1. Introduction
Grammaticalization has been an important research topic for some time now. Grammaticalization theory has made it possible for researchers to shed new light upon a variety of phenomena from many languages, as can be seen from the work reported in reference works such as Narrog & Heine (Reference Narrog and Heine2011). However, no analysis is immune to criticism. Grammaticalization proceeds in small steps, and it is not always clear if, or to what extent, an element or a construction should be considered grammaticalized.
A popular topic in the grammaticalization literature has been so-called pseudocoordination with posture verbs in Mainland Scandinavian. An example is (1), from Norwegian (constructed, like all examples without a source annotation; unless indicated otherwise, all the numbered examples in this paper are from Norwegian).
(1)
The standard view in the literature is that this is a grammaticalized progressive construction, in which the posture verb is decategorized and has a bleached meaning. In this article, I will follow up some critical voices from recent years, and argue that the traditional arguments for grammaticalization are not substantiated. However, I will also discuss new evidence for early grammaticalization.
It is first necessary to say something more general about pseudocoordination in Mainland Scandinavian. In Scandinavian grammar, the term pseudocoordination is used of sentences such as the Norwegian (1)–(5).
(2)
(3)
(4)
(5)
A pseudocoordination contains two verbs with the same inflectional form and the conjunction og ‘and’ between them. It might look like a coordination of two verb phrases, but its grammatical properties are different from those of regular coordination (see Section 2). The first verb can be a posture verb, as in (1), but some verbs of other types are also possible, as shown in (2)–(5).
The set of first verbs that allow pseudocoordination in Scandinavian is rather heterogeneous. Pseudocoordination with different first verbs shows different grammatical behavior (Lødrup Reference Lødrup2002, Reference Lødrup, Butt and King2017). Pseudocoordination has been reported in several languages (Ross Reference Ross, Pratas, Pereira and Pinto2016), but there is no assumption here that pseudocoordination must have the same grammatical analysis across or within languages.
I take the traditional position that pseudocoordination is not coordination. This is the analysis of the great Danish grammarians Jespersen (Reference Jespersen1895) and Diderichsen (Reference Diderichsen1957:156). The second verb phrase is ‘an infinitive in disguise’ (Jespersen Reference Jespersen1895:170, original wording ‘en forklædt infinitiv’), which is the complement of the first verb (Lødrup Reference Lødrup2002, Reference Lødrup, Helland and Salvesen2014a). Pseudocoordination is most often a control construction in which the subject of the first verb controls the PRO subject of the second verb.Footnote 1
In some languages, there are several well established cases of posture verbs grammaticalizing into a progressive marker (see e.g. Heine & Kuteva Reference Heine and Kuteva2002, Newman Reference Newman and Newman2002). Arabic is often mentioned in this connection; the example in (6) is from Kuwaiti.
(6)
(Camilleri & Sadler Reference Camilleri, Sadler, Butt and Holloway King2017:171)
Notice how the meaning of the sentence makes it impossible to interpret the posture verb literally.
The article is based mainly on Norwegian data. However, Danish and Swedish are only minimally different from Norwegian in relevant respects, and I will also draw on these languages. I will not examine Icelandic and Faroese pseudocoordination, which is discussed in Jóhannsdóttir (Reference Jóhannsdóttir and Puig-Waldmüller2007) and Heycock & Petersen (Reference Heycock, Petersen, Braunmüller and Gabriel2012). The term Scandinavian in this article is therefore to be understood as Mainland Scandinavian.
For Mainland Scandinavian, it is a common view that pseudocoordination with posture verbs is a grammaticalized progressive construction. This is the view of Heine (Reference Heine1993:37–39), Ebert (Reference Ebert and Dahl2000), Kuteva (Reference Kuteva2001), Hopper & Traugott (Reference Hopper and Traugott2003:206–207), Hilpert & Koops (Reference Hilpert and Koops2008), Hesse (Reference Hesse2009), Hansen & Heltoft (Reference Hansen and Heltoft2011:988), Hilpert (Reference Hilpert2011), Höder (Reference Höder, Gregersen, Parrott and Quist2011:75), Biberauer & Vikner (Reference Biberauer, Vikner, LaCara, Moulton and Tessier2017). It should be noted, however, that the claims about grammaticalization vary in strength. In some cases, an author just states his or her view in passing (e.g. Höder Reference Höder, Gregersen, Parrott and Quist2011:75), while in other cases, there is a thorough treatment (e.g. Hesse Reference Hesse2009). Furthermore, some of the authors listed above primarily give a summary of the work of others, implicitly accepting it, but without having researched the topic themselves.
There is also a question of the degree of grammaticalization. Grammaticalization is a gradual process, and some authors make it clear that they assume grammaticalization to some extent only. For example, Kuteva (Reference Kuteva2001:46) says that ‘we cannot speak of a very high degree of grammaticalization of this structure’. Tonne (Reference Tonne1999: 157) says that it is ‘grammaticalized to some extent’, but she also uses the term ‘pre-grammaticalization’ (Tonne Reference Tonne1999, Reference Tonne2001). Biberauer & Vikner (Reference Biberauer, Vikner, LaCara, Moulton and Tessier2017:80) state that the first verb (in Danish and Afrikaans) is a ‘minimally grammaticalized element’. Others are even more critical toward the idea of grammaticalization, raising doubt if there is grammaticalization at all (Behrens et al. Reference Behrens, Flecken and Carroll2013, Ross & Lødrup Reference Ross and Lødrup2017).
The question of whether or not something is grammaticalized must be an empirical question. Research must take grammaticalization theory as the point of departure, see what predictions it makes for the phenomenon in question, and investigate if they hold – or rather to what extent they hold, because grammaticalization is a gradual phenomenon. Two important predictions of grammaticalization theory concern bleaching of meaning and decategorization. Grammaticalization affects the meaning of the unit in question; this is usually called bleaching of meaning, or desemanticization (e.g. Heine Reference Heine1993:54). If pseudocoordination with posture verbs is a grammaticalized construction, it is expected that the original meaning of the posture verb is bleached. Another ingredient in grammaticalization is decategorization (e.g. Heine Reference Heine1993:55; Hopper & Traugott Reference Hopper and Traugott2003:106–115). If the construction is grammaticalized, it is expected that the grammatical status of the posture verb is affected. It could change from a lexical verb to an auxiliary, or to a light verb, or to something in between (Cardinaletti & Giusti Reference Cardinaletti, Giusti, Corver and van Riemsdijk2001).
In the literature on pseudocoordination, both bleaching of meaning and decategorization are traditional assumptions. I will argue that bleaching of meaning is a general property of posture verbs, which is independent of pseudocoordination. When it comes to the question of category, I will argue that the posture verb is usually lexical. There is, however, a somewhat marginal option for using the posture verb as a light verb in a complex predicate construction. I will also discuss the standard assumption that the construction is progressive. My conclusion is that it is not, following Blensenius (Reference Blensenius, Anna, Holm, Lundin, Rahm and Tronnier2014, Reference Blensenius2015).
The structure of the article is as follows: Section 2 shows how pseudocoordination is grammatically different from coordination. Bleaching of meaning is discussed in Section 3, and decategorization in Section 4. The question of grammaticalization is discussed in Section 5. Section 6 takes up the question of progressive aspect, and Section 7 concludes.
2. Criteria for pseudocoordination
Before discussing grammaticalization of pseudocoordination, it is necessary to identify some grammatical criteria for distinguishing pseudocoordination from regular coordination. The basis for assuming pseudocoordination is the fact that some sentences that look like cases of verb or VP coordination have grammatical properties that are different from those of coordination. Discussions of pseudocoordination in Scandinavian usually mention some differences (see Kvist Darnell (Reference Kvist Darnell2008:86–92) for an overview with references). It is not unproblematic, however, what should be considered sufficient and necessary conditions for assuming that a sentence is a pseudocoordination. The sequence posture verb and verb is in principle ambiguous. Posture verbs can of course be coordinated with other verbs, and example (7) is not a pseudocoordination (see (i) below).
(7)
We will now discuss five properties of pseudocoordination to see to what extent they can be used as a sufficient condition for assuming pseudocoordination.
(i) A finite second verb does not follow the first verb to the verb second (V2) position, as shown in (8).
(8)
Two verbs that are coordinated appear as a unit in the V2 position, as in example (7) above. When pseudocoordination is not coordination, there is no reason the two verbs should move there together. Not allowing the second finite verb in the V2 position is a necessary condition to assume pseudocoordination. It cannot be a sufficient condition, however, because of sentences such as (9). (Example (9) and many other example sentences in this article are from the Norwegian Internet pages, either directly or indirectly through the NoWaC (Norwegian Web as Corpus; http://www.hf.uio.no/iln/om/organisasjon/tekstlab/prosjekter/nowac/index.html) corpus of web texts. Some orthographic corrections have been made silently.)
(9)
(NoWaC)
Sentences such as (9) are VP coordinations in which a second finite verb does not follow the first verb to the V2 position. The corresponding construction in German has been discussed as a kind of asymmetric coordination, see e.g. Höhle (Reference Höhle, Mascaró and Nespor1990).
(ii) The Coordinate Structure Constraint (Ross Reference Ross1967:89–114), which prohibits movement out of a coordination, is not active in pseudocoordination. Even so, a violation of this constraint is not a sufficient condition to assume pseudocoordination. Lakoff (Reference Lakoff1986) pointed out that the constraint may be violated under certain conditions. One of his examples is (10):
(10)
(iii) A complement of the first verb can follow the first or the second verb, as seen in the Swedish (11):
(11)
(from Teleman, Hellberg & Andersson Reference Teleman, Hellberg and Andersson1999:904)
Teleman et al. (Reference Teleman, Hellberg and Andersson1999:904) point out that i trädgården ‘in garden.def’ must be selected by stå ‘stand’. It would be very strange to say ?Han glodde i trädgården ‘he stared in garden.def’. When both the PP and the VP are complements of the posture verb, it is not unexpected that they do not have a fixed order. This option of letting a complement of a first verb follow a second verb is not possible in regular VP coordination, such as (12).
(12)
It thus seems to be a sufficient condition for assuming pseudocoordination.
(iv) A negation with the first verb negates the whole pseudocoordination, as in (13).
(13)
This fact follows from both subordination and complex predicate analyses of pseudocoordination. The option of negating the whole pseudocoordination seems to be a sufficient condition to assume pseudocoordination.Footnote 2
(v) The presentational focus construction is possible with all first verbs in pseudocoordination, except ta ‘take’. An example is (14):
(14)
The presentational focus construction has an expletive subject and realizes its argument as an object (see e.g. Platzack Reference Platzack and Karlsson1983, Lødrup Reference Lødrup1999). It represents a rather general option with one-place verbs in Mainland Scandinavian, independently of pseudocoordination. When two verb phrases are coordinated, the presentational construction is not possible with the first verb. Example (15) must be a regular coordination, because the first verb arbeide ‘work’ does not allow pseudocoordination.
(15)
The verb arbeide ‘work’ usually allows the presentational focus construction. Even so, (15) is ungrammatical. The reason is that a subject in a coordinated sentence can only be dropped when it is identical to the subject of the preceding sentence. This is different in a pseudocoordination such as (14) above (Cormack & Smith Reference Cormack and Smith1994:81; Lødrup Reference Lødrup2002). The analysis assumed here (Section 1) predicts it to be grammatical. The PRO subject of the second verb, which is usually controlled by the preceding subject, is controlled by the preceding object in presentational focus sentences. The option of the presentational focus construction with the first verb can be considered a sufficient condition for pseudocoordination.
We see, then, that some grammatical criteria for pseudocoordination can be identified. This is necessary to allow us to decide if a sentence can be considered a pseudocoordination, independently of one’s assumptions about their more controversial properties.
3. Bleaching of meaning
Bleaching of meaning is an important component in grammaticalization. It is clear that the first verb in a pseudocoordination can have a bleached meaning, when posture verbs are put aside. Examples are ta ‘take’ (Vannebo Reference Vannebo2003), and drive ‘carry on’ (Lødrup Reference Lødrup, Butt and King2017), as in examples (5) and (3) above. There are also first verbs that could hardly be assumed to have a bleached meaning, such as ringe ‘call’ in (4) above.Footnote 3
It is a common assumption that posture verbs have bleached meanings in a pseudocoordination (e.g. Vannebo Reference Vannebo1969:64; Hansen & Heltoft Reference Hansen and Heltoft2011:994). I agree that this is often the case. However, I will argue that this is not relevant to the analysis of pseudocoordination, because posture verbs often have bleached meanings independently of pseudocoordination.
The idea of bleached meanings with posture verbs in pseudocoordination is older than modern grammaticalization theory; Kvist Darnell (Reference Kvist Darnell2008:44) sees it as a traditional standard observation. The literature is somewhat vague and ambiguous on this point, however. It has been claimed that the subject of the pseudocoordination must have the orientation denoted by the posture verb (e.g. Teleman et al. Reference Teleman, Hellberg and Andersson1999:904; Kuteva Reference Kuteva2001:46; Bylin Reference Bylin2013:73). However, this is denied by others (e.g. Braunmüller Reference Braunmüller1991:103–104; Faarlund, Lie & Vannebo Reference Faarlund, Lie and Vannebo1997:648). It has also been claimed that the bleaching effect is weak (e.g. Tonne Reference Tonne2001:110; Behrens et al. Reference Behrens, Flecken and Carroll2013). To clear this confusion, it is necessary to consider the use of posture verbs more generally, independently of pseudocoordination. Posture verbs are used in rather different ways across languages, as shown by an interesting literature (e.g. Newman Reference Newman and Newman2002, Ameka & Levinson Reference Ameka and Levinson2007, Viberg Reference Viberg2013). Some languages do not have posture verbs, e.g. French. Other languages, including the Scandinavian languages, use them in ways that extend their basic meaning.
One might say that the lexical meaning of the posture verb is often defocused (or backgrounded, Kinn Reference Kinn, Coussé, Andersson and Olofsson2018:86) in pseudocoordination. However, this option of defocusing is very general in Scandinavian. Posture verbs are often used to locate the subject (as in many other languages, see Ameka & Levinson Reference Ameka and Levinson2007). Blensenius (Reference Blensenius2015:35–36) shows that using a posture verb is much more frequent than using the copula (in Swedish) in locational sentences such as (16)–(18), adapted from Blensenius (Reference Blensenius2015:36).
(16)
(17)
(18)
In Scandinavian, posture verbs are not only used of the posture of a person, but also with inanimate and abstract subjects, in various metaphorical uses, and more or less fixed expressions (e.g. Hansen Reference Hansen1974, Holm Reference Holm2013, Berthele et al. Reference Berthele, Whelpton, Næss and Duijff2015). For example, a beer ‘stands’ in the fridge, a ring ‘sits’ on the finger, a city ‘lies’ on the coast, a word ‘stands’ in the dictionary, an ability ‘sits’ in the brain, options ‘lie’ ahead, etc.
A common way of arguing for bleaching of meaning in pseudocoordination can be found in e.g. Faarlund et al. (Reference Faarlund, Lie and Vannebo1997:648) and Hansen & Heltoft (Reference Hansen and Heltoft2011:994). The latter gives the sentence in (19) as an example, noting that the subject ‘they’ cannot literally have spent months sitting around a table.
(19)
However, the posture verb can be used in the same way outside of pseudocoordination, as in (20).
(20)
Examples (21)–(30) below are intended to illustrate how the use of posture verbs in pseudocoordination has parallels outside of pseudocoordination.
(21)
(NoWaC)
(22)
(23)
(NoWaC)
(24)
(25)
(26)
(27)
(NoWaC)
(28)
(29)
(sv-se.facebook.com › Platser › Hammerfests kommun › Brandstation)
(30)
Each pseudocoordination is followed by a non-pseudocoordination which is selected to be as parallel as possible. The examples show that independently of pseudocoordination, a feeling ‘sits’ inside us, as in (21)–(22), those who govern us ‘sit’ while doing so, as in (23)–(24), food ‘stands’ while being prepared, as in (25)–(26), the temperature ‘lies’ at a certain level, as in (27)–(28), and we ‘lie’ in a lane when driving on the freeway, as in (29)–(30).
Even some fixed expressions with a posture verb can be used with or without pseudocoordination. An example is her sitter jeg ‘here sit I’, as in (31)–(32), which is used colloquially for ‘as for me’.
(31)
(32)
Posture verbs can also have a secondary subjective use, with what Biberauer & Vikner (Reference Biberauer, Vikner, LaCara, Moulton and Tessier2017:83) call ‘speaker-perspective-related meanings’. In (33), the speaker expresses a negative attitude toward what is said:
(33)
This subjective use has a parallel in other languages. It is discussed by Breed (Reference Breed2017) for Afrikaans pseudocoordination, and by Lemmens (Reference Lemmens2005:185) for Dutch posture verbs with infinitives.Footnote 4 Again, the relevant use of the posture verb can be found independently of pseudocoordination in Norwegian, as in (34).
(34)
Our point of departure in Section 3 was the traditional assumption that a posture verb has a bleached meaning in a pseudocoordination. We see now that this is a misrepresentation of the facts. Posture verbs can have bleached meanings, but this is a general phenomenon that does not depend upon pseudocoordination. If there is a connection between bleaching of meaning and pseudocoordination for posture verbs, one could speculate that it goes the other way (as proposed by an anonymous reviewer). It could be that the bleached semantics of posture verbs has opened the way for their use in pseudocoordination. This development must have taken place in prehistoric times, because pseudocoordination with posture verbs is well established in Old Norse (Vannebo Reference Vannebo1969:66, 84–86).
For Norwegian, there seems to be no way of using a posture verb in a pseudocoordination that does not have a parallel outside of pseudocoordination.Footnote 5 However, Danish gives an example of such a case. Danish, or variants of Danish, has a use of ligge ‘lie’ that is special for pseudocoordination. Korzen (Reference Korzen2016:29–30) says that ligge can intensify the action, especially in contexts where irritation or desperation is expressed. One of her Internet examples is (35). This use of ligge does not seem to exist in Norwegian or Swedish.
(35)
This use of ligge in Danish is interesting. It seems to be an instance of a posture verb that can have a specialized meaning in pseudocoordination. It is not clear, however, that this can be related directly to grammaticalization.
4. Decategorization
If pseudocoordination with posture verbs is a grammaticalized construction, it is expected that the grammatical status of the posture verb is affected. It could change from a lexical verb to an auxiliary or a light verb, or to something in between.
There are several claims in the literature that the grammatical status of the posture verb is reduced to that of an auxiliary or a light verb in a complex predicate construction. The idea that the posture verb is a kind of auxiliary is independent of modern grammaticalization theory; it can be found e.g. in Thorell (Reference Thorell1973:190). Explicit claims of auxiliary status can be found in Heine (Reference Heine1993:37–39), Hesse (Reference Hesse2009) and Bylin (Reference Bylin2013:72–74). It is also hinted at in Faarlund et al. (Reference Faarlund, Lie and Vannebo1997:534) and Teleman et al. (Reference Teleman, Hellberg and Andersson1999:902). A light verb analysis can be found in Wiklund (Reference Wiklund2007) and Kjeldahl (Reference Kjeldahl2010).
If posture verbs are decategorized in pseudocoordination, it would provide an important argument for grammaticalization. The prediction is then that their grammatical properties are affected, such as their syntactic selection and their ability to undergo valency-changing rules. It will be argued that the posture verbs are usually not decategorized. Even so, there is a somewhat marginal option for using the posture verb as a light verb in a complex predicate construction.
4.1 Selectional properties
If the posture verb has changed its categorial status to an auxiliary or a light verb, it could affect its syntactic selection. Hesse (Reference Hesse2009:51) says that only a posture verb without a locative can be considered a part of a pseudocoordination. However, it is clear that a sentence can satisfy the criteria for pseudocoordination that were discussed in Section 2 even if the posture verb takes a locative. An example is (14) above, repeated as (36), whose first verb takes part in the presentational focus construction (see the discussion under (v) in Section 2).
(36)
Even if posture verbs in pseudocoordination can take a locative complement, they often don’t. This might seem to be an important difference in syntactic selection from posture verbs in other uses. However, a closer look at the facts shows that this is not the case. A posture verb outside pseudocoordination often requires a complement. The exception is the agentive ‘maintain position’ meaning of Levin & Rappaport Hovav (Reference Levin and Rappaport Hovav1995:127). Locative complements are common, but it is important that various types of complements can satisfy this requirement, as exemplified in (37). (There is a gradual transition to fixed expressions.)
(37)
In pseudocoordination, one might think that there is no requirement for the posture verb to take a complement. However, with the analysis assumed here, the pseudocoordinated VP is a complement to the posture verb, which must be specified in its lexical entry. Posture verbs have a relatively meager meaning, and this is probably the reason that they usually require a complement. The posture verbs could be compared to the verb bo ‘live, reside’. This verb also requires a complement of some kind. There is often a locative complement, but other complements can also satisfy the requirement, as shown in (38):
(38)
We must conclude, then, that the selectional properties of posture verbs in pseudocoordination are not, in principle, very different from those of posture verbs outside of pseudocoordination. Pseudocoordination does not reduce their complement taking options.
4.2 Valency-changing rules
If the posture verb has developed into an auxiliary or a light verb, it is not expected to undergo valency-changing rules. However, posture verbs in pseudocoordination keep their regular options in this respect (Lødrup Reference Lødrup2002, Reference Lødrup, Helland and Salvesen2014a). They can passivize, as in (39), and they can take part in the presentational focus construction (discussed in Section 2), as in (14) above, repeated as (40).
(39)
(40)
The presentational focus construction does not affect the second verb of the pseudocoordination in any way.Footnote 6 This gives a reason for not analyzing the first verb as an auxiliary, because auxiliaries are not subject to valency-changing rules. The second verb can also show its independence. For example, it can passivize without affecting the posture verb, as in (41):
(41)
We see, then, that both the first and the second verb can undergo valency-changing rules independently of each other. This lends support to the argument that the two verbs do not constitute a complex predicate, because a complex predicate undergoes this kind of rule as a unit.
Another argument against a complex predicate analysis is given by sentences such as (42) with a periphrastic future and (43) with a periphrastic perfect (Lødrup Reference Lødrup, Helland and Salvesen2014a).
(42)
(43)
(NoWaC)
It is a general rule that complex predicate formation is not possible when the second verb has independent tense, i.e. when it refers to a time that is independent of the time of the first verb (Wurmbrand Reference Wurmbrand2001:79–99). This gives an argument against assuming complex predicates in (42)–(43). (Note that (42) has time adjuncts with incompatible time references.)
4.3 Pseudocoordination with a complex predicate
There are, however, two phenomena that give evidence for a complex predicate analysis of pseudocoordination with posture verbs in some cases. This was pointed out by Lødrup (Reference Lødrup, Helland and Salvesen2014a, b), who did not discuss grammaticalization. In examples such as (44)–(45), the two verbs have passivized as a unit. The passive subject corresponds to an argument of the second verb: the object in (44) and the oblique object in (45).
(44)
(45)
In cases such as (44)–(45), passivization gives evidence that that the two verbs must be analyzed as a complex predicate in a monoclausal construction. This kind of passive sentence is often called long passives. They can also be found in Swedish, illustrated in (46), and Danish, illustrated in (47).
(46)
(47)
Long passives can also be found with pseudocoordination verbs that are clearly grammaticalized, namely ta ‘take’ and drive ‘carry.on’ (Lødrup Reference Lødrup, Helland and Salvesen2014a, Reference Lødrup, Butt and King2017):
(48)
(NoWaC)
(49)
Norwegian also allows this kind of passive outside of pseudocoordination with a small number of verbs (Lødrup Reference Lødrup2014b). An example is (50):
(50)
Long passives such as (44)–(47) are interesting in that they indicate that a pseudocoordination with a posture verb can restructure to a monoclausal complex predicate construction. That said, examples such as (44)–(47) seem to be somewhat marginal and infrequent.Footnote 7
There is, however, another phenomenon that might also suggest restructuring to a complex predicate (mentioned in Lødrup Reference Lødrup, Helland and Salvesen2014a). In presentational focus sentences with pseudocoordination, the object in some cases follows the second verb, as in (51), and not the first verb, as in (14) above.
(51)
This could be taken as an indication that the two verbs constitute a complex predicate that has undergone the presentational focus rule as a unit.Footnote 8 Norwegian also allows presentational focus sentences with complex predicates outside of pseudocoordination in some cases (Holmberg Reference Holmberg2002:122; Lødrup Reference Lødrup2014b). An example is (52):
(52)
(Holmberg Reference Holmberg2002:122)
It is a marked option to let the object follow the second verb in a pseudocoordination such as (51) above. Examples do not abound in texts. The option seems to be subject to the condition that the first and the second verb form a collocation, such as ligge og flyte ‘lie and float’ in (51) above.
We see then that there are two independent phenomena – long passives and presentational focus sentences – that indicate that a pseudocoordination can be restructured to a monoclausal complex predicate construction in some cases. The first verb is then used as a so-called light verb. This should be seen as a case of grammmaticalization: a light verb merges its argument structure with that of the following verb, and no longer has its own clausal domain. It has been claimed, however, that the emergence of light verbs represents a rather minimal case of grammaticalization. Butt & Lahiri (Reference Butt and Lahiri2013) argue that there is a tight connection between a light verb and its corresponding main verb.Footnote 9 This makes light verbs different from auxiliaries, which lose the connection to the verb they are derived from. There seems to be no (further) bleaching of meaning when the posture verbs are used as light verbs. Example (44) above is illustrative in this respect. As the textual context (at forum.kvinneguiden.no/index.php?showtopic=404086) makes clear, its point is that one should eat one’s kebab sitting, not walking.
An important fact is that many sentences with pseudocoordination have grammatical properties that are incompatible with a complex predicate analysis, e.g. (40)–(43) above. It is therefore impossible to assume that the posture verb is always a light verb in a pseudocoordination. This situation is parallel to that of non-pseudocoordination verbs which can be used as light verbs. For example, forsøke ‘try’ allows restructuring, as seen in (50) and (52) above. Even so, it is easy to construct sentences with properties that are incompatible with restructuring, such as (53), in which the infinitival complement of forsøke ‘try’ is realized as a passive subject.
(53)
The parallel is also interesting in that there seems to be no clear bleaching of meaning effect with the restructuring verbs that do not allow pseudocoordination (such as forsøke ‘try’).
5. The question of grammaticalization
The emergence of pseudocoordination with posture verbs could be considered a case of grammaticalization. It seems natural to assume that its origin is regular coordination, as in Vannebo (Reference Vannebo1969:66) (see also Heine Reference Heine1993:37–39; Kuteva Reference Kuteva2001:67–73; Ross Reference Ross, Pratas, Pereira and Pinto2016:229–230). We do not know how this happened in Scandinavian, however. Pseudocoordination is well established in Old Norse (Vannebo Reference Vannebo1969:66, 84–86), and it seems to be impossible to follow it further back in time.
It is clear that pseudocoordination can be grammaticalized. A construction that has come far in this respect is pseudocoordination with the verb ta ‘take’, as in example (5) above (Vannebo Reference Vannebo2003). Pseudocoordination with the verb drive ‘carry on’, as in example (3) above, is also grammaticalized to some extent (Lødrup Reference Lødrup, Butt and King2017). On the other hand, there are cases of pseudocoordination that could hardly be thought of as grammaticalized, e.g. those with ringe ‘phone’ as their first verb (example (4) above; see also note 3).
Is pseudocoordination with posture verbs a grammaticalized construction, as is usually assumed in the literature? Sections 3 and 4 have argued against the traditional assumptions about bleaching of meaning and decategorization. There seems to be no evidence for an established grammaticalized construction to which all cases of pseudocoordination with posture verbs belong. However, modern grammaticalization theory sees grammaticalization as a gradual process (Roberts Reference Roberts2010, Traugott & Trousdale Reference Traugott and Trousdale2010a). The discussion has also shown that there is one phenomenon that seems to indicate early grammaticalization, namely the somewhat marginal option of using a posture verb as a light verb in a complex predicate construction (Section 4.3). This option has not been mentioned as an argument for grammaticalization before, apart from a hint in Ross & Lødrup (Reference Ross and Lødrup2017). However, the emergence of light verbs represents a rather minimal case of grammaticalization, according to Butt & Lahiri (Reference Butt and Lahiri2013). They argue that light verbs are tightly connected to their corresponding main verbs, and do not represent a first stage of further grammaticalization. Whether or not this is correct for the present case remains to be seen (see note 9).
6. Progressive aspect
The assumption that pseudocoordination with posture verbs is an expression of progressive (or durative) aspect can be found in e.g. Ebert (Reference Ebert and Dahl2000), Tonne (Reference Tonne2001), Wiklund (Reference Wiklund2007) and Hesse (Reference Hesse2009). Kvist Darnell (Reference Kvist Darnell2008:44) says that there is agreement on this point in most modern descriptions. The assumption is older than modern grammaticalization theory (see e.g. Vannebo Reference Vannebo1969), but it fits well with a grammaticalization analysis. However, Blensenius (Reference Blensenius, Anna, Holm, Lundin, Rahm and Tronnier2014, Reference Blensenius2015) argues that pseudocoordination with posture verbs differs from progressives in several ways, and this section is partly based upon his work.
Scandinavian has no systematic opposition between progressive and non-progressive forms. A simple verb form can have a progressive interpretation. In actual language use, pseudocoordination with posture verbs is seldom used to express progressivity. It has been shown that the predicate following the posture verb is usually atelic (see Tonne Reference Tonne2001:77 on Norwegian, Blensenius Reference Blensenius2015:14, 44 on Swedish). Typical predicates are e.g. ‘read’, ‘talk’, and ‘wait’. These predicates would get a progressive interpretation independently of the posture verb.
The question then is what happens in the few cases where the predicate following the posture verbs is telic. It is often claimed in the literature that the posture verb changes the aspectual meaning from completed to incomplete. Faarlund et al. (Reference Faarlund, Lie and Vannebo1997:649) compare examples (54) and (55). Without pseudocoordination, as in (54), the action is understood as completed, while (55), with pseudocoordination, is understood as incomplete.
(54)
(55)
This effect is not surprising, because a posture verb is in a sense progressive in itself (Blensenius Reference Blensenius2015:38). However, Teleman et al. (Reference Teleman, Hellberg and Andersson1999:905) point out that the interpretation of a telic second predicate is not necessarily affected by the posture verb. They give Swedish examples such as that in (56):
(56)
It is not difficult to find other examples in which the interpretation of a telic predicate is not affected by the posture verb, such as the Norwegian (57):
(57)
There is also no aspectual effect when a posture verb takes a stative second verb (Tonne Reference Tonne2001:80–81), as in (58). With regular progressives, stative verbs are typically not used.
(58)
(Østlendingen, 14 September 2013, page 9)
It is interesting to compare to pseudocoordination with drive ‘carry on’ as a first verb, which is found only in Norwegian (Lødrup Reference Lødrup, Butt and King2017). This verb always has an aspectual effect, giving a progressive, habitual or iterative interpretation. (Alternatively, it could be described as pluractional, see Lødrup Reference Lødrup, Butt and King2017.) It often takes a telic second predicate, which gets an incomplete or iterative interpretation, as in (59)–(60). The Norwegian verb drive is different from the posture verbs in that it always has an aspectual effect in sentences with a telic second predicate.
(59)
(60)
Posture verbs in pseudocoordination do not consistently have an aspectual effect, and the construction should not be considered progressive. This conclusion is strengthened by studies of how Norwegian pseudocoordination with posture verbs is translated into languages with a grammaticalized progressive construction (Tonne Reference Tonne2001, Ebeling Reference Ebeling, Ebeling and Hasselgård2015, Hareide Reference Hareide, Ji, Hareide, Li and Oakes2016). A pseudocoordination is not always translated into a progressive. For translations to Spanish, Hareide (Reference Hareide, Ji, Hareide, Li and Oakes2016) shows that the progressive construction with estar ‘be’ + gerund is seldom used – only in 12.5% of the cases in her material. In translations to English, Ebeling (Reference Ebeling, Ebeling and Hasselgård2015) says that it is common to keep the posture verb, often adding locative there, and using the –ing-form of the second verb. Among her examples is (61), a sentence from a Norwegian novel, and its published English translation (62).
(61)
(62)
(Bergljot Hobæk Haff, Shame. Translated by Sverre Lyngstad. London: Harvill Press, 1999; page 176.)
We see that posture verbs in pseudocoordination do not consistently have an aspectual effect. The aspectual perspective seems to have been misleading. A better alternative could be to follow Tonne (Reference Tonne2001:121), Behrens, Flecken & Carroll (Reference Behrens, Flecken and Carroll2013:129–130), and Blensenius (Reference Blensenius2015:35–42) in seeing the construction as locational. (A better term might be situative, as suggested by an anonymous reviewer.) The function of the construction is characterized in the following way in Behrens et al. (Reference Behrens, Flecken and Carroll2013:129–130):
[T]he posture pseudocoordinations in Norwegian are expressions of temporary localizations in space (stand–sit–lie) that contribute to the interpretation of the description of a situation in two ways: They “localize” the subject referent in space (statively), and their temporary state denotation adds temporariness to the situation expressed by the main verb.
7. Conclusion
The idea that pseudocoordination with posture verbs is a grammaticalized progressive construction has been popular and influential. In this article, I have argued against the traditional assumptions, following up some critical voices from recent years’ research. The construction was argued not to be aspectual, but rather locational or situative. The traditional assumptions about bleaching of meaning and decategorization were found to be unsubstantiated. I also discussed new evidence for grammaticalization, namely the use of posture verbs as light verbs in complex predicates (Section 4.3). This is an ‘extra’ option for some sentences with pseudocoordination, which must be considered a rather minimal case of grammaticalization.
In Section 1, it was observed that pseudocoordination in Scandinavian is a rather heterogeneous phenomenon. This observation extends to the degree of grammaticalization shown by the different types. The extremes are represented by pseudocoordination with ta ‘take’, which could be considered completely grammaticalized (Vannebo Reference Vannebo2003), and pseudocoordination with ringe ‘call’ (see note 3), which can hardly be grammaticalized at all. Between these extremes we find pseudocoordination with drive ‘carry on’ (Lødrup Reference Lødrup, Butt and King2017). This is also where we find the posture verbs, leaning a bit more to the side of ringe ‘call’.
Acknowledgements
For valuable input to this work, I would like to thank the participants at the Workshop on Pseudo-coordination and Multiple Agreement Constructions (Ca’ Foscari University of Venice, May 2017), and at a departmental seminar (Oslo, September 2017). I would also like to thank my colleagues Ingebjørg Tonne and Janne Bondi Johannessen for discussion. Special thanks are due to Daniel Ross (University of Illinois), co-author of Ross & Lødrup (2017). I am also grateful to the anonymous NJL reviewers, who made me reconsider some of my views. Special thanks to one of them for exceptionally thorough and constructive feedback.
Corpus
NoWaC (Norwegian Web as Corpus)
http://www.hf.uio.no/iln/om/organisasjon/tekstlab/prosjekter/nowac/index.html.
