Root participles: directive, commissive, expressive and representative participles in Germanic root configurations

Abstract

The present paper investigates participial root configurations, i.e. participial clauses that are grammatically independent of a host clause. Unlike previous work, which has focussed on either directive or (non-directive) performative uses of so-called past participles (i.e. participles that have passive and/or perfect(ive) interpretations), the present paper establishes a typology of ‘root participles’ in Germanic and contrasts the properties of four main types: (1) directive (RP^dir), (2) expressive (RP^exp), (3) commissive (RP^com), (4) representative root participles (RP^rep). The main claim with respect to the properties of these distinct types is that they differ in terms of whether they include a verbal or an adjectival (passive) participle. In fact, arguments based on argument structure, orientation, aspect, and adverbial modification are presented to substantiate the claim that types (1) and (2) are formed with verbal and types (3) and (4) with adjectival participles. Additionally, the distinct types will be shown to differ in their status of either being non-sentential (i.e. structurally different from potential clausal counterparts) or merely elliptical (just phonologically reduced): types (1) and (3) can be shown to be non-sentential and hence receive a dedicated syntactic analysis, where special attention is paid to the contribution of the (imperative vs. declarative) left periphery.

1 Introduction

In contrast to periphrastic and adnominal uses of perfect and passive participles, the occurrence of so-called past participles¹ as the core of non-finite root clauses, i.e. without a clausal host to attach to, is still notoriously understudied. In fact, apart from some works which merely treat them in passing (most notably Fries 1983), there have only been very few attempts in the recent past to shed light on these uses: Rooryck and Postma (2007), Coussé and Oosterhof (2012) as well as Heinold (2014) study directive participles, and Ørsnes (2020) focusses on non-directive performative participles in root configurations. Prototypical examples of these two kinds in German and Dutch are provided in (1) and (2).

(1)	a.	Abmarschiert!		a’.	Afgemarcheerd!
		away.marched			away.marched
		‘Go away!’			‘Go away!’
	b.	(HSV-Fans)	aufgepasst!	b’.	(Ajax-fans)	opgepast!
		(HSV-fans)	out.watched		(Ajax-fans)	out.watched
		‘(HSV-fans) pay attention!’			‘(Ajax-fans) pay attention!’
(2)	a.	Versprochen!		a’.	Beloofd!
		promised			promised
		‘I promise!’			‘I promise!’
	b.	Abgemacht!		b’.	Afgesproken!
		agreed			agreed
		‘It is settled!’			‘It is settled!’

While the cases in (1) are used as commands and intended to commit the addressee to a course of action, those in (2) allow the speaker to perform commissive speech acts. This suggests that the two are associated with distinct (imperative vs. declarative) force-operators in the C-domain, which may but need not be selectionally restricted to non-finite participial phrases. Immediate support for this claim comes from the alternation of directive participles with bare infinitives like Aufstehen! (‘Stand up!’) and Aufpassen! (‘Pay attention!’). Although these are subject to fewer restrictions than their participial counterparts, they likewise trigger imperative readings (see Fries 1983). This availability does not carry over to infinitival counterparts of commissive cases like *Versprechen! (lit. promise.inf) and *Abmachen! (lit. agree.inf), which are ungrammatical.

Participles denoting directive and commissive speech acts are available in German and Dutch, but not in the other Germanic languages (consider, for instance, *Pencils sharpened! and *Promised! and their Norwegian counterparts *Blyanter skjerpet! and *Lovet!). Additionally, Dutch—unlike German and the other Germanic languages (consider *Thanked! and *Congratulated! in English and *Tackat! And *Gratulerat! in Norwegian)—allows for the use of participles in root configurations for expressive speech acts, as in (3).

(3)	a.	*Gegrüßt!	a’.	Gegroet!
		greeted		greeted
		‘Greetings!’
	b.	*Gedankt!	b’.	Bedankt!
		thanked		thanked
		‘Thanks!’

Although this contrast raises questions concerning the parameterisation of root participles, what the Germanic languages share is that they never allow for declarations to be conveyed with the help of root participles: German cases like Hiermit… *Getauft! *Verheiratet! *Gesegnet! *Gefeuert! *Festgenommen! (lit. herewith… baptised! married! blessed! fired! arrested!) are utterly ungrammatical (cf. Ørsnes 2020, 343).² Another common characteristic is that all Germanic languages also give rise to declarative root participles with assertive and verdictive readings, as in the examples in (4), once more from German and Dutch, and (5), taken from Norwegian, all of which are also available in English.³

(4)	a.	Erledigt!		b’.	Gedaan!
		done			done
		‘Done!’			‘Done!’
	b.	Gut	gemacht!	b’.	Goed	gedaan!
		well	done		well	done
		‘Well	done!’		‘Well	done!’
(5)	a.	Problem	løst!	b.	Godt	utført!
		problem	solved		well	executed
		‘Problem	solved!’		‘Well	executed!’

Given that German and Dutch are most flexible with respect to the range of possible root participles, we will mostly focus on these two Germanic languages. However, in the discussion of representative root participles, we will also rely on data from English (and occasionally also mention Scandinavian languages where relevant).

Whereas previous work has simply focussed on either directive cases, as in (1), or (non-directive) performative uses of past participles, as in (2), the present paper starts off by establishing a typology of ‘root participles’. This term is used in analogy to so-called root infinitives like Mary go and Sie schlafen (lit. she sleep.inf) from child language (cf. Kupisch and Rinke 2007) as well as various types of infinitival forms like Noch einmal Champagner schlürfen (‘I would that I could sip champagne one last time.’) and Alle Kinder ins Bett gehen (‘All kids, go to bed now!’) in root uses (cf. Lohnstein 2019, 35). Based on their force specifications and the speech acts they convey, we can distinguish four main types of root participles: (i) directive, (ii) commissive, (iii) expressive, and (iv) representative (bare, assertive, and verdictive) participles. These will be shown to differ along the lines of whether they are proper non-sentential structures (structurally different from their potential clausal counterparts) or rather just elliptical configurations (structurally fully present, but subject to phonological deletion). Most importantly, however, they will be argued to differ in terms of being formed on the basis of either verbal passive participles or (resultative) adjectival participles. In fact, directive and commissive root participles will be shown to be non-sentential, based on verbal as opposed to adjectival participles. Root participles in directive configurations will be shown to be introduced by an imperatival functional head, which plays a crucial role in licensing a so-called quantificational subject that serves as the overt external argument (which otherwise remains implicit in passive configurations). The adjectival participles in commissive configurations, on the other hand, lack an external argument altogether and are introduced by a performative C-head that ties the impoverished participial event to the speaker. Unlike these two, expressive and representative root participles are argued to boil down to elliptical passive periphrases, on the one hand, and adjectival passives (also called stative passives), on the other. The flexibility of the latter type with respect to orientation is traced back to the interplay of a reduction in argument structure and informativeness, whereas the robust speaker- as opposed to hearer-orientation of directive and commissive root participles hinges on the presence of dedicated functional heads. Since these are argued to head a structurally reduced domain in both cases, we will zoom in on the syntax of these non-sentential structures and differentiate them from the elliptical types of root participles. This will be done against a minimalist (anti-lexicalist) backdrop, although the analysis laid out below can be translated into other frameworks in a straightforward fashion.

To lay out a principled analysis of the distinct types of root participles, Sect. 2 first provides a typology of root participles by focusing on distinctions along the lines of force specifications and speech acts in Sect. 2.1. The bulk of Sect. 2 is devoted to the properties of the distinct kinds of root participles: argument structure (Sect. 2.2), orientation (Sect. 2.3), aspect (Sect. 2.4), and adverbial modification (Sect. 2.5). In Sect. 2.6, these are shown to unequivocally point to the traditional dichotomy of verbal and adjectival participles. Section 3 critically evaluates previous approaches and turns to questions of ellipsis. Based on this discussion, a novel approach to the syntax of root participles is presented in Sect. 4. Finally, Sect. 5 draws some conclusions and points to some open questions.

2 The typology of root participles

The present section will provide an overview of the grammatical properties of the four main types of root participles: (i) directive (RP^dir), (ii) expressive (RP^exp), (iii) commissive (RP^com), (iv) representative (bare, assertive and verdictive) root participles (RP^rep). The discussion of force specifications and speech acts (Sect. 2.1), the individual argument structure (Sect. 2.2), orientation (Sect. 2.3), aspectual (Sect. 2.4), and modification (Sect. 2.5) properties culminates in an overview of the intricate distinctions between these types (Sect. 2.6). This will serve as the basis for the principled distinction into root participles that are based on verbal (passive) participles and those based on adjectival (resultative) participles, which in turn forms the core of the analysis to be laid out below.

2.1 Force specifications and speech acts

Apart from syntactic and semantic contrasts along the lines to be laid out in the following subsections, a central dimension of difference that allows us to distinguish root participles into several subtypes in a principled way are their force specifications (mood) and the speech acts they may serve to perform.

Imperative force may be encoded by three distinct morphological means in the Germanic languages. English only makes use of a default form of the verb (Run!) in the context of an imperative head encoding 2nd person properties. German, Dutch and the Scandinavian languages either employ a dedicated finite form that is overtly inflected for imperative force (consider, e.g., Rök inte! in Swedish and Rauch(e)/Raucht nicht! ‘Don’t smoke!’ in German) or resort to a bare infinitive (Inte röker! and Nicht rauchen! ‘Don’t smoke!’). The latter shows that force specifications may also be associated with non-finite verb forms. A third option that crops up in German and Dutch and is the one that is of main interest for the purposes of the present paper is employing so-called past participles to convey directive speech acts (cf. Heinold 2014, 314).

(6)	a.	Aufgepasst!	Weggetreten!	Abmarschiert!		Hingesetzt!
		pay.attention.ptcp	step.away.ptcp	off.march.ptcp		sit.down.ptcp
	b.	Opgelet!	Ingerukt!	Afgemarcheerd!	Niet	geklaagd!
		look.up.ptcp	in.pull.ptcp	off.march.ptcp	not	complain.ptcp
		‘Pay attention!	Dismissed!	March off!	Sit down!	Do not complain!’

This use has often been taken to be very restricted. Hoeksema (1992), for instance, assumes participial imperatives in Dutch to be a lexical, idiomatic quirk and, in a similar vein, Aikhenvald (2010, 283) claims for German that “‘participle commands’ are restricted to a few motion and posture verbs”. As has more recently been shown in Rooryck and Postma (2007), Coussé and Oosterhof (2012) and Heinold (2014), however, participles with a directive function are actually fairly productive and subject to less restrictions than previously suggested. While comparing the frequency of bare infinitives and their participial counterparts shows that the latter are indeed more restricted (cf. Wunderlich 1984, 98), this should not hastily be exaggerated. In fact, Heinold (2014, 332) even suggests that there are no grammatical restrictions whatsoever in German, although simplex forms (say *Gesetzt! rather than Hingesetzt! ‘Sit down!’ and *Geleistet! rather than Widerstand geleistet! ‘Put up a fight!’) seem to be disfavoured. Coussé and Oosterhof (2012, 51f.) show that the main restriction in Dutch is morphological in nature, namely that participial imperatives be weakly conjugated (and hence *Opgestaan! and *Hiergebleven! are ungrammatical, unlike their German counterparts). Differences in frequency could eventually be due to the large degree of ambiguity that the participle exhibits in comparison to other modes of encoding imperative force (cf. Heinold 2014, 332). Whenever forming an imperative root participle is possible, it may be uttered out of the blue, but is restricted to the here and now (cf. Rooryck and Postma 2007, 285) and directed towards addressees who are currently present (cf. Zifonun et al. 1997). In the literature, these root participles are usually analysed as non-sentential configurations based on a verbal passive (Rooryck and Postma 2007, 273f.) or perfect participle (Fries 1983, 236).

The other types of root participles are clearly not associated with imperative force. Rather, all of the root participles in (7) and (8), adopted from Ørsnes (2020, 345) and Fortuin (2019, 31), as well as (9) and (10) may be argued to be associated with declarative force, despite giving rise to different performative and representative speech acts.

(7)	a.	Versprochen!	Abgemacht!	Akzeptiert!
		promise.ptcp	settle.ptcp	accept.ptcp
	b.	Beloofd!	Afgesproken!	Geregeld!
		promise.ptcp	settle.ptcp	accept.ptcp
		‘I promise!	That is settled then!	I accept!’
(8)		Bedankt!	Gegroet!	Gecondoleerd!
		thank.ptcp	greet.ptcp	condole.ptcp
		‘Thank you!	Greetings!	My condolences!’
(9)	a.	(Hausaufgaben) erledigt!	(Einspruch) abgelehnt!	Problem gelöst!
	b.	(Homework) done!	(Objection) overruled!	Problem solved!
(10)	a.	Goed gedaan!	Schlecht gespielt!
	b.	Well done!	Badly played!

The root configurations in (7) and (8) are performative: uttering them triggers Searle’s (1976) commissive speech acts in the former (where speakers commit to future actions) and expressive speech acts in the latter (where speakers express psychological states).

A prototypical case for the proper use of commissive participles in a suitable discourse context may be found in (11).⁴

(11)	A:	Papa,	gehen	wir	morgen	in	den	Zoo?
		dad	go	we	tomorrow	in	the	Zoo
		‘Dad,	will we	go	to the zoo tomorrow?’
	B:	Versprochen/Abgemacht,	mein	Sohn!
		promise.ptcp/settle.ptcp	my	son
		‘I promise, my son!’

The root participle denotes the father’s promise of going to the zoo (on the next day). This promise is overtly directed towards the son, who appears as a (structurally independent) vocative. In the absence of a vocative, the immediate discourse context—in which the root participle is tightly anchored with the help of a propositional variable that is resolved anaphorically (cf. Ørsnes 2020: 377)—makes it clear that the hearer is the one who the speaker makes the promise to.⁵ Fries (1983, 236) simply analyses such root participles as elliptical stative (i.e., adjectival) passives, whereas Ørsnes (2020, 385) suggests that performative participles are non-sentential configurations based on a verbal passive participle.

A third type of root participle with performative import is the expressive type, which is available in Dutch but not in German, as observable in (12) (cf. Rooryck and Postma 2007, 274).

(12)	a.	*Gedankt!	*Kondoliert!	*Gegrüßt!	*Beglückwünscht!
		thanked	condoled	greeted	congratulated
	b.	Bedankt!	Gecondoleerd!	Gegroet!	Gefeliciteerd/Gelukgewenst!
		thanked	condoled	greeted	congratulated
		‘Thanks!	Condolences!	Greetings!	Congratulations!’

Dutch is quite flexible with respect to the formation of expressive root participles, whereas German does not allow for root cases and only allows for merely partially reduced cases like Sei gegrüßt! (‘Be greeted!’) or Ihm sei gedankt! (‘He shall be thanked!’). Similar cases optionally arise in Dutch, where the fixed form Wees gegroet! (lit. be greeted) (cf. Coussé and Oosterhof 2012, 34) shows that we are dealing with an imperative form rather than a subjunctive form here, as Dutch lexically distinguishes the two forms of be (wees vs. zij), unlike German (sei in both cases). Discourse anchoring seems to be achieved primarily via the presence of a covert addressee, whom the performative is directed to. Rooryck and Postma (2007, 273f.) claim—in passing—that expressive root participles are merely elided variants of eventive passives.⁶

The configurations in (9), e.g. Done! and Objection overruled!, and (10), e.g. Well done! and Badly played!, above, on the other hand, are primarily associated with representative speech acts in that an achievement is either just reported or evaluated verdictively. Even though some representatives like Overruled! seem to trigger performative acts as well, this could also be traced back to merely a statement about observing the law (cf. Dunn 2003, 514). Rather than proper declarations like I hereby pronounce you husband and wife, which bring about a given state of affairs immediately, the resultative statement is all there is with these root participles. In a similar vein, examples like (Class) dismissed! and (Meeting) adjourned!—conventionalised in the sense that they are typically restricted to official contexts like military situations, judicial settings as well as work or school contexts—cannot be analysed as imperative, but rather may be grouped with representative root participles. This becomes clear once we consider the verbal semantics of the underlying predicates, which do not direct somebody to dismiss or adjourn, but rather the hearer may be part of the gatherings denoted by the respective (implicit or explicit) internal argument. In other words, what is being communicated is a resultative decision as reached by somebody in charge in an official context; i.e., nobody is actively commanded to do something with the help of an imperative utterance, but rather somebody simply decides that the meeting is adjourned or the class is dismissed.⁷ The interchangebility between cases with (Objection overruled!) and without (Overruled!) a bare nominal item exemplifies that such root participles may more or less strongly be tied to the discourse. In both cases, the participial result (anaphorically) relates to some entity of relevance in the discourse before, though.

The root participles in (10), while closely resembling the assertive cases in (9), crucially differ in terms of evaluating an event. The presence of an evaluative adverb is crucial for obtaining this interpretation. In contrast to proper declarations, which are not able to surface as root participles (*Verheiratet! *Getauft!), these verdictives may also be seen as carrying out representative rather than performative speech acts. In contrast to their assertive counterparts, the situation which is attributed to the hearer (rather than the speaker) is evaluated by the speaker with the help of an adverbial modifier. With respect to their force specifications, there are two potential alternatives to merely analysing verdictives as declarative. Rett (2012) distinguishes exclamatives from sentence exclamations. The latter (e.g., John bakes delicious desserts) semantically introduce the negation of an expectation. Exclamatives (e.g., in What delicious desserts John bakes) in turn introduce degree properties and denote that an expectation has been surpassed. Neither of these meaning contributions clearly arises from verdictives: a verdictive like Gut gekocht! neither necessarily asserts that a (scalar) speaker-expectation has been surpassed nor that it needs to be rejected entirely. In fact, the example given can be uttered even if the cook is famous for her tasty food and the quality of the cooking (hence the quality of the food that results from it) does not surpass any expectations held by the speaker. Accordingly, verdictives are more closely related to sentence exclamations than exclamatives but cannot be equated with either of the two: a proposition is asserted that is objectively evaluated by the speaker, without the necessity for the negation of an expectation or it being surpassed. Both types of representative root participles, i.e. those denoting assertive and those giving rise to verdictive speech acts, have been analysed as elliptical active perfect configurations in the literature (see Wunderlich 1984, 114fn14; Rapp and Wöllstein 2009, 167).

The distinct types of root participles, as exemplified in (6) to (10) will henceforth be termed directive (RP^dir), commissive (RP^com), expressive (RP^exp), and representative (RP^rep) root participles. The central contrasts between these in terms of different force specifications and speech acts are summarized in Table 1.

Table 1 The force specifications of root participles

While the force distinction arguably needs to be encoded by distinct functional heads in the C-domain, the specific type of speech act need not necessarily be hard-coded syntactically. Rather, as in their clausal counterparts, it is derived pragmatically in the case of RP^exps and RP^reps. With RP^dirs and RP^coms, which will be argued to be non-sentential, the specific type of speech act may be traced back to an imperative or performative head, associated with a hearer- or speaker-orientation, respectively.

2.2 Argument structure

As we will see in the following sub-sections, the distinct types of root participles strongly differ with respect to the licensing of external arguments (Sect. 2.2.1), but also exhibit some intricate contrasts concerning the licensing of internal arguments (Sect. 2.2.2).

2.2.1 External arguments

As we can see by comparing the German, Dutch and English examples in (13) with those in (14), all root participles allow for the introduction of structurally independent vocatives. Only RP^dirs, however, license nominal entities that serve as proper external arguments. Unlike vocatives, which may not be analysed as arguments but just (further) specify the addressee, these are prosodically and syntactically integrated in the clause (cf. Fries 1983).⁸

(13)	a.	Aufgepasst,	Alex!	Der	Unterricht	beginnt.
		pay.attention.ptcp	Alex	the	lesson	begins
		‘Pay attention, Alex! The lesson is starting.’
	b.	Bedankt,	Robert!	Het	was	een	mooie	avond.
		thank.ptcp	Robert	it	was	a	nice	evening
		‘Thanks, Robert! It was a nice evening.’
	c.	Versprochen,	Anja!	Ich	bin	um	18	Uhr bei dir.
		promise.ptcp	Anja	I	am	at	18	o’clock at you
		‘I promise, Anja! I will be with you at 6 o’clock.’
	d.	(Homework) done, mommy! Now I can play.
	e.	Well done, Lena! Keep it up.

(14)	a.	Alle	(Teilnehmer)/	HSV-Fans	aufgepasst/	aufgestanden!
		all	participants	HSV-fans	pay.attention.ptcp	stand.up.ptcp
		‘All participants/HSV-fans, pay attention/stand up!’
	b.	*Ik/*We bedankt voor een mooie avond!
	c.	*Ich/*Wir versprochen/abgemacht! So machen wir es.
	d.	*I/*We (homework) done!
	e.	*You well played!

The declarative cases in (14b) to (14e) do not permit the use of a pronominal (attuned to the speaker-/hearer-orientation of the specific type of root participle) as an external argument (henceforth EA). While this is not surprising in the case of performative speech acts like the RP^exp in (14b) and the RP^com in (14c), as the EA is simply identified with the speaker, it crucially also holds for the RP^reps in (14d) and (14e). Imperative root participles (RP^dirs), in turn, license overt EAs. These clearly differ from vocatives, since they are not structurally independent (as marked by the lack of an intonational break) and may even occur alongside these, as shown in (15) where the vocatives are Robert, du/je (‘you’), and Peter.

(15)	a.	Robert *(,) aufgestanden!	Du *(,) aufgepasst!	Alle aufgepasst, Peter!
	b.	Robert *(,) ingerukt!	Je *(,) opgepast!	Allemaal opgepast, Peter!
		‘Stand up, Robert! Pay attention, you! Everybody pay attention, Peter!’

The EAs in question (alle and allemaal) are overtly licensed and seem to refer exclusively to addressees that are currently present (cf. Zifonun et al. 1997). The same EAs are available in bare infinitival counterparts like Aufpassen! and Oppasen! (‘Pay attention!’). As shown by Reis (1995, 148ff.) for such infinitival cases, they are not proper subjects but rather quantify over addressees, which is why only quantifiers and bare plural nominals are allowed to occur (see also Fries 1983, 52; Rapp and Wöllstein 2009, 168; Gärtner 2013, 204; Heinold 2014, 316). The occurrence of the latter, as in (16) below, either with or without an accompanying overt universal quantifier, shows that these cannot just be reduced to floating quantifiers. This suggests that the addressee is a quantified sum, whose subset-properties may be determined by the noun.

(16)	Alle	HSV-Fans/Teilnehmer/Frauen	zugehört!
	all	HSV-fans/participants/women	listen.up.ptcp
	‘Listen up, all HSV-fans/participants/women!’

In fact, apart from the universal quantifier (alle in German and allemaal in Dutch), other members of the class of floating quantifiers also occur as such EAs, but the same holds for quantificational subjects that do not surface as floating quantifiers, as exemplified in (17).⁹

(17)	(Ein)	jeder	zugehört!	*(Alle) beide	aufgepasst!	Niemand/Keiner hingesetzt!
	a	everyone	listen.up.ptcp	all both	pay.attention.ptcp	nobody sit.down.ptcp
	‘Everybody listen up! Both of you, pay attention! Nobody sit down!’

Crucially, the restrictions concerning the range of possible EAs pertain if the quantification subject remains implicit, as in (18).

(18)	Aufgestanden!	Aufgepasst!	Zugehört!	Hingesetzt!
	stand.up.ptcp	pay.attention.ptcp	listen.up.ptcp	sit.down.ptcp
	‘(Everybody) stand up/pay attention/listen up/sit down!’

An additional argument in favor of the presence of EAs in the argument structure of RP^dirs comes from the use of By-phrases.

(19)	a.	Von	allen	(Teilnehmern)	zugehört!	Von	allen	weggetreten,	jetzt!
		by	all	participants	listen.up.ptcp	by	all	step.away.ptcp	now
		‘(Everybody) listen up/step away now!’
	b.	*Von	allen	(Teilnehmern)	zuhören!	*Von	allen	wegtreten!10
		by	all	participants	listen.up.inf	by	all	step.away.inf

Whereas infinitival items that serve to form bare imperatives fail to introduce BY-phrases due to the lack of passive morphology, their participial counterparts may introduce EAs in the form of agentive BY-phrases, but only if they meet the restrictions to quantificational subjects. In other words, all and only those EAs that surface as overt quantificational subjects also lend themselves to occur in BY-phrases. Accordingly, non-quantificational vocatives like Robert or a 2nd person pronoun in (15) do not appear in such adjuncts. This points to verbal passive properties (demoting the EA in the sense that it may be existentially quantified over)¹¹ but also suggests that the imperative ingredient affects the range of possible interpretations. Unlike in finite imperative counterparts, however, it does not just restrict the range of EAs to 2nd person, but requires it to be a plural set of entities.

A second kind of root participle that allows for the licensing of BY-phrases is the expressive type. Even though these do not allow for the overt realization of a (quantificational) subject in contrast to their imperative counterparts, external arguments readily occur as adjuncts, as in (20). While there are no restrictions as to the range of possible EAs, a proper performative is restricted to entities that are construed as relating to the speaker, as in (20a). If the expressive speech act is supposed to be carried out by proxy, as in (20b), the adjunct is not agentive (door vs. van).

(20)	a.	Smakelijk bedankt	door een tevreden klant.	Gefeliciteerd door ons!
		deliciously thank.ptcp	by a satisfied customer	congratulate.ptcp by us
		‘Thanks from a satisfied customer! We congratulate you!’
	b.	Gegroet van Robert-Jan.	En gefeliciteerd	van mijn moeder.
		greet.ptcp by Robert-Jan	and congratulate.ptcp	by my mother
		‘Greetings from Robert-Jan! And congratulations from my mother!’

The other declarative root participles, i.e. RP^coms and RP^reps, typically do not occur with agentive BY-phrases, but the claim that this is an absolute grammatical restriction is challenged by the occasional occurrence of examples like those in (21).¹²

(21)	a.	?Von	der	Kanzlerin	versprochen!	?Auch	von	mir	versprochen!
		By	the	chancellor	promise.ptcp	also	by	me	promise.ptcp
		‘The chancellor promises that! This is also promised by me!’
	b.	i.	?Well	played by	him!
		ii.	?Gut	gemacht	von	deinem	Schützling!
			well	done	by	your protégé
		‘This was well done by your protégé!’
	c.	?Hausaufgaben	auch	von	mir	erledigt!
		home.work	also	by	me	finish.ptcp
		‘The homework was also done by me!’

According to Ørsnes (2020, 353), the first example in (21a) may be an instance of delegated speech, but the limited grammaticality of the second example shows that by-phrases are generally marked with RP^coms. However, such root participles seem to lose their performative impetus, which is not surprising if we consider that “speaker restriction is a defining characteristic of performatives” (Ørsnes 2020, 372). In fact, such cases seem to be more closely related to assertive RP^reps like (21c). What should keep us from prematurely treating this as unequivocal evidence for the presence of an implicit EA in declarative participles, however, is that all of these cases can be transposed into adjectival passives.

(22)	a.	Das	ist	von	der
		that	is	by	the
		Kanzlerin	versprochen	Das	ist	auch	von	mir	versprochen.
		chancellor	promise.ptcp	that	is	also	by	me	promise.ptcp
		‘The chancellor promises that! This is also promised by me!’
	b.	i	The ball is well played by him.
		ii.	Das	war	gut	gemacht	von deinem	Sohn.
			that was well make.ptcp				by your	son
			‘That was well done by your son.’
	c.	Die	Hausaufgaben	sind auch		von	mir	erledigt.
		the	homework.pl	are	also	by	me	finish.ptcp
		‘The homework was also done by me!’

This is relevant because these are canonically taken to be copular constructions based on adjectival participles (see Kratzer 1994, 2000; Rapp 1996; Maienborn 2007, 2009, 2011), formed with the copula sein (‘be’) rather than the passive auxiliary werden (‘become’) in German. One of the central properties that sets them apart from verbal participles concerns the role of the EA: adjectival participles either grammatically lack an EA altogether (see Kratzer 1994, 2000) or the EA is semantically licensed and hence subject to semantic constraints like other pieces of event-related modification (see Gehrke 2015). The cases in (22), and by implication also those in (21), may be treated as instances of the latter, modifying an event kind of the state that is introduced by the adjectival participle rather than instantiating the EA of a fully instantiated event token (see Gehrke 2015, 900). In a similar vein, we could also claim that only those modifiers are acceptable that describe the result state rather than the verbal event (see McIntyre 2015, 941). In either case, the restricted occurrence of BY-phrases with these types of declarative root participles does not challenge the claim that these lack a proper EA.

The claim that RP^dirs and RP^exps, on the one hand, differ from RP^coms and RP^reps, on the other, in that no EA is grammatically represented in declarative root participles is corroborated by purpose clauses and the reduced accessibility of the subject in discourse continuation. The former is observable in comparing directive and expressive cases like those in (23) to the commissive and representative instances in (24). The two sets of examples exhibit a contrast in acceptability that goes beyond the reluctance of root participles to embed adverbial clauses.

(23)	a.	?Daher	aufgepasst,	um	nichts	zu	verpassen.
		so	pay.attention.ptcp	in.order	nothing	to	miss
		‘So pay attention in order not to miss anything.’
	b.	?Aufgegessen,	um	später	keinen	Hunger	zu	haben.
		eat.up.ptcp	in.order	later	no	hunger	to	have
		‘Eat up so that you are not hungry later.’
	c.	?Gegroet	om	u	vaarwel te			zeggen.
		greet.ptcp	in.order	you	goodbye to			say
		‘I greet you in order to say goodbye to you.’
	d.	?Gefeliciteerd	om	u	te	helpen	uw	succes
		congratulate.ptcp	in.order	you	to	help	your	success
		te	realiseren.
		to	realize
		‘I congratulate you in order to help you realise your success.’

(24)	a.	*Versprochen,	um	die	Bürger	zu	beruhigen.
		promise.ptcp	in.order	the	citizens	to	soothe
		‘This is promised to soothe the citizens.’
	b.	*Well played, in order to win.
	c.	*Hausaufgaben	erledigt,	um	mich	anderen	Dingen	zu	widmen.
		homework	done	in.order	me	other	things	to	commit
		‘The homework is done so that I can turn to other things now.’

Against the backdrop of the assumption that purpose clauses include an implicit subject that may be controlled by a higher EA, the RP^dirs and RP^exps in (23) may be taken to license an implicit EA. The RP^coms and RP^reps in (24), in turn, do not allow recourse to an EA and hence cannot license purpose clauses. As the diagnostic value of purpose clauses for the grammatical status of an EA has recently been put into question (cf. Bruening 2014, 381),¹³ additional evidence should be adduced.

The discourse continuations in (25) further exemplify the degraded status of the EA.¹⁴

(25)

a.

A:

Verprochen!

B:

Nein,

das

ist

es

nicht. /

*Nein,

das

promise.ptcp

no

that

is

it

not

no

that

hast

du

nicht.

have

you

not

‘A: This is a promise, then! B: No, it isn’t!’

b.

A: Gut

beobachtet!

B:

Nein,

das

ist

es

nicht! /

*Nein,

das

hast

well

observe.ptcp

no

that

is

it

not

no

that

have

du

nicht.

you

not

‘A: This is well observed! B: No, it isn’t!’

c.

A: Mission accomplished! B: No, it isn’t. / *No, you did/have not.

(26)

a.

A: Zugehört!

B:

*Nein, das

ist

es

nicht! /

Nein,

das

mache

ich

nicht.

listen.up.ptcp

no that

is

it

not

no

that do

I

not

‘A: Listen up! B: No, I won’t!’

b.

A: Gefeliciteerd!

B: *Nee,

dat

is

het

niet. /

Nee,

dat

doe

je

niet.

congratulate.ptcp

no

that

is

it

not

no

that

do

you

not

‘A: I congratulate you! B: No, you do not!’

Rather than permitting speaker B to take up the EA of speaker A’s utterance with the help of an overt subject-related pronoun, the more natural way of responding is by using a subject-less predicative expression taking up the result of the original utterance. This is different with RP^dirs and RP^exps, as observable in (26).

In a nutshell, RP^coms and RP^reps lack (implicit) EAs, unlike RP^dirs and RP^exps, the former of which may even overtly license the form of quantificational subjects.

2.2.2 Internal arguments

A principled distinction between RP^dirs/RP^exps and RP^coms/RP^reps may also be made on the basis of licensing the internal argument (IA). The two former types readily permit the overt licensing of referential IAs. In fact, whenever there is an IA in the θ-grid of the underlying predicate in RP^dirs, the IA needs to be licensed or else a root participle cannot be formed. Accordingly, (27) presents examples from German and Dutch, where such cases seem to be less frequent but do appear as well (see van der Wurff 2007: 53f.), based on transitive verbal predicates.

(27)	a.	Den	Müll	rausgetragen!	Den	Bleistift	gespitzt!
		the	trash	take.out.ptcp	the	pencil	sharpen.ptcp
		‘Take out the trash! Sharpen the pencil!’
	b.	Nu	de	kleren	opgeruimd!	De	boterhammen	opgegeten!
		now	the	room	clean.ptcp	the	sandwich	eat.up.ptcp
		‘Tidy up your room now! Eat up the sandwich now!’

Arguments marked with dative case in German may (and have to) be realized overtly as well. Dutch has lost its dative case and hence does not distinguish types of IAs anymore. Reflexives, in turn, are bound to remain implicit.

(28)	Ihm	den	Rücken	zugekehrt!	Dem	Verkäufer	das	Geld	gegeben!
	him	the	back	turn.to.ptcp	the	vendor	the	money	give.ptcp
	‘Turn your back on him! Give the vendor the money!’

(29)	a.	(*Sich)/	Alle	mal	hingesetzt!	(*Sich)	jetzt	aber	mal
		self	all	ptc	sit.down.ptcp	self	now	but	ptc
		zusammengerissen!
		pull.together.ptcp
	b.	Setz	dich	hin!	(*Dich)	hinsetzen!	Reiß	dich
		sit	yourself	down	yourself	sit.down.inf	pull	yourself
		zusammen!	(*Dich)	zusammenreißen!
		together	yourself	pull.together.inf

As the examples in (29) show, the ban on overt reflexives carries over to infinitival imperatives¹⁵—and so does the licensing of dative objects, e.g. in infinitival counterparts to (28) like Ihm den Rücken zukehren! (‘Turn your back on him!’).

A peculiar property of the licensing of IAs in RP^dirs is that they surface either with accusative case (acc) marking, as in (27a) above, or with nominative case (nom). As shown in (30), there is a principled distinction based on whether the EA is licensed as a quantificational subject or appears in an adjunct by-phrase.¹⁶

(30)	a.	Alle	Schüler	den/	*der	Müll	rausgetragen!
		all	pupils	the.acc/	.nom	trash	take.out.ptcp
		‘All pupils, take out the trash!’
	b.	Von	allen	Schülern	*den/	der	Müll	rausgetragen!
		by	all	pupils	the.acc/	nom	trash	take.out.ptcp
		‘All pupils, take out the trash!’

(31)	a.	Alle	Schüler	den/	*der	Müll	ssraustragen!
		all	pupils	the.acc/	nom	trash	take.out.inf
		‘All pupils, take out the trash!’
	b.	*Von	allen	Schülern	den/	der	Müll	raustragen!
		by	all	pupils	the.acc/	.nom	trash	take.out.inf

If the RP^dir overtly licenses a quantificational subject, the IA appears with acc-case, as in (30a). If, however, it is accompanied by an agentive by-phrase (demoting the EA to an adjunct), as in (30b), the IA carries nom-case. Infinitival imperatives, which never license agentive by-phrases, accordingly never give rise to nom-ias, as can be seen in (31). This immediately points to dependent case (see Marantz 1991) and its downward assignment of dependent acc in the presence of a properly licensed argument position (nom-acc vs. nom; Marantz 1991, 26). It is also compatible with Burzio’s (1986) generalisation if we assume that there is an implicit quantificational subject in (27), say , whereas the EA is existentially bound without being syntactically realised in cases like (30b). Given that participial and infinitival imperatives pattern alike in this respect (consider Den Müll raustragen! ‘Take out the trash!’ as an infinitival counterpart to (27a)), the assignment of dependent case seems to hinge on the licensing of a quantificational subject.¹⁷

Expressive instances of root participles also permit the licensing of an overt internal argument, as observable in (32). However, unlike RP^dirs, they only do so optionally and are not subject to any flexibility in terms of case assignment, but rather always carry nom-case.

(32)	a.	(Je)	bedankt/	gefeliciteerd!	(Jullie)	gecondoleerd!
		you.sg	thank.ptcp	congratulate.ptcp	you.pl	condole.ptcp
		‘Thank you! Congratulations! Condolences!’
	b.	Wim	en	Irma/	Winnaars	hierbij
		Wim	and	Irma	winners	hereby
		gefeliciteerd!
		congratulate.ptcp
		‘Congratulations to Wim and Irma/the winners!’
	c.	Ook	u	gefeliciteerd	met	uw	rijbewijs.	Een ieder
		also	you	congratulate.ptcp	with	your	driver’s.license	a everyone
		van	u	gecondoleerd	met		dit	enorme verlies.
		of	you	condole.ptcp	with		your	enourmous loss
		‘Also, congratulations on your driver’s license! Condolences to every one of you for your enormous loss!’

Thus, RP^exps pattern with the RP^dirs in (30b) rather than (30a), i.e. with those cases that are clearly associated with a passive configuration and cannot license their EAs as part of the argument structure. This falls out naturally against the backdrop of the claim that there is a close relation between properly licensing the EA as an overt quantificational subject in RP^dirs without by-phrases and assigning acc-case to the object: since the EA is bound to be existentially quantified over, only nom-case can be assigned to the IA.

In contrast to the RP^dirs and RP^exps, the other declarative root participles, i.e. RP^coms and RP^reps, cannot license overt IAs. In fact, there is only one sub-type of these that permits this, namely assertive ones like (33c).

(33)	a.	*Der	Zoobesuch	versprochen!	*Der	Plan	abgemacht!
		the	zoo.visit	promise.ptcp	the	plan	settle.ptcp
	b.	*The ball well played!	*Die	Ausgangslage	gut		erfasst!
			the	initial.situation	well		capture.ptcp
	c.	Hausaufgaben	erledigt.	Mission	erfüllt!		Einspruch abgewiesen!
		homework	finish.ptcp	mission	accomplish.ptcp		objection overrule.ptcp
		‘Homework done! Mission accomplished! Objection overruled!’

The ban on overt IAs in commissive configurations like (33a) may be attributed to the special status of the IA, an implicit propositional IA. Likewise, verdictive RP^reps, as in (33b) as well as bare assertive RP^reps like Done! are tied to the discourse on the basis of their implicit IA.¹⁸ The assertive cases in (33c), on the other hand, at first sight seem to be the odd man out¹⁹ with respect to potentially licensing an IA: most bare assertive root participles freely alternate with an assertive variant that introduces a nominal referent that corresponds to the object in active clausal counterparts (Ich habe die Hausaufgaben erledigt ‘I have done the homework.’). However, as exemplified in (34), the only licit manifestation of an IA in these configurations is a bare (singular or plural) noun or the universal quantifier.

(34)	a.	(*Die) Hausaufgaben erledigt! (*Die) Mission erfüllt! (*Der/*Den)	Einspruch
		abgewiesen! Alles erledigt! Alles vermerkt!
	b.	(*The) homework done! (*The) Mission accomplished! (*The) Objection
		overruled! All done! All noted!

These bare nouns and quantifiers (have to) fulfil an anaphoric function in that all they do is take recourse to something that is part of the discourse context already, typically some type of a to-do. Accordingly, rather than really serving as a proper universal quantifier, the quantifier all- is restricted by the discourse context (as in ‘all of what has been mentioned before’ or ‘everything that is part of my to-do’). The same pertains to bare nouns like objection, which does not introduce a new referent but rather just refers to an argument or plea that is already part of the discourse context. This suggests that the IA is either not properly licensed as a referential nominal entity introduced as part of the argument structure of the underlying predicate or it is deleted as a consequence of this being an elliptical structure. In either case, the reason for the restrictions on the IA are due to the assertive configuration. This is corroborated by the fact that such assertive configurations also introduce bare nominal referents if they are based on predicates that are not derived from verbs, as the non-participial assertives in (35).²⁰

(35)	a.	Haus leer! Datenlage/alles korrekt! Spiel zu Ende!
	b.	House empty! Data/all correct! Game over! Assignment complete!

Further evidence in favour of a close connection between participial and non-participial assertive configurations comes from the observation that both give rise to copular counterparts establishing the same type of predicative relation. In such cases, as in (34) above and (36) below, a fully-fledged DP may (and in fact, has to) be licensed, though.

(36)	a.	Die Hausaufgaben sind erledigt! Der Einspruch ist abgewiesen! Das Haus ist leer!	Das Spiel ist zu Ende!
	b.	The homework is done! The objection is overruled! The house is empty!	The game is over!

Unlike in their root counterparts, the assertive cases in (36) feature referential subject DPs, which makes them self-contingent (rather than tying them closely to the discourse). Eventually, then, regardless of whether ellipsis in terms of D-deletion or a substantially reduced structure is responsible for the restriction on the IA, only RP^dirs (obligatorily) and RP^exps (optionally) but not RP^coms and RP^reps license overt referential IAs.

In conclusion, the argument structural properties of the distinct types of root participles are summarized in Table 2.

Table 2 The argument structural properties of root participles

2.3 Orientation

In addition to their argument structural properties, root participles also differ along the lines of their orientation. In fact, considering the situation denoted by the underlying verbal predicate, there is a clear split between RP^coms, RP^exps and assertive RP^reps, on the one hand, and RP^dirs as well as verdictive RP^reps, on the other. The former types attribute the situation brought about by the predicate to the speaker: the speaker either commits to what is denoted by the predicate (e.g., making a promise), expresses a psychological state (e.g., gratitude), or reports an achievement (e.g., having done the homework). The remaining types, RP^dirs and RP^reps, in turn, are oriented towards the hearer, who is supposed to carry out an event or has carried out an event that is evaluated by the speaker.²¹

(37)	a.	Aufgepasst,	du/*der	Idiot!	Aufgepasst,	hinter	dir/ *hinter dem Mädchen!
		paid.attention	you/the	idiot	paid.attention	behind	you behind the girl
		‘Pay attention, you idiot! Pay attention, behind you!’
	b.	Well played, you/*the genius! #Well played, you Messi-haters!

As we can see in (37a), the hearer-orientation is robust and likely to be encoded as part of the imperative contribution in RP^dirs, which are also restricted to 2nd person vocatives. The verdictive RP^rep in (37b) seems to behave in an analogous fashion: it is restricted to 2nd person, although it is irrelevant whether the hearer (say Lionel Messi) is in earshot or not.²² The entity in the vocative is the one responsible for bringing about the situation in question (which is why the second variant of (37b) cannot mean that Messi played the ball well despite what some Messi-haters would predict). In a similar fashion, RP^exps, assertive RP^reps, and RP^coms like those in (38) are unequivocally associated with a speaker-orientation, hence with a 1st person reading.

(38)	a.	Bedankt! Gecondoleerd!
		‘Thanks! My condolences!’
	b.	(Homework) done! Mission accomplished!
	c.	Versprochen! Abgemacht!
		‘I promise! This is settled!’

The question that remains to be answered, though, is whether these properties are robust and might even have to be hard-coded in the C-domain with these cases, as they arguably are in RP^dirs. Question contexts like those in (39) seem to support this claim, as the answers can only be attributed to the speaker.

(39)	a.	A: Did you finish the job? B: Done!
	b.	A: Did Mary finish the job? B: *Done!
	c.	A: Gehst du mit in den Zoo? B: Versprochen!
	d.	A: Geht Mama mit in den Zoo? B: *Versprochen!
		‘Will you/mommy accompany us to the zoo? I promise!’

Transforming the 1st person RP^coms and assertive RP^reps into questions, however, transposes the speaker-orientation into a hearer-orientation, as would be the case in fully clausal counterparts.

(40)	a.	Versprochen? Abgemacht?
		‘Do you promise this? Is this agreed upon?’
	b.	Erledigt? Notiert?
		‘(Task) done? Noted?’

This suggests that there is some flexibility in terms of orientation (and person properties) after all.²³ This claim is corroborated by the fact that there is room for 3rd person construals with these types of root participles. This shines through on the basis of the limited accessibility of by-phrases we have seen in (21) above as well as in headlines, where the root participles are not closely tied to a specific discourse.

(41)	a.	?Von	der	Kanzlerin	versprochen.
		by	the	chancellor	promise.ptcp
		‘This is promised by the chancellor!’
	b.	Gefeliciteerd	van	mijn	moeder.
		congratulate.ptcp	by	my	mother
		‘Congratulations from my mother!’
	c.	i.	?Well played by him.
		ii.	?Gut	gemacht	von	deinem	Schützling.
			well	done	by	your	protégé
			‘This was well done by your protégé!’
	d.	?Hausaufgaben	auch	von	ihr	erledigt.
		home.work	also	by	her	finish.ptcp
		‘The homework was also done by her!’

(42)	a.	Gaspreisbremse	versprochen.
		gas.price.cap	promise.ptcp
		‘The gas price cap has been promised!’
	b.	Vorschlag	abgelehnt.
		proposal	rejected.ptcp
		‘The proposal has been rejected!’

In all of these cases, rather than a 1st or 2nd person construal, the participial situation is attributed to an entity of the 3rd person. What is interesting about this is that the commissive impetus is lost in (41a) and the same potentially holds for the expressive performative properties in (41b). This shows that the dedicated 1st person properties are not encoded independently in the C-domain, but rather follow from the performative contribution (as introduced by some performative head, for instance). Accordingly, it is not surprising that they tend to vanish with a decrease of discourse anchoring. In fact, restrictions in terms of speaker- or hearer-orientation typically rather follow straightforwardly from issues of informativeness: assertive RP^reps are bad on a 2nd person reading, as the addressee would know that they performed the act in question, whereas adding an adverb additionally communicates the evaluation of the action, which adds relevant information (see Goldberg and Ackerman 2001 on obligatory adjuncts).²⁴ A 3rd person reading in turn is often ruled out because the strong discourse anchoring pragmatically restricts the interpretation of the reduced utterances to the immediate interlocutors. Similarly, questioning supposed RP^exps and RP^coms transforms them into RP^reps, as rather than asking for the performative speech act, only the result of the participial event can be put into question.

While the orientation of RP^dirs, RP^exps and RP^coms seems to be fixed with the help of the respective imperative or performative contribution, RP^reps are more flexible in that they are mainly restricted by how tightly they are anchored into the discourse. Accordingly, the orientation of RP^reps in Table 3 below is to be taken with a grain of salt, as there is merely a strong preference for a 1st or 2nd person construal that vanishes if the discourse anchoring is weak,²⁵ giving rise to a 3rd person interpretation.

Table 3 The orientation of root participles (if strongly anchored to discourse)

2.4 Aspect

In contrast to RP^dirs and RP^exps, RP^coms and RP^reps denote results. Accordingly, the RP^com in (43a) as well as the verdictive and assertive RP^reps in (43b) and (43c) only allow for readings that entail the completion of the event in question.

(43)	a.	A:	Wir	könnten	eine	Gaspreisbremse	vorschlagen. B:	Abgemacht!
			we	could	a	gas.price.cap	propose	agree.ptcp
		‘A: We could propose a gas price cap. B: This is settled, then!’
	b.	A:	Jemand	sollte	festhalten, dass die Klausur im Juni stattfindet. B: Notiert!
			somebody should write.down that the exam in June takes.place note.ptcp
		‘A: Somebody should write down that the exam takes place in June. B: Noted!’
	c.	A: You do not seem to be happy with the decision. B: Well observed!

The RP^com in (43a) conveys that a proposition is agreed upon, where the commitment to the situation at hand is strengthened by transposing it into an irreversible result. The assertive RP^rep in (43b) denotes that the action of noting down the date has been successfully completed. With the verdictive RP^rep what is being verdictively evaluated is a situation that has come to an end: Well observed! in (43c) can only be uttered if the addressee has made an at-issue observation before and Well sung! may be uttered after some specific part (e.g., a very difficult one) has been sung or the singing in its entirety has come to an end (but cannot be used to evaluate something that is still in progress).²⁶

Accordingly, active present tense counterparts of the examples in (44) are not interpretively equivalent.

(44)	a.	A:	Wir	könnten	eine	Gaspreisbremse	vorschlagen. B:	Das	machen	wir ab.
			we	could	a	gas.price.cap	propose	this	make	we up
		‘A: We could propose a gas price cap. B: We will agree on this!’
	b.	A: Jemand	sollte	es	festhalten.	B:	Ich	notiere	das.
		somebody	should	it	write.down		I	note this
		‘A: Somebody should write it down. B: I will note this down!’
	c.	A: You do not seem to be happy with the decision. B: You observe this well.

The finite present tense responses in these cases lack a resultative component and are thus considerably weaker than their root participial counterparts.²⁷ This is quite unlike the clausal adjectival passive counterparts in (45), which are interpretively equivalent and arguably based on the same underlying participial forms.

(45)	a.	Das	ist	(hiermit)	abgemacht!
		this	is	hereby	agree.ptcp
		‘This is (hereby) settled, then!’
	b.	Das	ist	notiert!
		this	is	note.ptcp
		‘This is noted!’
c.		This is well-observed!

This leaves RP^dirs and RP^exps. The former point to events that are supposed to be carried out by the addressee(s). Accordingly, apart from just resorting to a finite active, the only way of forming an imperative configuration is by means of employing a verbal passive. Transposing a directive root participle into an adjectival passive, if not ruled out by the ban on stative impersonal passives, on the other hand, leads to the loss of the imperative impetus, as shown in (46b).

(46)	a.	Es	wird	aufgepasst!	Jetzt wird	zugehört! Die Stifte werden angespitzt!
		it	becomes	paid.attention	now becomes	listened.up the pencils become sharpened
	b.	*Es ist	aufgepasst!	*Jetzt	ist	zugehört!	#Die	Stifte	sind
		it is	pay.attention.ptcp	now	is	listen.up.ptcp	the	pencils	are
		angespitzt!
		sharpen.ptcp
		‘Pay attention! Listen up now! Sharpen your pencils!’

While personal variants like Die Stifte sind angespitzt are grammatical if they refer to the result of the underlying event (‘The pencils are sharpened!’), as in (46b), a directive reading may only be obtained from the eventive passive counterpart in (46a).

Likewise, RP^exps pattern more closely with eventive passives and finite actives than with resultative stative passives. In fact, the clausal and the root variant in (47a) are interpretively identical and they are also closely related to finite actives in (47b), as there is no resultative contribution.

(47)	a.	(Je wordt)	bedankt/	gefeliciteerd/gelukgewenst	/gecondoleerd!
		you become	thank.ptcp	congratulate.ptcp	condole.ptcp
		‘Thanks! Congratulations! Condolences!’
	b.	Ik dank u! Ik feliciteer jou! Ik condoleer je!
		‘I thank you! I congratulate you! My condolences!’
	c.	?Je	bent	bedankt/	gefeliciteerd/gelukgewenst	/gecondoleerd!
		you	are	thank.ptcp	congratulate.ptcp	condole.ptcp
		‘Thanks! Congratulations! Condolences!’

The adjectival passive in (47c), on the other hand, is not interpretively equivalent, as it attributes a resultative state to the subject rather than introducing an ongoing event. While both variants are viable for a performative speech act, the stative variant is less productive and degraded (apart from the fixed expression Wees gegroet! ‘Be greeted!’, featuring an imperative copula).

Finally, some further evidence for an aspectual contrast between RP^coms and RP^reps, on the one hand, and RP^dirs and RP^exps, on the other, comes from tautologies like those in (48) and (49).

(48)	a.	Versprochen	ist	versprochen.	Beloofd	is	beloofd.
		promise.ptcp	is	promised.ptcp	promise.ptcp	is	promise.ptcp
		‘A promise is a promise.’
	b.	Erledigt ist erledigt. Done is done.
(49)	a.	*Hiergeblieben	ist	hiergeblieben.
		here.stayed	is	here.stayed
	b.	*Gefeliciteerd	is	gefeliciteerd.
		congratulate.ptcp	is	congratulate.ptcp

The former types readily allow for insertion into tautological configurations, which underline the resultative contribution of the participle, as well as regular adjectival passives, whereas RP^dirs and RP^exps do not, as we can see in (49).

Table 4 presents the aspectual contrasts between the distinct types of root participles.

Table 4 The aspectual properties of root participles

2.5 Adverbial modification

Another property that distinguishes the distinct types of root participles is their ability to combine with adverbial modifiers. Directive root participles readily lend themselves to being modified by event-related adverbials. This is observable in (50a).

(50)	a.	Den	Mülleimer	mit	der	Heugabel	geleert!	Den	Bleistift
		the	trash.can	with	the	pitchfork	empty.ptcp	the	pencil
		vorsichtig/langsam	angespitzt!
		carefully/slowly	sharpen.ptcp
		‘Empty the trash can with the pitchfork! Sharpen the pencil carefully/slowly!’
	b.	*Der	Mülleimer	ist	mit	der	Heugabel	geleert.	*Der Bleistift	ist
		the	trash.can	is	with	the	pitchfork	empty.ptcp	the pencil	is
		langsam	angespitzt.
		slowly	sharpen.ptcp
	c.	Der	Mülleimer	wird	mit	der	Heugabel	geleert.	Der	Bleistift wird
		the	trash.can	becomes	with	the	pitchfork	emptied	the	pencil becomes
		langsam angespitzt!
		slowly sharpen.ptcp
		‘The trash can is (being) emptied with the pitchfork! The pencil is (being) sharpened carefully/slowly!’

The root participles in (50a) pattern with the fully clausal eventive passives in (50c) in terms of readily combining with event-related modifiers like carefully, slowly or with the pitchfork. Together with the ban on adverbial modification in adjectival passive counterparts like (50b), this points to the verbal nature of the underlying participles, as a standard diagnostic for verbal vs. adjectival participles is whether or not the participial forms combine with such modifiers (see Rapp 1996). However, there is more to this, as there is occasionally room for seemingly event-related modifiers: the State Relevance Hypothesis claims that modifiers are compatible with adjectival participles if they describe facts that can be inferred by inspection of the result state (see McIntyre 2015). Accordingly, sorgfältig (‘meticulously’) and liebevoll (‘tenderly’) are compatible with the second example in (50b). Adjectival passives are subject to specific restrictions on modifiers and these also shine through with the root participial configurations that are based on adjectival participles, but not with those that are based on verbal participles.

The same holds for RP^exps, as can be seen in (51). These readily allow for adverbial modifiers that appear to be related to the underlying event.

(51)	a.	Hartelijk/wel/	van	harte
		cordially/much/from	from	heart
		gefeliciteerd!	Oprecht	gecondoleerd!
		congratulate.ptcp	sincerely	condole.ptcp
	b.	??Je bent hartelijk/wel/van harte gefeliciteerd! Je bent hartelijk gecondoleerd!
		you are cordially/much/from heart congratulated you are sincerely condoled
	c.	Je	wordt	hartelijk/wel/van	harte	gefeliciteerd!	Je
		you	become	cordially/much/from	heart	congratulated	you
		wordt	oprecht	gecondoleerd!
		become	sincerely	condole.ptcp
		‘Heartfelt/many congratulations! My sincere/heartfelt condolences!’

The fact that event-related adverbial modification is possible with the root participles in (51a) just like in the eventive passives in (51c) suggests once more that the underlying participle is verbal rather than adjectival. Adjectival passives like those in (51b), in turn, are degraded. Accordingly, we conclude that there is evidence (not just from adverbial modification, but also from argument structure) in favour of RP^exps to be based on eventive passives.

With RP^coms and assertive RP^reps, event-related adverbial modification is out, as the examples in (52) show.²⁸

(52)	a.	*Vorsichtig/*Zögerlich/*Widerwillig/*Mit Nachdruck versprochen.29
		carefully/hesitantly/reluctantly with conviction promise.ptcp
	b.	*Hausaufgaben	widerwillig	erledigt!
		homework	reluctantly	done
		*Mission easily done!
	c.	*Gewissenhaft erledigt! *Passionately rejected!
		conscientiously done

However, there is room for adverbial modification with assertive RP^reps, as we can see in (53a).

(53)	a.	(Aufgabe) vollständig erledigt! (Mission) fully accomplished! (Verdict) duly noted!
	b.	Completely ready! House fully empty! Game completely over!

While these at first sight appear to be event-related modifiers, the fact that they also occur in non-participial assertive configurations like those in (53b) shows that they do not attach to an underlying verbal event, but rather just describe result states. Furthermore, the participial cases readily give rise to adjectival passive counterparts like those in (54).

(54)

Die Aufgabe ist vollständig erledigt!

The mission is fully accomplished!

The verdict is duly noted!

This is in line with the State Relevance Hypothesis. Eckardt (1998) refers to such modifiers as degree-of-perfection adverbs (see also Gehrke and Castroviejo 2019).

This leaves verdictive RP^reps, which require an overt adverb to allow for a proper evaluation.³⁰ These also appear to be event-related at first sight, but this is likewise put into question by their ability to occur in adjectival passives.³¹

(55)	a.	Gut gespielt! Überzeugend dargelegt!
		Well played! Convincingly argued!
	b.	Der Ball ist gut gespielt! Diskussionspunkt ist überzeugend dargelegt!	The ball is well played!
		The talking point is convincingly argued!

Rather, verdictive modifiers may also be subsumed under the heading of degree-of-perfection adverbs that are exempt from the restrictions on forming adjectival participles because they primarily modify the result state (in line with the State Relevance Hypothesis). For the present purposes, however, we will treat those adverbs of this type that provide a qualitative evaluation as a dedicated type of adverbial modifier to account for the fact that they are a necessary ingredient for licensing verdictive RP^reps.

Eventually, then, what we are dealing with in the case of these exceptions are specific subtypes of adverbials that are not event-related in the sense that they need to modify an event token (cf. Gehrke 2015 on modifiers of event kinds). Rather, what is modified is the impoverished remnants of an event structure that is suppressed by an adjectival ingredient.³² In fact, the modifiers in question share that they primarily modify the result, as predicted by the State Relevance Hypothesis.

Table 5 presents an overview of whether adverbial modification is possible in the distinct types of root participles.

Table 5 The adverbial modification of root participles

2.6 Taking stock

The previous sub-sections were devoted to empirically motivating a typology of root participles into four main types, only one of which is associated with a force specification other than declarative, namely imperative root participles. The grammatical properties of the different types, as summarized in Table 6, may be used to motivate a principled distinction into an underlying verbal as opposed to an adjectival participle.

Table 6 The properties of root participles

As has been touched upon in the previous sections already, what speaks out in favour of a split along the lines of RP^dirs and RP^exps as verbal and RP^coms as well as RP^reps as adjectival participles is that (i) only RP^dirs and RP^exps introduce a (c)overt EA and may also license an overt referential IA (argument structure), (ii) only RP^coms and RP^reps convey resultative properties, (iii) only RP^dirs and RP^exps allow for event-related adverbial modification. These properties fall into line with the traditional claim that the main contrast between the participles employed in eventive passive, on the one hand, and stative (i.e., adjectival) passives, on the other, is that only the former introduce a fully verbal structure, including an EA (to be existentially bound) and introducing an actual event (instantiated as a token; see Gehrke 2015). Adjectival passives, in turn, are not formed with an auxiliary but with a copula which selects for an adjectival participle. The adjectival contribution of this form imposes a resultative interpretation but also forces the verbal characteristics to be reduced: neither implicit EAs, as taken up in the form of agentive by-phrases, nor modifiers that make recourse to the verbal event are licit. This was traditionally reflected by a structurally reduced verbal domain, e.g. lacking a passive VoiceP or an agentive verbal head, in syntactic approaches to the distinction at hand (see Kratzer 2000). However, both agentive by-phrases as well as event-related modifiers occasionally show up with adjectival participles as well, as observable in (56), taken from Maienborn (2007, 2009).

(56)	a.	Das	Manuskript	ist	von	Chomsky	zitiert.
		the	manuscript	is	by	Chomsky	cited
		‘The manuscript is cited by Chomsky.’
	b.	Der	Brief	ist	mit	roter	Tinte	geschrieben.
		the	letter	is	with	red	ink	written
		‘The letter is written with red ink.’

Exceptions of this kind led syntacticians to reintroduce passive VoicePs or agentive vPs into the structures of adjectival participles (see Alexiadou et al. 2014; Sleeman 2014), substantially diminishing the structural differences between the two types. In Gehrke’s (2015) seminal work on adjectival passives, the contrast is traced back to the adjectival contribution of existentially binding the event variable: “adjectival passives involve event kinds rather than event tokens [and] event-related modifiers modify an event kind” (Gehrke 2015, 918). This line of reasoning may also be applied to the occasional occurrence of exceptions to the ban on agentive by-phrases and supposedly event-related adverbial modifiers in the context of root participles. As we have seen, the State Relevance Hypothesis (McIntyre 2015) suggests that modifiers of adjectival participles describe the result state rather than the verbal event. Substantial support for the claim that none of this goes against the adjectival nature of the root participles in question comes from the observation that adjectival passive counterparts of the possibly exceptional cases could be shown to also be licit.

With this fundamental distinction in place, let us now turn to how the distinct types of root participles are analysed in previous approaches, before laying out a novel approach that does away with the shortcomings to be uncovered. Apart from the question of which kind of participle the distinct types of root participles are based on, this is where questions of ellipsis come in. Are root participles merely elliptical variants of full clauses or are there arguments in favour of analysing them as dedicated (non-sentential) structures? In addition to considerations regarding interchangeability with clausal counterparts, this is also where the issue of orientation comes back in.

3 Previous approaches and questions of ellipsis

Although root participles have mostly just been considered in passing and the number of dedicated analyses is very limited, we will now critically examine these one after the other. In doing so, special attention will be paid to the question of whether the root configurations in question are just phonologically reduced variants of clausal counterparts (ellipses) or whether they are rather structurally reduced with the potential for dedicated material to occur that is not part of a potential clausal counterpart (non-sentential; see Progovac 2013).³³

Imperative RP^dirs cannot just be elliptical variants of fully clausal counterparts, a view that is broadly acknowledged in the literature (see Fries 1983; Wunderlich 1984; Rooryck and Postma 2007). It finds immediate support in the observation that participial RP^dirs which include quantificational subjects, like the one in (57a), cannot give rise to verbal passive counterparts like (57b). The periphrastic perfect in (57c), in turn, is also not a licit counterpart since it does not convey a directive interpretation (cf. Fries 1983, 237; Wunderlich 1984, 114fn15).

(57)	a.	Alle	Teilnehmer	aufgepasst!
		all	participants	pay.attention.ptcp
		‘All participants, pay attention!’
	b.	*Alle	Teilnehmer	werden	aufgepasst!
		all	participants	become	pay.attention.ptcp
	c.	#Alle	Teilnehmer	haben	aufgepasst!
		all	participants	have	pay.attention.ptcp
		‘All participants have paid attention!’

There are three strands of approaches to the syntax of RP^dirs in the literature. Fries (1983, 239)—in the first approach to the syntax of German RP^dirs – proposes that participles employed in imperative configurations are part of an IP whose specifier position may be filled overtly, as sketchily laid out in (58a). Rapp and Wöllstein (2009, 168), on the other hand, treat participial imperatives as configurations consisting of a functional head triggering the imperative reading. This head, as we can see in (58b) allows for the introduction of a quantificational subject. Rooryck and Postma (2007, 281ff.), in turn, suggest a more complex derivation that involves (phrasal) movement, as sketched out in (58c).

(58)	a.	[ipalleI°+PRF [vp den Aufzug benutzt]]
	b.	[fp [specf Alle Teilnehmer [f‘ [f° [vp aufgepasst]]]]]
	c.	[cp [partp opgepast/opgerot]i c° [ip ti]]

These structural analyses differ in three fundamental respects: (i) the presence of an imperative head in (58b) and (58c) but not in (58a), (ii) the presence of a TP/IP in (58a) and (58c) in contrast to the absence thereof in (58b), (iii) the lack of movement in (58a) and (58b) as opposed to phrasal movement of the participial phrase in (58c). With respect to issue (i), a sensible syntactic means to account for availability of an imperative reading is by introducing a dedicated functional head. This ingredient may be held responsible for imposing restrictions on the range of possible subjects (to 2nd person) and attracts the finite verb in imperative clauses. The counterpart that combines with participial and infinitival verbal items, in turn, paves the way for licensing quantificational subjects, as in Alle aufstehen! and its participial counterpart Alle aufgestanden!, which both cannot trigger anything but the directive reading ‘Everybody stand up!’. Whether this dedicated imperative head is a proper C-head, as in (58c), or a somewhat impoverished F-projection as in (58b) is not conclusively settled yet, though (see Gärtner 2013 for discussion).³⁴ In either case, the properties of that functional head should be defined properly, which is not to sufficiently done in Rapp and Wöllstein (2009, 168).

Issue (ii) boils down to the presence of an inflectional head denoting perfect tense in (58a), which is highly dubious given that participial imperatives denote requests for carrying out events in the immediate future rather than situations that have concluded (where there is also no ingredient that is likely to encode a future change of state). Moreover, this is not justified by Fries (1983), but rather seems to solely be based on the apparent need for active properties associated with perfect rather than passive participles.³⁵ However, this is at odds with the lack of perfect semantics. In fact, even if we assume that a perfect interpretation allows for imperfectivity, there is no reason to believe that the event that is supposed to be carried out has already begun (as should be the case if a Perfect Time Span is involved, see Iatridou et al. 2001). In (58c), there is also an IP, but one that is not further specified. In fact, Rooryck and Postma (2007, 283) claim that “the sentential structure between C° and Part° […] is truly empty” but still include an IP in their structure as the source of the PartP, probably to license temporal adverbs like nu (‘now’).³⁶ We will simply assume that it is the lack of T that binds the non-finite command to the moment in which it is uttered: root infinitival structures cannot be shifted through time but are related to the here and now (i.e., the immediate discourse situation), which is why only certain temporal adverbs can appear (cf. Rooryck and Postma 2007, 285).

This leaves issue (iii), i.e. the question of whether there is a necessity for movement to the C-domain in the formation of participial imperatives. Although Rooryck and Postma (2007, 281) hold a C-head responsible for the introduction of imperative properties in finite imperatives and participial imperatives alike, they claim that the verbal element undergoes head-movement to C in the former, whereas some participial projection is moved into the specifier domain of CP as part of phrasal movement in the latter. That there really is no head-movement involved here can be seen on the basis of data like (59) and (60).

(59)	a.	Den	Müll	rausgebracht!
		the	trash	out.taken
		‘Take out the trash!’
	b.	Rausgebracht *(,)		den Müll!
		out.taken		the trash

(60)	a.	Bring	den	Müll	raus!
		take	the	trash	out
		‘Take out the trash!’
	b.	*Gebracht	den	Müll	raus!
		taken	the	trash	out

Under the assumption of head-movement, we would expect the IA to be left behind in the verbal domain, but as we can see in (59), the IA has to precede the imperative participle unless there is extraposition (as marked by a prosodic break) of the former. Additionally, as we can see in (60), complex verbs like rausbringen (‘take out’) can (and have to) split up in finite imperatives, but not in the context of participial imperatives. While this could simply be traced back to participle formation prohibiting any split (although the availability of Raus hat er den Müll gebracht! ‘Out, he has taken the trash.’ suggests otherwise), the observation that infinitival imperatives behave alike (*Bringen den Müll raus!) is telling. Hence, either the participle remains in place or the whole participial phrase moves. According to Rooryck and Postma (2007, 283) bare infinitives pattern with participial imperatives: “in both cases, an extended projection of VP is moved to the Specifier of a C° with imperative force”. Their main motivation for this comes from adverbs like nu (‘now’) and PPs like voor de hond (‘of the dog’). These follow the participle in examples like Even opgepast nu/voor de hond (‘Now pay attention for a second.’ or ‘Beware of the dog for a second.’), indicating that the participle (together with adverbs like even ‘for a second’, unlike with finite imperatives) moves to a higher position: the participial phrase moves past adverbs like nu (‘now’) while PP complements are displaced (supposedly via scrambling) out before movement.

What is problematic about this approach is the lack of a formal motivation for displacing the PP (due to the questionable status of scrambling) and the weak empirical basis in the sense that nu (‘now’) optionally occurs before the participle and it is not entirely clear if its postverbal positioning is properly integrated into the RP^dir or not (as it appears to be subject to the insertion of an intonational break). This holds for German in cases like Eben aufgepasst jetzt! and Jetzt eben aufgepasst! (‘Just pay attention now!’) as well, where the former seems to include an apposition (supporting the imperative reading) rather than properly integrating the item into the clause. Moreover, the initial positioning of jetzt (‘now’) is not just triggered by eben (‘for a second’), but rather also works with other adverbs, e.g. mal (‘once’) or schnell (‘quickly’), as in Jetzt (aber) mal aufgepasst (lit. now but ptc out.watched, ‘But pay attention now!’) or Jetzt (noch) schnell bestellt! (lit. now ptc quickly ordered, ‘Order quickly now!’) in German. Additionally, an approach based on phrasal movement does not in any way account for the role of quantificational subjects, which are not taken into consideration in Rooryck and Postma (2007). In fact, there seems to be an incompatibility of an approach based on phrasal movement to Spec, C and the assumption that quantificational subjects are also introduced in Spec, C, as in (58b). Although it is of course possible to make room for multiple specifiers in the C-domain (as part of a rich CP-periphery; see Rizzi 1997), the question remains of whether this structural elaboration is justified and whether it is supported by considerations regarding word order. The fact that the quantificational subject readily appears before the participle as well as its modifiers in examples like Alle HSV-Fans jetzt aber mal eben aufgepasst! (‘All HSV-fans, please briefly pay attention now!’) suggests that it is not. If the participial phrase were subject to phrasal movement, the quantificational subject should be introduced in Spec, C before (and hence end up lower than) the participle and its modifiers, contrary to fact. This shows that there is no principled reason for neglecting the traditional claim of (58a) and (58b); i.e., the assumption that the participial phrase does not undergo movement.

Finally, the approaches in (58) all remain too vague with respect to the syntactic interaction between the quantificational subject and the participle. To be precise, neither the approach (apparently) based on active perfect participles in (58a) nor the approach in (58b), which also does not state whether the participle bears passive properties, is explicit about whether there is any relation at all. While it is not surprising that the appearance of quantificational subjects has led to a reluctance to analyse the participial forms as passive, the occurrence of by-phrases provides unequivocal evidence in favour of this. However, this demands a principled account of how the imperative contribution (in terms of interpretive restrictions on the range of possible subjects) goes along with the passive contribution of introducing an implicit EA. This will be laid out in the next section as part of a novel approach to the syntax of RP^dirs.

Turning next to RP^exps, the properties we have uncovered in the previous section point to an underlying verbal passive participle. In fact, the examples in (61) all have a natural verbal passive counterpart, as shown in (62).

(61)	a.	(Hierbij)	gefeliciteerd	met	Je	nieuwe	neefje!
		hereby	congratulate.ptcp	with	your	new	nephew
	b.	(Hierbij)		bedankt	voor	je	input!
		hereby		thank.ptcp	for	your	input
	c.	(Hierbij)	van	harte	gelukgewenst	met	jullie
		hereby	from	heart	congratulate.ptcp	with	your
		prachtige	prestatie!
		lovely	performance

(62)	a.	Je	wordt	(hierbij)
		you	become	hereby
		gefeliciteerd	met	je	nieuwe	neefje!
		congratulate.ptcp	with	your	new	nephew
	b.	Je	wordt	(hierbij)
		you	become	hereby
		bedankt	voor	je	input!
		thank.ptcp	for	your	Input
	c.	Je	wordt	(hierbij)	van	harte
		you	become	hereby	from	heart
		gelukgewenst	met	jullie	prachtige	prestatie!
		congratulate.ptcp	with	your	lovely	performance
		‘Congratulations on your new nephew! Thanks for your input! Heartfelt congratulations on your lovely performance!’

This is in line with Rooryck and Postma’s (2007, 273f.) analysis of expressive root participles as reduced verbal passives: the subject and the auxiliary are just elided. That such root participles are not just elliptical variants of adjectival passives can be seen in the degraded status of regular adjectival passive counterparts (especially if there is adverbial modification), as shown in (63).³⁷

(63)	a.	??Je	Bent	(hierbij)	Gefeliciteerd	met	je	nieuwe	neefje!
		you	are	hereby	congratulate.ptcp	with	your	new	nephew
	b.	??Je	bent	(hierbij)	bedankt	voor	je	input!
		you	are	hereby	thank.ptcp	for	your	input
	c.	??Je	bent	(hierbij)	van	harte	gelukgewenst	met jullie prachtige prestatie!
		you	are	hereby	from	heart	congratulate.ptcp	with your lovely performance

Based on these observations, RP^exps are best analysed as full clausal configurations which are subject to phonological deletion (ellipsis). Accordingly, they differ strongly from the dedicated root configurations that are RP^dirs.

This leaves RP^coms, which we turn to next, and RP^reps. Fries (1983, 236) simply analyses both types of non-directive root participles as reduced adjectival passives, whereas Ørsnes (2020, 385) suggests that performative participles (our RP^coms) are non-elliptical verbal (passive) participles. These views are sketchily represented in the labelled bracketing structures in (64), where the rounded brackets in (64a) indicate phonological deletion.

(64)	a.	[IP (Das) [PredP (ist) [AP abgemacht]]]
	b.	[IProot [VPpass Versprochen]]

The analysis in (64a) at first sight properly accounts for the properties of this kind of root participle, as laid out in the previous sections. However, it raises the question of whether there are convincing arguments against an analysis based on ellipsis rather than an impoverished (non-sentential) structure (see Progovac 2013). Such a structure is proposed in (64b), where the participle introduces a 1st person null EA and a null pronominal propositional IA and is embedded under IProot. This serves to account for the fact that such participles do not allow for structural embedding (cf. Ørsnes 2020, 385). Accordingly, rather than ellipsis, a proper root configuration is proposed here, i.e. the performative participles in question “are not associated with a reduced clausal structure” (Ørsnes 2020, 355). The underlying participle is explicitly assumed to be a verbal passive on the basis of its supposed licensing of hiermit (‘hereby’), manner adverbials (ungern ‘reluctantly’, leider ‘reluctantly’, selbstverständlich ‘self-evidently’, hoch und heilig ‘high and holy’) and recipient IAs (allen Subaru-Freunden ‘all Subaru-friends’). However, all of these ingredients readily appear with adjectival participles (and hence in stative/adjectival passives) as well: the manner adverbs are sentential adverbs rather than modifiers of event tokens and self-sufficient recipient IAs carrying dative case even occur prenominally (consider the DP die den Subaru-Freunden versprochene Veranstaltung ‘the event that was promised to the Subaru-friends’). Likewise, hiermit (‘hereby’) is not in any way telling with respect to the verbal status of the participle, as it readily occurs in adjectival passives as well, as in (45) above. A central problem of the claim that the underlying participle is a verbal passive is that these cannot impose resultative properties—which are acknowledged by Ørsnes (2020, 349)—without external help, unlike adjectival participles.

Returning to the question of whether there are arguments against an analysis based on ellipsis and hence a structure along the lines of (64a), Ørsnes (2020) presents two compelling arguments in favour of a non-sentential analysis. First, there is a clear contrast in the range of available interpretations: adjectival passives allow for assertive as well as performative (commissive) readings, whereas RP^coms are restricted to the latter (cf. Ørsnes 2020, 366).

(65)	a.	Das	ist	(hiermit)	abgemacht/versprochen!
		this	is	hereby	settle.ptcp/promise.ptcp
		‘So it has been agreed/is promised!’ (assertive) vs. ‘That is a deal!’ (commissive)
	b.	Abgemacht/Versprochen!
		settle.ptcp/promise.ptcp
		‘That is a deal!’ (commissive)

Accordingly (65a) can be used assertively to answer a polar question (Kommt die Kanzlerin denn überhaupt heute? ‘Will the chancellor be here today at all?’) as well as performatively to accept an agreement (Die Kanzlerin kommt heute ‘The chancellor will be here today.’). As pointed out by Ørsnes (2020, 366), the RP^com in (65b), on the other hand, is bound to denote a performative reading in both cases. Unlike what we would expect under an analysis based on ellipsis, the two variants thus do not always give rise to the same range of interpretations.

The second argument brought forth by Ørsnes (2020, 385) is that there are not always one-to-one correspondences of RP^coms and adjectival passives. In fact, there are verbs like sich einigen (‘to reach an agreement’) and sich anschließen (‘to fall into line’) that give rise to performative participles, but not to adjectival passives, as we can see in (66), and there are verbs like erinnern (‘remind’), which form adjectival passives but no RP^coms, as shown in (67) (cf. Ørsnes 2020, 363).

(66)	A:	Man	könnte	tatsächlich	das	Beispiel	löschen.
		one	could	indeed	the	example	delete
	B:	Geeinigt!
		agree.ptcp
	B‘:	Darüber	?sei/	*ist	sich	geeinigt!
		thereover	be.sbjv/	.ind	self	agree.ptcp
		‘A: You could indeed delete the example. B: I agree!’

(67)	a.	Der	Teufel	steckt	im	Detail,	daran	sei	erinnert.
		the	devil	lies	in.the	detail	thereof	be.sbjv	remind.ptcp
		‘I remind you, the devil is in the detail.’
	b.	*Der	Teufel	steckt	im	Detail,	erinnert.
		the	devil	lies	in.the	detail	remind.ptcp

The fact that there is no proper interchangeability between adjectival passives and their supposedly elliptical counterparts in this context adds to the claim that we are dealing with non-sentential structures that are based on adjectival participles in the case of RP^coms.

Finally turning to RP^reps, Wunderlich (1984, 114fn14) and Rapp and Wöllstein (2009, 167) argue that verdictive and assertive RP^reps, respectively, are reduced active perfects, as sketchily represented in (68a) and (68b). Fries (1983, 236), as mentioned above, treats these as reduced adjectival passives, which could be structurally accounted for as in (68c).

(68)	a.	[IP (Du) [AuxP (hast) [VPperf (das) gut gemacht]]]
	b.	[IP (Ich) [AuxP (habe) [VPperf verstanden]]]
	c.	[IP (Der Befehl) [PredP (ist) [AP ausgeführt]]]

The analyses in (68a) and (68b) account for the resultative properties of RP^reps by claiming that the underlying participle is an active perfect participle. What immediately speaks out against an analysis of verdictive RP^reps as ellipses is that this is not compatible with the word order of English: consider *You have well done that as opposed to You have done that well, only the former of which would sensibly lend itself to ellipsis resulting in Well done!, unless additional movement processes are involved.³⁸ This problem does not arise in (68c), whose verdictive counterpart could be an elliptical variant of the adjectival passive The penalty was well saved! Furthermore, what speaks out against an analysis of RP^reps as elliptical active perfect configurations is that assertive root configurations also surface without participles, as we have seen before: consider Game over! and Haus leer! (‘House empty!’). These are likewise mirrored in copular configurations, though: The game is over! Das Haus ist leer! (‘The house is empty!’). What is striking about these is that they are subject to the same D-deletion resulting in a bare noun in the root variants. A final piece of evidence against elliptical active perfects is that these do not account for the restrictions on licensing an EA and introducing proper event-related modifiers. The analysis in (68c), an elliptical adjectival passive, is not subject to any of these objections and also naturally accounts for D-deletion as a consequence of ellipsis.³⁹ The only potential argument in favour of a non-sentential analysis of RP^reps is that they are typically subject to specific person properties (1st person in the case of assertive RP^reps and 2nd person in the context of verdictive RP^reps). However, as laid out in Sect. 2.3. The syntax of root participles, with a decrease in discourse anchoring, the restriction to specific person properties vanishes. In a headline like Mission accomplished: 15 years of peacekeeping success in Liberia, for instance, what is conveyed is a neutral 3rd person interpretation that attributes the participial situation neither to the hearer nor to the speaker.

In conclusion, both verbal as well as adjectival variants of root participles come as non-sentential as well as elliptical configurations. This is summarized in Table 7.

Table 7 Non-sentential vs. elliptical and verbal vs. adjectival root participles

The expressive and resultative speech acts are simply derived from the sentential declarative configurations in the case of RP^exps and RP^reps, as they are in their non-elliptical counterparts. The non-sentential RP^dirs and RP^coms, however, denote directive and commissive speech acts by means that differ substantially from their potential clausal equivalents. These are the configurations that we will hence focus on in the next section.

4 The syntax of root participles

Against the backdrop of the main properties of the distinct types of root participles (cf. Table 6) in Sect. 2 and the discussion of previous approaches and questions of ellipsis (cf. Table 7) in Sect. 3, let us now turn to a novel approach to the syntax of root participles. In doing so, we will mostly focus on the non-sentential RP^dirs in Sect. 4.1 and RP^coms in 4.2, as these structurally differ from potential clausal counterparts. The syntax of elliptical RP^exps and RP^reps, in turn, will only be considered in passing in Sect. 4.3 and Sect. 4.4.

4.1 The syntax of directive root participles

In the previous sections, we worked out the main properties of RP^dirs: (i) they are based on verbal passive participles (including an implicit EA, licensing IAs and readily combining with event-related modifiers), (ii) they include dedicated functional material in the C-domain which encodes imperative force, contributes to licensing quantificational subjects and does not allow for clausal embedding, (iii) neither their participial head nor the participial phrase moves into the left periphery. While we may simply take over the structural properties of verbal passive participles from the literature, we have to model (a) how the imperative ingredient restricts the range of possible EAs, and (b) how it contributes to licensing quantificational subjects.⁴⁰ Accordingly, we should focus on the contribution of the imperative C-domain.

As the examples in (69a) show, finite imperative configurations take 2nd person subjects and the finite verb, morphologically the base form of the verb, exhibits singular or plural agreement based on the number properties of the subject, regardless of whether it is realised overtly or remains covert. Accordingly, even if the subject remains inaudible, it is syntactically active. Given the absence of finite inflectional morphology in the non-finite counterparts, no number distinction may be encoded overtly in (69b) and (69c).

(69)	a.	Lauf(t)	weg!	Setz(t)	dich/	euch	hin!	Pass(t)	auf!
		run	away	sit	yourself/	yourselves	down	watch	out
	b.	Weglaufen!	Hinsetzen!	Aufpassen!
		away.run	down.sit	out.watch
	c.	Weggelaufen!	Hingesetzt!	Aufgepasst!
		away.run	down.sat	out.watched
		‘Run away!	Sit down!	Watch out!’

Nevertheless, restrictions are imposed as to the range of possible EAs: infinitival cases like those in (69b) are directed towards the general public and tied to a specific place (hence often found on signs, e.g. Vorsichtig fahren! ‘Drive carefully!’ in front of an elderly home or Nicht berühren! ‘Do not touch!’ near a wall with fresh paint), whereas participles as those in (69c) are directed towards specific addressees that are currently present (cf. Brinkmann 1971; Aikhenvald 2010). In both cases, there is a set of addressees, which is in line with the observation that what may occur as an overt subject in German needs to quantify over potential addressees: Alle/Schüler (*,) hinsetzen/hingesetzt (‘Sit down, everybody/pupils!') vs. Lena/ein Teilnehmer *(,) hinsetzen/hingesetzt (‘Sit down, Lena/one participant!'). The requirement for an intonational break in the latter example shows that Lena/ein Teilnehmer merely serves as a vocative, which directs the utterance to a specific addressee and hence picks out one entity from the undefined quantity introduced by the imperative. As there is no (inflectional) evidence in favour of the presence of an argument in the T-domain, or a T-head (with specific φ-features) to be structurally present for that matter, the restriction to plural subjects seems to be encoded as part of the imperative C-domain. In fact, what is encoded is plural number and 3rd person, which serves to restrict the set of referents to addressees of the directive speech act, unlike in the case of finite imperatives where the number specification is flexible but there is a morphological restriction to 2nd person.

Following Jensen (2003) and Bennis (2006), we may assume that there is a dedicated imperative head which interpretively restricts the range of possible subjects. More specifically, we follow Zanuttini et al. (2012) in assuming that such a dedicated head in the C-domain is responsible for the imperative restrictions in all types of configurations associated with imperative force. In finite imperatives, this functional head—referred to as Jussive-head to cover ground beyond imperatives—combines with T, with which it forms a complex head via bundling its person feature with T’s number and case features (cf. Zanuttini et al. 2012, 1246). This combination “contains an interpretable second person feature, semantically binds the subject and Agrees with it, thus licensing the null pronoun in subject position” (Zanuttini et al. 2012, 1246). This is structurally represented for an imperative with a plural subject in (70), as taken from Zanuttini et al. (2012, 1246), where the T- and the Jussive-head have bundled to form a single head (resulting from head-to-head movement).

While the imperative restrictions are thus based on the interaction with the T-head in the domain of finite imperatives, their non-finite counterparts do not have a means to impose φ-featural restrictions on the range of possible subjects with the help of (combining with) T. Rather, the imperative head directly combines with the participial domain with no intervening T-head. Additionally, the imperative head apparently contains no interpretable 2nd person feature, but is specified for plural number and 3rd person. This leaves the question of how these interact with the implicit EA that is expected to be introduced by passive participles to the effect of either leaving the subject implicit or introducing it overtly as a quantificational subject.

In order to tackle this question, let us consider another context in which the implicit argument of a passive participle needs to be licensed explicitly: the licensing of the EA of passive participles in the have-perfect. In fact, the core claim of identity approaches to passive and perfect (often misleadingly subsumed under the term ‘past’) participles is that the active perfect is derived from a passive configuration whose implicit EA is associated with an overt argument that is licensed by the perfect auxiliary have. In other words, passive participles license acc-objects if the EA is overtly realised by have (see Roberts 1984, 218f.). Technical means to grasp this are based on θ-absorption (see Baker 1985, Jaeggli 1986 and Åfarli 1989) and θ-merger (see Ackema 1999 and Ackema and Marelj 2012): the P(assive)M(orpheme) absorbs the external θ-role (as if it was an argument itself) and there is a mechanism that allows have to assign this absorbed role, which would otherwise just be existentially bound, to be assigned to an overt argument. This is sketched on the basis of θ-merger in (71), taken from Ackema and Marelj (2012, 235).

To illustrate this in some more detail very briefly, we could start out with a verbal predicate like kiss which involves a θ-grid (<Ag, Th>) that leads to the licensing of a transitive argument structure in active clauses (say Mary kisses John). Once the participial morpheme attaches to the verb, it absorbs the θ-role associated with the EA (Ag) in the sense that it prevents this role from being assigned to an overt argument and triggers existential binding at LF (λx∃x), where an adjunct by-phrase could alternatively be used to specify the implicit EA (John was kissed by Mary). If, however, the perfect auxiliary have is introduced into the structure, it serves to take up the absorbed role by means of including an empty θ-role that is in need of specification in languages exhibiting auxiliary alternation (have vs. be) or specified optionally whenever there is an external role in those that do not (have-only). The mechanism of θ-merger ensures this by establishing the identity of the absorbed role with the empty role. This allows have to effectively license an overt EA (Mary has kissed John), which in turn prevents a by-phrase from being introduced or existential binding to apply.

Although we can adopt many of the core ingredients of Ackema and Marelj’s (2012) approach to imperative root participles if we assume that the imperative head also includes an empty θ-role, we crucially need to account for the central differences as well: (i) restricting the range of potential referents to 3rd person plural which interpretively singles out the set of addressees, and (ii) doing so either by means of realizing an overt argument or by leaving the (restricted) EA implicit (in which case it may be introduced in an adjunct by-phrase). None of these restrictions are part of the perfect auxiliary in (71) (see also (75) below), which hence allows for the full range of possible EAs, but they need to be incorporated as a core ingredient in the case of RP^dirs. The representation in (72) does so, exemplifying the structure of an imperative case like Alle den Stift weggelegt! (‘Put away the pencils now, everybody!’).⁴¹

The syntax of eventive passive participles as employed here is based on standard assumptions from the non-lexicalist literature: participles are derived from verbs by means of introducing a combination of a passive VoiceP and an aspectual AspP (see Embick 2004; Alexiadou et al. 2014; Sleeman 2014). Accordingly, rather than assuming a primitive passive morpheme that directly attaches to the verb, as in (71) above, we assume that the passive contribution stems from a Voice-head that combines with the verbal domain and paves the way for existential binding by ‘absorbing’ the θ-role via entering a local relation with an eventive v-head. This follows naturally in approaches in which the EA is syntactically ‘severed from the verb’ (see Kratzer 1996). However, due to the presence of an empty θ-role on the Imp(erative) head, θ-merger paves the way for overtly licensing an EA by identifying the empty role with the external role of the underlying verb.⁴² By means of assigning the empty role to its specifier, Imp thus provides a DP-host for the suppressed θ-role of the passive VoiceP. Unlike in the case of the perfect auxiliary have, which is not associated with any φ-features unless it combines with a finite T, this Imp-head carries interpretable φ-features. The subject DP enters the derivation without person and number values but attains these from the Imp-head, as has been claimed for person features in the case of Zanuttini et al.’s (2012) Jussive-head as well. This serves to restrict the range of potential subjects to 3rd person plural quantificational subjects.

Pending a discussion of case assignment (acc vs. nom), this accounts for examples like those in (73).

(73)	a.	(Jetzt	aber	mal	bitte)	alle	den	Stift	weggelegt!
		now	but	just	please	all	the.acc	pencil	away.put.ptcp
	b.	(Jetzt	aber	mal	bitte)	den		Stift	weggelegt!
		now	but	just	please	the.acc		pencil	away.put.ptcp
		‘Please just put away the pencils now, everybody!’

The argument in Spec, Imp is licensed as the overt quantifier alle (‘all’) in (73a) and it could even merge with a nominal complement like Teilnehmer (‘participants’). This accounts for alle and alle Teilnehmer (‘all participants’). The item that is valued for 3rd person plural may, however, also remain covert: Teilnehmer (‘all participants’). In the absence of a nominal complement, the empty variant of (‘all’) also accounts for (73b). The adverbial modifiers in (73) may be taken to modify the Imp-head after the insertion of the quantificational subject.⁴³

Alternatively, a nominal argument need not necessarily be introduced in Spec, Imp, in which case Imp contributes 3rd person plural properties via its interpretable φ-features and undergoes θ-merger with the role introduced by passive Voice but does not introduce a nominal argument. This is why a by-phrase, which is also interpretively restricted via the contribution of the Imp-head, is introduced as a modifier of VoiceP, as in (74a), or existential binding kicks in as a last resort, as in (74b).⁴⁴,

(74)	a.	(Jetzt	aber	mal	bitte)	von	allen	der	Stift
		now	but	just	please	by	all	the.nom	pencil
		weggelegt!
		away.put.ptcp
	b.	?? (Jetzt	aber	mal	bitte)	der		Stift	weggelegt!45
		now	but	just	please	the.nom		pencil away.put.ptcp
		‘Please just put away the pencils now, everybody!’

This leaves the curious distinction in terms of case assignment: cases with an overt or covert universal quantifier like (73) involve an aCC-IA, whereas those with an implicit EA that is subject to existential binding like (74) only manage to assign nom to the IA. The pattern in (73) is the one that also shines through in the structurally reminiscent hAVE-perfects, as represented in (75) below, which translates the basic assumptions behind (71) into the framework at hand.

Unlike in the imperative case above, the auxiliary have does not introduce any φ-features, but only supplies its (formerly empty) θ-role before moving to T and taking up the tense inflection.⁴⁶ Accordingly, the EA is not subject to any φ-restrictions. Apart from this contrast, the argument-licensing of RP^dirs and the have-perfect is the same: a DP introduced with the help of higher functional material is thematically associated with VoiceP, which would fail to license an EA by virtue of being passive. In the have-perfect, the Aux-head necessarily assigns its (formerly empty) θ-role to its specifier, and in the directive root configurations, the Imp-head may either (i) introduce an overt or covert quantifier (alle/$-\!\!\!-\!\!\!-\!\!\!\!\!\!\!\!\!\!{alle}$), or (ii) not assign its θ-role to a specifier. In the former two variants, the presence of a syntactic argument carrying the external role leads to the availability of an acc-ia, whereas its absence in the latter variant leads to nom-ia (and an existentially bound ea or a by-phrase). This is in line with Burzio’s Generalization, where “[a]ll and only the verbs that can assign a θ-role to the subject can assign accusative case to an object.” (Burzio 1986, 178), and the Dependent Case Theory (see Marantz 1991), where (dependent) accusative case can only be assigned to an object if there is a higher (unmarked) argument. If there is an EA carrying the external θ-role (be it overt or covert), the IA receives acc (from Voice or due to the presence of a higher DP carrying unmarked case) and the higher EA receives nom by default. If, on the other hand, there is no syntactically projected EA, the IA can only get nom (as Voice cannot assign acc or given the absence of a higher DP carrying unmarked case).

4.2 The syntax of commissive root participles

The second non-sentential type of root participle which deserves a dedicated syntactic analysis is the commissive one. As laid out in Sect. 2, RP^coms are based on adjectival participles which are introduced by a performative C-head that conveys declarative force.

As hinted at above, traditional approaches to the grammar of adjectival (as opposed to verbal) passives trace their peculiarities back to a reduction in verbal material due to the introduction of adjectival properties, most prominently the lack of a VoiceP (or vP). More recent approaches primarily emphasise the role of the adjectival contribution of binding the event variable as the source for the central restrictions in terms of reducing the verbal denotation to an event kind rather than an event token (see Gehrke 2015). The structural representation in (76) acknowledges a reduced verbal domain (accounting for the lack of an EA and a ban on event-related modifiers) and suggests that the main contribution of adjectival participles consists of a resultative Asp-head that is embedded under an A-head.

The Asp-head existentially binds the event variable on v and presents a (result) state to the A-head, which attributes this state to the IA, by moving up a λ-operator into Spec, A. This allows the IA to be externalised in the sense of interpretively relating it to an external entity. This approach adopts (a version of) λ-abstraction (abstracting over the category to form a predicate) for the purposes of IA-externalisation⁴⁷ from McIntyre (2013) and Bruening (2014) but deviates from these by means of (i) not assuming a VoiceP to be present in adjectival participles (unlike Alexiadou et al. 2014 and Sleeman 2014, for instance), and (ii) dissociating adjectivisation from the means to derive a result in the first place.⁴⁸ Additionally, it sides with Embick (2004, 366) and McFadden and Alexiadou (2010, 412) in assuming that the resultative Asp-head can only combine with a ‘fientive’ verbal domain, i.e. a (simple) change of state (CoS).⁴⁹ This, in turn, deviates from Gehrke (2015, 918), who assumes that there is no AspP, as this would instantiate the event (as a token). Not exploring this any further at this point, the main logic behind this devation is that the specific participial Asp-head simply picks out the state that it is presented by the reduced verbal predicate, which is not sufficient to instantiate the event in any way. This accounts for the properties of adjectival participles (as employed not just in RP^coms, but also in adjectival passives and their elliptical counterparts that we will turn to below): the absence of an EA, a resultative state and its attribution to some (external) entity.

Turning now to the specific properties of the RP^com, the claim that we are dealing with an adjectival participle sheds new light on Ørsnes’ (2020, 377) insight that these configurations include a (propositional) null pronominal that is subject to anaphoric resolution “[s]ince the propositional argument […] is fully interpretable but cannot be overtly realized”. Given the absence of a PredP that could introduce an overt referent that is attributed the result state, the IA is bound to remain covert. This is resolved on the basis of the tight anchoring of RP^coms to the discourse, which readily supplies a referent in the form of what it is that is supposed to be promised or agreed upon, etc. This is supported by the presence of a performative C-head, which carries declarative force, but also conveys a speaker-orientation.⁵⁰ This arguably suffices to attribute the source of the participial result to the speaker despite the absence of an EA in the argument structure of the underlying predicate.

4.3 The syntax of expressive root participles

Unlike RP^dirs (which exhibit properties that are not found in any potential clausal counterparts) and RP^coms (which are not always interchangeable with adjectival passive counterparts), RP^exps have been argued to be derived from fully clausal counterparts via ellipsis.⁵¹ In fact, the expressive use of root participles may always be traced back to fully clausal counterparts, as illustrated by examples like hartelijk gefeliciteerd met je nieuwe neefje! (‘I hereby congratulate you on your new nephew!’). At the core of the RP^exp is a regular eventive passive that allows for discourse ellipsis of the auxiliary, the performative adverb and the subject, not just as a whole, but also partially: in Eline en jij hartelijk gefeliciteerd met deze belangrijke mijlpaal! (‘I congratulate Eline and you on this important milestone!’) only the auxiliary is omitted and the appearance of the coordination structure shows that there may be more complex material than a simple 2nd person pronoun in the subject position. Based on the analysis of eventive passive participles, which has also been employed as the core of RP^dirs in Sect. 4.1 above, the eventive passive syntax of RP^exps is represented in (77).

The claim that RP^exps simply boil down to elliptical verbal passives is supported by the overt licensing of an IA (like jij or je), the lack of an aspectual contribution, the availability of event-related adverbs (like hartelijk or oprecht), and the fact that we can derive expressive speech acts from declarative force. In other words, there is nothing that particularly points to the necessity for a non-elliptical structure which would arguably demand dedicated functional material. The requirement for a 1st person subject (and a 2nd person object) pragmatically stems from the performative use of such elliptical structures, when tightly anchored in a discourse.⁵²

4.4 The syntax of (bare) assertive and verdictive representative root participles

In contrast to RP^exps, which are elliptical verbal passives, RP^reps are elliptical adjectival passives and hence have adjectival participles at their structural core. Accordingly, the syntax of the participles involved in these configurations is similar to the one proposed for the non-sentential RP^coms above. Rather than simple participial APs merged with performative C, they are introduced as the complement of PredP, which is headed by the copula be and establishes a predicative relation to a nominal entity in its specifier position. This is structurally represented in (78) for bare assertive RP^reps.⁵³

This structure accounts for the resultative properties, the lack of an EA and the reduced combinability with adverbial modifiers. While degree-of-perfection modifiers like completely and fully may be introduced, they merely modify the adjectival state and hence need no fully instantiated event (token) to attach to. The same structural configuration may be employed to account for assertive RP^reps which introduce an overt subject, say All done! or Objection overruled! In both cases, the speaker-orientation, attributing the participial result state to the speaker, comes about by implication and, as we have seen above, dissolves with a decreasing degree of discourse anchoring. The only contrast between the two is that the nominal item introduced in Spec, PredP is not just an elided pronominal establishing a relation to an entity mentioned in the discourse. Rather, a universal quantifier or a DP that is subject to D-deletion appears in these cases.⁵⁴ The latter is employed to establish an anaphoric relation with something that is part of the discourse already (rather than introducing a new referent), e.g. that the speaker was supposed to carry out a specific task or that there was an objection that the speaker needs to judge. This provides further evidence in favour of an analysis based on ellipsis. In fact, like RP^exps, RP^reps may be subject to distinct degrees of phonological deletion, which suggests that eliding the copula is decoupled from phonologically deleting the subject, for instance. Finally, evidence for the elliptical nature of RP^reps also comes from the appearance of non-participial assertive root configurations like Game over!, Data correct! and House empty! These can also be analysed along the lines of (80), with a proper adjective rather than an adjectivized participle. As discussed in Sect. 2, these also permit degree-of-perfection modification, which shows that these do not modify any underlying event, and are subject to D-deletion just like their participial counterparts (consider *The game over).

This leaves the verdictive type of RP^reps, which primarily differs in the presence of a verdictive adverb, but apart from this may also be accounted for on the basis of the shared structure in (78). Adverbial modification is not just an optional ingredient but a crucial requirement for the formation of verdictive RP^reps and the hearer-oriented readings they convey. The latter may however primarily be traced back to considerations of informativeness in cases like Well done/played/observed! Badly drawn! and Wonderfully served! While it is not informative for a speaker to evaluate their own achievement, just stating that some result state has been reached when addressing a hearer who must know that they are the source of the result in question is not either. Adding a verdictive evaluation, however, justifies doing the latter. While it is tempting to introduce adverbial modifiers that achieve this—a subset of manner adverbs (cf. Schäfer 2005)—externally (i.e., outside of the participial AP, e.g. as modifiers of the LambdaP), this is put into question by the availability of cases that do allow for overt subjects, i.e. verdictive counterparts of assertive RP^reps like Job well done! And Money well spent! (or Alles gut gemacht!, ‘All well done!’). These suggest that the adverbs attach to the verbal domain, but are not restricted to modifying an event token, quite unlike proper event-related modifiers like slowly and carefully. Once again, while the hearer-orientation is determined by informativeness and a strong discourse anchoring, this diminishes with a decrease in the latter. Accordingly, fully-fledged adjectival passives, which do not explicitly lend themselves to a strong discourse anchoring, permit more flexibility and even allow a more neutral 3rd person reading.

This concludes the discussion of the syntax of root participles, where we have now been able to distinguish four major types: (i) non-sentential RP^dirs which consist of an Imp-head that embeds a verbal passive participle, (ii) non-sentential RP^coms which are formed by combining a performative C-head with an adjectival participle, (iii) RP^exps which boil down to elliptical (verbal) passive periphrases, and (iv) (bare) assertive and verdictive RP^reps which may be analysed as elliptical adjectival passives.

5 Conclusion

The present paper investigated participles in root configurations. It started off by establishing a typology of four basic types of root participles: RP^dirs, RP^coms, RP^exps and RP^reps. Then, on the basis of a principled discussion of their basic properties, it proposed that there is a split both in terms of whether the distinct types are formed on the basis of verbal or adjectival participles (RP^dirs and RP^exps vs. RP^coms and RP^repS) as well as along the lines of whether they are non-sentential or elliptical (RP^dirs and RP^coms vs. RP^exps and RP^reps).

The type of root participle that could be shown to differ most strongly from its potential clausal counterparts was the verbal RP^dir, which was argued to consist of an imperative functional head that introduces a directive interpretation and has the potential to license the otherwise suppressed EA of the underlying passive participle (by virtue of carrying an empty θ-role and 3rd person plural φ-features). This could be shown to have interesting repercussions for the licensing of either an acc-ia in the context of a quantificational subject (in analogy to the have-perfect in identity approaches to past participles) or a nom-ia, if the quantificational subject is demoted to an adjunct. The second non-sentential variant of root participle, RP^com, in turn, was shown to be structurally reduced in the sense of comprising a performative C-head that selects for a resultative participial AP, which prevents the insertion of an overt subject (due to the absence of PredP) and assures that a commissive speech act is carried out. The two types of RP^exp and RP^rep, on the other hand, could be shown to be fully clausal eventive and adjectival passives, respectively, yet subject to ellipsis.

What is left for future research are technical questions like those pertaining to the need of a dedicated mechanism like θ-merger for the licensing of an EA in the context of a passive VoiceP. Another technical issue concerns the mechanisms behind ellipsis: most pressingly, is there a syntactic reality to eliding specific items in the context of RP^exps and RP^reps? Additionally, the parameterisation of the distinct types of root participles deserves further attention: why are RP^dirs and RP^coms only available in Dutch and German but not in the other Germanic languages? Can this simply be traced back to functional heads in C-domain that come equipped with different sub-selectional requirements (e.g., selecting AP or AspP)? What are the precise licensing conditions for RP^dirs in Dutch as opposed to German? And why are the elliptical RP^exps only available in Dutch in contrast to the elliptical RP^reps, which are available across Germanic? Finally, fruitful insights may also be gathered from a comparison of root participles with participles that are only very loosely attached to a host clause: absolute clauses like Den Einspruch abgewiesen, fuhr er mit dem Kreuzverhör fort (‘(With) the objection rejected, he continued with the cross-examination.’), which occasionally also exhibit acc-ias.

In addition to raising many novel questions and presenting routes for future research, what the present paper could show on the basis of teasing apart the various types of root participles is that the grammatical system may actively exploit the underspecification of participial items in terms of their argument structure and aspectual properties to encode distinct speech acts on the basis of reduced structures.