3.1 The aspectual and modal meaning of the English simple tenses

In this section, I summarize the basic tenets of the semantic analysis of the English simple tenses introduced more elaborately in De Wit (forthcoming: chapter 4). More specifically, I argue that we need to assume that we are dealing with a present perfective tense, in order to (a) account for the curious observation that the English simple present appears to be used for ‘anything but the present time’ (Langacker Reference Langacker2001: 251; italics in the original) and (b) to offer an explanation for the noteworthy exceptions to this general restriction. This analysis in terms of perfectivity will be shown to have important epistemic implications, in the sense that a global perspective on a situation entails full identifiability of this situation. I will furthermore demonstrate that this analysis is also relevant for the English simple past.

An account that interprets the simple present as having a perfective meaning goes against the more standard view according to which present-tense constructions are imperfective by definition (see, e.g., Bybee et al. Reference Bybee, Perkins and Pagliuca1994; Michaelis Reference Michaelis2004, Reference Michaelis, Aarts and McMahon2006, Reference Michaelis2011). However, there are a number of reasons why an analysis of the simple present in terms of pure imperfectivity is hard to maintain. For instance, if we apply Vlach's (Reference Partee1981) when-test, we see that the simple present can only yield perfective (sequential) readings, while an imperfective (incidental) interpretation is excluded:

1. (15) When I look up, she walks away.

Observations such as this have incited researchers like Brinton (Reference Brinton1988), Smith (Reference Smith1997: 110–12, 185–6), Williams (Reference Williams2002a: 128–66) and Vanden Wyngaerd (Reference Vanden Wyngaerd, Dikken and Tortora2005) to assume that at least part of the aspectual semantics of the English simple present is perfective. In De Wit (forthcoming), I propose that the simple present in English is perfective in all of its uses. In order to appreciate this analysis, we first need to provide a definition of perfective aspect. The literature on grammatical aspect has put forward a variety of related concepts to analyze the basic distinction between perfective and imperfective aspect: totality versus lack of totality (Comrie Reference Comrie1976: 16; Dahl Reference Dahl1985: 78; Smith Reference Smith1997: 3, 66), boundedness versus unboundedness (Chung & Timberlake Reference Chung, Timberlake and Shopen1985; Smith Reference Smith1997: 301–2), external viewpoint versus internal viewpoint (Michaelis Reference Michaelis1998), foregrounding versus backgrounding (Hopper Reference Hopper and Givon1979, Reference Hopper and Hopper1982; Partee Reference Partee1984), temporal definiteness versus temporal indefiniteness (Dickey Reference Dickey2000). In my view, these meanings can be regarded as more specific instantiations of a basic epistemic opposition. That is, I assume that a perfective viewpoint entails that a situation is fully verifiable by the conceptualizer: she can conceive of its beginning, its endpoint and all the intermediate phases of which the situation consists. Imperfective aspect, on the other hand, involves but a partial conceptualization of a situation: parts of the situation are out of focus and need not be known by the speaker. This epistemic approach to aspect is crucial to account for the use of perfective-aspect constructions with stative verbs, which refer to unbounded situations (thus failing to comply with the bounded viewpoint that is typically said to be imposed by perfective constructions). As noted by Bybee et al. (Reference Bybee, Perkins and Pagliuca1994: 77) and De Wit (forthcoming), it is not uncommon cross-linguistically for stative verbs to combine with perfective aspect, and this is also the case in English, in which the simple present combines naturally with stative verbs to refer to present-time states. The critical property of states that distinguishes them from events is the so-called contractibility property (Langacker Reference Langacker1987: 258–62). According to this property, states, which do not change over time, are fully identifiable on the basis of one random sample, which is representative for the state in its entirety. In that sense, perfective aspect and states are perfectly compatible: the viewpoint imposed by a perfective construction samples a part of a state that, due to the contractibility property of states, allows verification of the state as a whole. A similar analysis can be proposed for the use of the simple present in habitual expressions. Just like states, habits are unbounded and they do not change over time, hence they are contractible. Thus, a present-perfective construction such as the English simple present can sample a representative part of a habit at the time of speaking.

An analysis in terms of perfectivity is generally considered more straightforward for non-stative uses of the English simple present.Footnote ¹¹ In the case of performative expressions, the speech event in fact constitutes the reported event – e.g., by saying I promise, you are performing the act of promising (Austin Reference Austin1962) – and this event is therefore by definition fully known at the time of speaking. Play-by-play sports commentaries often involve situations that are actually just past at (or at least not exactly coincident with) the time when the speaker is reporting them, yet for reasons of vividness they are often construed as occurring in the present (Langacker Reference Langacker, Patard and Brisard2011: 60). Since the events in sports announcements also normally involve fairly stereotypical, scripted situations (Langacker Reference Langacker2001: 265), they follow from the structure of the world (of sports commentaries) (Brisard Reference Brisard and Brisard2002: 286). Hence, play-by-play announcements typically involve events whose beginning, concise development and endpoint are fully conceptualized at the time of their report. A similar analysis can be proposed for running commentaries accompanying demonstrations, which are fully (epistemically) controlled by the speaker.

An analysis in terms of perfectivity is more widely accepted for the simple past than for its present counterpart. As illustrated in (16), the simple past cannot normally be used to report events that are ongoing at some past reference point:

1. (16) He told me she ?had / was having a bath, so she couldn't come to the phone.

Michaelis (Reference Michaelis2004: 53) shows that, in cases such as (17), the simple past does appear to be yielding both perfective (sequential; (a)) and imperfective (incidental; (b)) readings:

1. (17)

   1. (a) Thick smoke filled the corridor. [In a matter of minutes, we could no longer see the exit signs.]

   2. (b) [I opened my door and looked out.] Thick smoke filled the corridor.

However, in examples such as (17b), the speaker has sufficient information at the time of reference to conceive of the filling event in its entirety. In fact, as pointed out by Michaelis (Reference Michaelis2004: 53), fill is given a stative construal here. Hence, a perfective viewpoint, including its requirement of full identifiability, is again felicitous.

The challenge is now to align this analysis of the English simple tenses as perfective constructions with the function of full-verb inversion, so as to identify the conceptual motivation underlying the anti-progressive constraint. Let us, therefore, first zoom in on some illuminating previous proposals addressing the function of full-verb inversion.

3.2 The function of English inversion: the cognitive organization of discourse and the ‘deictic effect’

My study does not intend to offer a genuine alternative for previous comprehensive accounts of inversion in English, as proposed in Drubig (Reference Drubig, Klegraf and Nehls1988), Birner (Reference Birner1994), Dorgeloh (Reference Dorgeloh1997) and Chen (Reference Chen2003). Rather, in section 5, I will align some of their observations, notably Drubig's ‘deictic effect’ and Chen's ground-before-figure proposal, with my own analysis of the English simple tenses in terms of perfectivity, and in addition introduce the notions of canonicity and inevitability to account for the anti-progressive constraint. The discussion presented in the current section is confined to those observations that are directly relevant for the present analysis; for more comprehensive reviews of the extant literature on English inversion, see Dorgeloh (Reference Dorgeloh1997: chapter 3), Chen (Reference Chen2003: 7–32), Kreyer (Reference Kreyer2006: chapter 3) and Prado-Alonso (Reference Prado-Alonso2011: chapter 2). I will, more particularly, single out two main perspectives on inversion: its role in the cognitive organization of discourse, and the subjective viewpoint it entails. Birner (Reference Birner1994) and Birner & Ward (Reference Birner and Ward1998) argue that inversion in English is used to present discourse-old information (expressed by the preposed constituent) before discourse-new information (expressed by the postposed constituent), so as to connect this relatively new information with the previous context. Inversion is thus assumed to serve an information-packaging function. Chen (Reference Chen2003: 15–25) criticizes this discourse-oriented analysis by pointing to various examples in which the ‘discourse-old before discourse-new’ account runs into difficulties. As an alternative, he suggests a cognitive linguistic analysis of English inversion in terms of a ‘ground-before-figure organization’.Footnote ¹² In cognitive linguistics, it is assumed that spatial relations in language involve the location or motion of one entity (the figure) with respect to another entity, which constitutes the ground for that figure (Croft & Cruse Reference Croft and Cruse2004: 56–8). Chen (Reference Chen2003: 46–55) further argues that the ground is first anchored by means of a landmark, which directs the hearer's attention to the ground. In example (18), for instance, the ground (below the wings) is first situated by means of an explicit reference to the wings, i.e. the landmark, in previous discourse; the postverbal constituent is then located with respect to preposed ground.

1. (18) The huge engines were built into the wings. Below the wings was a pair of stubby sea-wings, which served to stabilize the aircraft when it was in the water. (Chen Reference Chen2003: 60)

Despite the differences between Birner's discourse-pragmatic and Chen's cognitive approaches, they have in common that they both analyze the entity referred to in the preposed constituent as given in some way. Their insights will prove to be important for my own analysis, yet in order to account for the apparent lack of progressives in full-verb inversion, I think their accounts need to be complemented with others that focus more on the viewpoint adopted in inverted constructions. Let me begin by pointing out the strong deictic character of presentational there-sentences (Lakoff Reference Lakoff1987; Drubig Reference Drubig, Klegraf and Nehls1988; Langacker Reference Langacker1993). By using these constructions, the speaker predicates the presence, absence or (dis)appearance of a figure within the deictic center (i.e. the speech event and its immediate circumstances). Both Bolinger (Reference Bolinger1977) and Drubig (Reference Drubig, Klegraf and Nehls1988) identify a similar presentative function for locative and directional inversion, which Bolinger (Reference Bolinger1977: 93–4) refers to as thereless presentationals, which present ‘something on our immediate stage’. As pointed out by, e.g., Drubig (Reference Drubig, Klegraf and Nehls1988: 86–8) and Kreyer (Reference Kreyer2006: 202–7), the use of inversion can trigger a ‘displaced speech effect’, whereby the addressee is invited to conceptualize a given situation from the viewpoint ‘designated by the referent of the NP in the fronted constituent’ (Dorgeloh Reference Dorgeloh1997: 103). This ‘displaced speech effect’ can be – and often is – metaphorically conceived of in terms of a camera, which visualizes a situation from a certain vantage point, and which draws in viewers to conceive of that situation from this vantage point (Dorgeloh Reference Dorgeloh1997: 103–5; Chen Reference Chen2003: 65–6; Partee & Borschev Reference Partee, Borschev, von Heusinger and Comorovski2007: 156). Consider the following example:

1. (19) One night there was a tap on the window. Mrs. Rabbit peeped through the window. Outside stood a little angel. (Birner, 1638)

As readers, we are invited to adopt Mrs. Rabbit's perspective on the angel, i.e. we are standing inside the house and looking at the angel standing outside. Drubig (Reference Drubig, Klegraf and Nehls1988: 88–90) goes on to argue that directional (as opposed to locative) inversion does not only trigger a ‘displaced speech effect’, but also a ‘deictic effect’ – a critical point which is largely ignored in subsequent analyses of inversion (though see also Dorgeloh (Reference Dorgeloh1997: 164–87) for similar arguments). The difference between the ‘displaced speech effect’ and the ‘deictic effect’ is that only in the latter case is the conceptualizer's vantage point placed within the location designated by the preposed constituent (e.g., in (19) we, as readers/listeners are not standing outside ourselves). Consider the following opposition, pointed out by Drubig (Reference Drubig, Klegraf and Nehls1988: 88):

1. (20) He opened the bedroom door and the cat walked in.

2. (21) He opened the bedroom door and in walked the cat.

While (20) is ambiguous with regard to the subject's and the cat's positions – e.g. it is unclear whether the subject and the cat end up being in the same room or not – there is no ambiguity in the case of (21): the subject is standing inside the bedroom, he conceives of the cat's walking into that bedroom from that perspective and it is this conception that we, as readers/listeners, are invited to adopt.Footnote ¹³ Contrary to Drubig (Reference Drubig, Klegraf and Nehls1988), I assume that directional inversion does not necessarily entail a deictic effect: e.g., in the case of directional inversion along an unbounded trajectory, as in (10) (repeated here in (22)), the conceptualizer's vantage point does not have to lie within the location designated by the preposed constituent (the arid desert in (22)):

1. (22) Across this arid desert drifted an occasional low cloud of red dust. (Birner, 1093)

Although this modification will prove critical, Drubig's observations, just like Chen’s, constitute essential building blocks for my own analysis presented in section 5. Bringing together Drubig's analysis with Chen’s, we could claim that full-verb inversion in general involves a shift to a more privileged vantage point which is part of the landmark that serves to locate the ground. Whenever the deictic effect arises (i.e. with some cases of directional inversion), we are invited to conceptualize a situation from within the ground itself which functions as the (shifted) deictic center.Footnote ¹⁴

4.1 Corpus data

Given the presumed scarcity of progressives in full-verb inversion, it is important to study as many data as possible in order to make any claims about the contexts that allow the use of the progressive (if any) and about the commonalities of these contexts. Since this calls for large samples of corpus data, I have collected cases of inversion in three different corpora: 190 examples were drawn from the Collins corpora of modern written and spoken text, 314 examples come from the Brown Corpus of Standard American English, and 678 examples from the OntoNotes Corpus, release 5.0.Footnote ¹⁶ This boils down to a total number of 1,182 instances, only one of which features progressive aspect:

1. (23) Out of the mouths of revolutionaries are coming words of moderation. (OntoNotes)

In addition, I have been granted the opportunity to consult corpus data gathered by Betty Birner and Carlos Prado-Alonso for their own research on inversion (for book-length presentations of their studies, see, respectively, Birner & Ward (Reference Birner and Ward1998) and Prado-Alonso (Reference Prado-Alonso2011)). Birner's collection of 1,778 present- and past-tense instances of full-verb inversion – collected on the fly, and excluding there-sentences – contains only two examples featuring progressive aspect.

1. (24) From the west were rolling pile after pile of fat, white, complicated clouds, and above the clouds was the clear and uncomplicated blue. (Birner, 1524)

2. (25) They have a great big tank in the kitchen, and in the tank are sitting all of these pots. (Birner, 1561)

Prado-Alonso's data collection is more systematic and corpus-based (all written and spoken examples of inversion were collected in a database of 1,804,000 words consisting of subcorpora from FLOB, FROWN, ICE-GB and CSPAE),Footnote ¹⁷ and he does take into account presentational there-sentences. On a total of 649 collected targets he counts only one progressive instance (Prado-Alonso pc):

1. (26) I’m sure you know that the Queen is standing beside the Duke of Edinburgh. In the middle and on her right is standing the Lord Mayor of London. She is wearing a lime green suit. . . (Prado-Alonso Reference Prado-Alonso2011: 133; pc)

Thus, to sum up, on a total of 3,609 tokens of full-verb inversion collected for the current investigation and for previous research, only four progressive tokens could be retrieved.

4.2 Native-speaker elicitations

On the basis of the corpus data presented in section 4.1, it is safe to say that the use of the progressive in full-verb inversion is extremely rare. Yet it is not entirely barred, as is also reflected in the attestation of Google tokens. This not only raises the question why the progressive is generally avoided in inversion – which is the main topic of this study – but also in which contexts this general restriction is defied, and how these exceptions tie in with an explanation for the strong preference for simple tense forms. Since the naturally attested progressive tokens are obviously too few to base any generalizations on, I have launched two native-speaker surveys on Mechanical Turk.Footnote ¹⁸ For the first survey, I created 44 sentences involving the various types of full-verb inversion listed in (6) to (13), featuring both simple present and progressive aspect, as well as similar sentences without inversion, and seven control sentences (three grammatical ones, and four decidedly ungrammatical ones). In each case in which both a simple present and a present progressive version of the same sentence were given (i.e. all cases, except for predicate inversion), I had to ensure that none of the informants could see both aspectual variants so as to avoid syntactic priming (e.g., after having seen the simple present sentence, the same sentence in the progressive might have been considered more acceptable). Therefore, I split up these sentences into two batches of 22 sentences, such that every single informant could only see one aspectual variant. In total, 40 native speakers rated the acceptability of the sentences in one batch, using a four-point scale ranging from ‘very bad’ to ‘perfectly fine’.Footnote ¹⁹ For every case, I investigated whether there was a significant difference in terms of acceptability between the simple present and the present progressive variant, so as to identify the types of inverted contexts that are more progressive prone. Despite the relatively coarse scale used in my study, significant differences were attested for directional inversion with a specified endpoint (27), presentational there (28) and quotative inversion (29).Footnote ²⁰

1. (27) I open the door and in comes the cat / *I open the door and in is coming the cat. (p = 1.132–9)

2. (28) There goes my bus / *There is going my bus. (p = 4.251–40)

3. (29) ‘Why are you crying?’ asks Jill / *‘Why are you crying?’ is asking Jill. (p = 4.940–20)

It is noteworthy that for pure locative inversion, for directional inversion along an unbounded trajectory and for directional inversion with a specified source, no significant differences could be attested. Thus, while (27) is clearly considered ungrammatical, (30) (31) and (32) were found to be much more acceptable:

1. (30) In the chair is sitting a beautiful man. (p = 0.819)

2. (31) Across the desert are drifting clouds of red dust. (p = 0.100)

3. (32) From below is coming a strange sound. (p = 0.349)

These first observations seem to indicate that types of motion of which the endpoint is envisaged are more progressive-resistant than other types of directional and locative inversion. This could suggest that there is a connection between a given endpoint (and thus, presumably, a full view on the development and ending of a state of affairs) and the obligatory use of the simple present. In order to verify this hypothesis, I conducted a second Mechanical Turk survey (using the same methodological precautions as with the first study), thereby specifically focusing on locative and different types of directional inversion. Again, simple-present sentences were contrasted with present-progressive ones, and a single informant only got to see one of the two. In total, 13 simple-present and 13 present-progressive sentences were judged by 28 informants. This second survey confirms the observation that the use of the progressive in locative inversion (examples (33) to (35)) and directional inversion along an unbounded trajectory, as in examples (36) to (38), are considered quite acceptable:

1. (33) In that house are living strange people. (p = 0.501)

2. (34) On top of the square block is lying another block. (p = 0.376)

3. (35) Upstairs are hanging my mother's clothes. (p = 0.129)

4. (36) Across that bridge are going hundreds of cars. (p = 0.020)

5. (37) Along the riverside path are walking several tourists. (p = 0.389)

6. (38) Down the beach are running cute puppies. (p = 0.443)

However, my study does not yield any uniform results for cases of directional inversion involving a specified source or endpoint: while sentences (39) to (41) are considered acceptable, sentences (42) to (44) are not:

1. (39) Up to the hilltop they are running. (p = 0.645)

2. (40) Onto the stage is stepping a beautiful woman. (p = 0.515)

3. (41) From our neighbor's house is coming the sound of piano music. (p = 0.100)

4. (42) *Out of the room is stepping an enormous man. (p = 0.0001)

5. (43) *She whistles and away he is going. (p = 0.0003)

6. (44) *In is coming the President. (p = 8.3468–5)

It seems, then, that explicit reference to an event's initial or final boundary does not preclude the acceptability of progressive aspect. Rather, this second Mechanical Turk study brings to the surface another parameter: as I will explain in section 5, it is the rise of the deictic effect that blocks the use of the progressive in cases such as (42) to (44).

5.1 Inevitability, canonicity and perfectivity

Let me now align the analysis of the simplex tenses in English in terms of perfectivity with the function of full-verb inversion in English, as outlined in section 3.2 in terms of givenness, ground-before-figure organization and the deictic effect.

As already hinted at in the introduction to this article, previous proposals to account for the restriction on progressives in full-verb inversion have been formulated by Lakoff (Reference Lakoff1987: 496) for presentational there-constructions and by Chen (Reference Chen2003: 180–3) for locative inversion. Lakoff states the following:

It should be noted that it is not true that there-sentences exclusively feature the simple present: my corpus data show that past-tense presentational there-sentences are occasionally attested as well:

1. (45) ‘Are you earning enough?’ ‘Of course I am.’ And so there went the theatre again. All this had happened in the autumn of 1964 . . . (Collins)

However, more important for the present analysis is Lakoff's claim that presentational there-constructions refer to an act of pointing which is instantaneous. In other words, what is designated by the construction is not so much a process of motion or location, but rather the momentaneous attestation of that motion/location at reference time. With regard to locative/directional inversion, Chen (Reference Chen2003) argues that ‘[t]he progressive, which forces the hearer to pay much attention to the verbal process, could be a misfit for the GbF [ground-before-figure] model, for the GbF requires that attention be given to the ground and the figure’ (181). Thus, when a certain ground is introduced, such as Into the room, the hearer will be concerned with who came into the room rather than with the way in which this person came into the room. Using the progressive forces the addressee to linger on the process referred to by means of the verb, thus delaying the identification of whoever is moving.Footnote ²¹ In other words, in line with Lakoff's analysis of presentational there, Chen believes that with locative/directional inversion it is the presence, absence or (dis)appearance of a figure in a ground that is profiled rather than the manner in which this figure is present or the movement that led to this presence.

While Lakoff's and Chen's proposals are seminal in many ways, they do not explain why the progressive is nevertheless used or considered acceptable in some cases, nor do they (explicitly) acknowledge the perfective value of the simple present.Footnote ²² I contend that, whenever the deictic effect arises, a perfective viewpoint is automatically entailed. Recall, first, that presentational there-sentences by definition convey a deictic effect and, secondly, that I assume that not all instances of directional inversion entail a deictic effect (pace Drubig Reference Drubig, Klegraf and Nehls1988). Now, given a configuration in which the conceptualizer is placed within a deictic center/ground, with respect to which, at the same time, a certain figure is located, the conceptualizer has a limited view. She can note the presence, absence or (dis)appearance of a figure in this ground, but since her viewpoint is anchored within the same ground, she cannot step out of it, so to speak, and zoom in on the process that leads to this presence, absence or (dis)appearance. Thus, whenever the deictic effect arises, she can only adopt a global – i.e. perfective – perspective, and progressive construals are excluded. In view of this configuration, which involves full knowledge about a situation and its boundaries, it is not surprising that full-verb inversion in English conveys a sense of inevitability with motion verbs, as illustrated in (46) to (49) (see also (8)):Footnote ²³

1. (46) There goes the neighborhood.

2. (47) Then boom! Along comes the Internet. (Collins)

3. (48) Up we go!

4. (49) Now comes the good part.

If the conceptualizer ‘waits’ long enough in the location in question, the figure will inevitably (dis)appear in each of these cases.Footnote ²⁴ As noted by Dorgeloh (Reference Dorgeloh1997: 76), the use of directional inversion is typically exploited to create a sense of tension. This sense of tension is naturally derived from the notion of inevitability: given a certain preposed ground, you can be certain that something is coming, but you do not necessarily know what yet. It is not surprising, then, that full-verb inversion should often be used in the context of play-by-play sports casting and demonstrations (Green Reference Green and Tannen1982), in which perfective marking is the rule rather than the exception. Note, incidentally, that it is the fronting of the ground, rather than the postposition of a figure that yields the inevitability effect: as is illustrated in (48), the effect also arises in non-inverted sentences with pronominal subjects. This reflects my assumption that it is the anchoring/givenness of the conceptualizer's vantage point that creates this effect.

In each of the examples listed in (46) to (49), the movement is construed as irrevocable, and thus, crucially, the conceptualizer can fully identify the development and endpoint of these motions. I claim now that such full knowledge is characteristic of all types of full-verb inversion, including static locative inversion and directional inversion along a trajectory, which do not trigger a deictic effect. In (7), repeated here in (50), the only ways in which a book can normally be on a shelf (apart from general ‘being’) is by lying or standing.

1. (50) On the shelf lies a book.

The relation between the book and the shelf is inferable on the basis of general knowledge about shelves and books (Birner Reference Birner1995); there is, in other words, a canonical relation between the figure and the ground. As extensively described by Birner (Reference Birner1995), locative inversion is restricted to verbs that are ‘informationally light’: they do not contribute any crucial information to the utterance, since the manner in which a figure is present in the ground could already be inferred from the nature of the ground and of the figure. Note that it is not the presence or absence of a figure within a ground as such that is canonical (cf. the frequently observed exploitation of full-verb inversion to create a sense of surprise). What is known, however, is the process (i.e. the situation designated by the verb) that relates the figure and the ground; one could claim that given a certain ground and a certain figure, the relation between them emerges naturally. Such canonical events typically trigger simple tense forms, whereas the use of the progressive is linked to more phenomenal, incongruous and possibly surprising situations, as is demonstrated in various analyses (Calver Reference Calver1946; Dowty Reference Dowty1975; Goldsmith & Woisetschlaeger Reference Goldsmith and Woisetschlaeger1982; Langacker Reference Langacker2001; De Wit & Brisard Reference De Wit and Brisard2014). Cases of directional inversion involving a deictic effect may be regarded as extensions of the prototypical full-verb inversion construction. The canonicity property characteristic of this basic construction is inherited by the deictic directional subtype, which in addition conveys a sense of inevitability.

In spite of the distinct features of quotative inversion compared to other types of full-verb inversion – analyses with reference to the givenness of the ground or to a deictic effect are clearly not applicable – the fact that it bars progressive aspect can equally well be explained in terms of the canonicity, and thus full identifiability, of the process referred to by the verb of speaking. The manner of speaking of the subject can typically be predicted on the basis of the preposed quote, and it does not contribute important information to the sentence. It would be anomalous to put focus on the verb of speaking, given the high prominence of the fronted quote.

Note, finally, that the acceptability of the present and past perfect in full-verb inversion follows naturally from the analysis presented in this section. Unlike the progressive, the perfect in English construes past events as completed, i.e. as a whole. In addition, these events typically still have some relevance at the time of speaking, in that the presence, absence or (dis)appearance of a figure within a given ground is construed as ‘fresh’.

5.2 Acceptability of progressive aspect: exceptions that prove the rule

I contend that in cases where full-verb inversion is used to convey a deictic effect, the use of the progressive is automatically barred for the conceptual reasons explained in the previous section. In cases where we do not get a deictic effect, the conceptualizer can zoom in on the denoted event, and the use of the progressive is not entirely excluded. This explains its relative acceptability with locative inversion (see (30) and examples (33) to (35) above). However, given the canonical information expressed by the locative (posture) verb, the use of the progressive remains highly rare. Lack of canonicity might account for the progressive occurrence in example (25), repeated here in (51):

1. (51) They have a great big tank in the kitchen, and in the tank are sitting all of these pots. (Birner, 1561)

Example (26), repeated here as (52), however, does involve a canonical relationship between the figure and the ground, yet it still features progressive marking:

1. (52) I’m sure you know that the Queen is standing beside the Duke of Edinburgh. In the middle and on her right is standing the Lord Mayor of London. She is wearing a lime green suit . . . (Prado-Alonso Reference Prado-Alonso2011: 133; pc)

My tentative explanation for this use of the progressive resides in syntactic priming: since the speaker has used the progressive is standing in the first sentence, she naturally repeats it, despite the inverted sentence structure. The progressive is also found acceptable in examples of directional inversion that do not involve a deictic effect, such as examples (36) to (40). Although in the latter two cases the endpoint of the motion is explicitly referred to in the preposed constituent, the conceptualizer's vantage point does not necessarily lie within that ground. Consider also example (24), repeated here in (53):

1. (53) From the west were rolling pile after pile of fat, white, complicated clouds, and above the clouds was the clear and uncomplicated blue. (Birner, 1524)

The use of the progressive here is, in my view, motivated by the gradual motion of the clouds; the more an event is stretched in time, the more natural it becomes to devote more attention to this process, which is consequently construed as less canonical.

There are two examples that appear to defy my analysis, one actually attested ((23), repeated in (54)), and one constructed yet considered acceptable ((41), repeated in (55)):

1. (54) Out of the mouths of revolutionaries are coming words of moderation. (OntoNotes)

2. (55) From our neighbor's house is coming the sound of piano music.

In these examples, the denoted events ‘end’, so to speak, in the conceptualizer's deictic center. Yet this is not an effect of the construction, but rather of the verb come, which by definition subjectively refers to the deictic center. The crucial difference between these examples and the ungrammatical sentences cited in (42) to (44) is that, in the latter cases, the deictic center coincides with the ground that serves to locate the figure. By contrast, in (54) and (55), the figure is located with respect to a ground that is not the deictic center. Consequently, the conceptualizer's viewpoint is not anchored within or with respect to this ground and there is no constructionally induced deictic effect. Non-deictic cases such as these may allow the use of the progressive under specific circumstances. In (54) and (55), we are again dealing with events that are temporally extended – in (54), the words are being uttered repeatedly, and in (55), the sound of the piano music obviously does not ‘arrive’ instantaneously – and thus progressive construals are licensed.

In the overwhelming majority of the cases of full-verb inversion though, the use of the progressive is either dispreferred (i.e. rarely used compared to the simplex tenses), because there is no need to focus on the canonical process, or it is entirely barred, whenever the deictic effect arises. Apart from these functional constraints, there may also be more formal factors playing a role, namely the presence of other progressive-like constructions. This brings me to predicate inversion, in which the present participle is part of the preposed constituent. In line with, among others, Dorgeloh (Reference Dorgeloh1997: 86–8) and Chen (Reference Chen2003: 134–48), I believe that this fronted present participle has lost its verbal qualities, and that it is therefore misleading to speak of progressive aspect here, as Birner & Ward (Reference Birner and Ward1992) do. For instance, as pointed out by Birner & Ward (Reference Birner and Ward1992) themselves, the following example does not necessarily refer to an ongoing situation, since it could also be used to report the loss of the main opposition party (i.e. a completed event):

1. (56) A free election was held in Czechoslovakia for the first time since the war. Losing the election was the main opposition party. (Birner & Ward Reference Birner and Ward1992: 1)

Ongoing readings are not excluded with predicate inversion, but, as amply demonstrated by Chen (Reference Chen2003: 138–46), the preposed participle is unambiguously adjectival in most cases, whereas unambiguously verbal instances are not attested (i.e. the participle can always be interpreted as giving more information about the referent of the subject NP and thus as an adjective). Predicate inversion may be exploited to construe the event of, in the case of (56), losing, as non-inferable, thus exempting the ‘real’ progressive from this function. Note that presentational there-constructions also have competing progressive-like constructions, as illustrated in (57):

1. (57) There’s someone stepping/jumping into the room.

Here, too, we are dealing with a present participle, modifying someone, yet again this construction could be exploited to construe the denoted motion as non-canonical, where full-verb inversion is a less likely candidate to fulfill this function. Compare, for instance, the following two examples:

1. (58) Look, there’s someone jumping into the room!

2. (59) ?Look, into the room is jumping / jumps someone!

As illustrated in (59), it sounds at least contrived to use full-verb inversion to emphasize the exceptionality of the designated event (jump). Thus, for all types of full-verb inversion, there appears to be a related construction that could block the use of the actual progressive. Note that the presence of these progressive-like constructions does not preclude, but rather complements my conceptual account.