Present Participle as Predicative with vara ‘be’ in Present Day Swedish

1 Introduction

This article focuses on a case of lexical restrictions on present participles with predicate vara ‘be’ in Swedish, specifically on the limits of such restrictions, the applicability of the structure with vara to test the participle’s character and a possible language change in progress. In standard Present Day Swedish (PDSv), the present participle can be used with vara only if the participle is no longer perceived to be a verbal form. Compare the acceptable fascinerande in (1) with the unavailable *skrikande in (2):

At first glance, there is no problem to be addressed here: fascinerade, despite its etymology, has become lexicalized and is now an adjective and thus allowed in the predicative position. Skrikande, on the other hand, exhibits more verbal than adjectival properties and for this reason is disallowed. Indeed, the reference grammars of standard PDSv use this very structure to test whether a given lexeme is an adjectival or a verbal participle (see also Malmgren 1990; Hedlund 1992). However, modern corpora of Swedish provide examples, in which a verbal participle is used as a predicative after vara, as shown in (3).

Taking such observations into consideration, we will reexamine the restrictions on the present participle use with vara across three large corpora accessible via Språkbanken: academic texts (Akademiska texter, henceforth AT), Swedish television news service (henceforth SVT) and blogs (Bloggmix, henceforth BM). They represent different registers: formal written language, relatively formal spoken language and informal register, closest to the spoken informal variant. The present participle in Swedish is formed by means of the suffix -ande or -ende (depending on the phonological properties of the stem). It can also be followed by another suffix, -s, of disputed origin (Kock 1898), which we will disregard in the present study. As has been pointed out in earlier studies, in standard Present Day Swedish the s-form is restricted to the verbal uses of the participle (Thurén 2008: 13).

The article is structured as follows. In Section 2 we present the notion of participle orientation and juxtapose it with thematic roles and properties of subjects. In Section 3, we delve into the challenges of establishing the categorial status of the present participle and survey the tests applied to the present participles in standard Present Day Swedish. This is followed by a description of the material and method in Section 4 and presentation of the findings in Section 5. In Section 6 we discuss the results and draw conclusions with respect to a potential ongoing language change and the reliability of the currently employed tests. Section 7 summarizes the study.

2 Present Participles and Orientation

The label ‘participle’ is widely used in linguistics; however, the term itself is not particularly well-defined (Shagal 2019: 14). Some authors apply the label to all non-finite verb forms (Hartmann and Stork 1972: 165), some to non-finite verb forms other than the infinitive (Crystal 2003: 337–338). Some refer to them as ‘verbal adjectives’, sharing properties of both verbs and adjectives, others focus on all participles’ having the distribution of adjectives and argue that participles with event structure also are adjectives (Lundquist 2013: 12). In the traditional approach, participles are regarded as verb forms that “behave like adjectives with respect to morphology and external syntax” (Haspelmath 1994: 152); Haspelmath (1994) disregards forms similar to converbs in his analysis. In (some) generative studies the participles are considered ‘neutralized’ entries (Chomsky 1981; Hoekstra 1984), whose categorial status is determined by the syntactic environment in which they appear. Svenska Akademiens Grammatik, the reference grammar of the Swedish Academy, chooses to categorize participles as a separate category, different from both verbs and adjectives (SAG 1999).

Even taking into account the vast body of research on participles, we may note that the research has so far mainly focused on the past participles (Wasow 1977; Levin and Rappaport 1986; Bresnan 1996; Doron 2000; Kratzer 2000; Embick 2004; Horvath and Siloni 2008 and others). Present participles have been the focus of less research, with notable exceptions including Brekke (1988), Borer (1990) and Meltzer-Asscher (2010), who all attempt to establish the categorial status of the present participle and characterize this class more precisely. There are also a number of studies devoted to the English gerund (e.g., Milsark 1988; Emonds 1991; Fonteyn 2019), which address mainly the verbal properties of the participle. In Swedish, the present participle has been studied by Thurén (2008) and past and present participles in Hedlund (1992), although the latter is mainly concerned with the past participle; Platzack (1980) considers only the past participle in Swedish. The issue of adjectivization of participles is addressed in Lindroth (1906), who, however, also focuses mainly on the past participle.

In cross-linguistic studies of participles, both present and past, are classified in terms of their orientation. The notion of orientation, adapted from Lehmann (1984: 152), was first used with respect to the participles in Haspelmath (1994: 153) and explained with both pure adjectival and participial examples. For participles, consider (4a–b).

Both participles are derived from the same verb, paint, which is a transitive verb taking a subject (Agent) and an object (Patient). Painted is oriented towards the object of the verb (‘a wall’ in (4a)), painting is oriented towards the subject (‘a man’ in (4b)). Such two participles are typically described as past and present respectively,^[1] and cross-linguistically are among the most common participles (Haspelmath 1994: 154–157).

The notion of orientation is further exploited in Shagal (2019), who treats participles as equivalents of relative clauses (e.g., a painted wall = a wall which was painted) and links the participial orientations to the range of relativizable positions, including subject and object (Shagal 2019: 53). In her approach, the subject of a transitive clause and the subject of an intransitive clause are treated separately. Following Comrie (1981), Shagal uses notation A for the agent-like participant of transitive clause, P for the patient-like participant of the transitive clause and S for the single participant of the intransitive clause (Shagal 2019: 58, see Haspelmath 1994: 154), we follow this practice here. In our study, we limit the distinction to S–A, as we only consider the present participles. Shagal (2019: 128) concludes that the S–A characteristics of present participles are based “on semantic properties of the participants, such as humanness and control”. For this reason and for a better organization of the analysed material, we include some properties of subjects, i.e., animacy and thematic roles. The use of participles seems to differ depending on these properties. Consider (5b) and (6b).

We follow Van Valin and La Polla (1997) and Toyota (2008) in distinguishing between Agent, Effector, Experiencer and Theme as the most likely thematic roles for a subject in an active clause, see Van Valin and La Polla (1997: 146). The roles form a hierarchy of sorts, with a receding degree of volition and control, compare with Toyota (2008: 10).

Agents are the arguments with most control and volition and are the typical thematic role for a subject in an active transitive clause, (8a). A lower level of control and volition is associated with Effectors, e.g., ‘malaria’ in (8b). Experiencer is the entity undergoing a state of being, an emotion or a perception expressed by the verb, (8c). Finally, Theme signifies a participant which is neither acting nor experiencing, but rather is an entity located somewhere or undergoing a change of location, (8d).

Thematic roles are to some extent a function of subject animacy. Animate subjects can take any thematic role, but inanimate subjects are excluded as Agents. The Agent typically denotes a sentient being, capable of instigating action, with a high degree of volition and control. This is only possibly with animate referents, (9a). Animate entities can also be Effectors, when the action is not purposeful or intentional, (9b). Similarly, to experience emotions or be able to perceive, an entity must be animate, however, there are verbs, such as tremble, which can take inanimate subjects (9c). Finally, Themes, can be animate or inanimate and are things that are located or undergoing a change of location (9d).

The present participle in Swedish is always oriented towards the subject, which means it can either be A-oriented (participles such as sparkande ‘kicking’, derived from the transitive sparka ‘kick’) or S-oriented (participles such as springande ‘running’, derived from the intransitive springa ‘run’). In Table 1 we present an overview of possible forms, showing how the characteristics of orientation (A or S) and thematic roles (Agent, Effector, Experiencer or Theme) may interact with each other. The participles included in the table are connected to verbs in the aforementioned example sentences (9). The lack of the interplay between the two features does not mean that it is the thematic role of the subject that restricts the formation of the present participle, we note however that some combinations are not possible, i.e., Theme cannot be a subject in the transitive clause. It is further difficult to form a present participle from transitive verbs with Experiencer subjects and without incorporated objects, such as gilla ‘like’ (compare the participles used in attributive positions: en sparkande häst ‘a kicking horse’, en dödande sjukdom ‘a deadly (lit. killing) disease’, ?en gillande man (?a liking man) versus musikgillande ‘(lit.) music-liking’, en springande pojke ‘a running boy’, en leende figur ‘a smiling figure’, en darrande hand ‘a trembling hand’, en sittande katt ‘a sitting cat’).

In this section we have explored the notion of participle orientation, i.e., the idea that participles are oriented towards one of the arguments of the verbs from which they are derived. Following Haspelmath (1994) and Shagal (2019) we have differentiated between present and past participles, oriented towards subjects and objects respectively and between present participles oriented towards the subject of a transitive clause (A-oriented) and towards the subject of an intransitive clause (S-oriented). Both Haspelmath and Shagal restrict their discussion of participles to such forms that can be used in adnominal modification (Haspelmath 1994: 153; Shagal 2019: 5). In our paper we consider the use of the present participle in the structure [SBJ vara PRS.PTCP] – a context traditionally reserved for true adjectives, in which verbal forms are disallowed (unlike the gerund in English). Nevertheless, we find this typology to be a useful tool for presenting our data and relevant for the present study.

We have further proposed to augment this typology with thematic roles Agent, Effector, Experiencer and Theme (Van Valin and La Polla 1997; Toyota 2008). An overview of the possibilities is given in Table 1. According to the standard grammatical descriptions of Present Day Swedish, none of these participles can be used as a predicative with copula vara ‘be’. The structure [SBJ vara PRS.PTCP] is therefore often employed as a test for the categorial status of the present participle in Swedish. We explore this further in the following section.

3 The Categorial Status of the Present Participle in Swedish

The grammar of the Swedish Academy (Svenska Akademiens Grammatik) emphasizes that the distinction between Swedish participles’ adjectival or verbal meaning is often gradual and that different language users’ notion of their verbal meaning in certain uses may vary (SAG 2 1999: 583). Hedlund (1992: 178) observes that the Swedish present participle “is very easily adjectivized, and therefore the present participle is often considered to always be an adjective.” Thurén (2006) on the other hand first proposes a distinction between three types of Swedish non-nominals with common surface form ending in -ande/-ende: verbal participles, adjectival participles and true adjectives, but in her later analysis she argues for a dichotomous distinction between present participles of verbal character and true adjectives. In this paper we adopt the perspective expressed in the reference grammar of Swedish, SAG (1999), and in Thurén (2006) and focus on the binary distinction between the words’ verbal and adjectival properties, taking into consideration the words’ etymology. The principles for the distinctions found in literature are based on semantic, syntactic, distributional and morphological criteria, which we discuss in the following parts of this section. The list of questions in Table 2 is based on surveys presented in Thurén (2005, 2006, 2008), who modelled her analyses using e.g., Wasow (1977), Borer (1990), Malmgren (1990), SAG (1999) and Embick (2003), sometimes adapting the distinctions described for English and making them valid for standard Present Day Swedish. The answers to these questions can be used to check whether the participle is of adjectival or verbal character.

Question #1. Regarding the first question, when the meaning of a participle can be fully derived from the meaning of the corresponding verbal stem and it denotes an event the same way as the corresponding verb, it can be described as having fully verbal character. Participles with adjectival character, on the other hand, often have a denotation closer to a property and they have more figurative, overall meaning which cannot always be directly derived from the meaning of the verbal stem (SAG 2 1999; Thurén 2008).

Many participles are used as adjectival participles only with the figurative meaning, e.g., strålande (‘brilliant’, from stråla ‘beam’). The meaning of some present participles can be lexicalized and derived from one or more meanings of a particular verb, e.g., rörande (‘moving, causing emotion’, from röra ‘move, cause to change position’ or ‘move, cause emotion’). The meaning of some participles can also be very distant from or not present in the verb they are derived from, e.g., träffande (‘appropriate’, from träffa ‘meet’) (SAG 2 1999: 584, 613). Sometimes the relation between a participle and the verb it diachronically refers to is even completely lost, so that they do not have any common meaning, e.g., spännande (‘thrilling’, from spänna ‘tighten’) (Hultman 2003: 79). Present participles can also be formed in compounding (e.g., regnbågsskimrande ‘iridescent’, bollspelande ‘ball playing’, bredvidliggande ‘adjacent’), such compound participles are also predominantly of adjectival character (SAG 2 1999: 616–617).

Prototypical adjectival participles denote more stable, prolonged states of being (SAG 2 1999: 584, 613). The interpretation of (10) tends to be that the situation in (a) causes a long-term state of concern, while the situation described in (b) refers to a more temporary state.

Such adjectival participles often do not specify who the experiencer of the action is (see example 11) and they have more generic interpretation (SAG 2 1999: 613).

For adjectival participles, the notion of the verb arguments and their semantic functions, e.g., agent and object, is often weakened or even completely absent (SAG 2 1999: 584, 612). Lundquist (2008: 213) points out that adjectival present participles are often derived from object-oriented so-called psych-verbs, e.g., skrämmande (‘frightening’), fascinerande (‘fascinating’), imponerande (‘impressive’), irriterande (‘irritating’). Brekke (1988) notes that what is characteristic of such kind of formations is the presence of an Experiencer NP in non-subject position – he regards them though as true adjectives, in contrast to (verbal) participles, as in (12), translated from English.

Question #2. When it comes to the syntactic criteria, the distinction lies in that participles of verbal character appear with verbal arguments and participles of adjectival character appear with adjectival arguments. The arguments of verbal participles retain the structure of the corresponding verbs (see example 13), while the arguments of adjectival participles may have a different structure or may not be obligatory (see 14), even if they are obligatory for the corresponding verbs (see 15) (SAG 2 1999: 584; SAG 3 1999: 618–624; Thurén 2006: 154).

Question #3. The next criterion refers to the suffix -s, which can attach to the present participle in spoken Swedish and informal, colloquial written Swedish. The meaning of such participles remains unchanged in relation to the unsuffixed variants, but according to the reference grammars of Swedish, they can be used only in free predicative or as complement to the verb komma (‘come’) in bound predicative (SAG 2 1999: 614–615; Hultman 2003: 156), but never in a preposed attributive in a nominal phrase (Thurén 2008). Only verbal present participles can appear suffixed -s (Thurén 2005, 2006, 2008).

Question #4. Only highly lexicalized adjectival participles can be prefixed with the negative o- ‘un-’ – with the exception of participles derived from some verbs already prefixed by o- (such as oroa – oroande ‘unsettle – unsettling’). The participle in (17a), liggande ‘lying’, is of verbal character, while sammanhängande ‘coherent’ in (17b) is of adjectival character.

Question #5. Another criterion is the possibility of modification by degree adverbials. Participles that have more verbal properties and a closer relation to the meaning of the corresponding verb are not gradable and cannot be used in comparisons with adverbials such as mer, ganska, mycket, väldigt (‘more’, ‘quite’, ‘very’, ‘extremely’), as opposed to participle having more adjectival properties (Thurén 2006: 156; Hultman 2003: 79).

Question #6. It is characteristic of verbal participles that they can be used in bound predicative with verbs komma (‘come’) and bli (‘become’/‘remain’), whereby the former mainly applies to participles derived from intransitive verbs of motion (or even denoting a sound accompanying the motion) and the latter mainly to participles derived from intransitive verbs of position. The function of the verbs komma and bli in such uses resembles the function of auxiliary verbs (Thurén 2008; SAG 2 1999; SAG 3 1999).

Question #7. Another syntactic criterion distinguishing between adjectival and verbal participles is the ability to appear as a bound predicative to vara ‘be’, which is only possible if the present participle has adjectival properties rather than verbal ones. Only some present participles have or can have adjectival properties and appear as complement of vara (SAG 2 1999).

Nevertheless, Thurén (2006, 2008) mentions the existence of a “marginal construction corresponding to English progressives”:

In the study we aim to reexamine the reliability of the construction [SBJ vara PRS.PTCP] as a test for the participle’s status, as we are confronted with ambiguous instances of [SBJ vara PRS.PTCP], both with participles used as adjectival predicatives and with complements of vara in form of participles that appear to have a rather verbal reading, see examples (3a–d). However, since the latter is said to be ungrammatical in standard Present Day Swedish and yet is found across different corpora, we wish to establish to what extent this is a widespread phenomenon, found across genres of different levels of formality. In the following section we present the corpora used in our study and the methodology before discussing the results.

4 Materials and Methods

In this section we present the corpus material analysed in this paper, in particular which corpora were selected and what search query was used.

The study is based on material extracted from three Språkbanken corpora. Språkbanken is an open-access research infrastructure for language data published by the Department of Swedish, Multilingualism, Language Technology at the University of Gothenburg. The infrastructure includes the largest corpus of the Swedish language called Korp (Borin et al. 2012). For the study, three smaller corpora have been selected, namely Bloggmix 2016 (blog texts from the category Sociala medier ‘Social media’, abbreviated to BM), SVT nyheter 2016 (from the category Nyhetstexter ‘News texts’, abbreviated to SVT) and Akademiska texter (‘Academic texts’, abbreviated to AT). These particular corpora have been selected to ensure the variation of registers in the language use. The construction [SBJ vara PRS.PTCP] is said to be ungrammatical in standard Present Day Swedish but seems to appear in informal, spoken language and in unedited texts. Therefore, we wish to compare the characteristics of the said constructions found across three registers spanning from the most informal (blog texts), through more formal (news texts), to the most formal written language (academic discourse). As the academic texts corpus includes only two relatively small subcorpora, it is selected in its entirety. The Bloggmix corpus contains subcorpora with blog texts written between 1998 and 2017 (with over 600 million tokens); similarly, the SVT nyheter (‘SVT news’) corpus contains texts from the period 2004–2023 (with over 240 million tokens). From these latter two, the corpora from 2016 are selected, respectively Bloggmix 2016 and SVT nyheter 2016, based on their relative similarity in size and year of data collection. Table 3 gives the size of each corpus selected for the study.

The search method in the Språkbanken corpora is fairly straightforward; our initial search query is given in (23). The first word in the string is the verb vara ‘be’ in all its forms (respectively, infinitive, preterite, supine and present). The second word (directly following the verb) is a word ending in -ande, -ende, which are all the possible endings of the present participle (without the suffix -s). We disregard the endings with the suffix -s (-andes and -endes), as the searches including these forms did not provide any relevant examples. There were no examples of [SBJ vara PRS.PTCP-s] in the AT corpus, only 5 examples in SVT (all with the participle ovetandes ‘unaware’) and 18 examples in BM (these, however, included only four distinct participles, all of which derive from positional verbs: liggandes ‘lying’, ståendes ‘standing’, sittandes ‘sitting’ and sängliggandes ‘bedridden’). The disappointing number of examples and their virtual lack of variation are the basis for the exclusion of the s-forms from the current study.

Additionally, we add several exclusion criteria to the search query. We exclude the class of proper names (egennamn) and full words ande and ende (the former means ‘spirit’ in Swedish, the latter is most likely a misspelling). We also exclude some words ending in -ande/-ende, which are lexicalized to such a degree that they no longer function (or, in some cases, have never functioned) as present participles. These include deverbal nouns formed with the suffix -ande/-ende (e.g., ordförande ‘chairman’, förhållande ‘relationship’, förtroende ‘confidence’, meddelande ‘message’, etc.) and highly lexicalized adjectives, adverbs, and prepositions (e.g., beroende ‘dependent on’, fortfarande ‘still’, beträffande ‘concerning’, gällande ‘regarding’, etc.). The choice of the excluded words was based on previous smaller pilot studies done with the three Språkbanken corpora.

We also exclude from the analysis examples of clearly misspelled constructions, in which the present participle is used instead of an adjective or past participle (e.g., Om USAs nära allierade i de arabiska oljediktaturerna säger han « Deras dagar är räknande! » ‘About USA’s close allies in the Arab oil dictatorships, he says «Their days are counting! »’ intended: räknade ‘numbered’; BM).

The remaining examples were annotated according to the animacy of the grammatical subject: animate and inanimate; and orientation: S where the grammatical subject was the single participant of the intransitive clause and A for the agent-like participant of transitive clause. For the A participles, the objects can be expressed explicitly (hämmande för studenter, ‘repressive for students’), incorporated (narcissistskapande ‘narcissist-creating’) or implicit (kontrollerande ‘controlling’).

In the next section we present the results of the corpus study. First, we turn to a revision of major types of participles found in the [SBJ vara PRS.PTCP] construction (Section 5.1), and second, we discuss the more challenging examples, including possible verbal participles (Section 5.2).

5 Results

The first search including all words which follow a form of the verb vara ‘be’, ending with either -ande or -ende returned 3456 hits in AT, 7734 in SVT and 3187 in BM. Many of the hits were duplicates of each other, in the SVT data because the same article was published twice and in the BM data because the same blog post can appear on several subpages of the blog; they were eliminated. Next step involved elimination of other forms that were deemed lexicalized based on their frequency in the chosen corpora (forms with an external Experiencer were especially frequent, e.g., oroande ‘worrying’, spännande ‘exciting’, frustrerande ‘frustrating’, as the SVT material contains many incident reports and opinion pieces and blog post from the BM data usually consists of emotionally charged content). We did not consult the grammatical information provided in the dictionaries by the Swedish Academy, SAOL, SO and SAOB: none of the existing dictionaries categorizes non nominal -ande/-ende-words as participles, they can be presented either as verbal inflection forms or adjectives, but the criteria for the choice of convention are not explicitly explained (see Holmer 2016, 2022). We excluded also forms which turned out to be modifiers or nouns, or forms which were a part of a construction different than [SBJ vara PRS.PTCP] (see (24), containing small clauses).

After the elimination process, the number of hits accepted for further analysis was reduced to 1916 in total. An overview of the results is given in Table 4.

Two comments are due with respect to the data. Firstly, we find that the A-oriented participles prevail. Secondly, the subjects of the [SBJ vara PRS.PTCP] construction are mainly inanimate. The group that is of main interest in the context of the present study includes A-oriented participles with animate subjects; it is within this group that we may expect to find verbal participles, rendering the whole construction reminiscent of the English progressive (type: The man is painting a wall.) rather than a copula with an adjectival complement (type: The book is fascinating). Before diving into these examples, we will briefly review the general characteristics of the participles attested.

6 Discussion and Conclusions

In our study we have looked at the present participles used with copula vara ‘be’ across three corpora of PDSv. The construction is traditionally used as a test of the present participle’s categorial status, as only the adjectival participles are said to appear in it while the verbal participles are disallowed (see Section 3). We have searched three PDSv corpora for all forms ending in -ande or -ende. Having excluded all lexicalized forms (e.g., nouns such as ordförande ‘chairman’) we have sorted the remaining examples into A- and S-oriented participles (see Section 2) and within each category according to the animacy of the subject of the underlying verb. The resultant sample consists of 1916 examples (see Section 5, Table 4). When analysing the examples, we have found some recurrent patterns, such as the presence of external experiencer (see also Brekke 1988) and incorporated objects or particles (Section 5.1). These have already been observed in earlier studies (Brekke 1988; Thurén 2008), but our investigation gives them a strong empirical foundation. The common denominator for all participles discussed in Section 5.1 is that they are close to adjectives in meaning, as they describe actions or events which happen regularly or all the time, becoming properties rather than descriptions of events. Furthermore, Thurén (2008: 14) points out that denotation closer to properties is even reflected in the English translation of the Swedish participles, which is also confirmed in our presentation of the examples. The participles discussed in Section 5.1 uphold thereby the reliability of the construction [SBJ vara PRS.PTCP] as a test to verify participle’s adjectival status.

We now turn to a discussion of such examples that seem to contradict the standard assumption that verbal participles are disallowed as predicatives with vara ‘be’ in standard PDSv (presented in detail in Section 5.2). They include, in particular, the participles with underlying verbs with agentive subjects, both S- and A-oriented. The following observations can be made.

For the A-oriented participles, it can be noted that the object of the underlying verb is either incorporated in the participle (see examples (34), (36), (37)) or reduced (see (30), (31), (35)). We have not found a single instance of the construction with the object given as a free lexeme (compare with the English gerund in The man was painting a picture). The reduced valency renders the participle less verbal (and more adjectival). Consider example (34), repeated here as (39).

Incorporation results in a non-specific interpretation of the incorporated nominal. In her study of incorporation across languages, Mithun (1984) notes that the predicate that is a result of incorporation must denote an institutionalized activity, meaning that the incorporated object loses its individual salience both semantically and syntactically. No longer referring to a specific entity it simply narrows the scope of the V (Mithun 1984: 856). Mithun’s study is concerned with incorporation proper (morphological noun incorporation), i.e., a process whereby an N stem is compounded with a V stem to yield a larger, derived V stem. In the case of our data, the noun is incorporated in a deverbal form, a participle. The discussion of the process in PDSv exceeds the scope of the present paper, however, we note an important parallel in the object incorporation with participles and incorporation proper. The loss of individual salience of the object renders the whole predicate non-specific, denoting a type of activity (e.g., renskötande ‘reindeer herding’) rather than a specific event (sköta renar/renarna ‘herd reindeer.PL / the reindeer.PL’). Note that PDSv allows for a syntactic noun incorporation, e.g., sköta ren ‘herd reindeer.0’, whereby the nominal denoting the object is used without determiners and with a different prosody (the verb and its object are part of the same phonological phrase, i.e., the stress is only on the object). As regards the semantics, the focus is only on the verb rather than on both verb and noun, which is interpreted non-specifically, in contrast to examples with determiners, where specific reindeer is the object (see also Asudeh and Mikkelsen (2000), an account of syntactic noun incorporation in a closely related Present Day Danish).

Another valency-reducing operation is a complete reduction of the object, illustrated in (35), repeated here as (40).

The complete reduction of the object renders the activity itself salient, rather than its object, resulting in a habitual or omnitemporal reading; the activity is viewed as pluriactive, executed on multiple objects (note the wording: får oss alltid att jobba ‘makes us always work’), which are not clearly defined and thus becomes a property of the Agent rather than a specific process. This reading is further enhanced by the use of alltid ‘always’ and the present tense; the activity is no longer temporally anchored to a specific moment in time but rather can be viewed as atemporal.

For the S-oriented participles, there are no objects to be incorporated or reduced. In fact, it is within this group that we have found the most unexpected examples, admittedly, of low frequency. Consider example (41).

When a larger context is taken into account, we find that the fragment Carlson var samtalande och lyssnande ‘Carlson was talking (to, it can be assumed, many people) and listening (again, to many people)’ is contrasted with ‘got what he wanted’. It seems that the repetitive reading of ‘talk and listen’ is made to illustrate, again, a property rather than an activity of the Agent. The Agent is shown to seem open to the suggestions of others, only to do as he wanted in the end.

However, this reading becomes excessively difficult with some of our examples. Consider (38a), repeated here as (42).

As we have discussed in Section 5.2, the context suggests a contrast between walking and using a wheelchair. Even if it is not a straightforward property, the participle patterns with the A-oriented participles discussed above, where we noted that the presentation of the activity as executed on many different objects (which was achieved by either incorporating or reducing the object) highlights the activity itself rather than its result on the Patient.

Another possible reading is to interpret the ambiguous -ande/-ende forms with vara as nominalizations, consider (43) and (44), where stridande can be read as either participle ‘fighting’ or noun ‘fighter’ and medföljande can be read as ‘accompanying’ or ‘accompanying person.’ Such nominal interpretation allows avoiding a verbal one, similar to the English progressive.

According to the reference grammars of PDSv and tests presented in Section 3, the participles in (43–44) are considered inadmissible as predicatives with vara. A number of examples discussed in the present paper, if seen in isolation, would be baffling. However, as we have seen, in the majority of cases, verbal reading is suppressed through either incorporation or reduction of the object and through the loss of temporal anchoring. In the absence of these strategies, we were still able to provide an adjectival interpretation for these examples. Does that mean that the construction [SBJ vara PRS.PTCP] is still a reliable test for the categorial status in PDSv?

Before we turn to this question, we would like to briefly consider the parallels between these examples and the English progressive, as illustrated by (45a–b) and referring to (38c).

At the same time, the Swedish example in (46a) also has its almost exact counterpart in English, (46b).

However, the structures are not the same, as the English verb be is a copula in (46b) but an auxiliary in (45b), i.e., in the progressive construction. In English, the progressive tense may have evolved from constructions such as (46b) and part of this evolution was auxiliarization of be (see e.g., Fonteyn 2019, see de Groot 2007 for an alternative account deriving the progressive from an absentive construction with a nominalization). For the Swedish data, however, we find that examples with vara ‘be’, and verbal present participles are relatively few. There is too little to suggest that a similar language internal process of auxiliarization is taking place in Swedish, i.e., we do not have enough data to suggest that Swedish is developing a progressive construction, with auxiliary vara ‘be’ and verbal present participle. Extensive contact with English cannot be excluded as the source of some of these examples; we note that some troublesome examples come from two genres which remain under a heavy influence of English, i.e., on the one hand, academic texts, where most of the relevant literature is in English and many writers use English when writing academic texts; and on the other hand, blogs, with their less formal language. We only found a few similarly baffling examples in the SVT corpus, which we may assume is more resistant to structural change following language contact. We may further note that similar examples are also found in the truly informal Flashback corpus, note the use of gråtande ‘crying’, levande ‘living’ and cyklande ‘cycling’ in example (3a–c) in Section 1, repeated here as (47), interestingly all of them are S-oriented participles with the -s-suffix.

The Flashback forum material is definitely an avenue to be pursued in further research. For the present paper, our goal was to establish whether the verbal participles can be found in [SBJ vara PRS.PTCP] construction across genres which differ in formality. This in turn resulted in the choice of smaller corpora; with its 4,53G tokens Flashback is at least four times as large as all three corpora used in the present study and for this reason it was discounted. Although our material didn’t show a sufficient number and variation of examples ending with the suffix -s (-andes and -endes), [SBJ vara PRS.PTCP-s] has a potential for further exploration.

Whether the examples we see are due to contact with English or a result of language internal development, we note that the construction as such overwhelmingly allows adjectival participles only. The verbal, or verbal-like, participles that we have found are sporadic, while the majority of examples can be given an adjectival interpretation. This suggests that if a present participle is placed in an adjectival predicative slot (with the verb vara ‘be’), the main motivation on the part of the speaker is to exploit the adjectival construction’s semantics, i.e., present the action as time-stable (or even atemporal), to the point of it becoming a property. As Langacker (1987) points out, this time-stable quality is further reinforced by the stative vara ‘be’ (Langacker 1987: 304, see also De Smet and Heyvaert 2011). Our analyses and comments on the possible readings are in line with observation included in the reference grammar of the Swedish Academy about the graduality and the language users’ individual perception of the distinction between participles’ adjectival and verbal meaning (SAG 2 1999: 583).

On the other hand, the mere fact that there are examples such as (42) (var gående ‘was going/walking’), (45a) (varit bloggande ‘have been blogging’), and (47) (är gråtandes ‘is crying’, är levandes ‘are living’, var cyklandes ‘was cycling’) suggests that there may be a real change underway. Once the adjectival reading becomes untenable (as in (47)), perhaps a progressive construction will have become part of the Swedish grammar. Existence of a marginal construction is acknowledged by Thurén (2006: 157), but only as a footnote mention, but the structure remains unnoticed in Blensenius (2009) who focuses on other structures in Present Day Swedish which correspond to the progressive construction, such as pseudo-coordinations, constructions with hålla på ‘keep on’ or even some participial constructions: directive constructions with komma, locative constructions with bli and additive constructions with various verbs.

7 Summary

The aim of this paper was to reexamine the Swedish present participle’s status, specifically in the construction with the verb vara, traditionally reserved for present participles with adjectival properties and disallowing verbal forms. We have presented an overview of the material from three corpora of Present Day Swedish representing different genre, including examples that appear to challenge this conventional view. The model of our analysis was based on the notion of participles’ orientation and complemented with some properties of subjects, namely animacy and thematic roles. Such design has proven to be relevant for the study, as both the character of the participles and the ability to appear in the construction in question differ depending on these properties – different patterns and different lexical restrictions are noted for A- and S-oriented participles derived from verbs with various types of subjects. Participles with a potential verbal interpretation appear occasionally among the S-oriented participles, derived from intransitive verbs, with Theme/Experiencer, and even Agent subjects. For A-oriented participles derived from transitive verbs there is a possibility to suppress the object, which leads to a more adjectival reading.

The construction seems also to vary across genres which differ in formality and in susceptibility to the English influence, the most striking examples noted in academic texts and blogs, which may constitute a starting point for further research, encompassing other corpora or even acceptability judgment studies. The most challenging examples are admittedly few. They allow, however, to draw a conclusion about the reliability of the construction [SBJ vara PRS.PTCP] as a test for the categorial status in Present Day Swedish, i.e., that the construction with the verb vara is on all accounts so clearly related to an adjectival complement that even if it appears with a participle of more verbal character, that the verbal reading, suggesting traces of a possible language change, is suppressed through other lexical restrictions and is avoided by an adjectival or even nominal interpretation.