2. Exclamative structures and sentence types in German

Michaelis (Reference Michaelis, Martin Haspelmath, Oesterreicher and Raible2001:1038-1039) lists a number of cross-linguistically valid definition criteria for exclamatives: They convey the speaker’s affective responses to situations, namely surprise, and they presuppose the coded proposition, often involving a scalar degree. Michaelis’s typological survey shows that the formal coding of exclamations is cross-linguistically recurrent. She lists co-occurrence with interjections, complementation structures involving factive epistemic matrix verbs, topic constructions, anaphoric degree adverbs, question words, NP complements, free NPs, and inversion as characteristics of exclamative constructions (Michaelis Reference Michaelis, Martin Haspelmath, Oesterreicher and Raible2001:1042-1048). Beijer (Reference Beijer2002:10) describes exclamatives as expressions of “deviations from certain norms, but also various feelings, such as astonishment, joy, surprise and disappointment in relation to this deviation”on the part of the speaker.

For a discussion of German exclamatives Fries (Reference Fries1988:4) invokes the notion of an affective–emotive stance towards the encoded proposition on the part of the speaker as the functional commonality. Acknowledging the difficulty of defining an exclamative sentence type by means of simple formal criteria, d’Avis (Reference D’Avis, Meibauer, Steinbach and Altmann2013:171) states that exclamatives are usually identified through their function, such as the non-propositional expression of divergence between the encoded state of affairs and the speaker’s expectation, which leads to astonishment. Indignation in this article is understood as a subtype of astonishment, one that highlights a negative emotional reaction to a perceived state of affairs. Acknowledging the overall fuzziness of the categorial status of sentence types insofar as a well-defined set of characteristics is only available for their core areas, whereas at the fringes there are overlaps and mixed types, Batliner (Reference Batliner1988:243) remarks that the category of exclamatives is particularly fuzzy. Schallert (Reference Schallert2019:479) points out that morphosyntactically exclamatives belong to an abundance of marginal form types in German, of which wh-sentences and a subclass of free-standing dass-initial sentences can be identified as the main types. Each of these can be differentiated from formally similar form types by prosodic marking, specific modal particles and gradable expessions, albeit not consistently depending on the usage context. Such class-building features (Schallert Reference Schallert2019:480) therefore do not seem to constitute sentence types as such, but rather contribute to their identification. Schallert (Reference Schallert2019:479) furthermore mentions declarative-like sentences that gain exclamativity through modal particles as well as exclamative nominal constructions consisting of a noun phrase and a relative clause. Repp (Reference Repp2016:4) reports that in German wh-exclamatives the position of the verb matters, which shows that certain formal coding differences can be directly linked to a change of illocutionary potential in exclamatives. According to her verb-second exclamatives of this type are more specific; they encode surprise about a gradable property, as in Wie ist er gewachsen! (‘He has grown so much!’). Verb-final wh-exclamatives such as Wie er gewachsen ist! (‘He has grown!’) on the other hand are semantically more versatile; they simply encode surprise about an unexpected circumstance. While some dass- and wh-initial exclamative sentences may also express speaker indignation, this specific function is not associable exclusively with these types. A dass-initial exclamative such as Dass dieser Typ die Zwiebeln brät! (‘How can this guy fry the onions!’) is an exclamation of surprise, which may be interpreted in terms of indignation in a given situation, but not necessarily so. As I discuss below, the form–function relation between certain zu-infinitives and speaker indignation is much tighter, which supports a constructionist analysis. In her cross-linguistic examination of root infinitives Nikolaeva (Reference Nikolaeva2014) proposes a constructionist approach that posits the existence of abstract, illocutionary “super constructions,” such as for instance the “abstract directive construction,” each associated with a family of morphosyntactic patterns. These inherit the use conditions of the super constructions, but have individual formal restrictions. Given the formal disparateness, such an approach also seems viable for exclamatives. Somewhat similarly, Chernilovskaya (Reference Chernilovskaya2014:4-5) remarks that there is no one-to-one mapping between certain morphosyntactically defined sentence types and exclamative speech acts per se. Acknowledging the formal heterogeneity in the realm of exclamativity, she therefore defines exclamatives in terms of their discourse properties.Footnote ³ Specifically formulated for wh-exclamatives, but open to the idea that a similar approach may also shed light on other types, she concludes that exclamatives cannot be denied by the speaker, they upgrade the common ground, and – most importantly – they show that the speaker undergoes an emotive event triggered by a situation or an utterance (Chernilovskaya Reference Chernilovskaya2014:141-144). Such properties are examined with respect to independent zu-infinitives in the following section.

3. Formal and functional properties of independent zu-infinitives

For a zu-infinitive to be used independently as a semantically constrained exclamative some necessary conditions must be met. Most strikingly at least one constituent in preverbal position seems obligatory. Independent bare infinitives, on the other hand, are grammatical and common without preverbal material.

Note that the condition is not limited to objects of transitive verbs, since the same form–function relation obtains in intransitives. In this case, however, either a particle, an adverbial or a phrasal adjunct may precede the verb. If a transitive verb is used without further material, indignant meaning may still be associated, but it is considerably less clear:

These observations suggest that the specific illocutionary force is not brought about by the mere use of a syntactically independent zu-introduced infinitive verb, but more likely results from a form–function relation that obtains between the structural pattern [X+zu+Inf] and the feature indignant(sp, prop), where X can be a variety of elements ranging from particles such as, for instance, the clausal negator nicht to a full-fledged DP or PP in either argument or adverbial function. As has been pointed out by Gärtner (Reference Gärtner, Meibauer, Steinbach and Altmann2013:205), nominative arguments that refer to the addressee(s) are excluded. They may, however, be encoded as an expression of direct address immediately preceding the construction (compare (4d) and (4e)). Note that this constraint does not apply to directive bare infinitives such as Alle die Hände vor den Mund halten! (‘Everyone cover their mouths with their hands!’), which are perfectly grammatical.

There are some further formal constraints. Consider the contrast in (5). While all three examples are grammatical as such, they differ in terms of illocutionary potential. The sentence in (5a) has an exclusive indignant reading. The utterance in (5b) with a bare NP argument, on the other hand, makes this reading impossible. The distinction DP versus bare NP apparently makes the difference here. The third utterance in (5c) can likewise be interpreted as an indignant exclamative, despite the non-determined NP argument.

Utterances of the type (5b) are used exclusively as expressions of offers. They commonly occur in asymmetric communication situations such as shouting from a market stall or in the written medium on signs or in ads. It remains to be investigated whether it is the asymmetry of the communication situation that disambiguates the illocutionary function or the constructional pattern itself. However, even the contrast shown here gives us a hint at how the usage context is essential: Typical circumstances of communication may contribute to the emergence and in some cases sedimentation of constructional patterns with specific non-compositional meanings. I can only speculate here, but quite possibly the constraint on undetermined NPs results from another, very specific form–function pairing that has emerged to convey the meaning of an offer, which may block the indignant reading, given appropriate circumstances or use conditions: The offer reading occurs only if the verb denotes market transaction of sorts. Note that with verbs outside this semantic realm indignant exclamative meaning obtains even despite bare NPs, as (5c) shows, though some degree of ambiguity remains. If one assumes two parallel constructions that differ only in one small detail syntactically, but have very different and equally specific illocutionary functions, this may help to explain why the presence of a determiner in (5d) leads to the exclusive exclamative effect of speaker indignation. It appears as if this exemplifies disambiguation in cases of formal similarity of constructional patterns.

The independent zu-infinitive shows the typical constituent order of verb-final subordinate clauses in German including infinitive complements, namely OV. Postposition of the additional constituent results in ungrammaticality. However, when compared with their embedded counterparts, it becomes obvious that postposition is perfectly possible (see example (6c)). If one further assumes that a preverbal position must be filled only for the indignant construction to be recognized as such, this is not surprising.

Beijer (Reference Beijer2002:10; also cf. Rosengren Reference Rosengren1997) points out the importance of marked intonation in bringing about exclamative meaning. However, in zu-infinitives a special intonational contour cannot be held responsible for it. Prosodic cues such as a marked intonational contour may enhance the exclamatory effect and also help to disambiguate by approaching intonational contours and accents approximating that of other, more established constructions to the effect of forcing either a directive or an exclamative reading in a given situation, but they do not seem to be an integral part of the patterns as such (cf. Schallert Reference Schallert2019:479). The contrast between bare infinitives such as Die Zwiebeln braten! and Die Zwiebeln zu braten! presented in section 1 makes this apparent. The bare infinitive may in principle be used exclamatively as well as directively. One can argue that the directive function is basic and the addition of an exclamative accent may serve to disambiguate, but even an accented version Die ZWIEbeln braten! is easily mistaken for a directive with contrastive focus (for the relation between exclamative and focus accent see Schallert Reference Schallert2019:487 and sources therein). The zu-infinitive, on the other hand, is always exclamative; it is unambiguous even without marked intonation, as long as it displays all the features that make up the constructional pattern proposed in this article. This is in accordance with Truckenbrodt’s (Reference Truckenbrodt2013:596-597; see also Batliner Reference Batliner1988:269) analysis of exclamative intonation as an extra-grammatical gesture that refers to salient propositions. The indignant zu-infinitive displays a more specific form–function relation and thus exerts exclamative force without context, which is not possible with bare infinitives. It should furthermore be noted be that even in written form only the zu-infinitive is almost exclusively exclamative, which is also evident in the corpus data.

The zu-infinitive construction proposed in this article matches the discourse properties formulated by Chernilovskaya (Reference Chernilovskaya2014): Its exclamative effect unfolds as an emotive reaction triggered by an utterance or event that causes indignation on the part of the speaker. The exclamative cannot be denied by the speaker. When uttered in interaction, also the common ground is updated insofar as the speaker’s indignation is now manifest in the discourse. In contrast to other structures that may be interpreted as exclamation of speaker indignation such as bare infinitives or independent non-directive dass-clauses, the independent zu-infinitive conveys this exclamative function more consistently. That is to say, if A hears B exclaim a zu-infinitive without immediately knowing in what context B makes the utterance, A will get the impression that B is negatively surprised by something. If A, on the other hand, hears B use another potential exclamative such as a bare infinitive, the impression of B’s indignation does not arise, if no context information is available to A.

Having outlined constraints that apply to zu-infinitives with unambiguous indignant reading, I now present and motivate an analysis in terms of a constructional approach. Given that there is no identifiable element that one can ascribe the specific attitudinal meaning of speaker indignation to, in order to capture this idiosyncratic meaning I suggest the following form–function relation for the independent zu-infinitive:

The index _-sub indicates that the construction is syntactically independent. This is assumed to be stored as a complex schematic construction which combines a particular syntactic pattern with a specific semantic structure and consequently has a very narrow illocutionary range. Even from the examples given so far it becomes clear that the pattern at hand is not idiomatized in the sense that there is a fixed collocation partially sensitive to morphosyntactic processes such as, for instance, to kick the bucket/kicked the bucket, but rather a schematic construction, which combines idiosyncratic properties of lexical signs with the productivity of syntactic constructions. In constructionist approaches all of these are viewed as pairings of form and meaning (see Goldberg Reference Goldberg1995, Stefanowitsch & Gries Reference Stefanowitsch and Gries2003, Fried Reference Fried2013, Hilpert & Diessel Reference Hilpert, Diessel and Schmid2017, Boas & Ziem Reference Boas, Ziem, Boas and Ziem2018) that form a spectrum with simplex elements such as morphemes or words at one end and structural configurations such as schematic constructions with variables or even sentence types at the other end.

The exclamative zu-infinitive construction as outlined in this article represents a systematic correspondence of morphosyntactic conditions and a specified illocutionary potential. As such, it is of course one of numerous exclamative form types that exist in German for related purposes. Whether this is reason enough to subsume it under a sentence type, however, is not clear. The question as to what actually constitutes a sentence type in German, whether it is definable in terms of primarily formal or semantic–pragmatic characteristics and also by what mechanism these two realms are matched systematically, is subject to debate (for a more detailed discussion see d’Avis Reference D’Avis, Meibauer, Steinbach and Altmann2023:173-174 and Schallert Reference Schallert2019:489). Basically, there are two major approaches: On the one hand, the correspondence approach (Altmann Reference Altmann1993; see also Meibauer Reference Meibauer, Meibauer, Steinbach and Altmann2013, Finkbeiner & Meibauer Reference Finkbeiner and Meibauer2016:8–12) proposes constitutive classes of formal characteristics that can be combined relatively freely. On the other hand, there is the derivational approach, which assumes that the communicative function of a sentence type is always the result of composition (Reis Reference Reis1999), that is to say its illocutionary use potential can be calculated from its component parts and the way these are structurally configured. While the first approach resembles the more holistic constructionist view, where linguistic expressions are signs of various size and complexity, the second is essentially generative in character.

Jacobs (Reference Jacobs2016:23) claims that projectionist solutions such as the aforementioned correspondence approach work as long as one can safely calculate the features of a complex expression from those of its component parts, otherwise one must resort to constructions. He furthermore distinguishes between substantial and schematic constructions, both defined as classes of expressions: Substantial constructions comprise morphemes, words, and idiomatic phrases, because their meaning is non-compositional, as it cannot be calculated from their component parts. Also their respective classes of expressions contain only one element each. Schematic constructions, on the other hand, may be instantiated by a potentially open set of expressions, while at the same time the pairing of form and meaning remains constant, resulting from a recurrent and easily recognizable configuration of constants and variables. The exclamative zu-infinitive is, in my view, best described as such a schematic construction, as it is defined in terms of a structural configuration of invariant elements and positional slots.

A case in point similar to the one discussed in this article can be found in verbless directives such as (7). Using these the speaker commands an addressee to bring about a desired change of location of the entity encoded in the NP-complement to the preposition mit.

This pattern combines a locative PP with another PP governed by mit. A full representation suggested by Jacobs (Reference Jacobs2016:27) looks like this:

Here the propositional meaning is that x moves to an endpoint along the way w, where α^prop represents the propositional content of α. This is in my view a neat and elegant analysis of the regular but unexcpected illocutionary property apparent in these structures.Footnote ⁴

For the semantic structure (Sem) of the constructional representation Jacobs (Reference Jacobs2016) assumes several levels, each responsible for certain meaning aspects. A construction may be semantically represented by the propositional, the illocutionary, and the attitudinal level or a combination thereof, each hosting features that determine the appropriate interpretation and use conditions. The illocutionary level is exemplified in the representation above for verbless directives: The presence of a feature dir(prop) limits the use of the contruction to directive illocutions.

Attitudinal meaning is commonly associated with specific items within the construction, such as gern (‘gladly’) in the following imperative sentence:

By inserting the word gern in this example the speaker assures the hearer that he has a positive attitude towards the hearer’s realization of the prompted proposition. Without this word the sentence would be perceived as attitudinally neutral or perhaps even rude, as the hearer would have no clue as to whether he really is welcome to call.

For this expression Jacobs assumes the semantic representation below:

Because the meaning of gern in this case expresses a positive speaker stance or attitude towards the proposition, it is incompatible with the negative version of the same imperative:

This instantiation of gern contrasts with one that expresses the positive stance towards the state of affairs on the part of the semantic subject (Jacobs Reference Jacobs2016:50), as in (9b). If such items have attitudinal meaning it must be compatible with the meaning of the larger structure. In (9a) the speaker disapproves of the state of affairs encoded in the predicate and consequently wants the addressee to prevent it, while at the same time attitudinal gern entails that he in fact embraces the realization of said state of affairs being brought about by the addressee, which results in a mismatch.

My claim is that speaker-attitudinal meaning is associable not only with simplex items like gern, but also with schematic constructions. Note that while the syntactic configuration is the same, there is a semantic difference between the sentences in (7) above and the ones in (10):

In contrast to those in (7) the expressions in (10) are exclamations driven by speaker attitude rather than true directives. While this differentiation is difficult to motivate in a projectionist approach, it can easily be captured in the semantic representation by assuming a different specification at the illocutionary levelFootnote ⁵ and also adding an attitudinal level to the schematic construction:

This way one can account for the fact that with the expressions in (10) the speaker strongly opposes the concept encoded in the NP complement to mit, regardless of there being an addressee or not. Arguably this more specified representation is the result of a further advanced constructionalization process, in the course of which a single feature in the layered semantic representation yields an entire class of expressions with similar use conditions. Indeed, the set of possible directional PPs is much smaller for expressions of the type (10) than for those of the type (7). Note also that the English translation of (10a) centering around the concept of ‘hell’ matches the German construction syntactically, while for all others a different construction seems more appropriate.

Opting for an analysis along these lines it can be said that in the case of the zu-infinitive the attitudinal level presumably hosts a feature indignant(sp, prop), which contributes the pragmatic–semantic component of speaker indignation with respect to the state of affairs encoded in the proposition. As seen in section 2, speakers can express such indignation in various ways in Modern Standard German, including independently used bare infinitives, so the feature cannot be simply attributed to the semantics of the infinitive particle. Rather it must be a feature associated with the complex configuration as a whole. For the construction syntactically characterized as

one can therefore assume the following semantic representation:

This way the attitudinal meaning component is associated with the entire schematic construction. There are no conflicts or mismatches, the construction is used exclusively with the specified attitudinal meaning. Note that so far the above representation contains merely an unspecified proposition. Using a construction characterized in this way seems appropriate wherever the speaker expresses indignation in regards to the state of affairs, irrespective of whether someone actually brings it about or not. However, indignation is often not targeted towards an abstract state of affairs, but rather towards the circumstance in which someone actually brings it about. The prop argument on the illocutionary level may thus more appropriately be understood in terms of a specified proposition, namely speaker indignation about the fact that there is in fact someone to realize the state of affairs (Joachim Jacobs p.c.). With this detail included a complete representation looks like this:

Here the attitudinal level provides this specification: With a construction represented this way the speaker not only exclaims an emotion towards a state of affairs, but more precisely expresses indignation about the circumstance that someone actually engages in it.

As could be seen so far there is a significant contrast in illocutionary potential between independently used bare and zu-infinitives. In the next section I turn to the question of how the functional specification of the latter has come about in relatively recent history of the Standard variety and propose likely emergence scenarios. Given that the regular environment for infinitives is subordinate predicates that carry no illocutionary force, the lack of a finite matrix clause gives the independent zu-infinitive a sense of incompleteness. As a situational phenomenon, expression of indignation is most likely to emerge in spontaneous dialogue. In dialogic discourse omission of constituents is most likely to occur on grounds of shared circumstantial and situational knowledge. In regards to the emergence of the specific pairing of form and function outlined so far, I would like to entertain the thought that it is here where a reduced utterance that allows the hearer to infer the intended illocution may become a preferred choice in recurrent situation types and eventually constructionalize over time. The emergence of a semantic feature such as indignant(sp, prop) may then be explained in terms of “blending” of the representation of such recurrent situations with the semantic structure of the expression in question. Frequent usage in certain contexts thus ensures that a construction acquires or loses new meaning components, similar to semantic changes in single words, such as meaning extension or bleaching.

4. Tentative emergence and entrenchment

The element zu has been in use as an infinitive particle (at least) since the Old High German period, having originated as a preposition. Since then the zu-infinitive has been increasingly used at the expense of bare infinitives in subordinate clauses (see Demske Reference Demske2001 and sources therein). The following example shows that in Old High German zu- and root infinitives were in more or less free variation:

Demske (Reference Demske2001:71-72; cf. also Brandner Reference Brandner2006:246) states that numerous verbs in Old High German could take both types of infinitival complement without any semantic difference. In later periods the infinitive is not only used as a complement, where the presence of desemanticized zu is dependent on the matrix verb class and governed by considerations of language processing, but also with adverbial or predicative function, where the semantic import of zu (denoting ‘purpose’) is relevant. Demske (Reference Demske2001:85) remarks that zu-infinitives generally enter into coherent constructions, that is a tighter connection between embedding and infinitival verb form as part of a complex predicate, not as readily as bare infinitives, but do so as complements of certain verb classes (modal and AcI-verbs), where they have replaced the bare infinitives in the course of language history. In Middle High German its distribution in comparison with the bare infinitive still is largely unsystematic, so that one must assume the emergence of the clear functional differentiation as proposed in this article in later periods. Schöbler (Reference Schöbler1969:314) points out that in Middle High German the zu-infinitive (encoded by the older form ze) is, generally speaking, still much rarer than in Modern German. A manual scrutiny of the FnhdC (Bonner Frühneuhochdeutschkorpus, Bonn Early New High German Corpus) confirmed this and also revealed that there is no syntactically independent use of infinitives.

What is most relevant for the present investigation, however, is that before the onset of the New High German period no correlation between zu-infinitives and indignant meaning could be found. Consequently, the next step was to find evidence for first usage of the form–function relation in later texts. To this end I utilized two archives of the DeReKo (Deutsches Referenzkorpus, German Reference Corpus, Institute for the German Language (IDS), Mannheim).Footnote ⁶ The historical archive (HIST) comprises corpora of various text types from the sixteenth century until the 1960s. The archive TAGGED-C2 documents current Standard German.Footnote ⁷ The historical corpora mainly consist of literary texts containing reported speech, which provides a window into conversational routines or at least an approximation thereof. This proved particularly helpful for tracing pragmatic phenomena, as one can manually analyze the respective co-text in order to detect indignant meaning. The corpus search was done using COSMAS II (corpus search and management system). Exclamation marks served as useful indicators for meaning components related to speaker stance, since that is their primary function in written language. The historic corpora of the DeReKo represent developments in an emerging standard variety, beginning with a leveling of various written varieties representing dialectal variation, followed by usage expansion of the new standard written variety and opening towards influences from spoken language (see Weiß Reference Weiß2005). The time of constructional emergence proposed in the present article falls within this period of expansion of the new Standard variety.

The findings can be summed up as follows: If the list is scrutinized chronologically the only examples of zu-infinitives with matching illocutions are embedded under a matching matrix, while independent zu-infinitives with indignant exclamative meaning only appear later. In my view this already supports an analysis of development from full, complex sentences with subordinate infinitives towards free-standing infinitives with limited illocutionary potential. A manual scrutiny of earlier texts in the HIST archives reveals several instances of free zu-infinitives, which clearly show speaker indignation. These occur first in poetic discourse, which is, however, often interactional and should therefore not be disregarded as a possible source (see Bade & Beck Reference Bade and Beck2017 for the benefits of lyrical texts for grammatical analysis). It is after 1730 that the picture changes and numerous matches from non-poetic discourse appear. While I am aware that the findings are scarce and any generalizations based on these are tentative, given the observable situation in Modern German, I believe the following claim is nevertheless justified: If the assumption formulated in the previous section is correct, a feature indignant(sp, prop) must have emerged and become exclusive at the attitudinal level before 1730.

Example (12) is one of the first few hits in the historical archives of the DeReKo. Here the exclamation can be characterized as disbelief:

A potential source construction is the following complex exclamative, where indignant speaker stance meaning is attributable to the matrix clause, while the subordinated chain of zu-infinitives specifies the state of affairs addressed in detail.

The exclamation mark clearly indicates that the matrix clause introduced by wie (‘how’) is in fact a typical wh-exclamative and not an interrogative.

After a lengthy gap in the corpus data numerous semantically clear examples of the construction appear from c. 1730 onwards:

It is worth noting that from around this time authors increasingly accommodated spoken language. According to Weiß (Reference Weiß2005), the emergence of a standard variety runs along the following lines: First a written variety arose as the result of the leveling of written varieties for effective communication across dialect boundaries, which underwent extension to more and more domains of communication. It then became permeable by influences from spoken language. An increasing number of speakers acquired passive competence in the new standard variety, which in due course led to active usage and ultimately to L1 acquisition by the following generations.

The form–function relation in (14) between free, independent zu-infinitives and the expression of speaker indignation is clearly evident from the co-text. Likewise the following exemplifies this:

Here the exclamation following the structure in question leaves no doubt about its indignant meaning. In (16) it is a string of derogatory terms for the prospective husband that makes a high degree of speaker indignation clear:

Note that there is no apparent restriction as to the verb in these examples. If this is compared with the synchronic situation as it presents itself in the DeReKo archives one can see that some collocations are way more frequent than others. Naturally emergence and initial development of the proposed schematic construction are speculative due to the apparent scarcity of solid data. Nevertheless, I am convinced that these collocations are a relatively recent innovation building on there being a somewhat conventionalized form–function relation, most likely by means of analogy of syntactic structure as a driving force (Diessel Reference Diessel, Bergs and Brinton2012; also see De Smet & Fischer Reference De Smet and Fischer2017 for an overview).

The first occurrence of the nowadays highly frequent collocation nicht zu glauben (‘unbelievable’) that clearly shows the targeted indignant meaning dates back to as late as 1828; the first syntactically independent usage does not occur until half a century later, as evident in (17) and (18), respectively:

As in (13) above, the exclamation mark identifies the matrix expression as a wh-exclamative. The example in (18) does without a matrix expression:

The crucial difference between the two examples, therefore, is that in the former the proposition that the speaker intends to express an attitude towards is encoded in a complement clause and as such is still part of the overall construction, while in the latter the relevant content is part of the situational context, which in this case is provided by the co-text immediately following the expression.

In contemporary Standard German expressions such as the following have a particularly high frequency of usage:

Here the condition that at least one constituent precedes the zu-infinitive is met by the presence of a particle (kaum, nicht). Arguably, such examples display a very high degree of conventionalization, perhaps only one step short of being analyzable as monolithic phraseologisms. Note that while they match the syntactic requirements prescribed by the schematic construction outlined above, their semantics are slightly different. In contrast to the examples discussed before, here the propositional argument of indignant(sp, prop) is not explicitly spelled out. Instead it follows from the context, just like the interpretation of the non-exclamative examples in (5) discussed in the previous section. Crucially, rather than expressing disbelief that someone else fails to grasp or believe something, the speaker exclaims indignation about some state of affairs that he or she perceives. This state of affairs, however, is not encoded within the construction itself: It can only be inferred from the context. The predicates in (19) are incomplete: There is no overt object, even though both verbs are transitive. In fact, they contain no arguments at all. As a result, the only possible use available for such expressions is an exclamation of general, unspecified indignation. Consequently, if speaker A hears speaker B utter one of the exclamations in (19), he will surely understand B’s indignation, but may feel the need to ask B what it is about.

How are such examples related to the indignant exclamative zu-infinitive constructions argued for in this article? Since the syntactic pattern of (19) exactly matches that of the schematic construction proposed for exclamative zu-infinitives, I assume that the latter must have been already available as a stored pattern at the time of first usage of the former and thus it could have served as the basis for analogic change (see De Smet & Fischer Reference De Smet and Fischer2017 for an overview). In other words, exclamative force is triggered by the constructional pattern, but there could be a slightly different diachronic source construction.

Some hint towards a possible source construction may be found in modal zu-infinitives. Holl (Reference Holl2010:9) discusses several types of modal infinitives with the matrix verbs sein (‘be’), which are passive in that the patient argument is syntactically the subject. Such sentences have two modal readings, which convey either necessity or possibility, as exemplified in (20). Functionally these passive infinitives bear some resemblance to the examples in (19) insofar as both types encode the speaker’s evaluation of a state of affairs, although here in terms of modality:

Note that in (20b), a modal passive construed so as to include the constituents of (19) given here for comparison, the possibility reading is somewhat similar to the examples in (19), albeit in uttering (20b) no exclamative force is being exerted. The difference lies in the fact that passive modal infinitives are in fact complex sentences: They contain an argument and the passive auxiliary sein. Such passive modal infinitives might well be the diachronic source structure of generic exclamative zu-infinitives, as long as they contain perception or cognition verbs that encode incredulity towards whatever is encoded in the matrix argument. In contemporary German leaving out Der Umstand ist (‘the circumstance is …’) in example (20b) invariably results in an exclamative expression of indignation directed towards the same circumstance.

With respect to the following examples Holl (Reference Holl2010:59) points out that with deletions such as in (21) passive modal infinitives can be used discourse-initially, that is not only as an elliptical response to a question:

Note that for (21b) and (21c) Holl (Reference Holl2010:60) provides non-linguistic contextual information. Here it shows that the examples are pieces of written communication. Such contexts make ellipsis of argument and auxiliary possible (just as in (5b) discussed in section 3), so that the residual zu-infinitive is correctly interpreted in the same way as the full structure. If no usage context is provided externally, through situational or medial cues, the remaining elliptical structure automatically receives an exclusive exclamative reading. This is the case in (21a), which contains the same type of highly conventionalized zu-infinitive that encodes an exclamation of generic indignation as in (19). In the case (21a) the object of disbelief – the state of affairs – is given in a subordinate clause, but for the purpose of expressing indignation this is entirely optional. Exclamativity, more specifically speaker indignation, emerges in elliptical structures of this sort as soon as no situational context is specified and the constructional criteria are met. For this reason, the passive modal infinitive construction as described by Holl appears to be a possible source construction for generic indignant zu-infinitives.

Table 1 shows the respective frequencies of the collocations zu fassen and zu glauben in the HIST and TAGGED-C2 archives as yielded by the search query [zu + V + !]. As exclamatives of indignation (and possible related concepts), both are largely identical semantically and very common in contemporary German.

Table 1.

Frequencies of the collocations zu fassen and zu glauben in two archives of the DeReKo (German Reference Corpus)

zu fassen!: x² = 13,475/p = 0,001, zu glauben!: x² = 29,46/p = 0,001

The results indicate that in the course of a roughly 200-year timespan since the first occurrence of these collocations there has been a notable increase in indignant speaker stance as an associated meaning component. Not surprisingly, all instances of independent indignant zu-infinitives with glauben and fassen are preceded by one of the two negative particles kaum (‘hardly’) or nicht (‘not’).

What can be seen so far is that first an erstwhile subordinate structure is being increasingly used as a syntactically independent but context-sensitive expression taking on specific meaning expressing speaker stance. Then at a later stage even further idiomatized structures with the same pairing of form and function arise that are applicable more generally, because they may leave the state of affairs that the speaker wants to comment on unspecified. How can the emergence of such schematic constructions be accounted for adequately? Some hints come from the emergence of a different structure. Discussing the diachrony of copular constructions between Old English and early Middle English, Petré & Cuyckens (Reference Petré and Cuyckens2009:340) make the following observation:

• Formal as well as semantico-pragmatic properties of schematic constructions may change. A change originates in instances on the utterance level, but may be represented as a feature of the schematic construction itself in due course. A change represented at the schematic level can in turn impact lower-level constructions that originally did not participate.

• Once the schematic construction is sufficiently entrenched in the minds of the speakers, one or more slots may begin to be productive, so that new lexemes become associated with the schematic construction.

Emergence is thus a matter of entrenchment, a gradual process, whereby each instantiation of a construction strengthens its mental representation (Diessel Reference Diessel, Bergs and Brinton2012; also cf. Schmid Reference Schmid2017). If individual speakers activate a linguistic concept or construction with sufficient frequency, it becomes increasingly recognizable and is likely to spread to the speech community, where it becomes routinized. According to Diessel (Reference Diessel, Bergs and Brinton2012), entrenchment is the result of habituation and a prerequisite for analogy. This means that formal similarity influences the likely selection of expressions for similar communicative purposes, which is essentially the situation found for exclamative zu-infinitives:

• First a meaning component erstwhile associated with a matrix clause becomes established with a given syntactic pattern on grounds of recurrent selection by speakers in similar situations (see example (14) above).

• This gives rise to further idiomatized patterns that are syntactically and lexically more restricted, but applicable in a wider range of situations so that a higher frequency can be expected.

Based on a typological pilot study, Evans (Reference Evans, Nikolaeva and Plank2007:367) introduces the concept of insubordination as “the conventionalized main-clause use of what, on prima facie grounds, appear to be formally subordinate clauses.” According to Evans such structures figure most prominently in three functional domains: interpersonal control, modality, and signaling that material is presupposed. In regard to interpersonal control, he also points out that insubordinate structures typically convey attitudinal and interactive meaning. Schallert (Reference Schallert2019:479; see also König & Siemund Reference König, Siemund, Meibauer, Steinbach and Altmann2013: 863) concurs that exclamatives in general are cross-linguistically typical cases of insubordination, which he refers to as syntactic disintegration. For German wh- and dass-sentences as well as other exclamative types he deems insubordination a likely development (in Jäger Reference Jäger2021 I present a diachronic emergence scenario for independent dass-sentences as a type of prohibitive, distinct from dass-exclamatives). Some authors such as Gras & Sansiñena (Reference Gras and Sansiñena2015) even go as far as to say that exclamativity is a feature of all insubordinated structures. Van den Stock et al. (Reference Van den Stock, Astrid De Wit and Gras2023) further elaborate on the cross-linguistic association of insubordinate structures and discourse-organizing functions as well as exclamativity as attitudinal meaning. Free-standing zu-infinitives fall into this category. The expression of attitudinal meaning in interaction described by these authors as cross-linguistically recurrent matches precisely the situation found with zu-infinitives used systematically as markers of speaker indignation. It exemplifies Traugott’s (Reference Traugott1989:35) notion of subjectivization, a semantic-pragmatic process, by which “meanings become increasingly based in the speaker’s subjective belief state/attitude toward the proposition, in other words, towards what the speaker is talking about,” a diachronic process for which she provides evidence from the history of English. A construction or elements therein thus acquire new semantic components that convey speaker attitudes towards the encoded state of affairs (also cf. De Smet & Verstraete Reference Hendrik and Verstraete2006:365 in Margerie Reference Margerie2011:124).

As Van den Stock et al. (Reference Van den Stock, Astrid De Wit and Gras2023) point out, what remains unclear in any analysis of insubordination is how exactly the transfer of illocutionary force from a matrix expression to an erstwhile non-illocutionary subordinate structure takes place before it becomes conventionalized. At this point an idea by Traugott (Reference Traugott2017) proves helpful. Drawing on the uniformitarian principle, which postulates that the same means are available to speakers at all times (cf. Labov Reference Labov1972:101), she suggests that “incomplete” structures and phrasal fragments are used as increments of interactive discourse freely and commonly employed wherever speakers have the necessary information to reconstruct the intended whole structure in a given situation (also cf. Ford, Fox & Thompson Reference Ford, Fox and Thompson2002). She furthermore states that while this happens continually at any given time, some of these fragments may acquire idiosyncratic meaning when used in appropriate situations and constructionalize over time. That is to say they begin the long path towards schematic constructions along the lines described above, or even move on further to become fully idiomatic expressions insensitive to grammatical operations. Emergence of patterns is thus dependent on the amount of inference available to a speaker–hearer in a significant number of recognizably similar communicative contexts.

This is essentially what I hope to have shown is true of exclamative zu-infinitives in Modern Standard German: Originating as subordinate clauses they have acquired an idiosyncratic meaning component that is easily recognizable, but not attributable to any individual part involved in the construction. This was possible due to sufficiently frequent omission of the matrix expression in dialogic discourse. The free-standing exclamative zu-infinitive therefore seems best described in terms of a schematic construction. Moreover, mechanisms of change such as advocated by Evans (Reference Evans, Nikolaeva and Plank2007), Diessel (Reference Diessel, Bergs and Brinton2012), and Traugott (Reference Traugott2017) also help to understand the emergence of further idiomatized structures such as Nicht zu fassen!: Given sufficient exposure to zu-infinitives in situations where a speaker evaluates a specific circumstance that he or she dislikes and includes it in the utterance, the constructional pattern becomes entrenched. Structural analogy, that is the recognizability of the structural pattern as being associated with speaker indignation, then paves the way for exclamative expressions of non-specific indignation without additionally encoding the circumstance that the exclamation is directed towards, even though the syntactic source structure may be different, such as modal passives. The observation that particular experiencer verbs such as glauben (‘believe’) or fassen (‘grasp’) figure prominently in such generic exclamatives also suggests that the semantic class of the verb also contributes to the direction a constuctionalization process takes, as could be seen in the effect of market transaction verbs versus other verb types in example (5) above.

5. Cross-linguistic comparison

Given that the empirical base for such a claim is not extensive, in order to gauge its likelihood and further support the idea let me now briefly discuss similar observations in other languages.Footnote ⁸ A pattern that strikingly resembles the one discussed in the preceding paragraphs is found in Modern English, which also has not received a great deal of attention in regard to the constructionalization of pragmatic meaning: the independent to think-exclamative.

It can be argued that in this case too a feature such as indignant(sp, prop) has undergone some degree of entrenchment as part of a schematic construction. The examples in (23) show that if this is the case, it is subject to language-specific structural constraints slightly different from those in German. Firstly, whereas the German zu-infinitive construction contains a variable for the predicate, its English counterpart only allows a single verb of cognition, think, which suggests a further advanced stage of conventionalization. Secondly, for an indignant reading a clausal complement encoding the state of affairs that speaker stance is directed towards is obligatory:

As evident in the examples above, phrasal adjuncts are not sufficient for independent use and neither are verbs other than think.

The English to think-exclamative may thus be represented by the simplified constructional schema below:

There is, however, some indication that the apparent restrictiveness is the result of a diachronic process quite similar to the one outlined for the German zu-infinitive. Indignation as an exclusive function of free-standing to-infinitives can be traced back to the fourteenth century (Kerkhof Reference Kerkhof1982). The expressions in (24) from Chaucer exemplify this:

This shows variation as to the verbs that enter into a collocation. The most obvious structural parallel to German is the arguably obligatory preverbal constituent. This pattern, however, has since become unproductive. As can be seen in Modern English, on the other hand, only think is a permissible verb and the state of affairs must be encoded in a compulsory subordinate clause following it.

Visser (Reference Visser1966:1046) notes that while in Middle English variation between bare and to-infinitives is rather free, there is a strong affinity between to-infinitives in particular and preceding interjections such as the ones shown in (24). He further remarks that there are some examples found in Middle English where free infinitives encode indignation, but also related notions such as disbelief and astonishment as well as deprecation, longing, disapproval, and concern. Starting with an apparently rather broad association between independently used infinitives and the expression of emotional involvement or exclamativity, the specific form–function relations found in contemporary English therefore must have come into existence and developed in the course of the Modern English period. Having emerged probably as increments of interactive discourse, this type of expression has become more specific in regard to the range of attitudinal meaning and at the same time more constrained, syntactically as well as lexically.

In Modern Dutch there are various syntactically independent infinitives showing different stages of constructionalization or even idiomatization (cf. IJbema Reference IJbema2001). The following utterances display the range of associated illocutions as well as the degree of optionality with respect to the particle te (‘to’).

Furthermore, there exists a schematic construction with a very specific pairing of form and meaning of speaker indignation:

This strongly resembles the to think-construction in English in form and function. Its lexical specification is different but no less idiosyncratic. The syntactic pattern may be represented as follows: [en dan te bedenken [X_{SUBORDINATE CLAUSE}]]_-sub.

Both English and Dutch show at least similar tendencies in regard to free-standing particle infinitives as part of schematic constructions for the expression of speaker indignation. However, the available structural patterns are formally more constrained in both languages, so that constructionalization is further advanced than in German. Taking into account these cross-linguistic parallels, the question arises as to whether speaker indignation is a category that easily lends itself to processes such as constructionalization or even grammaticalization in some way or another. Wolfram (Reference Wolfram, Kortmann and Schneider2004:218), for instance, notes that the use of come as an auxiliary with the specialized meaning of expressing indignation on the part of the speaker is unique to African American Vernacular English (also cf. Spears Reference Spears1982):

This observation alone supports the idea that speaker indignation as a category of speaker stance meaning constitutes a likely domain for grammaticalization as such. However, the data discussed so far further suggest a particular significance of non-finite verbs for such processes cross-linguistically, which warrants further investigation involving a broader empirical basis, both diachronically as well as typologically (also cf. Al-Bataineh Reference Al-Bataineh2023).

The Latin infinitivus indignantis, which is a colloquial expression, is a case in point. Here the expression of indignation has been identified as one of the main functions of the use of infinitives without subordinating verbs of emotion (Hofmann Reference Hofmann1965:365; also cf. Palmer Reference Palmer1954:75):

According to Hofmann (Reference Hofmann1965:366), all the Latin free infinitives are colloquial reductions that are used affectively where the narrative is hasty. He refers to this characteristic as an “averted grammaticalization” (Hofmann Reference Hofmann1965:366). Moreover, he specifically points out that a matrix verb of emotion, to which such infinitives once had been subordinated, are not thought of by the speaker employing the construction. With respect to another independent infinitive that encodes directive function he further points out that while its origin lies in spoken discourse, at some point the construction occurs in written texts, primarily those closely representing colloquial language. Overall the usage frequency increases over time. These diachronic observations essentially exemplify Evans’s (Reference Evans, Nikolaeva and Plank2007) idea of insubordination outlined in the previous section. Expression of speaker indignation has likewise been identified as a function of a specific exclamative construction in Homeric Greek, as Smyth (Reference Smyth1920:448) points out. Here the infinitive collocates with an accusative second-person subject. Russian too uses a specific infinitive construction systematically to encode speaker indignation (Maurice Reference Maurice1996:81).

The agent in this structure may be marked as either nominative or as dative with no noticeable functional distinction. However, unlike the former the latter cannot be employed as a question. Indignant meaning is also thought to be encoded at least partly in the constructional pattern itself (Maurice Reference Maurice1996:81). Last but not least Chintang (Tibeto-Burman, Kiranti) has matrixless infinitives that productively encode events that happen instead of other events originally intended by the speaker (Schikowski Reference Schikowski2010):

By using this particular structure, the speaker expresses disappointment about the fact that the subject referent is performing an expected action in an obviously unorthodox manner. A structure that combines a mirative marker and an infinitive specifically encodes the speaker’s indignation regarding this circumstance.

These few examples indicate that not only is there a cross-linguistic propensity for speaker indignation to be encoded systematically in schematic constructions, but also that infinitives figure prominently in such structures.

6. Concluding remarks

In this article I have argued for a schematic construction with an attitudinal level of semantic representation characterized by the feature indignant(sp, prop) in order to account for free-standing syntactically independent zu-infinitives and their difference from independent bare infinitives in Modern Standard German in regard to illocutionary potential.

Drawing on somewhat scarce diachronic corpus data, I demonstrated that such a feature emerged in independent zu-infinitives in the decades before 1730, leading to the schematic construction one finds today, which is associated with the specific function of expressing indignant speaker stance. This form–function relation became entrenched early in the New High German period. This has since led to individual even further conventionalized, almost idiomatic constructions through structural analogy, based on the entrenchment and consequently the recognizability of the pattern.

On the basis of similar observations in a selection of other languages, I was able to show that indignation as a feature of speaker stance meaning is likely to be grammaticalized across languages, regardless of genetic affiliation, in often idiosyncratic ways, often involving syntactic disintegration or insubordination. A cross-linguistically recurrent association between speaker indignation and infinitives as part of schematic constructions was attested.

Abbreviations

acc: accusative, conj: conjunctive, dat: dative, def: definite, erg: ergative, excl: exclamative, hon: honorific, inf: infinitive, mir: mirative, neg: negative, part: participle, pl, plural, prop: propositional, refl: reflexive