as a postposition in examples such as Fyrtårnene kan findes øen rundt ‘The lighthouses can be found around the island’ and Turen øen rundt tager to timer ‘The trip around the island takes two hours’. Rundt expresses spatial relations by introducing spatial entities as NP-complements and relating them to a verb and a noun, respectively.
In terms of semantics, an adposition may impose certain semantic restrictions on its complement. The Danish preposition imellem ‘between’ imposes on its complement that it is semantically plural (Nguyen Reference Nguyen, Andersen and Jensen2021a). In the same vein, rundt can be seen as an adposition that imposes on its complement either i) that it expresses a spatial entity that has a contour or a periphery (Nguyen Reference Nguyen, Andersen and Jensen2021a), or ii) that it expresses a “spatial range over which a denoted action/state extends” (Kobayashi Reference Kobayashi2014: 21). In (2), the interpretation ‘circularly around the island’ presupposes that the island has a contour or a periphery, along which one can sail, whereas the interpretation ‘all over the island’ presupposes a spatial range over which the action of sailing extends.
In terms of syntax, adpositions are heads of adpositional phrases, and these phrases presumably have the same syntactic distribution, whether they are PrePs or PostPs. More specifically, if the string ‘NP + rundt’ is a PostP, it is expected to have the same distribution as a PreP (Nguyen Reference Nguyen2020a, Reference Nguyen2020b, Reference Nguyen, Goldshtein, Hansen and Hougaard2021b). For instance, it is predicted that the string ‘NP + rundt’ can i) be fronted, ii) function as an adjunct, see (4), and iii) coordinate with a PreP, see (5):
Thus, given that ‘NP + rundt' occurs in the same syntactic contexts as PrePs, it is reasonable to consider whether strings of ‘NP + rundt’ are PostPs. If constructions like (4) and (5) were ungrammatical, the postposition analysis would have been falsified. However, the fact that the postposition analysis is not falsified does not in and of itself prove the postposition analysis. The data in (4) and (5) are consistent with other analyses: Other phrases than adpositional phrases can i) be fronted, ii) function as adjuncts, and iii) coordinate with PrePs. Perhaps one could view the predictions and the data in the following way: The more the postposition analysis is tested, and the more the analysis withstands such tests, the more the analysis is indirectly supported. If the above-mentioned three syntactic facts do not falsify the postposition analysis of rundt but rather indirectly support it, and if rundt has the same sort of semantic and functional properties as adpositions, it is reasonable to treat rundt as a postposition.
The postposition analysis would indirectly receive more support if the data ruled out all other analyses but the postposition analysis itself. It is by exclusion that the postposition analysis is deemed more likely, because i) it is consistent with the data and ii) all other analyses have been falsified. This squib follows this line of thinking, and in the next section, I attempt to falsify some alternative analyses.
3. Falsifying alternative analyses
One possible alternative analysis of rundt is that it is an adjective. Offhand, the adjective analysis seems plausible because rundt may be used as an adjective, agreeing with the gender of a noun. In (6), the adjective rund agrees with bygning ‘building’, and in (7), it agrees with vindue ‘window’:
Alternatively, one might regard rundt as an adverb, reasoning that it belongs to none of the other word classes (the adverb word class being negatively defined; see discussion in e.g. Hengeveld Reference Hengeveld and Lier2023). As an adverb, rundt is not inflectable.
Rundt, then, might be an adjective or adverb. In the next two subsections, I consider two types of analyses: i) rundt as a postmodifying adjective or adverb inside an NP (see Diderichsen Reference Diderichsen1976), and ii) rundt as the adjective or adverb head of the construction ‘NP + rundt’, i.e. the construction is an AdjP or an AdvP.
3.1 Rundt as a postmodifying adjective or adverb inside an NP
Consider an analysis where rundt is not the head of ‘NP + rundt’ but rather an adjective or an adverb functioning as a kind of postmodifier within the NP. Under such an analysis, it is expected that rundt could be omitted but the preceding NP could not. This runs counter to what the data suggest:
By contrast, these facts are compatible with a postposition analysis of rundt: Prepositions in spatial adverbials can generally not be omitted (as in (8)), and complements of prepositions can sometimes be omitted (as in (9)).
Consider also the fact that predicates like bebo ‘inhabit’ select only NPs, not PrePs, for example, and that such predicates cannot select strings of ‘NP + rundt’:
The data in (10)-(12) indicate that ‘NP + rundt’ is not an NP, and that rundt is therefore neither an adjective nor an adverb inside an NP (though rundt could still be the head of an AdjP or an AdvP containing the NP, see section 3.2). Note that (12) is not ill-formed for semantic reasons: Example (10) is near-synonymous with (12), and (10) is well-formed – as opposed to (12).
In contrast to predicates such as bebo ‘inhabit’, other predicates select for instance PrePs but not NPs. This is the case for the predicates ligge ‘lie’ and bosætte sig ‘settle REFL’, meaning ‘settle (down)’:
In (13)-(14), it is shown that these predicates do not select NPs but PrePs, and in the naturally occurring examples in (15)-(16), it is shown that these predicates can select strings of ‘NP + rundt’. In other words, strings of ‘NP + rundt’ are not NPs. Subsequently, rundt is not a postmodifying element inside such (non-existent) NPs, regardless of whether it is an adjective or an adverb. The grammaticality of (15) and (16), on the other hand, would follow under the postposition analysis.
In sum, the data in this section falsify the idea that rundt is an adjective or an adverb inside an NP but are consistent with the idea that rundt is a postposition, selecting the preceding NP as its complement.
3.2 Rundt as an adjective or an adverb heading an AdjP or an AdvP
It is possible to conceptualize the adjective and the adverb analyses differently so that rundt is an adjective or an adverb heading strings of ‘NP + rundt’. Accordingly, such strings would be AdjPs or AdvPs. Considering the AdjP analysis first, there are at least two cases in which the string ‘NP + Adj’ is conceivably an AdjP: special constructions with subject predicate, as in (17) (see also Aa. Hansen Reference Hansen1967: II: 404–405), and regular constructions with object predicate, as in (18).
In the special subject predicate construction, the adjective either agrees with the subject (voksne) or the NP (voksent), or it does not agree with anything at all (voksen). In the regular object predicate construction, the adjective agrees with the NP-object (blåt). One may suggest that these ‘NP + Adj’ strings are possibly AdjPsFootnote 4 and that strings of ‘NP + rundt’ are to be analyzed in the same way, i.e. as AdjPs. But this must be rejected: Whereas adjectival rund does occur as the inflected form rundt (the singular neuter form), rundt in its postpositional usage is not inflectable and therefore does not agree with any element, see for instance (16). Thus, in examples like (16) rundt is not an adjective – at least not an adjective in a construction such as those in (17) and (18).
This leaves the analysis where rundt is an adverb heading an AdvP in ‘NP + rundt’. I have not been able to falsify this analysis because spatial AdvPs and PrePs have the same syntactic distribution. Offhand, the fact that adverbs (unlike adpositions) do not seem to select NPs speaks against the AdvP analysis. It remains to be seen whether the AdvP analysis is tenable, but answering this question is beyond the scope of this squib.
4. Summary
In this squib, I have provided arguments in favour of the view that Danish rundt is plausibly a postposition. I have done so by discussing some of rundt's adpositional properties and by falsifying some competing analyses without falsifying the postposition analysis itself. It is an open question whether the postposition analysis stands up to scrutiny, and whether more plausible analyses can be brought to bear. I leave this for future research.
