We are grateful to all those involved in the building of the text corpora we have used; our research would not have been possible without their efforts. We thank Stefan Th. Gries for advice on the use of his R script. Thanks also to the audiences at Gramino 2 (Oslo, May 2018) and ICCG 10 (Paris, July 2018) for useful comments. Finally, we are grateful for the constructive criticism from one of the anonymous reviewers, which has helped us improve this text at several important points.
- 1 In examples, the verbs of pseudocoordinated phrases are in italics. The parentheses after each exam (...)
1Pseudocoordination (PC) in the Mainland Scandinavian languages (Danish, Norwegian, and Swedish) has been the subject of much research but is still rather poorly understood. PCs consist of two verbs and their phrases (V1 and V2) with a common subject and joined with a coordinator meaning ‘and’. The construction resembles canonical verb phrase coordination, but has a number of distinguishing characteristics. The best-known type of PC is posture verb pseudocoordination (PVPC), with V1s denoting a human posture. Central posture verbs (PVs) are those meaning ‘lie’, ‘sit’, and ‘stand’. In Mainland Scandinavian, the verbs for ‘lie’ are Danish/Norwegian ligge and Swedish ligga (jointly referred to as LIE) ; for ‘sit’, they are Danish sidde, Norwegian sitte, Swedish sitta (SIT) ; and for ‘stand’, Danish/Norwegian/Swedish stå (STAND). Examples (1)–(3) illustrate PVPC in the three languages:1
(1)
|
Om
|
ett
|
par
|
minuter
|
låg
|
Olle
|
i
|
djup
|
sömn
|
och
|
snarkade
|
(Sw.)
|
|
about
|
a
|
couple
|
minutes
|
lay
|
Olle
|
in
|
deep
|
sleep
|
and
|
snored
|
|
|
‘Within a couple of minutes, Olle was snoring, sound asleep’
|
(2)
|
Ved
|
kaffebordet
|
sitter
|
presten
|
Hald
|
og
|
leser
|
avisen
|
(No.)
|
|
at
|
the.coffee.table
|
sits
|
the.priest
|
Hald
|
and
|
reads
|
the.newspaper
|
|
|
‘At the coffee table, Hald, the priest, is reading the newspaper’
|
(3)
|
En
|
gammel
|
kone
|
står
|
og
|
kigger
|
på
|
en
|
skotte,
|
der
|
|
an
|
old
|
woman
|
stands
|
and
|
looks
|
on
|
a
|
Scot
|
that
|
|
spiller
|
sækkepibe
|
(Da.)
|
|
|
|
|
|
|
|
|
plays
|
bagpipe
|
|
|
|
|
|
|
|
|
|
‘An old woman is watching a Scot who is playing the bagpipe’
|
- 2 It may be noted that both Hansen & Heltoft and Nielsen (2011) use the term kongruenskonstruktion (‘ (...)
2General treatments of Mainland Scandinavian PC may be found in Hansen & Heltoft (2011: ch. VIII),2 Faarlund et al. (1997: 534–535, 647–651), and Teleman et al. (1999, vol. 4: 902–909). The present contribution focuses on constructions with V1s denoting the three central human postures. Much previous literature has had the same focus. This literature has tended to discuss the aspectual properties of such constructions and the extent to which the PV has undergone bleaching (desemanticization) towards an aspectual auxiliary, i.e. grammaticalization (see Section 2.1). Closely tied to the view that PVPC involves auxiliation is a frequently repeated claim that PCs in general (not only the PV variant) allow only a restricted number of different V1s, i.e. evidence of paradigmatization and grammaticalization (see Section 2.2).
3However, Nielsen (2011) argues that most types of PC in Danish have open paradigms of V1s, and Kinn (2018) documents that a large number of fully lexical verbs occur as V1s of atelic-motion PCs in Norwegian (e.g. ‘run around’), showing that paradigmatization and bleaching are not characteristics of PC in general (but possibly of some variants). Further, Blensenius (2015) argues that the localizing function of PVs in Swedish PVPC is more fundamental and characteristic than that of aspectualization. Nielsen (2011) sees the common property of all PCs in Danish in a V1–V2 meaning relation of ‘possibilification’ (see Section 2.4), while Kinn (2018) argues that ‘facilitation’ is more appropriate. A widely, but not generally, shared understanding is that PCs have single-event meanings ; this is discussed in Section 2.3.
- 3 PC-like expressions of phase and related meanings, which tend to be limited to imperatives (see Tel (...)
4The present contribution develops the notion of facilitation and applies it to PVPCs (see Section 3). PVPC is the constructional variant where the notion seems least amenable and where understandings focusing on aspectuality and auxiliation have appeared most fruitful. If it can be shown that facilitation is a key to understanding this variant, it can probably be generalized to all variants of PC in Mainland Scandinavian.3
5With some exceptions (notably Kvist Darnell 2008 ; Hilpert & Koops 2008 ; Andersson & Blensenius 2018, forthc. ; Kinn 2018), previous work on PC has been intuition-based and wanting in empirical weight. To give our analysis a sound empirical foundation, we build on comprehensive corpus studies of Danish, Norwegian (specifically: Norwegian Bokmål), and Swedish. We perform distinctive collexeme analyses (Stefanowitsch & Gries 2003) and identify V2s that distinguish the postures from each other. The results reveal with great clarity how different postures are construed as facilitating different types of V2 events (activities, states, etc.).
6It is not our primary aim in this contribution to argue against an understanding of PVPCs as having an aspectual function. Rather, we explore an alternative avenue. We analyse the semantics of PVPCs in the light of facilitation and drawing on notions provided by Frame Semantics. We argue for the centrality of facilitation by postural and locative stability – and on this background, we proceed to illuminate the motivation of developments that do involve bleaching and the possible beginnings of PV auxiliation.
7Central previous contributions and notions are outlined in Section 2. The notion of ‘facilitation’ is explicated in relation to the posture frame in Section 3. Section 4 describes our materials and methods. In Section 5, we present quantitative results and interpretations. Section 6 is a concluding discussion.
8PVPC mainly involves the verbs for LIE, SIT, and STAND. Related verbs in Mainland Scandinavian are those meaning ‘hang’ (intransitive) and ‘be (elsewhere)’. Danish also has holde ‘stand’, used about vehicles, cf. Nielsen (2011: 112). PVPC is but one of several kinds of PC (see below), i.e. it is a subschema under a more general PC schema (cf. Traugott & Trousdale (2013: 13–17) for these constructional concepts). Similar constructions are found in several other languages (see Ross 2016), but we will largely confine our discussion to literature on Mainland Scandinavian.
9All PC subschemas involve two verb phrases which appear to be coordinated. It is a subject of long-standing discussion, however, whether PC should be regarded as coordination or some other kind of construction, as indicated by the prefix pseudo-. See e.g. Johnsen (1988) ; Lødrup (2002, 2014a) ; Bjerre & Bjerre (2007) ; Wiklund (2007) ; Kjeldahl (2010) ; Ross (2016). We do not consider it essential to take a firm stand on this issue ; given that constructions may have prototype structure and change over time, it is to be expected that some variants are less typical and exhibit properties that set them apart from canonical instances. In this light, PC may be regarded as a noncanonical type of verb phrase coordination.
10The verb phrases refer to events which share a participant, viz. the subject referent. These events are subevents of a single complex event (see Section 2.3) – i.e. parts of a whole. In canonical coordination, there is a symmetric relation between parts ; the conjuncts are informationally similar and may switch places. In PC, however, V1 is informationally secondary to V2, and the VPs cannot be interchanged. There are a number of other properties distinguishing PC from canonical VP coordination and indicating asymmetry between V1 and V2, see e.g. Teleman et al. (1999, vol. 4: 903) ; Hesse (2009: 13–16) ; Hansen & Heltoft (2011: 981–985) ; Kinn (2018).
11We assume here that the asymmetry of VP subevents in PVPC is closely related to the facilitating role of the V1 posture with respect to the V2 event. The posture accommodates the activity, and not the other way around. While the various kinds of evidence for asymmetry have sometimes been taken to show that PC is not “real” coordination, asymmetry may be rather more typical in coordination of VPs, clauses, and sentences than the canonical symmetry. For example, if sentences are coordinated as in Susan turned the key and the engine started, generalized conversational implicatures strongly tend to enrich (buttress) the conjunction with sequentiality (‘Susan turned the key and then the engine started’), causality (‘Susan turned the key and as a result the engine started’), and intentionality (‘Susan turned the key intending it to be the case that the engine started’) (Bezuidenhout 2002: 261 ; Levinson 2000: 117). On such a background, the asymmetry of PC seems less exceptional. Facilitation, as we shall see, has some commonalities with the presumptive meanings (cf. Levinson 2000) created by implicatures.
12It is often asserted categorically (e.g. as recently as Biberauer & Vikner (2017: 82) on Danish) that V1 and V2 must have the same morphological form. In the normal case they do. However, Kinn (2014, 2017) discusses a systematic exception to this in Norwegian (where V1 present participles combine with V2 infinitives), and Blensenius (2015, Appendix III: 34) exemplifies a type of PCs in Swedish which formally combine present and past tense (but which appear to have one topic (reference) time). Thus, ideas of copying of inflectional features (e.g. Wiklund 2007) do not seem to be tenable. V1–V2 “agreement” is probably better understood in terms of semantics than of morphology.
13The above-mentioned asymmetries have motivated analyses of PC in terms of grammaticalization (Kuteva 1999 ; 2001: ch. 3 ; Hilpert & Koops 2008 ; Hesse 2009): V1s – especially PVs – are understood as having to some extent developed into aspectual (progressive) markers in a diachronic process of auxiliation. Works arguing for or against PVPCs in Mainland Scandinavian being viewed in this light (or just assuming such an understanding) include Vannebo (1969) ; Andersson (1979) ; Josefsson (1991) ; Heine (1993: 37–39) ; Tonne (1999, 2001) ; Bertinetto et al. (2000) ; Ebert (2000) ; Hopper & Traugott (2003: 206–207) ; Henriksson (2006: 64, 140ff.) ; Bjerre & Bjerre (2007) ; Hilpert & Koops (2008) ; Hilpert (2011) ; Behrens et al. (2013) ; Ebeling (2015) ; Blensenius (2015, 2016), Biberauer & Vikner (2017). Other kinds of PC have also been discussed in this light, involving V1s such as Swedish vara ‘be’ (Ekberg 1983), Swedish gå ‘go, walk’ (Wiklund 2009 ; Josefsson 2014 ; Andersson & Blensenius 2018, forthc.), Norwegian drive ‘carry on’ (Lødrup 2002, 2017 ; Kinn forthc.), Swedish and Norwegian ta, Danish tage ‘take’ (Ekberg 1993 ; Vannebo 2003 ; Wiklund 2009 ; Nielsen 2011), and Swedish hålla på ‘carry on’ (Blensenius 2013).
14Researchers arguing that PVPCs are aspectual still often note that grammaticalization has not gone far. The PVs have retained their lexical meanings without much bleaching (e.g. Kuteva 2001: 45). Blensenius (2015, 2016) presents arguments against regarding PVPC as a primarily progressive construction. He shows that, differently from canonical progressives, PVPCs are compatible with all situation types including states. Further, they do not clearly alter the aspect of the sentence, and, finally, they can be combined with another more clearly progressive construction.
15Another common claim about PC in general is that only a limited number of verbs are used as V1 (e.g. Josefsson 1991: 133 ; Wiklund 2007: 93 ; Biberauer & Vikner 2017: 78). However, as shown by Teleman et al. (1999, vol. 4, 902–903), there are a number of different types of PC in addition to those involving posture V1s: atelic motion (e.g. ‘jump around’), telic motion (e.g. ‘crawl in’), communication (e.g. ‘phone’), etc., and Lødrup (2002) mentions verbs of assuming a posture (e.g. ‘sit down’). Both Teleman et al. and Hansen & Heltoft (2011: 979–990) give examples with around 30 different V1s. Ahlberg et al. (2015) explore automatic identification of PCs in a Swedish corpus and find new subschema candidates. Kinn (2018), in a corpus study of PC of atelic motion in Norwegian, finds no less than 167 different V1s. He further points to previously unnoticed types of possible PC: with V1s of halted motion (e.g. ‘stop’), body-part motion (e.g. ‘poke about’), visual orientation (e.g. ‘look around’), etc. This research has shown that the number of V1s in the PC schema is clearly not small, although some subschemas may have small V1 paradigms. This further means that V1s cannot in general be regarded as auxiliaries or light verbs, as also noted for English by Weisser (2015: 135–136), arguing against Wiklund’s (2007) analysis.
16Since PVPC is one subschema among several under the general PC schema, it is reasonable to look for its meaning and function in the context of the meanings and functions of other subschemas of PC – similar forms often have similar meanings. One commonly observed property is that of single-event construal, e.g. Josefsson (1991) ; Lødrup (2002) ; Wiklund (2007) ; Kinn (2018): Although composed of two predicates describing potentially separate subevents, PCs are understood as referring to one single complex event. This is reflected in V1 and V2 sharing certain grammatical operators. In (4), for instance, epistemic modality (expressed with the adverb naturligtvis) and negation (inte) have scope over both V1 and V2, and the topic times of the conjuncts are the same, expressed by the verbs (satt, väntade).
(4)
|
Jag
|
satt
|
naturligtvis
|
inte
|
helt
|
passivt
|
och
|
väntade
|
på
|
ett
|
|
I
|
sat
|
naturally
|
not
|
wholly
|
passively
|
and
|
waited
|
on
|
a
|
|
samtal
|
(Sw.)
|
|
|
|
|
|
|
|
|
|
talk
|
|
|
|
|
|
|
|
|
|
|
‘Naturally, I wasn’t just passively waiting for a talk/call’
|
17The details of shared meaning elements across conjuncts are complex and cannot be treated fully here. But single-event-hood crucially involves temporality. Against a single-event meaning it might be objected that e.g. constructions with telic motion, as in (5), involve two consecutive subevents:
(5)
|
Jeg [...]
|
gikk
|
på
|
apoteket
|
og
|
fikk
|
utlevert
|
|
I
|
went
|
on
|
the.pharmacy
|
and
|
got
|
handed.out
|
|
sovetablettene
|
(No.)
|
|
|
|
|
|
|
the.sleeping.pills
|
|
|
|
|
|
|
|
‘I went to the pharmacy and got the sleeping pills’
|
18The event times in (5) are different ; going comes before getting. However, the topic times of V1 and V2 are and must be the same. If a temporal adverbial is inserted in V2 (e.g. deretter ‘thereafter’) and causes the topic time of V2 to be different from that of V1, the unity of the complex event is dissolved and the expression is no longer an instance of PC (as can be determined by applying well-known independent criteria for categorization as PC).
19In PVPC, the event times typically overlap, but Lødrup (2014b) gives examples of V2s with event times before and after the event time of V1. However, in such examples, too, the topic times of V1 and V2 are the same. (6) and (7) are from our materials:
(6)
|
Her
|
står
|
du
|
og
|
har
|
møtt
|
din
|
for
|
lengst
|
|
here
|
stand
|
you
|
and
|
have
|
met
|
your
|
for
|
longest
|
|
henfarne
|
bror
|
(No.)
|
|
|
|
|
|
|
|
deceased
|
brother
|
|
|
|
|
|
|
|
|
‘Here you are (standing), having met your long since deceased brother’
|
(7)
|
da
|
han
|
står
|
i
|
døren
|
og
|
skal
|
gå
|
(Da.)
|
|
when
|
he
|
stands
|
in
|
the.door
|
and
|
shall
|
go
|
|
|
‘when he’s (standing) in the door, about to leave’
|
20In (6), the topic time of V2 is the same as that of V1 ; the subject referent is said to be in the post-state resulting from the event (‘meeting ...’). In (7), the topic times are also identical, and the subject referent is said to be in the pre-state leading up to the event (‘leaving’). The descriptive details may vary depending on one’s preferred framework for tense and aspect, in particular the treatment of relative tense and viewpoint aspect (cf. discussion in Bohnemeyer 2014). V2 in (6) may be regarded as expressing present + anterior tense, while V2 in (7) exhibits present + posterior tense. Alternatively, V2 in (6) has present tense + retrospective viewpoint aspect, while V2 in (7) has present tense + prospective viewpoint aspect. Given an understanding in terms of retrospective/prospective aspect, PVPC may satisfy the criteria for the macroevent property as defined by Bohnemeyer et al. (2007), a property characteristic of constructions where complex events are construed as single events.
- 4 (8) appears to involve verb coordination and not verb phrase coordination; the stranded preposition (...)
21Another piece of evidence for single-event meaning is the fact that only the coordinators meaning ‘and’ can be used, not those meaning ‘or’ or ‘but’, since these latter require that one construes the subevents as alternatives or being in some kind of contrast, rather than forming a close-knit unit. Further, the ‘and’ coordinator cannot be combined with words meaning ‘both’ or ‘respectively’, since these too highlight complexity rather than unity (see also de Vos 2007). For example, (8) with både ‘both’ cannot be interpreted so that ‘we’ were simultaneously sitting and sleeping.4
(8)
|
en
|
seng[e]benk,
|
som
|
vi
|
både
|
satt
|
og
|
sov
|
i
|
(No.)
|
|
a
|
bed.bench
|
that
|
we
|
both
|
sat
|
and
|
slept
|
in
|
|
|
‘a “bed bench” that we both sat (on) and slept in’
|
22The exact delineation of the notion of ‘single event’ in relation to PCs will have to be deferred to future research ; for our present purposes, it seems justified to assume that the notion provides a fruitful basis for understanding PCs in general and PVPCs in particular.
23Another way to approach PC meaning is to investigate the semantic relation between the subevents referred to with V1 and V2. Lakoff (1986) notes that telic go and V in English denotes natural courses of events. Kvist Darnell (2008: 121ff.) argues that in Swedish PVPCs the V1 meaning forms a background for the V2 meaning. In most other PC subschemas (notably telic motion), the meaning of V1 is one of goal-directedness (målriktning) towards that of V2.
24Nielsen (2011) appears to be the first to propose a fully schematic V1–V2 meaning relation for PC. He argues that all PCs in Danish express a V1–V2 relation of possibilification (Danish muliggørelse). Possibilification is understood as a conventionally construed V1–V2 relation: The event of V1 is a (sufficient but not necessary) factor that contributes to making the event of V2 possible (Nielsen 2011: 107). Nielsen argues that in PVPC, the location of the agent holding the posture is crucial ; he uses the term situative about this subschema. This view resonates with Blensenius’s (2015, 2016) argument that locative meaning is of primary importance in PVPCs.
25Kinn (2018) argues that Nielsen’s (2011) notion of ‘possibilification’ is too strong and should be replaced by a weaker causal concept, the related notion of ‘facilitation’. This is especially true for PVPCs. In this section, we sketch the semantics of posture by drawing on FrameNet’s definition of the posture frame (https://framenet.icsi.berkeley.edu) and Holm’s (2013) findings on the use of PVs in Norwegian. The notion of ‘facilitation’ is then worked out in some detail and applied to PVPCs.
26To explicate the facilitating role of postures, we draw on notions independently developed in Frame Semantics, viz. those of the posture frame. Frames are built up of a number of core and non-core elements. We will refer to five of those listed on FrameNet for the posture frame: Agent, Location, Point_of_contact (written this way in FrameNet), Duration, and Time ; see Section 3.1.1.
27Newman (2002) discusses the prototypical and extended uses of posture verbs of lying, sitting, and standing. Lemmens (2002) is a study of the three central posture verbs in Dutch, and Jociuvienė (2008) investigates the metaphorical extensions of German posture verbs. Building on these works and on Hansen (1974) and Kvist Darnell (2008), Holm (2013) analyses the uses of the Norwegian posture verbs ligge, sitte, and stå. The explications in Sections 3.1.2–4 are based on Holm’s work, employing FrameNet terms. There may be some minor differences between Norwegian and the other Mainland Scandinavian languages (one is mentioned in Section 3.1.2), but the description based on Norwegian is largely valid for Danish and Swedish, too.
28The posture frame and the relevant frame elements are understood as follows (quotes from the FrameNet web page (accessed March 1, 2018) ; abbreviations spelled out):
29- Posture: “An Agent supports their body in a particular Location. The [lexical units] of the frame convey which body part is the Point_of_contact where the Agent is supported, what orientation the body is in, and some overall arrangement of the limbs (especially the legs) and the torso.”
-
Agent, a core element: “The Agent is the individual whose body is in a particular posture. The Agent [frame element] is generally expressed as the external argument of verbs” – that is, the subject.
-
Location, a core element: “A description of the position of the Agent. The Location of the Agent is frequently expressed and generally occurs as a PP Complement” ; we consider this element to be quite central to the meaning of PVPCs, as also argued by Blensenius (2015, 2016).
-
Point_of_contact, a core element: “The part of the body that provides support to the Agent’s body in the posture.”
-
Duration, a non-core element: “The length of time for which the Agent holds the posture.” The Duration does not have to be long, but a certain bodily stability over time is necessary for the posture to count as a posture.
-
Time, a non-core element: “This [frame element] indicates the Time interval during which the Agent held a certain posture.”
- 5 The reason that the numbers do not add to 100 % here and for the other PVs is that there were some (...)
30Holm (2013) is a corpus-based study of Norwegian PVs in all syntactic constructions. Holm finds (for 300 randomly chosen tokens) that the subject referents of Norwegian stå ‘stand’ distribute as follows: 52 % human, 21 % physical objects, 24 % abstract phenomena.5
31A standing Agent is prototypically extended vertically, her feet being the Point_of_contact. The Agent spends a fair amount of effort to stay up and in balance, but balance may also be maintained by leaning on something. The Agent’s head is in a high position, she has a good field of vision and is easily seen. She is able to exert force on things, and she is less vulnerable in this posture than in others. It is a good posture for speaking to many people and communicating verbally or bodily over some distance. A standing Agent is ready for movement and other actions. She tends to stand for fairly short Durations. She is clearly alive and conscious.
- 6 In Danish, one often uses the verb holde about parked vehicles, rather than stå.
32Physical objects that are said to stand are typically possible to move, and they tend to be vertically oriented oblong objects, including containers that open at the top end. Vehicles like cars also stand when they are parked.6 Certain abstract phenomena are also said to stand, especially texts with a view to being read.
33Holm (2013) finds (for 300 tokens) that the subject referents of Norwegian ligge ‘lie’ distribute as follows: 29 % human, 41 % physical objects, 27 % abstract phenomena. Note the large share of physical objects ; this is reflected in the distinctive collexemes of LIE in PC (see Section 5).
34When an Agent is lying, she is prototypically extended horizontally, the whole length of her body being the Point_of_contact. She does not need to spend any energy on balance. Her head is in a low position, she has a limited field of vision and may be hidden to others. She cannot exert much force, and she is quite vulnerable. She is typically resting rather than working, and she is not ready for movement. She tends to lie for extended Durations at a time. She may be asleep, unconscious or even dead.
35Physical objects that are said to lie are primarily (a) movable flat or long objects (rugs, sheets, ropes, pipes etc.) or many small distributed objects ; (b) locations like countries, towns, buildings, rooms ; (c) masses like fog, water, minerals etc. Various abstract phenomena may be said to be lying, including texts with a view to where they can be found.
36Holm (2013) finds (for 300 tokens) that the subject referents of Norwegian sitte ‘sit’ distribute as follows: 87 % human, 2 % physical objects, 3 % abstract phenomena. Note the strong dominance of human subjects.
37When an Agent is sitting, it is prototypically the case that her upper body is extended vertically while at least her upper legs are positioned horizontally, her buttocks and thighs being the Point_of_contact. She may spend some effort to balance her upper body, but balance is often maintained by leaning on something. Her head is at an intermediate height, and vision and visibility are also intermediate. So are the possibility of force exertion and vulnerability. She may be working with her hands or eating. She may be resting for a short time, often in the company of others. It is a good posture for talking with people more intimately. Durations of sitting tend to be longer than those of standing and shorter than many Durations of lying. The Agent is normally alive and conscious.
38Physical objects are not often said to sit – differently from Dutch, see Lemmens (2002) – ; if they are, they are typically not movable. For example, the roots of teeth sit in the jawbone. Abstract phenomena are not frequently said to be sitting ; internally experienced emotions, pain etc. may be said to be sitting somewhere in the body.
39As discussed above, the various PC subschemas in Mainland Scandinavian exhibit certain formal commonalities, which motivates their subsumption under one schema. Assuming that it is legitimate to recognize PC as a construction with a set of formal properties, the question arises what the meaning of the construction is. After all, it has become a standard approach in construction grammars to “define a construction as a form-meaning pairing” (Traugott & Trousdale 2013: 11). If a construction is recognized from the formal perspective, one expects the construction to have some identifiable meaning too – possibly very schematic. We have mentioned single-event meaning (see Section 2.3) as something that appears to hold for all subschemas of PC. But this seems somewhat meager. We explore the potential of the notion of ‘facilitation’ as a semantic characteristic of PVPC, thereby also attempting to clarify the semantics of Mainland Scandinavian PCs overall.
40‘Facilitation’ is a theoretical construct aimed at accounting for the commonalities of V1–V2 meaning relations found among different PC subschemas. As mentioned, the notion was introduced by Kinn (2018) in an analysis of PCs of atelic motion, as a weakened version of Nielsen’s (2011) ‘possibilification’. We conceive of facilitation as involving both family resemblances and prototypicality. In PVPC, V2 is (proto)typically facilitated by the Agent’s staying in a posture at a certain Location for a certain Duration at a certain Time (see further Section 3.3). (Note that it is not the posture alone that is the facilitating factor ; Location and Duration are central.) In PC of atelic motion, V2 is typically facilitated by the subject referent’s going from place to place ; in PC of telic motion, V2 is typically facilitated by the subject referent’s moving from one place to another ; in PC of communication, V2 is typically facilitated by the subject referent’s establishing a communicative channel for the mediation of a message ; etc. Thus, each PC subschema is assumed to involve a variant of facilitation. These variants may have prototype structure, but the extent to which particular expressions may deviate from the prototype varies among the subschemas. In PVPC, there is often considerable deviation, as we shall see. The subschematic meaning variants exhibit family resemblances and together constitute the meaning category of ‘facilitation’ at the schema level.
- 7 Our study does in some ways resemble that in Newman & Rice (2008) of English V and V constructions (...)
41A full explication of ‘facilitation’ and evaluation of its appropriateness for PC at the schema level should be based on empirical analyses of all known PC subschemas. Much research remains to be done in this field, and our present study of facilitation in relation to PCs aims to cover the largest uncharted territory, viz. that of PVPCs. Of necessity, then, the explication below cannot be definitive.7
42Facilitation is easiest to explain in relation to the subschema of telic motion. (9) exemplifies PC of telic motion ; ‘they’ move from indoors to outdoors, and the activity of V2 (clearing away snow) takes place outdoors. It is only by moving to the goal location that the subject referent can engage in the activity there. In this case, Nielsen’s ‘possibilification’ would also be a fitting description.
(9)
|
[Da]
|
får
|
de
|
selv
|
gå
|
ut
|
og
|
måke
|
(No.)
|
|
then
|
get
|
they
|
themselves
|
go
|
out
|
and
|
clear.away.snow
|
|
|
‘Then they’ll have to go out and clear away the snow themselves’
|
In this subschema, facilitation involves:
-
Location: The goal location of the telic motion of V1 and the location of the event of V2 are identical.
-
Time: The motion of V1 takes place before the event of V2 (the event times are sequential), but V1 and V2 must have the same topic time, as discussed in Section 2.3.
-
Motion: In the motion of V1, the subject referent takes herself from one place to another, the latter being the location of V2.
-
Intentionality: This is a presumptive meaning: If it is possible to assume that the subject referent moves as in V1 in order to be able to act as in V2, then that interpretation is chosen.
Nielsen (2011: 97–98) treats intentionality as a diagnostic of this and related subschemas, but the subject referent may actually be nonintentional, as in (10):
(10)
|
Enda
|
en
|
grein
|
har
|
falt
|
ned
|
og
|
blokkert
|
vannrenna
|
(No.)
|
|
yet
|
a
|
branch
|
has
|
fallen
|
down
|
and
|
blocked
|
the.trough
|
|
|
‘Another branch has fallen down and blocked the trough’
|
PC of atelic motion is illustrated in (11). Here, the bears’ motion from place to place helps them look for food:
(11)
|
Isbjørnene
|
rusler
|
rundt
|
på
|
isen
|
og
|
leter
|
etter
|
|
the.polar.bears
|
stroll
|
around
|
on
|
the.ice
|
and
|
search
|
after
|
|
mat
|
(No.)
|
|
|
|
|
|
|
|
food
|
|
|
|
|
|
|
|
|
‘The polar bears are strolling around on the ice looking for food’
|
43In examples like (11), facilitation amounts to the following: The distributed location is common to V1 and V2, the subevents have the same topic times, and the temporal durations of V1 and V2 are portrayed as identical. In addition, intentionality is assumed whenever possible.
44Not all PC subschemas involve facilitation that is based on the locative relation between V1 and V2. (12) exemplifies the subschema for PC of communication. Here, a central element is the establishment of a communicative channel in V1 which mediates the message as described in V2.
(12)
|
Den
|
unge
|
dommeren
|
skrev
|
og
|
ba
|
om
|
tillatelse
|
|
the
|
young
|
the.judge
|
wrote
|
and
|
asked
|
about
|
permission
|
|
til
|
å
|
trekke
|
seg
|
(No.)
|
|
|
|
|
to
|
to
|
draw
|
himself
|
|
|
|
|
|
‘The young judge wrote and asked for permission to withdraw’
|
45The PC in (13) illustrates yet another subschema, where V1 ascribes a state to the subject referent, and it is this state that paves the way for the event of V2. In this specific case, it is the subject referent’s good fortune that allows him to remain dressed.
(13)
|
Hamid
|
er
|
heldig
|
og
|
får
|
beholde
|
klærne
|
på
|
(No.)
|
|
Hamid
|
is
|
lucky
|
and
|
gets
|
keep
|
the.clothes
|
on
|
|
|
‘Hamid is lucky and gets to keep his clothes on’
|
46As we see, then, facilitation takes different forms in the various PC subschemas. This is a kind of variation within a category (PC in general) that we believe is best regarded as family resemblance. Section 3.3 addresses facilitation in PVPCs.
47To get an idea of how facilitation works in PVPC, consider (14).
(14)
|
Jag
|
har
|
mest
|
legat
|
på
|
balkongen
|
och
|
solat
|
(Sw.)
|
|
I
|
have
|
mostly
|
lain
|
on
|
the.balcony
|
and
|
sunbathed
|
|
|
‘I have mostly been sunbathing on the balcony’
|
48The speaker is the Agent of the posture and also the subject referent of V2. The Point_of_contact in lying is the whole body. The verb sola ‘sunbathe’ does not require the subject referent to be lying, but since the PV and V2 are simultaneous, the posture and its Point_of_contact are superimposed on the meaning of V2. The Locations of PV and V2 are one and the same (‘on the balcony’), and the same holds for Duration (indicated vaguely with mest ‘mostly’) and Time.
49Prototypical facilitation in PVPC thus consists of the following:
-
Location: V1 and V2 have identical Locations.
-
Time: V1 and V2 have identical topic times. This is shown in Section 2.3.
-
Duration: V1 and V2 are portrayed as having the same duration. This need not be long, but long enough for the posture to be regarded as a posture.
-
Posture: The posture is suitable for the V2 event (activity, state, etc.), i.e. better than other postures. This will be shown in Section 5.
-
Intentionality: The Agent (subject referent) has intentionally assumed the posture in order to perform the V2 activity. (This is merely a tendency and will not be discussed further.)
50In brief, V1 typically facilitates V2 through postural stability ; the Agent holds a posture in a certain Location for a certain Duration and performs the V2 activity there and then. Note again that several frame elements are involved, not just the posture itself. As far as we can tell, characteristics b) and c) are not subject to variation, whereas a), d), and e) are. The absence of characteristic d) most clearly removes the expression in question from the prototypical core of facilitation, while bleaching of characteristic a) appears to be needed for auxiliation of the PV to truly get under way. Auxiliation, then, involves a clear move away from the prototype of postural facilitation, eventually to a degree where there is no longer any construed facilitation.
51(14) is a clear example of the posture facilitating the V2 subevent, but (15) is a case where there is a much less obvious facilitative relation between posture and V2 event:
(15)
|
Vi
|
lå
|
længe
|
og
|
smilede
|
til
|
hinanden
|
(Da.)
|
|
we
|
lay
|
long
|
and
|
smiled
|
to
|
one.another
|
|
|
‘We lay for a long time, smiling at one another’
|
52‘We’ is the Agent and common subject referent of PV and V2. The posture and Point_of_contact are similar to what we saw for (14), but the smiling of V2 has no clear relation to these. The Location must be sought in the context (a bed, maybe) ; this does not seem pertinent to the smiling, either. The Duration of the posture is construed as long, and this is superimposed on the smiling ; the notion of smiling is malleable with respect to duration. There is one common Time. The specific posture and Location seem rather circumstantial to the smiling of V2. What is important here is the combination of posture, Location, and Duration. The stability of posture and Location described for (14) above is present here, too. The continued mutual smiling is construed as made easier by ‘our’ remaining there like that for some time ; indeed, without V1, V2 would more naturally be understood as referring to a brief smile. We see here the aspectual potential of PVPC.
53Thus, we do not claim that the posture of PVPCs is always such that it facilitates the V2 subevent. This aspect of facilitation is part of the typical meaning of the subschema, but is not evident for every single token. As observed by Nielsen (2011) in his discussion of possibilification, PVPC is the kind of PC least amenable to an understanding in terms of that notion. The same goes for facilitation as understood here, even though this notion involves a weaker causal relation in the combination of PV and V2. Showing that facilitation is typically at work even for PVPC therefore strengthens the view of it as central in PC in general.
54The posture frame elements of Time and Duration are always shared with V2, and so is normally Location. When PVs are used in other constructions than PC, the Location is typically overtly expressed. PVPCs are less strict in this respect. For instance, the example in (16) is fine without a specified Location, but just och han hade suttit would have sounded quite odd.
(16)
|
och
|
han
|
hade
|
suttit
|
och
|
talat
|
vid
|
den
|
sjuke
|
(Sw.)
|
|
and
|
he
|
had
|
sat
|
and
|
talked
|
by
|
the
|
sick
|
|
|
‘and he had been (sitting) talking with the sick man’
|
55As Holm (2013) observes (see Section 3.1), PV meanings (especially those of LIE) may be extended from prototypical human Agents to nonhuman Agents. Such extensions are also found in PVPC. These do not involve any radical changes to the subschema meaning, but there is bleaching of ‘posture’ towards ‘being located’, thus, the meaning of Location seems to become more prominent at the cost of posture.
56However, we will see in Section 5.3 that there are also cases where both the posture meaning and the Location meaning are weak or absent. This means that the Duration frame element becomes all the more salient, and this, we argue, is how certain subschemas of the PVPC subschema come to develop mostly aspectual meaning. Thus, the relative salience of the frame elements helps us explore the beginning aspectual auxiliation of PVs in PC, where the meanings of posture and Location are bleached. Kinn (2018) shows that the presence and complexity of atelicizing adverbials in Norwegian PCs of atelic motion are tied to bleaching of the motion V1: Short adverbials and (even more so) the absence of an adverbial tend to cooccur with the motion verb gå ‘go, walk’ used with a meaning approaching ‘general activity, existence’.
57Through a large-scale corpus study of the three Mainland Scandinavian languages and the use of distinctive collexeme analysis we will show that there is a very clear relation between the postures referred to with the PVs and the events of V2.
- 8 The corpora may not be fully comparable in terms of genres and text types. Since our study is not p (...)
58The empirical foundation of the present study is corpora of written Danish, Norwegian, and Swedish. For Danish, we used KorpusDK. This is a balanced corpus of just over 50 million words of written Danish from 1990–2000. For Norwegian, we used the Lexicographic Corpus for Norwegian Bokmål (LBK ; Knudsen & Fjeld 2013). This is a balanced corpus from 1985–2013 of about 100 million words. For Swedish, we used the Korp corpus tool (Borin et al. 2012) and included corpora intended to correspond in genres and size to the LBK, with a total of about 100 million words (non-fiction prose 45 %, fiction 35 %, newspapers 10 %, and miscellany 10 %).8
- 9 As observed by a reviewer, there might be differences in behaviour between the inflectional forms. (...)
- 10 Our study does not contrast expressions with differing amounts of linguistic material between V1 an (...)
59In each corpus, we searched for examples of LIE, SIT, and STAND followed by the coordinator meaning ‘and’ and a verb in the same inflectional form as the PV (or, in Norwegian, a V2 infinitive after a PV present participle, see below).9 We allowed for up to four words intervening between the PV and the coordinator but none between the coordinator and V2. In this way, we found examples with short subjects and/or adverbials between the PV and the coordinator, but not examples with longer intervening constituents. Thus, the searches did not find all examples of PVPC in the corpora, but the limitation was considered necessary in order to keep the amount of irrelevant hits tractable. Not allowing any words between the coordinator and V2 will have excluded only a small number of examples with an adverbial constituent in that position.10
60Examples with coordination at a higher syntactic level than the PV were not included. This involves two kinds of structures: (a) In Danish, in expressions where a supine PV is embedded under the perfect auxiliary have ‘have’, there may be a V2 requiring instead the auxiliary være ‘be’, and it is the auxiliaries that are coordinated in such cases. There were few such examples. (b) In Danish and Swedish, when a present participle PV is embedded under blive/bli ‘become, be’ (by far the most common use of the present participle in connection with PC), V2 appears to be coordinated with blive/bli and agrees with it. In Norwegian, the PV present participle in such cases exhibits a kind of quasi-agreement with a V2 infinitive. Therefore, Norwegian but not Danish and Swedish examples involving PV present participles were included. See further Kinn (2014, 2017) about constructions with present participles.
61The corpus hits were imported into a spreadsheet. For each PV in each language, we used the software to make a random selection of 3,000 hits, i.e. 27,000 hits in all. For each example, it was decided whether it was an instance of PVPC and should be included. (Examples with dynamic readings of PVs were excluded, e.g. Norwegian stå opp ‘stand up’.) All examples were examined by two researchers. Problematic cases were discussed by all three researchers before a final decision was made.
- 11 Some of the verbs are often or always used with reflexive objects (e.g. Norwegian vri seg ‘writhe’) (...)
62Further, all included examples were categorized for their V2, labelled with the infinitive form and distinguishing between homonymous verbs (e.g. Norwegian være ‘be’ vs. være ‘scent’).11 For polysemous verbs, we decided not to separate meaning nuances, since it would have been extremely difficult to decide consistently on cut-off points for such a varied inventory of verbs in three languages. We considered hidden polysemies to be less problematic than trying to differentiate meaning variants.
63To analyse the relations between individual PVs and V2s, we performed distinctive collexeme analyses, see Gries & Stefanowitsch (2004). This method let us compare the postures by finding V2s that are significantly attracted to each PV relative to the other two. (The same method has been used by Lemmens (2017) to reveal characteristics of Dutch liggen/zitten/staan te Vinfinitive (lit.: ‘lie/sit/stand to V’) compared to aan het Vinfinitive zijn (lit.: ‘be at the V’), both aspectual constructions.) For instance, we contrast Danish PCs with ligge ‘lie’ with those with sidde ‘sit’ and stå ‘stand’ (pooled) – with respect to the V2s that they cooccur with. This is done for each PV and each language separately, so that, for all nine language–PV pairs, we identify V2s that tend to be used with the respective PV. To perform the calculations, we used Stefan Th. Gries’s R script for collostructional analysis (Gries 2014).
64Tables 2–4 below list 183 V2s, viz. the (at least) 20 most strongly attracted distinctive collexemes for each of the PVs LIE, SIT, and STAND in each of the languages Danish, Norwegian, and Swedish. Perusals of these lists give a rough idea of which activities etc. are facilitated by lying, sitting, and standing, respectively. However, a more systematic treatment of these inventories was needed to bring out the picture more clearly. The cut-off point at rank 20 is necessarily arbitrary, and including lower-ranked V2s would of course have changed the details of the picture. But the less distinctive the collexemes are, the less relevant they must be assumed to be for the interpretation of the V1–V2 relation for the individual postures.
65To group the attracted V2s, we started by using VerbNet’s classification scheme for English verbs, which can be accessed from the Unified Verb Index web page (verbs.colorado.edu/verb-index). VerbNet’s classes are based on those of Levin (1993) and extended with a number of classes from Korhonen & Briscoe (2004) ; see Kipper et al. (2006). The Scandinavian verbs were classified as their closest English equivalents could have been. The VerbNet scheme was very helpful, but having completed the classification of the 183 verbs, we decided that it was necessary to adapt the result to our purposes. For instance, while verbs meaning ‘read’ and ‘write’ are in quite separate VerbNet classes, we considered it plausible to group such verbs together in a class of ‘written communication’. Thus, our verb classes do not correspond directly to the VerbNet classes, but were “tailor-made” for our analyses.
66After exclusion of irrelevant corpus hits, the total number of examples of PVPC was 23,250, distributed over languages and PVs as shown in Table 1.
Table 1. Number of examples of PVPC in the corpus study
|
lie
|
sit
|
stand
|
All verbs
|
Danish
|
2,426
|
2,854
|
2,274
|
7,554
|
Norwegian
|
2,535
|
2,810
|
2,605
|
7,950
|
Swedish
|
2,364
|
2,782
|
2,600
|
7,746
|
All languages
|
7,325
|
8,446
|
7,479
|
23,250
|
- 12 For each PV, the estimated corpus percentage was calculated as (h × e × 100) / (c × 3000), where h (...)
67We estimated the shares of PC for the individual PVs in Norwegian.12 For ligge ‘lie’, PCs account for about 7.6 % of all uses of the PV ; for stå ‘stand’, the share is 12.9 % ; and for sitte ‘sit’, as large as 22.1 %. These numbers indicate that PC is a fairly frequent construction for these verbs, especially the most human-specific one, viz. sitte (see Section 3.1.4). For Dutch, Lemmens (2005: 188) found similar constructions (liggen/zitten/staan te Vinfinitive, lit. ‘lie/sit/stand to V’) to account for much smaller shares: for liggen ‘lie’, 1.1 % ; for staan ‘stand’, 1.8 % ; and for zitten ‘sit’, 2.4 %. In both languages, the constructions in question make up the greatest share for ‘sit’ and the smallest for ‘lie’.
68Below, we first present quantitative results summarily in Section 5.1. The main part is Section 5.2, where we discuss semantic classes of V2s that appear as distinctive collexemes of LIE, SIT, and STAND, in the light of the notion of facilitation. Section 5.3 addresses some indications of PV bleaching and grammaticalization in its early stages as well as (briefly) lexicalization of the ‘V1 and V2’ complex.
69This section contains listings of the 20 most strongly attracted distinctive collexeme V2s for the nine language–PV pairs, with V2 frequencies and collostructional strengths. In some cases, more than 20 verbs are listed because of tied collostructional strengths. The next section discusses these findings based on the semantic classification of V2s.
70Table 2 shows the 20 most highly ranked distinctive V2 collexemes of the PV LIE for each of Danish, Norwegian, and Swedish, i.e. as distinguished from SIT and STAND. The most frequent V2s after LIE are shown with absolute frequencies in Table 9 in the Appendix ; not all frequent V2s are distinctively attracted (some are even repulsed). Tables 3 and 4 below and Tables 10 and 11 in the Appendix show corresponding results for SIT and STAND.
Table 2. Most distinctive collexemes of lie. CS = collostructional strength measured by log-likelihood.
Danish
|
|
Norwegian
|
|
Swedish
|
V2
|
Meaning
|
N
|
CS
|
|
V2
|
Meaning
|
N
|
CS
|
|
V2
|
Meaning
|
N
|
CS
|
sove
|
‘sleep’
|
197
|
313.46
|
|
sove
|
‘sleep’
|
239
|
485.55
|
|
sova
|
‘sleep’
|
226
|
431.33
|
flyde
|
‘float’
|
84
|
175.33
|
|
flyte
|
‘float’
|
48
|
110.35
|
|
sola
|
‘sunbathe’
|
30
|
53.55
|
lure
|
‘lurk’
|
41
|
48.36
|
|
vri
|
‘writhe’
|
44
|
65.61
|
|
skräpa
|
‘make a mess’
|
19
|
45.22
|
ulme
|
‘smolder’
|
16
|
36.42
|
|
slenge
|
‘make a mess’
|
22
|
50.42
|
|
dö
|
‘die’
|
23
|
42.27
|
rode
|
‘make a mess’
|
37
|
31.77
|
|
kave
|
‘flounder, struggle’
|
20
|
45.83
|
|
gro
|
‘grow’
|
15
|
35.68
|
dø
|
‘die’
|
20
|
30.04
|
|
snorke
|
‘snore’
|
18
|
41.23
|
|
snarka
|
‘snore’
|
14
|
33.30
|
vente
|
‘wait’
|
200
|
23.98
|
|
duppe
|
‘float, bob’
|
20
|
33.95
|
|
vila
|
‘rest’
|
29
|
31.91
|
vende
|
‘turn’
|
16
|
22.16
|
|
ulme
|
‘smolder’
|
14
|
32.06
|
|
vrida
|
‘writhe’
|
22
|
27.62
|
vride
|
‘writhe’
|
20
|
22.07
|
|
sole
|
‘sunbathe’
|
16
|
29.80
|
|
flyta
|
‘float’
|
11
|
26.15
|
dase
|
‘doze’
|
9
|
20.47
|
|
hvile
|
‘rest’
|
24
|
29.57
|
|
ruttna
|
‘rot’
|
14
|
22.69
|
lyse
|
‘light’
|
17
|
18.97
|
|
døse
|
‘doze’
|
19
|
25.33
|
|
guppa
|
‘float, bob’
|
11
|
19.99
|
være
|
‘be’
|
87
|
18.22
|
|
tenke
|
‘think, ponder’
|
78
|
21.80
|
|
pyra
|
‘smolder’
|
8
|
19.01
|
snorke
|
‘snore’
|
13
|
14.11
|
|
skyte
|
‘shoot’
|
9
|
20.60
|
|
skava
|
‘chafe’
|
8
|
19.01
|
hulke
|
‘sob’
|
6
|
13.64
|
|
råtne
|
‘rot’
|
15
|
17.85
|
|
be
|
‘pray’
|
12
|
13.44
|
klynke
|
‘whimper’
|
6
|
13.64
|
|
gløde
|
‘glow’
|
7
|
16.01
|
|
ha
|
‘have’
|
38
|
13.34
|
slange
|
‘be comfortable’
|
6
|
13.64
|
|
tørke
|
‘dry’
|
16
|
15.45
|
|
lura
|
‘lurk’
|
14
|
13.21
|
gispe
|
‘gasp’
|
10
|
13.48
|
|
slumre
|
‘doze’
|
9
|
14.86
|
|
gnaga
|
‘gnaw’
|
5
|
11.88
|
sole
|
‘sunbathe’
|
11
|
12.79
|
|
sprelle
|
‘wriggle’
|
9
|
14.86
|
|
sussa
|
‘sleep’
|
5
|
11.88
|
sprælle
|
‘wriggle’
|
8
|
12.68
|
|
lytte
|
‘listen’
|
86
|
14.72
|
|
gosa
|
‘be comfortable’
|
7
|
11.33
|
svømme
|
‘swim’
|
8
|
12.68
|
|
dirre
|
‘tremble’
|
6
|
13.73
|
|
trycka
|
‘lie still, press’
|
25
|
10.48
|
|
|
|
|
|
dø
|
‘die’
|
6
|
13.73
|
|
|
|
|
|
|
|
|
|
|
glitre
|
‘glitter’
|
6
|
13.73
|
|
|
|
|
|
Table 3. Most distinctive collexemes of sit. CS = collostructional strength measured by log-likelihood.
Danish
|
|
Norwegian
|
|
Swedish
|
V2
|
Meaning
|
N
|
CS
|
|
V2
|
Meaning
|
N
|
CS
|
|
V2
|
Meaning
|
N
|
CS
|
spise
|
‘eat’
|
74
|
112.64
|
|
skrive
|
‘write’
|
76
|
126.16
|
|
dricka
|
‘drink’
|
72
|
107.46
|
skrive
|
‘write’
|
57
|
102.51
|
|
spise
|
‘eat’
|
79
|
105.04
|
|
skriva
|
‘write’
|
64
|
82.14
|
drikke
|
‘drink’
|
68
|
66.51
|
|
drikke
|
‘drink’
|
74
|
70.66
|
|
äta
|
‘eat’
|
62
|
62.70
|
læse
|
‘read’
|
97
|
41.88
|
|
lese
|
‘read’
|
80
|
41.84
|
|
prata
|
‘talk’
|
102
|
30.45
|
snakke
|
‘talk’
|
85
|
31.60
|
|
sy
|
‘sew’
|
14
|
29.16
|
|
läsa
|
‘read’
|
94
|
29.27
|
fortælle
|
‘tell’
|
29
|
27.76
|
|
jobbe
|
‘work’
|
13
|
27.08
|
|
tiga
|
‘be silent’
|
11
|
16.56
|
tegne
|
‘draw’
|
18
|
24.00
|
|
snakke
|
‘talk’
|
80
|
24.10
|
|
snacka
|
‘talk’
|
28
|
16.43
|
sy
|
‘sew’
|
11
|
21.44
|
|
prate
|
‘talk’
|
38
|
21.93
|
|
röka
|
‘smoke’
|
34
|
15.29
|
strikke
|
‘knit’
|
14
|
20.89
|
|
bla
|
‘leaf’
|
27
|
21.81
|
|
samtala
|
‘converse’
|
11
|
13.17
|
spille
|
‘play’
|
33
|
17.64
|
|
spille
|
‘play’
|
28
|
21.55
|
|
peta
|
‘pick’
|
9
|
12.83
|
bladre
|
‘leaf’
|
8
|
15.59
|
|
kjøre
|
‘drive’
|
11
|
16.89
|
|
plugga
|
‘study’
|
6
|
12.29
|
sludre
|
‘chat’
|
10
|
13.73
|
|
tegne
|
‘draw’
|
11
|
16.89
|
|
berätta
|
‘tell’
|
17
|
12.11
|
diskutere
|
‘discuss’
|
19
|
13.39
|
|
studere
|
‘study’
|
23
|
14.69
|
|
spela
|
‘play’
|
28
|
11.85
|
ryge
|
‘smoke’
|
21
|
11.43
|
|
notere
|
‘take notes’
|
6
|
12.49
|
|
knapra
|
‘chew’
|
5
|
10.24
|
nikke
|
‘nod’
|
12
|
11.22
|
|
tvinne
|
‘twist, twiddle’
|
6
|
12.49
|
|
sy
|
‘sew’
|
5
|
10.24
|
klimpre
|
‘plunk’
|
5
|
9.74
|
|
nippe
|
‘sip’
|
10
|
11.75
|
|
åka
|
‘drive’
|
5
|
10.24
|
nippe
|
‘sip’
|
9
|
9.00
|
|
dingle
|
‘dangle’
|
5
|
10.41
|
|
tänka
|
‘think, ponder’
|
74
|
10.12
|
falde
|
‘fall’
|
7
|
8.56
|
|
strikke
|
‘knit’
|
5
|
10.41
|
|
stirra
|
‘stare’
|
95
|
9.98
|
forhandle
|
‘negotiate’
|
7
|
8.56
|
|
kjede
|
‘be bored’
|
7
|
9.41
|
|
rita
|
‘draw’
|
10
|
9.12
|
styre
|
‘steer, govern’
|
7
|
8.56
|
|
arbeide
|
‘work’
|
10
|
9.39
|
|
diskutera
|
‘discuss’
|
13
|
8.28
|
Danish falde ‘fall’ is here used in expressions like falde i søvn ‘fall asleep’, not about literal falling.
Table 4. Most distinctive collexemes of stand. CS = collostructional strength measured by log-likelihood
Danish
|
|
Norwegian
|
|
Swedish
|
V2
|
Meaning
|
N
|
CS
|
|
V2
|
Meaning
|
N
|
CS
|
|
V2
|
Meaning
|
N
|
CS
|
mangle
|
‘lack’
|
39
|
85.47
|
|
se
|
‘look, see’
|
505
|
135.54
|
|
se
|
‘look, see’
|
215
|
40.68
|
sige
|
‘say’
|
72
|
65.24
|
|
betrakte
|
‘watch’
|
62
|
35.07
|
|
stampa
|
‘stomp’
|
19
|
29.99
|
skulle
|
‘be going to’
|
40
|
44.62
|
|
kikke
|
‘look’
|
61
|
30.64
|
|
vinka
|
‘wave’
|
19
|
29.99
|
trippe
|
‘trip, tiptoe’
|
14
|
33.68
|
|
svaie
|
‘sway’
|
16
|
28.95
|
|
betrakta
|
‘watch’
|
49
|
29.25
|
vinke
|
‘wave’
|
23
|
30.08
|
|
henge
|
‘loiter, hang’
|
19
|
27.33
|
|
säga
|
‘say’
|
40
|
26.18
|
læne
|
‘lean’
|
18
|
25.30
|
|
trippe
|
‘trip, tiptoe’
|
16
|
24.79
|
|
hänga
|
‘loiter, hang’
|
29
|
25.40
|
sælge
|
‘sell’
|
10
|
24.04
|
|
peke
|
‘point’
|
11
|
18.48
|
|
väga
|
‘balance’
|
20
|
25.39
|
vaske
|
‘wash’
|
12
|
22.52
|
|
slå
|
‘beat’
|
14
|
15.38
|
|
vifta
|
‘wave’
|
16
|
24.06
|
kigge
|
‘look’
|
118
|
19.29
|
|
rope
|
‘yell’
|
25
|
14.85
|
|
skrika
|
‘yell’
|
31
|
16.93
|
se
|
‘look, see’
|
132
|
16.93
|
|
lene
|
‘lean’
|
16
|
14.73
|
|
ropa
|
‘yell’
|
20
|
16.77
|
svaje
|
‘sway’
|
7
|
16.82
|
|
si
|
‘say’
|
23
|
13.43
|
|
laga
|
‘repair’
|
10
|
15.97
|
spørge
|
‘ask’
|
9
|
15.85
|
|
stampe
|
‘stomp’
|
6
|
13.40
|
|
peka
|
‘point’
|
13
|
15.43
|
betragte
|
‘watch’
|
42
|
15.35
|
|
vinke
|
‘wave’
|
14
|
13.34
|
|
vänta
|
‘wait’
|
188
|
15.41
|
glo
|
‘stare’
|
39
|
15.00
|
|
holde
|
‘hold’
|
58
|
12.64
|
|
dansa
|
‘dance’
|
7
|
15.29
|
pege
|
‘point’
|
11
|
14.04
|
|
nøle
|
‘hesitate’
|
8
|
12.38
|
|
tveka
|
‘hesitate’
|
7
|
15.29
|
tage
|
‘take’
|
29
|
13.08
|
|
vaske
|
‘wash’
|
8
|
12.38
|
|
spana
|
‘look, watch’
|
14
|
14.80
|
pisse
|
‘piss’
|
5
|
12.01
|
|
vifte
|
‘wave’
|
8
|
12.38
|
|
sälja
|
‘sell’
|
9
|
13.98
|
ruske
|
‘shake (sth.)’
|
5
|
12.01
|
|
ta
|
‘take’
|
37
|
11.37
|
|
välja
|
‘choose’
|
9
|
13.98
|
råbe
|
‘yell’
|
21
|
11.33
|
|
male
|
‘paint’
|
9
|
11.26
|
|
le
|
‘smile’
|
19
|
13.73
|
tale
|
‘speak’
|
44
|
11.15
|
|
finne
|
‘find’
|
5
|
11.16
|
|
prova
|
‘try on/out’
|
6
|
13.11
|
|
|
|
|
|
|
|
|
|
|
sikta
|
‘aim’
|
6
|
13.11
|
71Our verb groupings (cf. Section 4) are shown in Tables 5–7 below ; we have divided this overview into several tables in order to make it more accessible. Table 5 first presents verbs denoting change of state, emission of energy (warmth or light), sound emission (mostly non- or semicommunicative human sounds), and (physical and psychological) abrasion. These classes are followed by verbs for various kinds of inactivity: rest and pleasurable relaxation, sleep, disorganization (meanings like ‘be/make a mess’), floating in or on water, waiting, and being supported in some way additional to the support afforded by the posture.
72Table 6 contains verbs of cognition, communication, and perception. The verbs of communication are subclassified into several groups. First come those that involve writing and reading, and then those of speaking. From the latter group we have extracted those that denote loud use of voice (for reasons to be made clear below) and one case of non-speaking. Finally there are verbs of nonverbal communication, above all pointing and waving.
73Table 7 contains verbs of consumption (eating, drinking, smoking), general work, various uses of hands (e.g. knitting and drawing), transfer (e.g. selling), and motion. Non-directed motions involve those where the Agent moves on the spot, while the directed ones combine posture with movement from one place to another. Finally, the table includes two residual groups of “other” activities and states.
Table 5. Classification of top-ranked distinctive V2 collexemes. Part 1: verbs with meanings of change of state, energy or sound emission, abrasion, and various kinds of inactivity. For verb meanings, see Tables 2–4.
|
lie
|
sit
|
stand
|
Change of state
|
Da. dø ; No. dø, råtne, tørke ; Sw. dö, gro, ruttna
|
|
|
Energy emission
|
Da. lyse, ulme ; No. glitre, gløde, ulme ; Sw. pyra
|
|
|
Sound emission
|
Da. gispe, hulke, klynke, snorke ; No. snorke ; Sw. snarka
|
|
|
Abrasion
|
Sw. gnaga, skava
|
|
|
Inactivity
– rest
|
Da. slange, sole ; No. hvile, sole ; Sw. gosa, sola, vila
|
|
|
– sleep
|
Da. dase, sove ; No. døse, slumre, sove ; Sw. sova, sussa
|
|
|
– disorganized
|
No. slenge ; Sw. skräpa
|
|
|
– on water
|
Da. flyde ; No. duppe, flyte ; Sw. flyta, guppa
|
|
|
– waiting
|
Da. lure, vente ; Sw. lura, trycka
|
|
No. nøle ; Sw. tveka, vänta
|
– support
|
|
|
Da. læne ; No. henge, lene ; Sw. hänga
|
74The first four verb classes in Table 5 are found only for LIE, and such meanings thus distinguish this posture from SIT and STAND: change of state, energy emission, sound emission, and abrasion. Most of these are nonagentive. Changes of state may involve positive development (e.g. Swedish gro ‘grow’), but verbs meaning ‘rot’ and ‘die’, denoting decay and death, are more typical. Such processes are clearly associated with lying, which does not require physical effort. Verbs of energy emission usually have inanimate subject referents, referring to some object or mass giving off or reflecting warmth or light, e.g. embers or snow. Verbs of sound emission, on the other hand, typically have human subject referents. The clearly most central sound is snoring, naturally related to sleep and the lying posture. Other verbs, represented in Danish, involve sounds of sorrow and pain (e.g. sobbing), which are partly involuntary but border onto communication. Verbs of abrasion are represented only in Swedish (e.g. skava ‘chafe’). These may have intentional or inanimate subject referents, but are often extended metaphorically to unpleasant thoughts. Thus, they approach the field of cognition, but with the stimulus as subject referent, rather than the experiencer.
75Not surprisingly, LIE attracts many verbs of inactivity. The classes involving rest and sleep include verbs with mostly human subject referents, and the facilitating role of the posture in such cases is evident: The Agent will naturally have assumed the lying posture in order to rest or sleep well. The class of disorganized inactivity mostly denotes inanimate subject referents lying about and in that sense resembles the classes of change of state and of energy emission. Disorganization is not strictly facilitated by lying ; rather, the alternative postures of sitting and standing require an amount of organization that clashes with the meanings of disorganization verbs. Verbs of inactivity on water (or in the surface) naturally cooccur with LIE rather than SIT and STAND.
76Verbs of waiting are represented as distinctive for both LIE and STAND. The verb meaning ‘wait’ (Danish/Norwegian vente, Swedish vänta) is actually in first, second or third place in the frequency lists for all language–PV pairs (see Tables 8–10 in the Appendix). We have classified waiting as inactivity, which is generally best facilitated by lying. But waiting also involves a readiness for (re)action when the awaited object or event occurs. This aspect of waiting is best facilitated by standing, since that posture allows the Agent to (re)act quickly in a variety of ways. Finally, inactivity involving support is represented with distinctive verb collexemes only for STAND. This, too, is quite as expected since the standing posture requires some amount of physical effort to maintain, more so than sitting and especially lying. In such cases, there is so to speak reciprocal facilitation between V1 and V2: Some degree of vertical posture is required for leaning to be possible, and leaning makes it easier to remain in that posture.
77No verbs from the classes collected in Table 5 are distinctive for SIT. Sitting, as will become apparent below, is strongly associated with human activity.
78It should be noted that all the listed distinctive collexeme V2s that are typically used with nonhuman subjects are found in Table 5, and all are distinctive for LIE. These are: all the verbs for change of state (except Danish/Norwegian dø ‘die’ and Swedish dö ‘die’), for energy emission, for abrasion, for disorganized inactivity, and to some extent those for inactivity on water. All other distinctive V2s in Tables 5–7, including all those for SIT and STAND, go well with human subjects. Recall from Section 3.1 that Holm (2013) found Norwegian ligge ‘lie’ to be much more frequently used with subjects referring to physical objects than are sitte ‘sit’ and stå ‘stand’ ; this corresponds well with our findings here.
Table 6. Classification of top-ranked distinctive V2 collexemes. Part 2: verbs with meanings of cognition, communication, and perception. For verb meanings, see Tables 2–4.
|
lie
|
sit
|
stand
|
Cognition*
|
No. tenke
|
Da. falde ; No. kjede ; Sw. tänka
|
No. finne
|
Communication
– written
|
|
Da. læse, skrive ; No. lese, notere, skrive ; Sw. läsa, plugga, skriva
|
|
– spoken
|
Sw. be
|
Da. diskutere, forhandle, fortælle, snakke, sludre ; No. prate, snakke ; Sw. berätta, diskutera, prata, samtala, snacka
|
Da. sige, spørge, tale ; No. si ; Sw. säga
|
– loud
|
|
|
Da. råbe ; No. rope ; Sw. ropa, skrika
|
– absent
|
|
Sw. tiga
|
|
– nonverbal
|
|
Da. nikke
|
Da. pege, vinke ; No. peke, vifte, vinke ; Sw. le, peka, vifta, vinka
|
Perception
|
No. lytte
|
No. studere ; Sw. stirra
|
Da. betragte, glo, kigge, se ; No. betrakte, kikke, se ; Sw. betrakta, se, spana
|
79Verbs of cognition (see Table 6) are not very prominent among the verbs distinguishing between the postures, but there is a certain tendency for such verbs to occur with SIT. Verbs of communication, on the other hand, are abundant. Except for the appearance of be ‘pray’ with LIE in Danish, all the communication verbs in Table 6 are distinctive of either SIT or STAND. Only SIT has distinctive collexemes involving reading and writing, and the facilitating role of the posture is evident (although reading in bed is also quite common in our materials). Turning to spoken communication (including loud and absent communication), the distinctive verbs are distributed over SIT and STAND, but not quite randomly. While all the verb meanings are quite compatible with either posture, the verbs that are distinctive for SIT denote modes of communication where the interlocutors are close to each other and generally speak quietly. Verbs denoting loud use of voice are distinctive only of STAND. Standing is clearly facilitative of communication at a distance, where loudness is needed and the upright position allows the voice to carry further and makes it easier for the interlocutors to see each other ; recall Holm’s (2013) analysis from Section 3.1. Interestingly, the verb for ‘say’ is distinctive of STAND in all three languages (Danish sige, Norwegian si, Swedish säga). This denotes one-way communication and appears not to be associated with the closeness between interlocutors typical of SIT. The verbs for nonverbal communication, too, tend to reflect a contrast in distance: Waving and pointing function at a distance, while nodding works better for interlocutors close to each other. Swedish le ‘smile’ is a counterexample to this tendency.
80Distinctive verbs of perception primarily denote vision. The one exception is lytte ‘listen’, which appears for LIE in Norwegian. The vision verbs are all distinctive of either SIT or STAND, with a majority clustering with the latter. The observation made for loud communication and nonverbal communication involving easily seen body movements is similarly valid here, too: The more upright the posture, the better the view. Thus, STAND generally facilitates vision the best, SIT somewhat less, and LIE clearly the least, cf. Section 3.1.
Table 7. Classification of top-ranked distinctive V2 collexemes. Part 3: verbs with meanings of consumption, work, use of hands, transfer, motion, and other activities and states. For verb meanings, see Tables 2–4.
|
lie
|
sit
|
stand
|
Consumption
|
|
Da. drikke, nippe, ryge, spise ; No. drikke, nippe, spise ; Sw. dricka, knapra, röka, äta
|
|
General work
|
|
No. arbeide, jobbe
|
|
Use of hands
|
|
Da. bladre, klimpre, spille, strikke, sy, tegne ; No. bla, spille, strikke, sy, tegne, tvinne ; Sw. peta, rita, spela, sy
|
Da. ruske, vaske ; No. holde, male, slå, vaske ; Sw. laga
|
Transfer
|
|
|
Da. sælge, tage ; No. ta ; Sw. sälja, välja
|
Motion
– non-directed
|
Da. sprælle, vende, vride ; No. dirre, kave, sprelle, vri ; Sw. vrida
|
No. dingle
|
Da. svaje, trippe ; No. stampe, svaie, trippe ; Sw. dansa, stampa, väga
|
– directed
|
Da. svømme
|
No. kjøre ; Sw. åka
|
|
Other activity
|
Da. rode ; No. skyte
|
Da. styre
|
Da. pisse ; Sw. prova, sikta
|
Other state
|
Da. være ; Sw. ha
|
|
Da. mangle, skulle
|
81All distinctive verbs of consumption (see Table 7) appear with SIT. For eating and drinking, the facilitating role of the posture is obvious ; see Section 3.1.4. This is less clear for the verb for ‘smoke’, but it appears with SIT both in Danish (ryge) and Swedish (röka).
82The fact observed above that SIT is associated with activity is shown by the verbs of consumption, but even more so by verbs meaning ‘work’ and those that specify an activity based on the use of hands in a certain limited space: in one’s lap or close to the front of the upper body. Examples include handiwork like knitting and sewing as well as drawing and playing an instrument or a card or board game, and some verbs for idle activities like twiddling one’s thumbs. No such verbs are found for LIE, and the verbs found for STAND are less restricted with respect to the space of hand activity than those found for SIT. A few verbs for transfer such as taking and selling are also found to be distinctive of STAND rather than LIE or SIT.
83We have divided distinctive verbs of motion into those denoting non-directed and directed motion, respectively. The non-directed ones found for LIE tend to denote restlessness or struggle to (re)gain body control, while those found for STAND denote motions related to (losing, keeping, regaining) balance (cf. Section 3.1.2) and to foot movement on the spot. Only one distinctive verb turns up for SIT, namely Norwegian dingle ‘dangle (one’s legs)’. All of these verb meanings of non-directed motion can be seen to be clearly tied to the respective postures, i.e. facilitated.
84In general, one would not expect verbs of directed motion to be distinctive of postures. But swimming (cf. Danish svømme) is combinable with an approximately horizontal body orientation in the water surface, and driving (a car ; cf. Norwegian kjøre, Swedish åka) is done while sitting.
85There are certain indications of beginning PV bleaching and grammaticalization in our materials, as could be expected given the amount of attention the aspectual properties of PVPCs have received in the research literature (cf. Section 2.1). To see how bleaching works, recall two central posture frame elements in addition to the posture itself: Location and Duration. It was shown in Section 5.2 that LIE has a number of distinctive collexeme V2s whose subject referents are typically nonhuman and nonsentient, with a physical organization far removed from the human posture, e.g. like ‘snow’ in (17) and ‘rubbish’ in (18):
(17)
|
Sneen
|
lå
|
i
|
driver
|
og
|
lyste
|
i
|
måneskinnet
|
(Da.)
|
|
the.snow
|
lay
|
in
|
drifts
|
and
|
lighted
|
in
|
the.moon.light
|
|
|
‘The snow lay in drifts, gleaming in the moonlight’
|
(18)
|
Så
|
trött
|
på
|
allt
|
skit
|
som
|
ligger
|
där
|
och
|
skräpar
|
(Sw.)
|
|
so
|
tired
|
on
|
all
|
shit
|
that
|
lies
|
there
|
and
|
is.messy
|
|
|
‘So tired of all the rubbish lying about in a mess’
|
86In such cases, Location and a certain Duration is what is crucially left of the original posture frame. Although the posture meaning may be regarded as bleached in the direction of mere ‘existence’ and ‘location’, there is little evidence of grammaticalization towards a primarily aspectual meaning in this usage. Thus, bleaching of posture is not sufficient for aspectualization.
- 13 The equivalents of stå og skulle are rather common in Norwegian and Swedish, too, but not those of (...)
87There are some other cases, however, where the beginnings of aspectual grammaticalization may be gleaned. As shown in Table 7 above, mangle ‘lack’ and skulle ‘shall, be about to’ appear as distinctive collexemes of STÅ in Danish. Most of the examples with stå og mangle ‘stand and lack’ and some with stå og skulle ‘stand and be about to’ have a strongly bleached meaning of Location, but the meaning of Duration remains.13
88An example with stå og skulle is given in (7) above (‘when he’s (standing) in the door, about to leave’). In that case, there is an actual standing posture. But (19) shows an example referring to a general life situation, of being ready for change:
(19)
|
da
|
jeg
|
i
|
1973
|
stod
|
og
|
skulle
|
ind
|
i
|
|
when
|
I
|
in
|
1973
|
stood
|
and
|
should
|
into
|
in
|
|
ordenspolitiet
|
(Da.)
|
|
|
|
|
|
|
|
|
the.uniformed.police
|
|
|
|
|
|
|
|
|
|
‘when in 1973 I was about to join the uniformed police’
|
89There is no posture in this example, nor a specific Location, but the frame element of Duration remains. Such examples may be regarded as evidence of incipient grammaticalization of stå ‘stand’ in the direction of an aspectual marker.
90Skulle indicates intention to act and therefore typically has a human subject referent. We have observed (Section 3.1.2) that readiness for action is tied to the standing posture, and in that light the use of stå for situations of ‘being about to’ seems motivated. Readiness for action is arguably also involved in cases with Danish stå og mangle ‘stand and lack’. The subject referent is normally human (including organizations and the like) and cannot do what she wants because of lack of something. An example is given in (20).
(20)
|
og
|
han
|
stod
|
og
|
manglede
|
en
|
lærling
|
(Da.)
|
|
and
|
he
|
stood
|
and
|
lacked
|
an
|
apprentice
|
|
|
‘and he was in lack of an apprentice’
|
91The actual posture of the Agent in such examples is rather irrelevant, and although humans necessarily have a Location, this too is of limited importance. The example refers to a situation of some construed Duration with the lack of an apprentice hampering the activities of the subject referent.
92Such examples are not limited to stå og mangle/skulle ; (21) is a Danish example with ligge ‘lie’ ; V2 rode refers to fumbling and unsuccessful activity. The described event is one of a general life situation, not actual lying.
(21)
|
Forældrene
|
må
|
så
|
selv
|
ligge
|
og
|
rode
|
med
|
|
the.parents
|
must
|
then
|
themselves
|
lie
|
and
|
mess
|
with
|
|
deres
|
vanskeligheder
|
(Da.)
|
|
|
|
|
|
|
their
|
difficulties
|
|
|
|
|
|
|
|
‘The parents must then struggle with their difficulties on their own’
|
93Similar examples may be found in Norwegian and Swedish, too. But Danish stå appears to be especially prone to bleaching. It is outside the scope of this text to quantify how far such developments have come. Also, we are not at the moment ready to suggest an explanation why certain verbs, like Danish mangle, should be involved as V2s rather than other verbs.
94But it does seem possible to sketch the evolution of auxiliation. We have observed two tendencies: Extension from humans to nonhumans (especially inanimates) does involve bleaching (or generalization) of the PV meaning. But as shown, this does not in itself lead to aspectual grammaticalization, as long as the Location is intact.
95The necessary kind of bleaching is independent of Agent type and involves loss of Location. If there is no Location, there cannot be any posture, human or other. What then crucially remains is the “Agent” and the Duration (at which point facilitation has become quite obscured). A Duration must necessarily be the Duration of something, and this something is found in the meaning of V2. These findings match fairly well the understanding of PV auxiliation argued by Lemmens (2014). Lemmens refers to Kuteva’s (1999) stages of PV auxiliation, where it is assumed that aspectual constructions with inanimate subjects develop before ones with animate subjects. Although our findings do not disprove such a view, they seem to indicate that expressions with animate (specifically human) subjects take the lead in the Mainland Scandinavian languages.
96As mentioned in Section 3.3, PVs used in other constructions than PC tend strongly to be used with a constituent expressing Location. When the Location in PVPC loses its salience and is not expressed, this also amounts to a change in the argument structure of the PV which may be interpreted as decategorialization and grammaticalization ; see Hilpert & Koops (2008).
97Apart from incipient grammaticalization of the PV, some combinations of PV and V2 can be seen to have acquired meanings and formal properties that differ from what might be expected in compositional terms – i.e. they are lexicalized (cf. Brinton & Traugott 2005). One example is Swedish ligga och skräpa ‘lie and be/make a mess’ (illustrated in (18) above): The verb skräpa is normally followed by the adverb ner ‘down’, but in PC with ligga it is most often used without the adverb. Cases where the V2 has partly different meanings in PVPC than when used elsewhere include for instance Swedish stå och hänga ‘stand and hang/loiter’. This issue deserves closer study, but cannot be pursued further here.
98We have explored the fruitfulness of the notion of ‘facilitation’ applied to PVPC, as an alternative and supplement to approaches focussing on aspectuality and grammaticalization. The concept has its roots in Nielsen’s (2011) notion of ‘possibilification’. The centrality of Location in the meaning of PVPC is present in that work and has been underscored further by Blensenius (2015, 2016). As an alternative to Nielsen’s ‘possibilification’, Kinn (2018) proposes ‘facilitation’ but does not develop it in any detail. We have refined it for PVPC by employing the frame-semantic posture frame. Some of the frame elements have been found essential to the understanding of how posture is construed as prototypically facilitating concomitant events: In addition to the posture itself, the Location and the Duration are essential. The facilitating factor in PVPC is the postural Location of the Agent over a certain Duration.
99Among the several subschemas of the Mainland Scandinavian PC schema, that of PVPC is clearly the one most amenable to the traditional aspectual approach as well as the one least amenable to our alternative facilitation approach. The usefulness of ‘facilitation’ having been demonstrated with regard to this subschema, its applicability to PC at the schema level is hardly in doubt, although further empirical studies of the various other PC subschemas are needed to map its properties in greater detail.
100We fully acknowledge that not all instances of PVPC involve facilitation of V2 by the posture itself. Importantly, facilitation in PVPC is like most other linguistic generalizing construals: It has a prototypical core but shades off into less central cases. Lying in bed is not the ideal position for writing, as is done in (22) ; sitting at a table would have been more practical (and, indeed, Norwegian skrive ‘write’ is a distinctive collexeme of sitte ‘sit’):
(22)
|
Jeg
|
ligger
|
i
|
sengen
|
og
|
skriver –
|
i
|
lyset
|
|
I
|
lie
|
in
|
the.bed
|
and
|
write
|
in
|
the.light
|
|
fra
|
en
|
lommelykt
|
(No.)
|
|
|
|
|
|
from
|
a
|
torch
|
|
|
|
|
|
|
‘I lie in bed writing – by the light from a torch’
|
101But even if the posture is not ideal, it provides the required stability of posture and Location for a certain Duration which allows the acticity, which means that some aspects of facilitation remain.
102It is well known that PVs are crosslinguistically common sources of aspectual auxiliaries (see e.g. Newman 2002). We have not intended to disprove the idea that PVPC in Mainland Scandinavian has in it the potential beginnings of aspectual auxiliation of LIE, SIT, and STAND (see references in Section 2.1). On the contrary, we have used the posture frame to develop an understanding of how such a historical process may unfold: When both posture and Location are bleached, there is hardly any facilitation any longer. There remains little but Duration, and if the Duration is to apply to anything, it has to be the event referred to with V2. In this way, the PV might over time evolve into an aspectual operator on V2. The clearest evidence that we have found for this in PVPC in Mainland Scandinavian is exhibited by the Danish collocation stå og mangle ‘stand and lack’, which is often void of specified posture and Location. V1s such as Norwegian drive ‘carry on’ (Lødrup 2002, 2017 ; Kinn forthc.), the verbs for ‘take’ in all three languages (Ekberg 1993 ; Vannebo 2003 ; Nielsen 2011), and Swedish hålla på ‘carry on’ (Blensenius 2013) appear to be clearer examples of grammaticalization under way.
103Distinctive collexeme analyses have shown that the overwhelming majority among the most distinctive collexemes of each of LIE, SIT, and STAND are easily understood in terms of facilitation and compatible with the analyses of PV meanings in Holm (2013). LIE typically collocates, for instance, with V2s for sleep, rest, inactivity, helplessness, deterioration, and death. SIT tends strongly to collocate with V2s for written and spoken communication, consumption, and manual activities of the types that are normally performed with the hands in front of one, e.g. on a tabletop or in one’s lap. STAND collocates with V2s of foot motion, maintaining or regaining balance, vision, communication, and more varied activities than those seen for SIT. V2s of perception are characteristic of PVPC, but are not evenly distributed over the postures. In particular, V2s of sight are typical of STAND, which is clearly motivated by the greater field of vision afforded by standing. V2s of communication are found with both SIT and STAND, but they tend not to be the same for the V1s: Communication verbs collocating with SIT tend to denote fairly quiet face-to-face communication, including nonverbal acts like nodding, while verbs collocating with STAND include ones of shouting and nonverbal acts like pointing and waving, which function better at a distance and when the Agent is standing.
104These results firmly establish the relevance of facilitation based on posture, Location and Duration in PVPC in Mainland Scandinavian. Such a strengthened empirical fundament provides a new and sounder point of departure for future studies of the extent to which PVPCs have actually undergone bleaching of posture and Location, thus paving the way for PV auxiliation.