In this article, it is proposed that different types of apparent “non-V2” arrangements in Present-Day German matrix clauses which are generally treated independently are similar in nature and derivable by means of a limited number of syntactic operations that do not challenge or put into question the classical account of German as a structural V2 language. The analysis reveals that an adequate formalization of all possible left-peripheral word orders must rest upon three basic assumptions: (i) V2 in Modern German main clauses can be neither movement to the head position whose specifier hosts a moved or base-generated XP nor (necessarily) movement to Force°, but can be generalized to raising of the Vfin to Fin°; (ii) German has a Split CP which is fundamentally similar, mutatis mutandis, to that of Romance languages; (iii) this language is subject to the bottleneck effect, which states that all movement into the CP passes through [Spec,FinP]. The theoretical approach pursued here attempts to account for left dislocation and other (frame-setting and non-frame-setting) topicalization phenomena by assuming that in German (differently from other Split-CP languages), XPs base-generated in the middle field move to their surface position by cyclical movement within the left periphery. This allows us to avoid ad hoc explanations, as well as violations of the bottleneck effect.
Present-Day German (PDG) has traditionally been regarded as a strict Verb-Second (V2) language in light of the assumption that its prefield (German Vorfeld), namely the left-peripheral area of the clause preceding the finite verb situated in C°, can only be occupied by one XP in matrix clauses. According to this principle, only one constituent can (and must) move to [Spec,CP] to satisfy an EPP-like feature carried by C that requires that the pre-C° position not be empty. This rule only applies to syntactically independent structures, since subordinate clauses introduced by an overt complementizer generally display a Verb-Final arrangement resulting from movement of the verb base-generated in the low-middle-field position V° to T°, where it surfaces at PF, to acquire the relevant inflectional traits. In this configuration, the access of the inflected verb to the CP head is blocked by the presence of the complementizer in the same position. For this reason, German is defined an “asymmetric V2 language”. The V2-matrix and the Verb-Final-embedded-clause constellations are exemplified in (1) (simplified illustrations):
That the V2-Verb-Final system exemplified in (1) constitutes the unmarked syntactic skeleton of PDG can certainly not be put into question. This pattern has been continuously attested since Old High German (approx. 750–1050 AD) and grammaticalized into the standard option to disambiguate main and embedded clauses from the Middle High German (approx. 1050–1350 AD) period (cf., among many others, Admoni 1990, Axel 2007, 2009, Axel-Tober 2012, 2015, Behaghel 1932, Dittmer 1992, Dittmer and Dittmer 1998, Greule 2000, Lippert 1974, Schlachter 2012, Szczepaniak 2013). Furthermore, this asymmetry does not seem to be in any way subject to change in PDG.
2 Four cases of multiply-filled prefield in PDG
In recent times, however, the descriptive idea of a rigid inviolability of the V2 constraint in main clauses has been challenged by the observation that the area of the clause immediately preceding the finite verb cannot consist of only one position. A wide range of phenomena, attested both in colloquial and Standard German and variously described and formalized in the syntactic literature (for an overview, cf. Müller 2003, 2005 or, more recently, Bunk 2020), seem to involve the activation of a larger portion of structure than the assumption of a single-position prefield would suggest.
Paradoxal as it may seem, V3-effects in which the finite verb seemingly surfaces in a non-second clause position are a general characteristic of Germanic V2-languages, as largely discussed in the literature (among many others, Kotsinas 1998 for Swedish, Quist 2008 for Danish, Freywald et al. 2013, 2015 for Norwegian, Haegeman and Greco 2018 for West Flemish, Sigurðardóttir 2019 for Icelandic, Meelen et al. 2020 for Dutch, Walkden 2017 for a comparative perspective). The notion of V2, indeed, is not to be understood in descriptive terms, but rather as the result of an obligatory syntactic operation moving the verb from a low (VP) position in which it is first-merged to a left-peripheral head (say, C° without committing to a more specific site in the Split-CP) in given clause-typing configurations, e.g., in declarative main clauses. In the present article, the general assumptions related to the V2 status of PDG will not be put into question. Rather, it will be shown that such V3 effects are derivable within a model in which only one constituent is moved into the left periphery of the clause.
In German, such phenomena include – but are not limited to – left dislocations (2) and embedded-clause topicalization (3) with resuming adverbs occurring in the pre-finite area of the CP, so-called post-initial topic particles associated with a fronted XP (4), multiple full XPs with framing function (5). Note that in all these cases, the preverbal elements must be located in a pre-C° (or, in more recent terms, pre-Fin°) projection:
|“I beat the living daylights out of that man”.|
|(J. Ritschel (2015), Barackencarlos, p. 52)|
|“If that were the case, I would not say ‘no’”.|
|(variant with dann: dasgelbeblatt.de, Jun. 23rd, 2015)|
|“This issue, indeed, is very problematic”.|
|(zeit.de, Feb. 9th, 2012)|
|“Yesterday, guardsmen with gats at the airport were shown on TV”.|
|(spiegel.de, Oct. 04th, 2001)|
Given that the elements that appear in the prefield of the examples in (2)–(5) are flagged with different information-structural features, the preverbal sequences in these clauses must correspond to different CP “allocations”. In each of the constructions illustrated above, two elements occupy the left-peripheral space before the inflected verb. Intuitively, it is very implausible that the German prefield may have two positions available, be they two projections or only one projection hosting one element in its specifier and one element in its head, to be randomly filled with whatever head or non-head XP surfaces in the clause. It seems, instead, more accurate to assume that the left periphery of the German clause contains as many projections as are necessary for the expression of the different information-structural categories (simultaneously) occurring in this area, and that these projections are hierarchically ordered, so that all possible configurations can be properly derived and some sequences are automatically excluded. The Split-CP model originally proposed by, inter alia, Rizzi (1997, 2001, 2004) for the syntax of Italian (6) lends itself optimally for the treatment of the above mentioned configurations, since it provides a number of left-edge projections to accommodate material realizing (or signaling the presence of material that instantiates) information-structural categories such as topic and focus:
|(6)||ForceP > TopP* > IntP > TopP* > FocP > Top* > ModP* > TopP* > FinP > (TP…)|
ForceP, the highest position in the left periphery, encodes clause-typing illocutionary force in matrix clauses and licenses the presence in the corresponding clause of any overt or covert features related to the expression of independent speech acts (i.e., the realization of modal particles, assertivity in declarative clauses, etc.). Its activation correlates with a speaker-oriented interpretation of the utterance. In Rizzi’s model, the head position of ForceP hosts the complementizer in embedded clauses introduced by an overt subordinating conjunction like that in declarative direct-object clauses. Four TopicP positions are assumed for a language like Italian in which different kinds of topic (in a left-to-right hierarchy of discourse givenness where high left-peripheral topics are shifting or contrastive, low left-peripheral topics are familiar, cf. Rizzi 2004, and Frascarelli and Hinterhölzl 2007) may surface. IntP is the position in which causal interrogative pronouns such as “why” are base-generated in English, Italian, Romanian and other languages (cf. Shlonsky and Soare 2011 for a slightly different view). In ModP, fronted modifiers (generally, modal adverbs) are accommodated that are non-focused and non-topical. This position is, according to Rizzi (2004), recursive just as TopP (also cf. Abels 2012, 231).
At this point, the question arises as to what extent this model can be applied to PDG and in particular to the phenomena illustrated above. In this respect, different proposals have been made.
2.1 Left dislocation
As for structures like (2), Catasso (2015) and Speyer and Weiß (2018) have recently assumed (in the spirit of Ross 1967, and Cinque 1977) that the left-dislocated constituent, which is generally taken to be a shifting or contrastive topic (cf., e.g., Breindl 2008, 38–9), is moved from its Merge position in the middle field to [Spec,TopP] and that the resumptive pronoun must therefore surface in a lower specifier, arguably [Spec,FinP] (for similar observations on German based on binding effects, cf. Frey 2005a, 2005b). Grewendorf (2002) proposes that this linearization results from movement of a “big DP” consisting of a full nominal expression and a d-pronoun (in (2) above, this complex would be [DP den Kerl [D° den]]) to [Spec,FinP] and successive raising of the left-dislocated constituent to [Spec,TopP], so that the resumptive pronoun is left behind in the specifier of the right-most CP position and the surface word order can be properly derived (also cf. Belletti 2005, Cecchetto 2000 for a view from Romance). In this analysis, the d-pronoun is a trace spell-out, which is witnessed, inter alia, by the fact that it bears the same case as the dislocated topic. Authors like Scheutz (1997), and Frey (2004a) do not share this operationalization. In particular, Frey (2004a) argues that such a derivation encounters a number of problems related, for instance, to the fact that one should assume movement out of an adjunct island to account for the grammaticality of structures like An seinem i Geburtstag, an dem arbeitet wahrscheinlich jeder Linguist i (lit. ‘On hisi birthday, on that works probably each linguisti’). Other authors (e.g., Hirschbühler 1975, Postal 1971, Riemsdijk and Zwarts 1997, Rodman 1974) propose that the left-dislocated constituent is first-merged in the left periphery. For reasons to become apparent below (Section 3.1.1), it is necessary to assume a movement rule for left dislocation in German, irrespective of the technical details of this operation.
2.2 Adverbial-clause preposing
The “big-XP” mechanism assumed to generate a structure like (2) might, mutatis mutandis, also be applied to constructions involving adverbial-clause (in this case, conditional-clause) topicalization with a resumptive element in preverbal position (3). As mentioned above, in German, adjunct clauses appearing in some left-peripheral position at PF can optionally be resumed by three classes of originally adverbial elements displaying different relations to the construction they immediately follow: (i) da (lit. “there”), which, in light of its original temporal-local meaning, reproduces exactly the semantic-deictic reference (sensu lato) of the preposed adverbial clause in terms of the localization of an event in space or time (7a); (ii) dann (lit. “then”), which introduces the apodosis in a material-conditional structure whose protasis is a topicalized if- or V1-clause. These originally adverbial elements, which convey a general “in-this-case”-semantics, basically replicate the conditional value of the fronted clause that they resume. It can, thus, be assumed that dann in (8) is also a trace spell-out like the resumptive pronoun den in (2) and, arguably, the adverb da in (7a); (iii) the originally only modal adverb so, which typically resumes conditional (9a) and concessive (9b) clauses in formal or written registers. Note that in structures like (9b), so cannot be treated as a concessive adverb with the meaning “however” or “nevertheless” responsible for an interpretation of the type “Although X, nevertheless Y”, since the main clause may – and does, in fact, often – contain explicit adverbs like trotzdem (‘anyway’) or adversative/mirative modal particles like doch. In previous historical stages of German and at least up to late New High German, so-resumption of other adverbial-clause types (e.g., causal and temporal clauses) was also possible and fully productive (cf. Catasso 2021):
|“When I was young, my favorite movie was Dirty dancing”.|
|“If I won the lottery, I would stop working”.|
|(variant with dann adapted from: S. Frank (2015), Ärztin ohne Gewissen, p. 47)|
|“If you don’t need the money now, you can save it for later”.|
|(G. Raspel (2015), Schatten über dem Enzianhügel, p. 23)|
|b.||…und wenn auch||die||Wege||im||Tierpark||etwas|
|“…and even if the paths in the zoo were a bit hilly, we nevertheless managed to walk them”.|
|(krusen.care.de, “Fahrt in den Tierpark Herford”)|
At least the optional resumption strategy by means of an originally local (“there”) and that by means of an originally temporal (“then”) adverb are a pervasive phenomenon in the Germanic V2-languages (cf. Catasso 2021, De Clercq and Haegeman 2017, Salvesen 2016). In this article, configurations like (7)–(9) will be treated as the adverbial counterparts of left dislocations involving cyclic movement of the clause via a left-peripheral specifier. Crucially, as will be contended below (Section 3.2), this operationalization does not imply a structural violation of the V2 constraint, i.e., that the verb is accommodated into a position lower than C°.
2.3 Topic markers
The third left-peripheral configuration to be reviewed here consists in the overt marking of topics by means of particles like nämlich (lit. “indeed”, cf. (4)), aber (lit. “however”), nun (lit. “now”), schließlich (lit. “after all”), jedenfalls (lit. “anyway”), übrigens (lit. “apropos”), as in (10). Note that these elements, which all appear to the immediate right of the corresponding topic, are all derived from originally adverbial or conjunctional items, but they perform quite a different function as topic markers (i.e., they are in some cases homophonous to other elements in the same lexicon, but they belong to a different morpho-syntactic class):
|“Peter, Maria and Hans were at the summer party like every year. Karl instead could not be there”.|
|“Maria originally intended to submit her postdoctoral dissertation this year, but her supervisor – predictably – threw a wrench in her plans”.|
Assuming Frascarelli and Hinterhölzl’s (2007) well-known typology of CP topics (cf. (11)), it is easy to observe that such particles systematically correlate with one of the higher left-peripheral topic types (i.e., contrastive or shifting topics), while familiar topics are too little salient in discourse (or display too low a grade of communicative dynamism, to use Bech’s 1998 notion), which is reflected by their lower position in the Split CP, and do not represent good candidates for topic-particle constructions.
|(11)||ForceP > ShiftP > ContrP > FocP > FamP* > FinP|
In (10a), the presence of the particle aber coincides with a contrastive topic in the left periphery of the clause. While Peter, Maria and Hans attended the fête mentioned in the first conjunct of this asyndetic coordination, Karl didn’t. The referent “Karl”, thus, is contrasted to the referents introduced in the first clause (which are presumably to be categorized as aboutness topics) with respect to the same content, namely their going to the summer party. This contrastive interpretation would also be possible if aber did not appear in the second clause, but the presence of this element (in light of the context into which the utterance is embedded) unambiguously forces the reading of “Karl” as a contrastive topic. In (10b), instead, the same particle signals that the DP which immediately precedes it is interpreted as a shifting topic. In this case, indeed, no contrast is introduced between the referent followed by the particle in the second clause and the topic(s) mentioned in the preceding sentence. In particular, the topic introducing the second clause, ihr Mentor (“her supervisor”), is not opposed e.g. to Maria with regard to the same event or circumstances, e.g., the submitting of a thesis, as would instead be the case if the second sentence read Peter aber wird erst nächstes Jahr abgeben (“Peter, instead, plans on submitting [his thesis] next year”). In the second clause, a new (in this case: contextually retrievable, but not familiar) topic is introduced which does not elicit a contrastive reading, and this shift is disambiguated by the presence of aber.
The syntax and information-structural correlates of this construct, in which a topic and a particle surface in the preverbal area of a matrix clause giving rise to a V3-effect, have been considered in a number of works, especially in the last two decades (cf. e.g. Breindl 2008, Breitbarth 2020, Catasso 2015, Métrich and Courdier 1995, Pasch et al. 2003, Speyer and Weiß 2018, Volodina and Weiß 2010). Volodina and Weiß (2010) assume that the particle is a head base-generated in the middle field in a position higher than the XP (12a). The XP moves then to [Spec,PrtP], and the whole complex is eventually raised to the left-peripheral specifier in which it surfaces at PF (12b), counting, thus, as one constituent.
More recently, it has been proposed that topic particles in German are first-merged in the Split-CP, thereby lexicalizing the head of the corresponding TopP (ShiftP or ContrP), to whose specifier the relative topic moves to occur in a Spec-head configuration with the particle and to obtain the correct linearization (cf. Catasso 2015, Speyer and Weiß 2018). Differently from what is implied in Volodina and Weiß’ (2010) proposal, in this derivation, the particle and the topic do not form a complex constituent assembled in some TP/VP position, but the relative linear order of XP and particle applies at PF, so that the visible syntax of the corresponding clauses does instantiate a “violation” of the V2 constraint (as we have seen, V3 is to be seen as a special case of V2). This idea follows from a number of arguments taken to show that a base-generation analysis for topic particles is more advantageous than a movement-based one, i.e., the fact that elements like aber cannot be reconstructed in a lower position with the same function (they always take scope over the whole VP if they appear in the middle field); the interaction between topic particles and complex DPs (nominal expressions modified by a relative clause, a genitive, etc.), which generates a limited number of possible structures occurring in the left periphery of the clause, also shows that a PF derivation is more plausible. The interested reader is referred to Catasso (2015) for the technical details. An additional argument in favor of a left-peripheral base generation of the topic particle is that apparently, such elements can also follow hanging topics in PDG (cf. (13)), which are standardly assumed to be first-merged in a very high position above the CP:
|“Respect for the father’s Herculean strength. As for the daughter, however… Well, tribunals and psychologists will have to decide about her future”.|
|(youtube.com, user’s comment, Sept. 30th, 2017)|
I will adopt this view in what follows. For the sake of convenience, no intermediate steps are represented in the derivation in (14), i.e., only the “big movement” of the XP from the middle field to the relevant specifier is sketched:
As will be shown in the next sections, this assumption reveals itself as particularly convenient when analyzing more complex structures that involve the interaction of a split-DP configuration, a topic particle and left dislocation in the CP of the same clause.
2.4 Multiple frames
The conclusions that can be drawn with respect to the derivation of structures like (5) (repeated in (15) for ease of reference), in which two phrases realizing frame(-like) information surface in preverbal position, are at the same time more simple and more speculative than those presented in the previous sections of this article.
|“Yesterday, guardsmen with gats at the airport were shown on TV”.|
They are, at least in principle, more simple because they result from the direct observation of a pattern that is fully productive in spoken and, to some extent, even written German ((5)/(15), for instance, is from a newspaper article), i.e., the ascertainment of the presence of two different phrases in this sentence is not theory-driven. It is uncontroversial that the adverb gestern and the PP im Fernsehen perspectivize the eventuality described here by making reference to two distinct categories, time and space, respectively, which in turn localize the situation in two different dimensions. These two pieces of information may, indeed, optionally be distributed in a more uniform way in the clause by generating a structure in which the first frame surfaces in the CP and the second frame in the middle field, irrespective of which frame appears in that position (16a)–(16b); alternatively, they can both surface in the middle field (17). Note that a frame is generally considered as such only if the corresponding phrase occurs in clause-initial – or at least in a left-peripheral – position:
This observation, however, is also quite speculative, since there does not seem to be any substantial empirical evidence that speakers interpret or, more importantly, produce the two frames as one complex or two separate pieces of information. From a theoretical point of view, the relevant question is whether gestern and im Fernsehen are both moved to the left periphery of the clause and, if so, whether they are base-generated in the middle field and merged into a complex frame-like constituent before being moved to the CP. In any case, if we take these two constituents to be first-merged in the middle field, we cannot ignore the fact that they can also be linearized discontinuously, i.e., another constituent may intervene between them, as in (18):
These examples suggest that two adjuncts like gestern and im Fernsehen can be moved independently into the left periphery. In the next section, however, it will be shown that there is evidence supporting an analysis in which both constituents appear in the CP in combination with topic particles and that originally autonomous adverbials like these, which have a similar reference in that they both localize one and the same event in the spatio-temporal continuum, can also be merged and interpreted as one “big XP” and raised into the preverbal area as a block. This formalization accounts for the empirical observations made so far and allows us to avoid a structural violation of the V2 constraint.
3 How “split” is the CP of PDG?
3.1 Evidence from contrastive topics and non-clausal frames
3.1.1 Contrastive topics
If the assumptions made above are on the right track, the left periphery of PDG must at least contain a TopP that hosts topics in its specifier and topic particles in its head (19a) and a FinP in which the EPP-like feature of constituents entering the CP is checked (19b). The head of FinP can also be taken to be the position into which the finite verb is moved in main clauses:
|(19)||a.||[TopP die Sache [Top° nämlich [FinP [Fin° ist [TP heikel]]]]].|
|b.||[TopP den Kerl [Top°] [FinP den [Fin° hab’] [TP ich grün und blau gedroschen]]].|
Note that, along these lines, arguing for finite-verb movement into the head of the projection in whose specifier the clause-initial constituent appears (as proposed e.g. in Fanselow 2009, 90, Frascarelli and Hinterhölzl 2007, 113, Frey 2004b, Frey 2005a) would be problematic, since the – inelegant and uneconomical – postulation should be accepted that the verb is raised to a different head position depending on whether e.g. in (19a) Top° is lexically occupied by a topic particle or not. The corresponding clause, indeed, would not mean something different if the particle were deleted. More importantly, the DP die Sache would still – at least potentially – be a topic, i.e., it would be linearized in [Spec,TopP] with or without the particle, as illustrated in (20):
|(20)||a.||[TopP die Sache [Top° nämlich [FinP [Fin° ist i [TP heikel [VP [V° ti] [T° ti]]]]]]].|
|b.||[TopP die Sache [Top° [FinP [Fin° ist i [TP heikel [VP [V° ti] [T° ti]]]]]]].|
Moreover, if Rizzi’s Split-CP is to be conceived of as an information-structurally-oriented system, then the question arises as to why the finite verb should check a [+top] feature in Top° after acquiring its phi-features in lower positions. In fact, V2 – intended as language-internal grammaticalized finite-verb movement to some C head – in main clauses is not sensitive to and does not depend in any way on the type of constituent that occupies the prefield. A theory in which the inflected verb has one and the same position (in this case, FinP, which is clearly related to finiteness) as its landing site in matrix clauses, therefore, appears to be descriptively more adequate.
Also note that if it is true that the single projections of a Split CP are endowed with specific information-structural features, the dedicated position hosting moved wh-phrases base-generated in some VP-internal position must be different from TopP. This assumption must be accepted even if there is no empirical evidence – as is the case for PDG – for sequences of the type “(topic) > whP > (topic)”.
If we further proceed on the premise that frames are elements that appear in a left-peripheral specifier hosting frame-setting topics (as proposed e.g. by Rizzi (1997), and Rizzi (2001) for adjuncts like domani “tomorrow” in Italian, as well as, inter alia, Jacobs 2001 for German), then we may assume, for the moment, that a second TopP is needed to accommodate the lower frame in a structure like (5)/(15) (formalization including the label “TopP” for the surface position of gestern and im Fernsehen adapted from Rizzi 1997, 295–6):
|(21)||[TopP gestern [Top°]] […] [TopP im Fernsehen [Top°]] [FinP [Fin° haben]] [TP sie …]|
These observations are corroborated by further evidence. As Speyer and Weiß (2018) show, left dislocation in German unambiguously corresponds to some kind of topicalization, as in the following example (Speyer and Weiß 2018, 79–80):
|“Peter, indeed, loves the danger”.|
In (22), the DP Peter is at the same time a topic and a left-dislocated phrase. Indeed, the same sentence without either the topic particle nämlich (“indeed”) or the resumptive pronoun der would convey the very same meaning. What is crucial here, however, is the syntactic derivation. Given that it is uncontroversial that der in [Spec,FinP] resumes Peter and that the particle nämlich disambiguates the topic status of the same element by appearing in a Spec-head configuration with it, it must be assumed that the left-peripheral position of Peter is not base-generated, but results from movement of this constituent from some middle-field position to TopP. The pronoun der, indeed, is a trace spell-out of Peter and witnesses its movement through [Spec,FinP]. Intuitively, it is to be excluded that Peter has been first-merged in FinP. A further, hitherto undiscussed piece of evidence in favor of a movement analysis of Peter in a structure like (22) comes from sentences like (23). This example is taken from a forum in which users share their personal experience with traveling with children and talk about the places they visited and the hotels they stayed in. The (pre-)context is given in italics:
|(23)||Die Besitzer waren, wenn ich mich recht entsinne, Bruder und Schwester. Die Schwester (Laure oder Laura) war sehr nett und entgegenkommend. Die haben wir sehr gemocht.|
|(“The owners were brother and sister, if I remember correctly. The sister (Laure or Laura) was really nice and cooperative. We liked her very much”.)|
|“However, we didn’t set eyes on Laura’s brother on the first day…”|
|(rund-ums-baby.de, May 22nd, 2018)|
In this example, which exemplifies a natural construction of conceptually oral (in Koch and Oesterreicher’s 1985 sense) German, the constituent der Bruder von Laura (lit. ‘the brother of Laura’) is discontinuous, and a topic particle intervenes between the DP der Bruder and its PP-modifier von Laura. This DP is clearly a contrastive topic and occupies therefore the specifier position of the higher TopP. As we have said, topic particles may be taken to be base-generated in the head position of the TopP hosting the relevant shifting/contrastive topic in their specifier, i.e., their position is not derived by movement (Catasso 2015). At the same time, it can be excluded here that this nominal expression, whatever the position of aber might be, is a hanging topic, given that it displays the same (non-default) case as the resumptive pronoun den. At this point, we have enough evidence to assume that contrastive topics (or at least contrastive topics like that in (23)) not only must be base-generated in the middle field, but may optionally undergo cyclic movement to their landing site by leaving a modifier in a deeper left-peripheral position which I assume is the lower TopP given the topic status of the whole DP and in the absence of another projection displaying the relevant features (also cf. the discussion in 3.2.3). If this line of reasoning is correct, the following important consequences ensue:
(i) at least this type of topic is not first-merged in the left periphery, but rather in some position below C, and then cyclically raised to its PF site, as schematically illustrated in (24) (for the sake of clarity, the intermediate movement of the topical constituent through [Spec,FinP] is not illustrated).
(ii) a violation of a freezing constraint must be assumed to account for the data presented above. Indeed, we are confronted with the inevitable assumption that, taken the DP den Bruder von Laura, it is possible to extract den Bruder even after the complex nominal has been moved. Note that the only way to avoid this problem would be, in principle, to postulate that the base-generation position of this constituent be the projection in which von Laura surfaces in (23). This position, however, is higher than FinP, which is, in turn, the projection in whose specifier we assume the raised DP to have left a trace lexicalized as a pronoun bearing the same phi-features as the nominal itself. This tentative solution, thus, would not be advantageous. Also note that only the contrastively interpreted part of the DP can be moved further to the higher TopP, while an operation in which den Bruder remains in the lower specifier and von Laura is raised to [Spec,ContrP], which generates a structure like *Von Laura aber den Bruder haben wir nicht gesehen, would lead to ungrammaticality. There does not seem to be any other plausible solution to operationalize the structure discussed above if one does not accept the idea – which would be based on very unsolid grounds – that topic particles are generated in the middle field and undergo, at some point of the derivation, VP/TP-to-CP movement. Although the labeling of the lower specifier in which the topic constituent may optionally leave a modifier must be speculative for the time being, it is advantageous to assume that it is the lower TopP proposed by Rizzi (1997).
Furthermore, note that the structure shown in (23) would also be possible with a range of other modifier types, e.g., relative clauses. In (25), the clause-initial nominal expression den Mann, for which a contrastive reading is assumed, is also followed by a topic particle and resumed by a CP-internal pronoun matching in case, number and gender with the N-head. Sticking to the “big-DP” idea that this constituent has reached its surface position by movement via [Spec,FinP] and assuming that the relative clause is part of the nominal group, one has to conclude that three left-peripheral specifiers are activated here as hosts for the syntactic operation that generates a PF arrangement like (25):
|“The man who received us I found very creepy”.|
This possibility is not limited to PDG. In (26), two examples from New High German (1650–1900) are given in which exactly the same pattern occurs. For the sake of consistency, in both cases the preverbal sequence consists of an inflected form of the pronoun d-jenige (“the one”), the topic particle aber, a restrictive relative clause (the pronoun d-jenige is obligatorily modified by a relative construction), and a resumptive d-pronoun. In (26a), the dislocate diejenigen bears nominative morphology just as the pronoun in [Spec,FinP]; in (26b), the CP-internal position of the topic is disambiguated by the fact that both the clause-initial pronoun and the resumptive in [Spec,FinP] display accusative inflection, which is characterized by an exclusive, non-syncretic n-termination in the singular paradigm of German. The configuration resulting from the insertion of a topic particle after denjenigen, the fact that the DP is case-marked and the presence of a “CP-internally stranded” modifier makes it very implausible that the constituent in first clause position is a hanging topic (but see Section 3.4 for an exceptional case). Crucially, the relative clauses modifying the DP in (26a)–(26b) are restrictive modifiers (which is further supported by the fact that the antecedent diejenigen “those”/“the ones” can only occur in combination with a defining relative clause) and are to be regarded as an extension (and thus as part) of the fronted (pro)nominal projection. For this reason, it can reasonably be excluded that they are parenthetically inserted into the structure:
|“Those who commit sins are enemies of their own lives”.|
|(J. Athias et al. (1712), Biblia pentapla, das ist: die Bücher der Heiligen Schrift des Alten und Neuen Testaments […], p. 233)|
|“The one whom he previously called his God in his wistful lamentation, he calls in his first and last word his father”.|
|(J. J. Rambachs (1751), Betrachtungen über das gantze Leyden Christi […], p. 76)|
At this point, the question arises as to whether – and, if so, how – frame-setting topics interact with particles like aber (in the function addressed here).
3.1.2 Non-clausal frames
As we have seen, deictic adverbs and PPs like gestern (“yesterday”) and im Fernsehen (“on TV”), respectively, may co-occur in the preverbal area, but they can also be raised independently. Now, cf. the example in (27) (the pre-context is given in italics):
|(27)||Es gab Zeiten, da wurden Mitarbeiter noch mit Weihnachts- und Urlaubsgeld, vielleicht einem Betriebskindergarten oder einem Dienstwagen, zumindest aber mit berechenbaren Aufstiegschancen geködert.|
|(“There was a time when employees could still be lured with Christmas bonuses and vacation allowances, perhaps even with a company-run kindergarten or a caboose, in any case at least with predictable opportunities for advancement”.)|
|“But it’s getting harder and harder to judge the attractivity of a company today – because of all the offers such as yoga in the lunchbreak, dog-sitting or ball pits in the lounge”.|
|(süddeutsche.de, Jul. 04th, 2018, “Berufsbild: Prügelknabe”)|
In this example, two XPs with frame-setting function, a temporal adverb and a PP, occur in the preverbal area. The pre-context shows, however, that only for one of these two constituents a contrastive reading is licensed. Heute is clearly opposed to Zeiten, da… in that: (i) the passage it opens up describes a situation that is supposed to be interpreted as diametrically different from the one depicted above; (ii) it is accompanied by the topic particle aber. The second constituent contains, by contrast, information that is presented in a stylistically very evoking way, but is not as salient in this utterance in that it is part of the (assumed) common ground shared by the journalist and the reader of this article.
As for constructions in which a frame-setting topic receives a contrastive reading, however, the structure in (27) is not the only licit configuration. It is, indeed, also possible to have two frames followed by a topic particle, as in (28). The pre-context is given in italics:
|(28)||Aber der eine, der Ich-Erzähler, den es nach Anklam verschlagen hat, ist Lehrer geworden, […], genießt nun seine Pension, […] und betreibt nebenher ein Antiquariat. Der andere mit dem Spitznamen “Euler” ist im Abfallgewerbe tätig, […] und verhandelt gerade mit der Stadt Anklam über eine ökologisch korrekte, topmoderne Deponie.|
|“But he, the I-narrator, who arrived in Anklam, became a teacher and is now enjoying his retirement and running a second-hand bookshop. The other one, who was called “Euler”, is currently working in the garbage industry and negotiating the realization of an eco-friendly, very modern landfill with the city of Anklam”.|
|“Back then in Munich, however, they always sat in the cafeteria of an insurance company”.|
|(tagesspiegel.de, Mar. 7th, 2011, “Muntere Sechziger”)|
This suggests that these two constituents, a temporal and a local deictic expression, respectively, must be operationalized together. For such constructs, I propose that at some point of the derivation, the two XPs, which are first-merged in the lower clausal domain, are moved into the left periphery as a unit, so that the whole complex serves as one “big frame” that refers to the same event and can be treated as a contrastive frame. In fact, the positive and solid state of affairs described in this sentence is clearly opposed – in its entirety – to the more unstable situation addressed in the pre-context. In consideration of the data illustrated in (27) und (28), it is thus plausible that these XPs merge into one complex constituent before being moved to the CP. This follows from the premise that although damals and in München formally refer to different dimensions (time and space), they may be assumed to build a complex unit in delimitating (in Krifka’s (2006) spirit) the domain within which the sitting took place in the depicted situation.
There is another argument that underpins this hypothesis. Even when two frame-setters, either separated by an intervening topic particle or not, surface in the prefield, an optional correlative adverb like da (lit. ‘there’) may resume the complex they build, as in (29b). Expectedly, the same element can also resume a left-peripheral frame in non-multiply-filled-prefield configurations (29a):
|“If I think about it now, it seems funny, but that’s just the way it was at that time”.|
|(faz.net, Jul. 6th, 2018, “Wer was erzählt, ist ein Verräter”)|
|“The European Union walked the right path to a good future back then”.|
|(n-tv.de, Mar. 25th, 2007)|
In (29b), the overt correlative element da, which could, in principle, refer to a temporal (as is the case in (29a)) or to a local frame in PDG, resumes the whole “big frame” damals in Berlin. An interpretation in which it refers only to one of the two constituents is not possible. Therefore, this sequence must be read as describing one frame that is deictically indexed to one and the same temporal-local situation in the past.
What is more, topic particles apparently do not exclude the resumption of a clause-initial frame by means of a resumptive of the da-type in [Spec,FinP], as shown in (30a). In this sentence, damals (“back-then”) is contrastively opposed to heute (“today”) (cf. the pre-context given in italics). Crucially, complex structures like (30b), in which two frame-setting XPs – here damals “back then” and an der Uni Potsdam “at the University of Potsdam” – followed by a topic particle and resumed by da in the specifier of FinP, are also possible. The sequence damals an der Uni Potsdam (“back then at the University of Potsdam”) functions as a contrastive frame-setting topic, as inferable from the context into which this utterance is embedded. Following the assumption that topic particles are located in the head position of the projection whose specifier is occupied by an XP which they linearly follow at PF, which is the projection arguably hosting aboutness/contrastive topics, it must be concluded that both damals and an der Uni Potsdam, two separate XPs at the beginning of this syntactic derivation, surface in the specifier of a left-peripheral projection as one constituent. Given that the complex XP damals an der Uni Potsdam is also resumed by da in FinP, it can be further assumed that this “big frame” is not first-merged in the left periphery of the clause, but has rather been base-generated in the middle field and has moved to the CP via movement through [Spec,FinP], thereby leaving a trace in that position that surfaces as da.
|(30)||a.||Heute kannst du ja Ringelblumensalbe und solche Dinge überall kaufen.|
|“Today, you can buy marigold unguents and similar things everywhere”.|
|“In the past, however, this old woman produced all these unguents and tinctures on her own”.|
|(Eva Bothe et al. (2016), Das wendländische Hexen-Volk. Hexen, Heiden, Heiler und Schamanen im Wendland - gestern und heute, p. 39)|
|b.||Ja, stimmt, jetzt ist alles bzw. man ist nur in Treskowallee. Nicht gerade der schönste Ort in Berlin […].|
|“Yes, it’s true, now all classes are at Campus Treskowallee. Not really the most beautiful place in Berlin”.|
|“Back then at the University of Potsdam, however, we were at Campus Griebnitzsee – beautiful!”|
|(studis-online.de, Jul. 26th, 2018, “Berlin – Campus HTW – Wirtschaftsinformatik”)|
In (30b), both frame-setting pieces of information occur to the left of the contrastive topic particle aber. Apparently, however, another preverbal sequence would also be possible in which the particle intervenes between the temporal and the local XP, as in (31):
|“Back then at the University of Potsdam, however, we were at Campus Griebnitzsee – beautiful!”|
In this sentence, parallel to example (27), only one of the two frame-setting adverbs appears in the specifier position dedicated to aboutness/contrastive topics, which is witnessed by the fact that the particle aber surfaces to the right of the temporal adverb but to the left of the locative PP. Given this configuration, indeed, an der Uni Potsdam is less salient than damals. If we compare the natural prosodic contour of (30b) and (31), we notice that in the first case the two XPs build a continuous, ascending prosodic unit, the pitch contour peaking on the first syllable of Potsdam (32a). The particle marks a sudden fall of the pitch contour; in the latter case, instead, only damals receives a contrastive intonation, while the left-peripheral material including the particle and the locative XP (which are generally separated by a very short phonological pause, just as damals and aber) displays a descending prosodic contour (32b):
Arguably, (30b) and (31) provide optimal continuations for slightly different pre-contexts. In the contexts in which (31) can possibly be uttered, the temporal localization of the event described in the clause is the information that is contrastively opposed to some deictic expression (e.g., “today” or “in the future”) occurring in the preceding utterances, while an der Uni Potsdam need not be part of the contrastive phrase in the higher CP projection. The specific interpretive implications of this construct are left to future research. The crucial point here is that both structures in (30b) and (31) are possible in PDG.
On the basis of the evidence presented here, I will contend that the CP domain contains (at least) two projections that may host frame-setting phrases. Following the current information-structurally-oriented terminology and for the sake of clarity, I will call both projections “FrameP” and assume they may be realized in these different positions of the left periphery of PDG, i.e., that they instantiate a recursive category (as is assumed e.g. by Rizzi and Bocci (2017, 4) for Italian, although they use the more general label “TopP”).
It seems that at least the frame-setting constituents at stake here are base-generated in the middle field, where they are perfectly reconstructable, build a syntactic unit before being raised to the left periphery, enter the CP via the specifier position of FinP, where the “big frame” leaves a trace that may optionally be spelled out as a correlative adverb, and may then surface in either a “compact” (as in (30b)) or in a “split” (as in (31)) configuration which becomes visible e.g. when a topic particle overtly intervenes between the two XPs.
As for the syntactic account of the arrangements in (30b) and (31), I assume this state of affairs not to be very different, mutatis mutandis, from the data illustrated in (23)–(26) above. The only relevant difference to be acknowledged between structures of the type in (23) (DP > topic particle > DP modifier > resumptive pronoun) and structures like (31) (frame > topic particle > frame > resumptive adverb) is that in the latter, the two XPs are less tightly bound to each other in the original computation of the clause. The fact that in (23) the resumptive element is a pronoun and in (31) it is an originally adverbial element is just a detail that correlates with the inherent properties of the fronted XPs. Given these premises, I will assume that the word orders in (30b) and (31) are derived by pretty much the same syntactic process as (23) (Den Bruder aber von Laura, den haben wir am ersten Tag gar nicht zu Gesicht bekommen), namely one in which the “big frame” cyclically moves through a lower CP-internal specifier specifier on its way to the position in which it surfaces (a higher FrameP). It can be postulated that the two frame-setting XPs are assembled in a separate workspace and are then raised to the CP as a unit. After the intermediate movement of this constituent through [Spec,FinP], the big frame raises to the higher FrameP via the lower FrameP, where the non-contrastive part of the XP can be left before the constituent further moves to the higher TopP specifier. Cf. (33):
The movement of the frame XP via the specifier position of the lower FrameP can be assumed to be obligatory in any case (i.e., irrespective of whether a part of this complex XP is stranded in the lower FrameP or not, or whether the frame-setting topic is made up of two XPs or not). The cyclic character of this derivation can possibly be explained in terms of specific features that need to be checked/acquired by the constituent before reaching its landing site. The determination of the nature of these features is left to future research.
Note that this derivation not only accounts for the fact that the temporal and the locative parts of this big frame can appear in a continuous or discontinuous configuration in the examples above, but also provides an explanation for the ungrammaticality – at least in Standard German (but see the discussion in 3.2.3) – of sequences like (34) if we assume an operationalization in which an adjunct and an argument are both moved into the left periphery:
|(int.)||“Yesterday I went to the doctor”.|
|(int.)||“Maria worked in Berlin for two years”.|
Interestingly, in varieties of German other than the standard, sequences like (34a) are fully productive. Kiezdeutsch, an emerging sociolect spoken particularly in some high-migration neighborhoods in Berlin (cf., inter alia, Wiese 2010, Wiese 2012, and Wiese 2013, Wiese et al. 2012, Freywald et al. 2013, 2015, Walkden 2015, Walkden 2017, te Velde 2017), is one of these varieties. In Kiezdeutsch, the sequence “frame > light pronoun > Vfin” is one of the features most prominently addressed in the literature, as illustrated in (35) (Wiese et al. 2012, 114):
|“Yesterday I was on the Kurfürstendamm. (= boulevard in Berlin)”|
In this variety, frames like gestern in (35) are plausibly first-merged in the left periphery (also cf. 3.2.3), so that the pronoun can surface in some low specifier specialized for familiar topics.
Moreover, the account proposed above for Standard German is advantageous in that it allows a similar explanation for two different phenomena – left dislocation (with or without an overt topic particle) and the occurrence of two frame-setting XPs in the CP area – leading to the same configuration, namely a multiply-filled prefield.
3.2 Evidence from preposed adverbial clauses
3.2.1 Genuine clausal frames
As shown in Section 2.2, adverbial clauses surfacing in the left periphery of the clause can also be resumed by a correlative element realizing a trace spell-out in a very low CP-internal position which I assume is the specifier of FinP. For such structures, a similar process can be taken to apply as for the constructions addressed in the previous sections. In sentences like (36a) and (36b), which are similar to the examples discussed in Section 2, the correlative adverb resuming the adverbial clause is not obligatory, but it is always possible:
|“But if I could choose, I would like to live a bit longer, of course”.|
|(variant with so from: M. Laue (2018), Das Schwert der Zentauren, p. 32)|
|“When I was a kid, there were only a couple of children’s programs”.|
|(deutschlandfunk.de, May 16th, 2014, “Alfred Jodocus Kwak ist mein Alter Ego”.)|
This configuration is also possible in cases in which the reference of the preposed clause is interpreted metaphorically. In (37a) and (37b), a conditional antecedent is resumed by originally local (and – as a derived form – temporal) da by virtue of the fact that the fronted constituent receives a “within-this-frame”-reading. After all, such adverbials refer to an event or state of affairs believed to be or to have been true in the actual or in a close-by possible world that constitutes the set of (deictic or conditional) coordinates in which the content described in the main clause is to be interpreted. Furthermore, there does not seem to be any apparent reason to assume that the frames in (37a) and (37b) on the one hand and (38a) and (38b) on the other hand, which are, respectively, clausal and non-clausal, should differ in any way affecting the semantic interpretation of the utterance if they refer to one and the same content:
|“If that happened to me, I would definitely call Hans”.|
|“In that case, I would definitely call Hans”.|
|“When Hans’ novel was published, I was living in Oslo”.|
|“In 2015/at that time, I was still living in Oslo”.|
In (37a), the wenn-clause expresses a condition which is necessary for the conclusion stated in the apodosis to follow. The same is true of in diesem Fall, which realizes the same content, but by means of a non-clausal construct. In (38a), the temporal clause localizes the event described in the matrix, and so do the adjuncts 2015 and damals in (38b). Preposed adverbial clauses – at least those with a non-clausal counterpart – should therefore be treated as (clausal) frames. Given that these structures can be resumed by a correlative element, it can be argued that they are moved to the CP and that the resumptive is a trace spell-out of their movement through the specifier of FinP.
In most cases, the raising of an adverbial clause to the left periphery does not allow any other XP to enter this area. Therefore, there are basically two possible configurations in main clauses involving adverbial-clause topicalization:
|(39)||a.||adverbial clause > Vfin|
|b.||adverbial clause > resumptive element > Vfin|
I assume (39a) and (39b) to be two instantiations of exactly the same structure, (39b) lexicalizing the movement of the embedded clause through [Spec,FinP] overtly. As for (39a), it can be assumed that the adverbial clause has left a trace in the same position, but this trace is not spelled out phonetically. Given that preposed adverbial clauses basically do the same job as non-clausal frames, it may be assumed that they occupy the higher FrameP specifier. Note that at least in spoken German, some adverbial clauses surfacing in clause-initial position can be interpreted contrastively – just as non-clausal frames –, and this reading can be disambiguated by the presence of an overt topic particle that may exclusively appear to the right of the clause and to the left of an optional correlative adverb. In the following example, in which a temporal clause is followed, in linearization terms, by the topic particle aber and the correlative adverb da, a pre-context is given to make the contrastive interpretation immediately retrievable:
|“Most children have an easy life today. When I was little, instead, everything was different”.|
In the run-of-the-mill case, thus, it seems that adverbial-clause preposing can be operationalized as in (41). Movement via [Spec,FinP] applies in any case, but the trace need not be pronounced. It goes without saying that, when the clause is followed by a topic particle in the left periphery, this particle is base-generated in that position (Frame°) just like in any other case:
As will be shown in what follows, however, it is not the case that all instances of adverbial clauses surfacing in a correlative structure in the left periphery of German may be derived via movement from the middle field.
3.2.2 Irrelevance, presupposed and biscuit conditionals
Irrelevance-conditional ob…oder-clauses of the type in (42a), in which the preposed adverbial structure is reprised by the adverbial resumptive so, do not seem to be compatible with a middle-field base generation, given that they cannot appear in that position (42b) and even lead to ungrammaticality if forced into a standard V2 configuration (42c), modeled on example (14) in D’Avis (2004, 147–8):
|“Whether it rains or snows, we’re going for a walk anyway”.|
|b.||*Wir gehen, ob es regnet oder schneit, doch spazieren.|
|c.||*Ob es regnet oder schneit, gehen wir doch spazieren.|
Resting on the cartographic assumption that left-peripheral adverbials may occupy different structural positions according to their interpretive features (cf. e.g. Munaro 2005, Munaro 2010), I assume (at least a subclass of) such conditionals to be positioned in the outer left periphery of the clause, i.e., above ForceP, at PF, but to be first-merged prefield-internally at LF, leaving a trace spell-out in the [Spec,FrameP] in which the clause is base-generated, as in (43). The fact that ob…oder-clauses may sometimes be resumed by a CP-internal so qualifies them as weakly integrated (but not fully disintegrated) constituents:
ob…oder-clause [ForceP [TopP[ + Frame]
The existence of weakly integrated adverbials becomes apparent if we consider such types of non-integrated (fully presupposed) pseudo-conditional clauses as (44), which are compatible – for independent reasons – with adverbial resumption (the correlative element generally being dann) and the presence of a prefield-internal wh-interrogative:
|“If you are really as good (as you say), then why didn’t you make it?”|
In (44), the resumptive adverbial dann can be spelled out to the left or to the right of the specifier in which the wh-interrogative warum surfaces, which I assume is [Spec,FocP] following Rizzi (1997). The projections immediately above and immediately below this specifier, respectively, are those dedicated to topical constituents, of which frame-setting (sensu lato) adverbial clauses are an instantiation. For this construction, in which the presupposed pseudo-conditional does not verbalize a premise for the interpretation of the utterance’s content, it can therefore be assumed that the wenn-clause is first-merged in one of these two specifiers and then obligatorily moved into a specifier higher than ForceP for interpretive reasons, as in (45). I will here abstract away from the relative position of the two types of “conditional” illustrated in (42) and (44), which arguably occupy two distinct sites above ForceP since they receive different interpretations, by stipulating that they both occupy a prefield-external position at PF:
This operationalization allows us to account both for the resumption of such “conditionals” in the left periphery – although it is all about structures interpreted differently from standard frame-setters – and the optional variability of the surface position of adverbial dann in (45). A further possible analysis, which also points to a special status of wenn-clauses like that in (44), would have the adverbial structure base-generated in a lower position and then “invisibly” moved into the left periphery. In other words, the structure would be insensitive to this special type of adverbial clauses by virtue of their semantics. This would account for the undeniable V3 configuration resulting from the dann-resumption illustrated above, but also the fact that some speakers accept (as a more marginal possibility) a placement of the resumptive in the middle field even when the semantics of the (pseudo-)conditional clause is fully presupposed (as in Wenn du so gut bist, warum hast du es dann nicht geschafft?). In any case and irrespective of the specific take on its syntacticization, it seems sensible to assume that the special interpretive features of this construction must correspond to a different derivation from that of standard conditionals – and at the same time to make sense of the optional presence of a resumptive.
Another piece of evidence in favor of a differentiation between correlative structures like (40) on the one hand and constructions like (42a) on the other hand comes from a third class of adverbial clauses, so-called “biscuit conditionals” (a term derived from Austin’s (1956) famous example There are biscuits on the sideboard if you want them). These clauses look like conditional clauses in that they are introduced by the standard subordinating conjunction meaning “if” (German wenn), but they do not verbalize a necessary condition for the interpretation of the situation described in the main clause. Rather, they behave like syntactically unintegrated (i.e., clause-external) adverbials that introduce a possible world in which a certain situation α is true, while in the main clause it is asserted that another situation β holds in the actual world, whereby β might be relevant if α were the case (Ebert et al. 2009, 276–7). In such cases, so-resumption is illicit in German, which indicates that the wenn-structure is presumably not merged in the matrix clause:
|b.||*Wenn du Durst hast, ist Bier im Kühlschrank.|
|c.||*Wenn du Durst hast, dann/so ist Bier im Kühlschrank.|
A comparison of examples (40), (42)–(44) and (46) shows that: (i) irrelevance and presuppositional on the one hand and biscuit conditionals on the other hand are not “equally disintegrated”, and; (ii) the corresponding structures imply a continuum of integration into the main clause. For constructions like (46), I will therefore contend (differently than for (42)) that the wenn-clause is base-generated clause-externally, as is usually assumed for hanging topics:
|(47)||[XP wenn du Durst hast, [ForceP Bier ist im Kühlschrank]].|
3.2.3 Frames are moved constituents in V2 configurations
As correctly pointed out by an anonymous reviewer, in the recent literature (e.g., in Hinterhölzl 2017, 212), it has been proposed that clausal and non-clausal frames must be assumed to be base-generated in their surface C-position in German on account of the apparent lack of Principle-C effects (48a) and pronoun binding (48b) in cases of preposed adverbials (adopting Frey’s 2005b take on left dislocation):
|“When Peter arrived home, he called his girlfriend”.|
|“When he called Maria, every student had already finished his exam”.|
Hinterhölzl (2017, 211) goes on to propose that all clause-internal full XPs surfacing above FocP, including framing, aboutness and contrastive topics in Frascarelli and Hinterhölzl’s (2007) spirit, are first-merged in their spell-out position. If this were the case, the assumptions made above about the cyclic derivation of complex frames would have to be ruled out. With respect to this objection, a premise must be made before briefly turning to the technical details of this intricate question: the ungrammaticality judgments referred to the examples in (34) (e.g., Gestern ich bin zum Arzt gegangen) pertain to standard PDG. In fact, structures like these do marginally occur in spoken interaction (cf. Bunk 2020, Breitbarth (to appear)), although they are less prominent than in Kiezdeutsch in the speakers’ representation of their own I-language. Cf., e.g., the following corpus attestation (adapted from Bunk 2020, 17):
|“Speech comprehension results from interactivity in the brain”.|
I find that Hinterhölzl’s proposal is very appealing (or better: the only possible option) if applied to V3 structures like (49), as well as to Kiezdeutsch and to all violations of the V2 constraint in which a genuine frame-setting constituent occurs in utterance-initial position. That this analysis in on the right track is corroborated by recent empirical work by Breitbarth (to appear) in which it is shown that the corresponding XPs are prosodically non-integrated in V3 arrangements. Also, I do not exclude that some frames may be base-generated in their spell-out position even in non-V3 clauses in German. However, I am not convinced that this analysis can be generalized to all frames. A number of facts appear to militate against a formalization in which all topics above FocP are first-merged in their surface position (for space reasons, I will just mention some crucial points here):
some local expressions that may function as frames, e.g., dort drin/da drin “in there” can optionally be split so that the actual referential element dort/da can appear either in the middle field together with drin (50a) or in a left-peripheral specifier (i.e., in a discontinuous configuration I which it forms a semantic unit with drin) (50b). Expectedly, the whole expression can be moved into the CP in a configuration in which the referential d-element arguably pied-pipes drin into that position (50c). Therefore, it is compelling to assume that such lexical elements, which may be frames given the relevant context, are base-generated in the middle field and partially or entirely moved into the left periphery:
(50) a. Sonst war nichts [dort / da drin]. otherwise be-pst.3sg nothing there there inside b. [Dort / da]i war nichts [[ti] drin]. there there be-pst.3sg nothing inside c. [Dort drin]i / [da drin]i war nichts ti. there inside there inside be-pst.3sg nothing “There was nothing (else) in there”.
In some (especially western and northern) varieties of German, pronominal adverbs, complex units typically consisting of referential da (lit. “there”) and a preposition, can optionally occur in split configurations similar to that illustrated in (50). These expressions, irrespective of their syntactic realization, may have, for instance, a contrastive interpretation. In the example in (51), in which the pre-context is provided for the reader’s convenience, the pronominal adverb dafür (lit. “for that”, in this context “in favor of that”) has the contrastive da-component, which resumes the predicate verbalized in the preceding sentence, surfacing in the left periphery. The other part of the word, für, instead, has remained in the middle field. The reading is disambiguated by the context (note that the alternative dagegen, lit. “against that”, is introduced in the following sentence): the speaker is in favor of bringing democracy to other countries, but against doing that forcibly. Given that in this sentence, da and für form a discontinuous constituent and that the “whole” form dafür could, in principle, appear as a unit either in the middle field (as is the case for dagegen in the following sentence) or in a left-peripheral specifier, it is quite clear that the lexical form dafür must have been base-generated in the IP/VP area. At the same time, it must be assumed, considering the interpretation of the utterance, that da in (51) must be placed in a specifier hosting contrastive elements (Spec,C-Topic in Hinterhölzl 2017, 211). Therefore, non-movement does not seem to be a conditio sine qua non for the individuation of a contrastive topic. In fact, this also accounts for the data discussed in 3.1, which include discontinuous CP constituents realizing contrastive topics.
(51) context Es ist eine ganz wunderbare Sache, allen Ländern … expl be-prs.3sg a-nom.sg very great-nom.sg thing all-dat.pl country-dat.pl Demokratie und Freiheit zu bringen. democracy and freedom to bring-inf “It is a great thing to bring democracy and freedom to all countries”. clause [Da]i bin ich [[t]ifür]! Aber ich bin there be-prs.1sg I-nom.sg for but I-nom.sg be-prs.1sg dagegen, das mit Waffengewalt zu tun. against-that that-acc.sg with armed-force to do-inf “I am in favor of that. But I am opposed to doing that by force of arms”. (derwesten.de, Sept. 9th, 2009)
So-called Satzverschränkungen (lit. “clause entanglements”) are marked syntactic constructions typical of spoken language involving fronting of a constituent base-generated (and interpreted) in a lower object clause into the prefield of the matrix clause (for an overview of the possible pragmatic implications of this construction, see e.g. Andersson 1981, Kvam 1983, Lühr 1988, Schwitalla 2003). This XP can be, e.g., a wh-element (52a) (Kvam 1983, 81) or a PP (52b) (Kvam 1983, 138), the latter coming to function as a (contrastive) frame here. Such structures are generally defined as cases of long-distance dependencies. In fact, it would be implausible for the constituent in first clause position in (52b) to be first-merged in that position, since it does not even appear within the clausal domain to which it refers. This also seems to suggest that contrastive/frame-setting constituents are not necessarily base-generated in some CP specifier.
(52) a. [Wen]i willst du, dass ich [ti] schicke? who-acc.sg want-prs.2sg you-nom.sg that I-nom.sg send-prs.1sg “Who do you want me to send in?” b. [In Köln]i weiß ich, dass es [ti] eines gibt. in Cologne know-prs.1sg I-nom.sg that expl one-acc.sg give-prs.3sg “I know that there is one in Cologne”.
With respect to the binding issues correctly raised by Hinterhölzl (2017), there is independent evidence that seems to point to the fact that these effects – although observable in contexts like (48a)–(48b) – do not incontrovertibly demonstrate that (clausal and/or non-clausal) frames are base-generated in the left periphery. In some cases, a personal pronoun occurring in a fronted clause with frame-setting function can be bound to a quantified expression in the matrix clause (53). This suggests that the ungrammaticality of (48b) above might have to be at least in part motivated by other factors. Moreover, binding between a pronominal anaphor in a preposed frame-setting PP and a quantified expression in the main clause seems to be licit at least in some cases (54). As this example shows, this phenomenon, which is referred to as “backward binding” in the literature, is not limited to psych-verbs (cf. Temme and Verhoeven 2017 for German):
|“Every Sevillian would love to live in the neighborhood in which they were born”.|
|(C. Schwab (2013), Texturen einer Stadt: Kulturwissenschaftliche Lektüren von Sevilla, p. 159)|
|“Every officer must present themselves at the commander’s office upon arrival”.|
|(K. B. Beaton 2001, A Practical Dictionary of German Usage, p. 38)|
In sum, I take the combination of these facts to indicate that frames are not (necessarily) first-merged into the left periphery in V2 clauses, but that this assumption is particularly compelling when applied to genuine V3 patterns of the “Kiezdeutsch type”.
In the next section, a configuration is addressed which apparently poses a problem for any generative analysis of surface non-V3 in PDG.
3.3 A more complex case: the sequence “DP > adverbial clause”
So far, cases of multiple filling of the prefield have been discussed that may be analyzed in compliance with the bottleneck effect (Cardinaletti 2010, Haegeman 1996, Hsu 2017, Roberts 2004, for an account of the bottleneck effect in German in terms of phase conditions, see Hinterhölzl 2017). This constraint claims that in V2 languages, while all projections in the Split CP are potentially activatable, only one constituent can be raised to the left periphery moving through [Spec,FinP] and making the movement of any other XP through that specifier – and, more in general, to the left periphery – impossible: (i) left-dislocated XPs have been argued to move to the higher [Spec,TopP] from the middle field; the fact that they may cyclically reach their surface position through the lower TopP does not violate the bottleneck effect; (ii) topic particles can be assumed to be base-generated in the CP area, and their presence in Top° does not result from movement; (iii) both frame-setting topics (of the clausal, as well as of the non-clausal type) and non-frame-setting topics (which include, inter alia, left-dislocated topics) are moved to their landing site. In these configurations, the moved element may (or must) leave an overt trace in [Spec,FinP], thereby giving rise to (linear, but crucially, not structural) “non-V2” word orders.
There is, however, at least one case in which the claims made above are problematic, namely the structure exemplified in (55), which has been mostly neglected in the literature so far:
|“Also my brother (once) said he had played with an old man when he was a kid”.|
|(mamiweb.de, May 13th, 2009, “Voll unheimlich”)|
In this sentence, two constituents occupy the preverbal area: the DP subject and a preposed temporal adverbial clause. At this point, one has to admit that the prefield of the clause in (55), which is a productive construction both in spoken and written language, contains two XPs. This structure has already been discussed in Catasso (2015, 350–1), where it is assumed that both phrases are moved to the left periphery, but the bottleneck effect is not considered. Different pieces of evidence support the view that mein Bruder and als er klein war occupy a CP-internal position, i.e., that neither the DP nor the CP appear in some specifier above ForceP. In the first place, mein Bruder is the subject of the clause, thus it cannot be taken to be an extra-sentential hanging topic, since it is not resumed by or co-indexed with a pronoun below C°. In fact, the DP in (55) performs the very same (syntactic) function that it would perform in the absence of the temporal clause. If this is true, then the embedded clause must necessarily be situated in a CP-internal position. Furthermore, the subject must be an aboutness or a contrastive topic in this construction, which can be disambiguated by adding a topic particle in the position immediately to the right of the corresponding constituent. This topic can be realized by a full DP or by a pronoun, but it can in no case be a familiar topic:
|“The parents instead only said he was too lazy […] when they talked to me”.|
|(nannynanny.blog, Oct. 23rd, 2015, “Enuresis…”)|
|“They instead only said he was too lazy […] when they talked to me”.|
At this point, the question arises as to the status of the adverbial clause. A possibility to consider would be to treat this constituent as a parenthetical structure, i.e., as a complex clausal constituent inserted at PF or base-generated in its surface position that does not affect the syntax of the sentence. If that were the case, we would envisage the element in first clause position to behave as any other topic (e.g., with respect to the possibility to be resumed by a correlative pronoun in [Spec,FinP], i.e., to be a left-dislocated topic), irrespective of the presence or absence of the adverbial clause. If the adverbial clause, instead, were to be analyzed as a constituent moved there from a lower domain of the clause, then we would have an undesirable violation of the bottleneck effect with which we would have to cope. In fact, the first, and not the second option seems to be given, at least in spoken German: the aboutness/contrastive topic in first position can also be a left-dislocated constituent co-indexed with a trace spell-out in FinP. In (57a), the topic die Maria bears nominative morphology and is resumed by the nominative pronoun die in the FinP specifier. In order to be sure that the DP in (57a) is not a hanging topic, an even more convincing example is given in (57b), where the initial XP is an accusative object resumed by a pronoun in the same case. Note that the grammaticality of the sentences in (57) provides two pieces of evidence in favor of a “regular” movement-based analysis of die Maria and den Hans: apart from the above-mentioned fact that the topic is taken up by a pronoun, at least (57b) provides unambiguous evidence that this constituent must have been assigned case in a lower (middle-field) position and then moved to the CP. The (only speculative) assumption that a topic in the accusative case resumed by a pronoun in FinP is base-generated in the left periphery would undoubtedly bring up more problems than it would solve. I will exclude this possibility.
|“Maria completely went nuts when she found out”.|
|“We beat the living daylights out of Hans when we found out”.|
Both in (57a) and (57b), the DP and the adverbial clause cannot be reasonably assumed to form a constituent in the middle field that is raised as a complex to the higher CP area via [Spec,FinP], since they have no relevant feature in common, and the adverbial clause is not a modifier of the DP. This implies that a “big-XP” analysis cannot be taken into consideration to account for this arrangement. At the same time, the (at least at first sight attractive) assumption must be dispensed with that the two constituents are moved independently of each other to the left periphery, since this would imply a violation of the bottleneck effect.
What is more, the adverbial clause in constructions like those in (57) can never be taken up by an adverbial resumptive in [Spec,FinP], which is, however, not the case in all other configurations in which an adverbial clause (or any other type of frame setter) with deictic reference surfaces in the CP. Consider (58), a variant of (57b) in which the pronoun has been replaced by a deictic correlative adverb co-indexed with the temporal clause. The result of this substitution is completely ungrammatical:
|“We beat the living daylights out of Hans when we found out”.|
The corresponding structure in which den Hans is not fronted (Als wir das erfahren haben, da haben wir den Hans grün und blau gedroschen) is, expectedly, perfectly grammatical. Thus, even if it is evident that two constituents with different functions surface in the left periphery of structures like those in (57), the assumption must be made – on the basis of the data presented above – that one of them has a parenthetical or base-generated nature (Altmann 1981, Müller 2003) in order to preserve the force of the bottleneck-effect approach.
A crucial piece of evidence for a formal categorization of the XPs surfacing in such configurations like those in (57) comes from the fact that while the nominal expression can always be resumed by a d-pronoun in [Spec,FinP], if [Spec,FinP] is occupied by an adverbial element resuming the adverbial clause as in (58), the grammaticality of the structure can be saved by inserting a resumptive (of any kind: a d- or a personal pronoun, or an epithet) into the middle field (also cf. fn. 23):
|“When Hans left his apartment, he/that idiot forgot to turn off the tap”.|
This suggests that in cases like (59), in which it is the adverbial clause in second position that is resumed in [Spec,FinP], the leftmost DP (here, der Hans) is arguably a hanging topic and not a left-dislocated topic. As a consequence, der Hans is base-generated in some clause-external position above ForceP, i.e., in a projection that does not influence the linear syntax of the clause. As we will see in the next section, this observation can also be applied to some exceptional cases in which the leftmost DP is case-marked.
As observed e.g. by Lühr (1985), Müller (2003, 33–4), and Speyer (2008, 456), the same linear position in which an adjunct clause may appear in the sentences discussed in this section can be occupied by non-clausal adverbials such as PPs. Cf. (60), adapted from Müller (2003, 33):
|“Some derivations in -e, besides those in -ung …, have developed into resultative or concrete nouns are are idiomatized”.|
In this sentence, the PP adjunct neben denen … Niederlegung occurs in second position to the right of the subject and immediately precedes the verb in Fin°. The adjunct clauses addressed above and this type of adverbials seem to only differ with respect to the fact that the former are fully-fledged clauses, while the latter are non-clausal. Syntactically, they do exactly the same job, and they have the same general information-structural function (i.e., they are frame-setters). Thus, if a base-generation analysis like the one proposed here for adverbial clauses in second position is on the right track, then it should also be applicable to non-clausal adverbials in second position.
Based on these facts and given the discussion about complex frames in 3.1.2 and in 3.2.2, I propose that the position in which clausal and non-clausal adverbials are base-generated and appear in the pattern addressed in this section is the lower of the two left-peripheral FrameP specifiers assumed above. Base-generation (or parenthetic insertion) into this position does not interfere with the movement of an XP to its landing site via [Spec,FinP] and accounts for three crucial facts: (i) such adverbials may appear between a wh-interrogative and the finite verb; (ii) they cannot be taken up by a resumptive in [Spec,FinP]; (iii) the structure can be even more complex and include a contrastive topic accompanied by a topic particle, an adverbial clause and a pronominal resumptive taking up the reference of the topic, as in (61) (modeled on (57a)):
|“Maria, however, completely went nuts when she found out”.|
|b.||[ForceP [FrameP [TopP Die Maria [Top° aber] [FrameP als-clause [FinP die [Fin° ist] [TP …]]]]]]|
Note that the operationalization assumed here does not clash with the assumptions related to V2, since in structures like (61), which may be assumed to at least approach the spatial limits of the German left periphery, only one constituent has been moved into the CP domain. The presence of base-generated material, as well as trace spell-outs does not jeopardize the validity of this constraint.
3.4 The strange case of “case-marked hanging topics”
A subset of data that has not directly been addressed so far concerns cases like the following, in which a case-marked DP in first clause position is followed e.g. by a wh-interrogative in the left periphery of the clause and taken up by a resumptive element in some lower (middle-field) position:
|(int.)||“When did you see Hans?”|
|(int.)||“Who has invited Hans?”|
Such (statistically infrequent) constructions, originally discussed by Scherpenisse (1983), have been traditionally considered as cases of left dislocation in the formal literature (cf. Boeckx and Grohmann 2005, Ott 2012). In both (62a) and (62b), den Hans exhibits accusative morphology and surfaces to the left of a wh-element (wann “when” and wer “who”, respectively). Given that the middle-field resumptive element, whose phi-features match those of the resumptive (middle-field den is also in the accusative case), is a d-pronoun, one may prima facie suppose that such constructions instantiate a left dislocation of the type illustrated in the previous sections of this article, especially under Frey’s (2004a) assumption that the resumptive of a left-dislocated phrase need not be located in [Spec,FinP]. However, they are in fact more akin to hanging topics than to left-dislocated constituents (cf. Samo 2019, 203). If we take a look at the distributional properties of the construction in (62), we notice that: (i) the leftmost XP is compatible with non-d-resumptives such as epithets and personal pronouns (63a), which is not the case in left dislocation; (ii) the co-referential (pronominal or DP) phrase in the middle field is always obligatory (63b). This possibly leads us to the conclusion that in cases in which a middle-field pronoun resumes a left-peripheral DP, the structure is not necessarily to be classified as a left dislocation, but is rather ambiguous – if the proper conditions apply – between a left dislocation in (Frey’s 2004a) sense and a (case-marked) hanging topicalization. Indeed, when a left-dislocated DP (i.e., an aboutness or contrastive topic) is resumed by a d-element surfacing in in [Spec,FinP], the latter is always optional without any relevant interpretive consequences (which is corroborated by the fact that left dislocation is limited to spoken language) (64):
|“Why did you kiss Hans/that idiot?”|
|b.||*Den Hans, warum hast du geküsst?|
|“I kissed Hans passionately”.|
|b.||*Den Hans, ihn habe ich mit Leidenschaft geküsst.|
For these reasons, I will assume Samo’s (2019) recent proposal that the structures observed above (in which a case-marked DP interacts with a wh-element in the left periphery and is resumed in the middle field) are, in fact, a subclass of hanging topicalization. These types of “case-marked” hanging topics differ from the standard cases in that they bear non-default case (i.e., they are not in the nominative case). In Samo’s approach, this is accounted for by assuming that this special type of hanging topicalization involves a (case-assigning) verb that is deleted at PF. Accordingly, the complex formed by the DP hanging topic and the verb is merged in a position higher than ForceP that is labeled “Hanging Topic” in Samo (2019, 211). The derivation of this construction is given in (65):
|(65)||a.||[Hanging Topic Den Hans
Samo’s (2019) proposal is convincing in that it accounts for two crucial facts: (i) that in such configurations, the co-indexed element in the middle field is not necessarily a d-pronoun, but can also be realized by a personal pronoun or an epithet (63a), and; (ii) that the resumptive element is not obligatory. What is more, it permits to circumvent a violation of the bottleneck effect. Indeed, if cases like (62) and (63a) were to be explained in terms of a “standard” left dislocation, one would have to postulate that it is occasionally possible to violate this constraint (i.e., to let two different XPs enter the CP domain without both passing through [Spec,FinP]). Such ad hoc violations, however, would be difficult if not impossible to operationalize in other terms. Alternatively, a bottleneck-effect-based non-hanging-topic derivation would imply that in a sentence like (62a), den Hans should be base-generated in a clause-internal left-peripheral specifier in which it may not receive case from the lower verb, thereby leaving unsolved the question of why this DP exhibits accusative morphology. Additionally, there is no evidence that the leftmost DP in such constructions receives a left-dislocation interpretation: proceeding from the assumption that left dislocation in PDG is generally licensed by an aboutness/contrastive-topic interpretation of the DP (as is the case e.g. in (64a)), in a sentence like (62a) this is not the case. On the contrary, den Hans is better interpreted as a clause-external element just as its default-case-marked counterpart in a structure like Der Hans, wann hast du den gesehen?. Of course, the middle-field element resumes only the hanging topic and not the complex “hanging topic + verb” in a derivation like (65). It may therefore be assumed that the resumption of a hanging topic applies post-syntactically.
Note that if we adopt this view, the structural parallelism between left dislocation and adverbial-clause resumption in PDG discussed in the previous sections of this article becomes even more apparent: in sentences like (66a)–(66b), resumption is absolutely optional, whereas in hanging topicalization, an overt element must appear in some position below the resumed constituent:
|“When I was little, everything was different”.|
|“I can’t stand Peter!”|
Note that what has been observed with respect to (58) above, in which the preposed DP is in the nominative case, can also be applied to cases in which the DP preceding a left-peripheral adverbial clause in preverbal position exhibits non-default-case morphology. In fact, the grammaticality of the sentence in (67) can be saved by resuming the accusative DP in the middle field e.g. by means of a d- or personal pronoun. This corroborates the idea that the nominal expression in first clause position is a hanging (and not a left-dislocated) topic despite its non-default case marking:
|“We beat the living daylights out of Hans when we found out”.|
Considering the remarkable amount of structural ambiguity that the differentiation between left dislocation and hanging topicalization implies, I thus propose that whenever the overt resumption of the fronted XP (e.g., DP or adverbial clause) e.g. by a d-pronoun is possible but optional, the derivation of the structure implies movement of the constituent from the middle field to the surface position via [Spec,FinP]. In all other configurations (to be assessed case by case), the construction is more akin to what is generally labeled “hanging topicalization”, in which the XP is base-generated above ForceP. This accounts for the grammaticality of sentences like (63a), since the hanging topic is first-merged in the position in which it surfaces and hence does not interfere with the syntactic derivation of the clause. Additionally, it has been shown that in the linear sequence “DP > adverbial clause”, the DP can always (optionally) be resumed by a pronoun in [Spec,FinP] and the adverbial clause need not have a co-indexed adverbial counterpart in the TP/VP, and in such cases the clause-initial DP is therefore to be classified as a left-dislocated topic, while the adverbial clause has parenthetical status; if, instead, it is the adverbial clause that is resumed in [Spec,FinP], then the DP preceding it is obligatorily resumed by a co-referential phrase in the middle field. In this case, the DP – which can exhibit default- or non-default-case morphology – is a hanging topic.
In this article, a unifying theoretical proposal was made to account for different types of constructions involving multiple filling of the prefield in PDG.
The main premises were that: (i) irrespective of whether the surface position of the elements involved in such structures results from movement to or base generation in the CP area, more than one item can be lexicalized in the left periphery of this language, which does not at all put into question the validity of the V2 parameter (intended as obligatory movement of the finite verb to Fin° in matrix clauses); (ii) Modern German, as a V2 language, is subject to the bottleneck effect, which claims that all fronting into the left periphery passes through [Spec,FinP] and leaves a trace in that position which blocks further movement into the CP; (iii) correlative pronouns and adverbs surfacing in [Spec,FinP], which are always optionally pronounced, can be taken to be trace spell-outs of XP movement into the left periphery.
The gist of the analysis proposed here is that non-familiar and frame-setting topics move cyclically to their canonical surface position, namely a high left-peripheral TopP/FrameP projection, via the specifier of FinP and an intermediate specifier which I proposed is a lower TopP/FrameP in compliance with the information-structural nature of the projections assumed to be present in the Split CP. For both frame-setting and non-frame-setting topics, it was shown that this postulation is advantageous in order to account for all possible preverbal sequences, including those in which two parts of a “big XP”, e.g., a DP and a DP-modifier (a genitival phrase, or a relative clause), are separated by an intervening topic particle, the latter being first-merged in the position it appears in (Top°). In this sense, left-dislocated phrases are nothing more than a possible instantiation of an aboutness or a contrastive topic whose trace spell-out is overtly realized in the specifier position of FinP. Moreover, it was argued that case-marked DPs in first clause position matching the phi-features and/or the reference of a resumptive surfacing in a lower specifier must sometimes be assumed to be non-canonical hanging topics (in Samo’s 2019 spirit), an assumption that calls for a revision of the widely accepted structural differentiation criteria between left dislocation and hanging topicalization.
As for the more controversial configuration in which e.g. a high-topical DP is followed by an adverbial clause in the preverbal area of a matrix clause, it has been shown that the most adequate formalization of this phenomenon consists in treating the adverbial clause as a parenthetical/base-generated construct. In all these cases, no violation of the bottleneck effect is incurred. An immediate consequence of the assumptions made above, however, is that XP movement into the left periphery is not subject to criterial freezing in German, which is possibly what distinguishes this language from other systems displaying a more “elastic” left periphery, i.e., a CP into which multiple constituents can be moved independently of each other.
Given that fronting can be a cyclical process and that more than one projection may be activated in a multiply-filled prefield, a further corollary is that V2 is to be understood neither as movement into the head position of the CP projection hosting a XP in its specifier nor as systematic raising of Vfin to Force°. The former option would make it difficult to determine the exact surface position of the finite verb in structures of the type Den Bruder aber von Maria habe ich heute getroffen (lit. “The brother.ACC however of Maria have I today met”), given that: (i) this sequence containing a topic particle intervening between a DP and its modifier cannot reasonably be derived as one complex constituent; (ii) there is no apparent difference between this sentence and the corresponding structure with a resumptive pronoun in [Spec,FinP] (Den Bruder aber von Maria, den habe ich heute getroffen). The latter option, instead, would imply that, irrespective of what kind of elements surface to the left of the finite verb in multiply-filled-prefield configurations, any of the complex arrangements discussed above would have multiple elements linearized before the finite verb moved to some extra-sentential position above ForceP. For instance, in the sentence Den Bruder aber von Maria, den habe ich heute getroffen, one should assume that the whole string den Bruder aber von Maria surfaces outside of the CP, which would, in turn, force us to answer the question as to how this word order has been obtained.
Thus, it seems that in the left periphery of PDG, a larger amount of projections may potentially be activated than the idea of a rigid one-specifier configuration would suggest. This basically corresponds to an extension of Frey’s (2004b) proposal that the PDG left periphery is a tripartite layer consisting of a CP, a ContrastiveP and a FinP.
The approach proposed in this article is a preliminary attempt at understanding the relative complexity of the German CP from a perspective in which the idea is adopted that different types of linear “non-V2 phenomena” can be accounted for assuming the same basic mechanisms of syntax. Of course, more research is needed to establish, for instance, whether categories like TopP, FrameP and FinP can really be taken to be the ones responsible for the above-mentioned arrangements, or new dedicated projections must be stipulated in order to make sense of these data.
Conflict of interest: The author states no conflict of interest.
Admoni, W. 1990. Historische syntax des deutschen. Tübingen: Niemeyer.Search in Google Scholar
Altmann, H. 1981. Formen der “Herausstellung” im Deutschen: Rechtsversetzung, Linksversetzung, freies Thema und verwandte Konstruktionen. Tübingen: Niemeyer.10.1515/9783111635286Search in Google Scholar
Andersson, S.-G. 1981. “Zum Problem der Satzverschränkung und damit funktional verwandter Konstruktionen im heutigen Deutsch.” In: Akten des VI. Internationalen Germanisten-Kongresses Basel 1980, Teil 2, eds. H. Rupp and H.-G. Roloff , p. 64–70. Bern, Frankfurt am Main & Las Vegas: Peter Lang.Search in Google Scholar
Austin, J. L. 1956. “Ifs and cans.” Proceedings of the British Academy 42, 109–32.Search in Google Scholar
Axel, K. 2009. “Die Entstehung des dass-Satzes. Ein neues Szenario.” In: Koordination und subordination im Deutschen (Linguistische Berichte, Sonderheft 16), eds. V. Ehrich , C. Fortmann , I. Reich and M. Reis , p. 21–41. Hamburg: Buske.Search in Google Scholar
Bech, K. 1998. “Pragmatic factors in language change: XVS and XSV clauses in old and middle English.” Folia Linguistica Historica 19, 79–102.Search in Google Scholar
Behaghel, O. 1932. Deutsche syntax. Eine geschichtliche Darstellung, Bd. 3: Satzgebilde. Heidelberg: Winter.Search in Google Scholar
Benincà, P. and Poletto, C. 2004. “Topic, focus and V2: Defining the CP sublayers.” In: The structure of CP and IP, ed. L. Rizzi , p. 52–75. Oxford: Oxford University Press.Search in Google Scholar
Benincà, P. 2001. “The position of topic and focus in the left periphery.” In: Current issues in Italian syntax: Essays offered to Lorenzo Renzi, eds. G. Cinque and G. Salvi p. 39–64. Amsterdam: Elsevier.10.1163/9780585473949_005Search in Google Scholar
Bhatt, R. and Yoon, J. 1992. “On the composition of comp and parameters of V2.” In: Proceedings of WCCFL X, ed. D. Bates , p. 41–52. Stanford: CSLI Publications.Search in Google Scholar
Boeckx, C. and Grohmann, K. K. 2005. “Left dislocation in germanic.” In: Focus on Germanic typology, ed. W. Abraham , p. 131–44. Berlin: Akademie Verlag.10.1524/9783050084336.139Search in Google Scholar
Breindl, E. 2008. “Die Brigitte nun kann der Hans nicht ausstehen. Gebundene Topiks im Deutschen. Erkenntnisse vom Rande. Zur Interaktion von Prosodie, informationsstruktur, syntax und Bedeutung.” In: Festschrift für Hans Altmann zum 65. Geburtstag, eds. E. Breindl and M. Thurmair , p. 27–49. Berlin: Schmidt.10.37307/j.1868-775X.2008.01.02Search in Google Scholar
Bunk, O. 2020. “Aber immer alle sagen das.” The status of V3 in German: Use, processing, and syntactic representation. PhD dissertation. Humboldt-Universität Berlin.Search in Google Scholar
Cardinaletti, A. 2010. “On a (wh-)moved topic in Italian, compared to Germanic.” In: Advances in comparative germanic syntax, eds. A. Alexiadou , J. Hankamer , T. McFadden , J. Nuger and F. Schaeffer , p. 3–40. Amsterdam & Philadelphia: John Benjamins.10.1075/la.141.02onaSearch in Google Scholar
Catasso, N. 2015. “On postinitial aber and other syntactic transgressions: Some considerations on the nature of V2 in German.” Journal of Germanic Linguistics 27(4), 317–65.10.1017/S1470542715000124Search in Google Scholar
Catasso, N. 2017. V2-Einbettung im spannungsfeld von hypotaxe und parataxe. Tübingen: Stauffenburg.Search in Google Scholar
Catasso, N. 2021. “Generalized and specialized adverbial resumption in Middle High German and beyond.” Journal of Historical Syntax 5(2), 1–38.Search in Google Scholar
Cinque, G. 1997. “The movement nature of left dislocation.” Linguistic Inquiry 8(2), 397–411.Search in Google Scholar
D’Avis, F. J. 2004. “In front of the prefield – Inside or outside the clause?.” In: The syntax and semantics of the left periphery, eds. H. Lohnstein and S. Trissler , p. 139–77. Berlin & New York: Mouton de Gruyter.10.1515/9783110912111.139Search in Google Scholar
De Clercq, K. and Haegeman, L. 2017. “The Typology of V2 and the distribution of Pleonastic die in the ghent dialect.” LingBuzz. https://ling.auf.net/lingbuzz/003791 (2 July, 2018).10.3389/fpsyg.2018.01342Search in Google Scholar
Dittmer, E. and Dittmer, A. 1998. Studien zur Wortstellung, Satzgliedstellung in der althochdeutschen Tatianübersetzung. Göttingen: Vandenhoek & Ruprecht . 10.13109/9783666203497Search in Google Scholar
Dittmer, E. 1992. “Die Wortstellung im althochdeutschen Tatian .” In: Althochdeutsch. syntax und semantik, ed. Y. Desportes , p. 245–58. Lyon: Centre d’études Linguistiques Jacques Goudet.Search in Google Scholar
Ebert, C. , Endriss, C. and Hinterwimmer, S. 2009. “A unified analysis of indicative and biscuit conditionals as topics.” In: Proceedings of Semantics and Linguistic Theory ( = SALT) XVIII, eds. T. Friedmann and S. Ito , p. 266–83. Ithaka, NY: Cornell University.10.3765/salt.v18i0.2473Search in Google Scholar
Fanselow, G. 2009. “Bootstrapping verb movement and the clausal architecture of German.” In: Advances in comparative germanic syntax, eds. A. Alexiadou , J. Hankamer , T. McFadden , J. Nuger and F. Schäfer , p. 85–118. Amsterdam & Philadelphia: John Benjamins.10.1075/la.141.05booSearch in Google Scholar
Frascarelli, M. and Hinterhölzl, R. 2007. “Types of topics in German and Italian.” In: On information structure, meaning and Form. Generalizations across languages, eds. K. Schwabe and S. Winkler , p. 87–116. Amsterdam & Philadelphia: John Benjamins.10.1075/la.100.07fraSearch in Google Scholar
Frey, W. 2004a. “Notes on the syntax and the pragmatics of German left dislocation.” In: The syntax and semantics of the left periphery, eds. H. Lohnstein and S. Trissler , p. 203–33. Berlin & New York: Mouton de Gruyter.10.1515/9783110912111.203Search in Google Scholar
Frey, W. 2004b. “The grammar-pragmatics interface and the German prefield.” Sprache und Pragmatik 52, 1–39.Search in Google Scholar
Frey, W. 2005a. “Zur syntax der linken peripherie im Deutschen.” In: Deutsche syntax: Empirie und Theorie (Göteborger Germanistische Forschungen 46), ed. F. J. d’Avis , p. 147–71. Göteborg: Acta Universitatis Gothoburgensis.Search in Google Scholar
Freywald, U. 2018. Parataktische Konjunktionen. Zur Syntax und Pragmatik der Satzverknüpfung im Deutschen – am Beispiel von obwohl, wobei, während und wogegen. Tübingen: Stauffenburg.Search in Google Scholar
Freywald, U. , Cornips, L. , Ganuza, N. , Nistov, I. and Opsahl, T. 2013. “Urban vernaculars in contemporary Northern Europe: Innovative variants of V2 in Germany, Norway and Sweden.” Working Papers in Urban Language and Literacies 113, 1–21.Search in Google Scholar
Freywald, U. , Cornips, L. , Ganuza, N. , Nistov, I. and Toril Opsahl . 2015. “Beyond verb second – A matter of novel information-structural effects? Evidence from German, Swedish, Norwegian and Dutch.” In: Language, youth and identity in the 21st century: Linguistic practices across urban spaces, eds. J. Nortier and B. A. Svendsen , p. 73–92. Cambridge: Cambridge University Press.10.1017/CBO9781139061896.006Search in Google Scholar
Gaumann, U. 1983. “Weil die machen jetzt bald zu.” Angabe- und Junktivsatz in der deutschen Gegenwartssprache. Göppingen: Kümmerle.Search in Google Scholar
Greule, A. 2000. “Syntax des Althochdeutschen.” In: Ein Handbuch zur Geschichte der deutschen Sprache und ihrer Erforschung, vol. 2, eds. W. Besch , A. Betten , O. Reichmann and S. Sonderegger , p. 1207–13. Berlin & New York: Mouton de Gruyter.Search in Google Scholar
Grewendorf, G. 2002. “Left dislocation as movement.” Georgetown University Working PAPERS in theoretical Linguistics 2, 31–81.Search in Google Scholar
Haegeman, L. and Greco, C. 2018. “West flemish V3 and the interaction of syntax and discourse.” Journal of Comparative Germanic Linguistics 21(1), 1–56.10.1007/s10828-018-9093-9Search in Google Scholar
Haegeman, L. 1996. “Verb second, the split CP, and null subjects in early Dutch finite clauses.” Geneva Generative Papers 4, 135–75.Search in Google Scholar
Haegeman, L. 2002. “Anchoring to speaker, adverbial clauses and the structure of CP.” Georgetown University Working Papers in Theoretical Linguistics 2, 117–80.Search in Google Scholar
Helbig, G. 2003. “Koordination vs. Subordination von Sätzen. Hauptsatz vs. Nebensatz.” In: Wort und Grammatik. Festschrift für Pavel Petkov anlässlich seiner emeritierung, eds. A. Dimova and H. E. Wiegand , p. 1–10. Hildesheim: Olms.Search in Google Scholar
Höhle, T. N. 2018. “Topologische felder.” In: Beiträge zur deutschen Grammatik. Gesammelte Schriften von Tilman N. Höhle, eds. S. Müller , M. Reis and F. Richter , p. 7–89. Berlin: Language Science Press.Search in Google Scholar
Hsu, B. 2017. “Verb second and its deviations: an argument for feature scattering in the left periphery.” Glossa 2(1), 1–33.Search in Google Scholar
Jacobs, J. 2001. “The dimensions of topic-comment.” Linguistics 39(4), 641–81.Search in Google Scholar
Keller, R. 1995. “The epistemic weil .” In: Subjectivity and subjectivisation. linguistic perspectives, eds. D. Stein and S. Wright , p. 16–30. Cambridge: Cambridge University Press.10.1017/CBO9780511554469.002Search in Google Scholar
Koch, P. and Oesterreicher, W. 1985. “Sprache der Nähe – Sprache der Distanz. Mündlichkeit und Schriftlichkeit im Spannungsfeld von Sprachtheorie und Sprachgeschichte.” Romanistisches Jahrbuch 36, 15–43.10.1515/9783112418307-002Search in Google Scholar
Kotsinas, U.-B. 1998. “Language contact in Rinkeby, an immigrant suburb.” In: Jugendsprache—langue des jeunes—youth language. Linguistische und soziolinguistische Perspektiven, eds. J. K. Androutsopoulos and A. Scholz , p. 125–48. Frankfurt: Peter Lang.Search in Google Scholar
Krifka, M. 2006. “Basic notions of information structure.” In: Interdisciplinary Studies on Information Structure 06, eds. C. Féry , G. Fanselow and M. Krifka , p. 1–46. Potsdam: University of Potsdam.10.1556/ALing.55.2008.3-4.2Search in Google Scholar
Krifka, M. 2008. “What do contrastive topics and frame setters have in common? The role of addressing and delimitation in information structure.” Paper Presented at the Conference on Contrastive Information Structure Analysis (CISA), University of Wuppertal, March 18th–19th, 2008.Search in Google Scholar
Kvam, S. 1983. Linksverschachtelung im Deutschen und Norwegischen. Eine kontrastive Untersuchung zur Satzverschränkung und Infinitivverschränkung in der deutschen und norwegischen Gegenwartssprache. Tübingen: Niemeyer . 10.1515/9783111633305Search in Google Scholar
Lippert, J. 1974. Beiträge zur Technik und Syntax althochdeutscher Übersetzungen unter besonderer Berücksichtigung der Isidorgruppe und des althochdeutschen Tatian. Munich: Fink.Search in Google Scholar
Lühr, R. 1985. “Sonderfälle der Vorfeldbesetzung im heutigen Deutsch.” Deutsche Sprache 13, 1–23.Search in Google Scholar
Lühr, R. 1988. “Zur Satzverschränkung im heutigen Deutsch.” Groninger Arbeiten zur Germanistischen Linguistik 28, 74–87.Search in Google Scholar
Meelen, M. , Mourigh, K. and Lai-Shen Cheng, L. 2020. “V3 in urban youth varieties of Dutch.” In: Syntactic architecture and its consequences II: Between syntax and morphology, eds. A. Bárány , T. Biberauer , J. Douglas and S. Vikner , p. 335–63. Berlin: Language Science Press.Search in Google Scholar
Métrich, R. and Courdier, G. 1995. “Die Zukunft des Menschen aber ist immer so lang wie seine Vergangenheit. Rand und band. Abgrenzung und Verknüpfung als Grundtendenzen des Deutschen.” In: Festschrift für Eugene Faucher zum 60. Geburtstag, eds. R. Métrich and M. Vuillaume , p. 29–38. Tübingen: Narr . Search in Google Scholar
Müller, S. 2003. “Mehrfache Vorfeldbesetzung.” Deutsche Sprache 31(1), 29–62.Search in Google Scholar
Müller, S. 2005. “Zur Analyse der scheinbar mehrfachen Vorfeldbesetzung.” Linguistische Berichte 203, 297–330.Search in Google Scholar
Munaro, N. 2005. “Computational puzzles of conditional clause preposing.” In: UG and external systems, ed. A. M. Di Sciullo , p. 73–94. Amsterdam: John Benjamins,10.1075/la.75.06munSearch in Google Scholar
Munaro, N. 2010. “Towards a hierarchy of clause types.” In: Mapping the left periphery, eds. P. Benincà and N. Munaro , p. 125–62. Oxford/New York: Oxford University Press.10.1093/acprof:oso/9780199740376.003.0005Search in Google Scholar
Pasch, R. 1997. “Weil mit Hauptsatz – Kuckucksei im denn-Nest.” Deutsche Sprache 25(3 ), 252–71.Search in Google Scholar
Postal, P. 1971. Cross-over phenomena. New York: Holt, Rinehart and Winston.Search in Google Scholar
Quist, P. 2008. “Sociolinguistic approaches to multiethnolect: Language variety and stylistic practice.” International Journal of Bilingualism 12, 43–61.10.1177/13670069080120010401Search in Google Scholar
Riemsdijk, H. C. van and Zwarts, F. 1997. “Left dislocation in Dutch and the status of copying rules.” In: Materials on left dislocation, eds. E. Anagnostopoulou , H. C. van Riemsdijk and F. Zwarts , p. 13–29. Amsterdam & Philadelphia: John Benjamins.10.1075/la.14.04rieSearch in Google Scholar
Rizzi, L. and Bocci, G. , 2017. “Left periphery of the clause: Primarily illustrated for Italian.” In: The Wiley blackwell companion to syntax, 2nd ed., eds. M. Everaert and H. van Riemsdijk , p. 1–30. Hoboken: John Wiley & Sons.10.1002/9781118358733.wbsyncom104Search in Google Scholar
Rizzi, L. 2001. “On the position of int(errogative) in the left periphery of the clause.” In: Current studies in Italian syntax, eds. G. Cinque and G. Salvi p. 287–96. Amsterdam: Elsevier.10.1163/9780585473949_016Search in Google Scholar
Rizzi, L. 2004. “Locality and left periphery.” In: Structures and beyond: The cartography of syntactic structures, vol. 3, ed. A. Belletti , p. 223–51. Oxford: Oxford University Press.Search in Google Scholar
Rizzi, L. 2006. “On the form of chains: Criterial positions and ECP effects.” In: Wh-movement: Moving on, eds. L. Cheng and N. Corver , p. 97–133. Cambridge, MA: MIT Press.Search in Google Scholar
Rizzi, L. 2010. “On some properties of criterial freezing.” In: The complementizer phase: Subjects and operators, ed. E. Phoevos Panagiotidis . Oxford: Oxford University Press.10.1093/acprof:oso/9780199584352.003.0002Search in Google Scholar
Roberts, I. 2004. “The C-system in brythonic celtic languages, V2, and the EPP.” In: The structure of CP and IP, ed. L. Rizzi , p. 297–328. Oxford: Oxford University Press.Search in Google Scholar
Ross, J. R. 1967. Constraints on variables in syntax. PhD dissertation. MIT, Cambridge.Search in Google Scholar
Salvesen, C. M. 2016. Resumptive particles and verb second. Unpublished manuscript. Oslo: Universitet i Oslo.Search in Google Scholar
Scherpenisse, W. 1983. “Topic, theme and the German initial field.” In: Topic, focus and configurationality, eds. W. Abraham and S. de Meij , p. 277–93. Amsterdam: John Benjamins.Search in Google Scholar
Scheutz, H. 1997. “Satzinitiale Voranstellungen im gesprochenen Deutsch als Mittel der themensteuerung und Referenzkonstitution.” In: Syntax des gesprochenen Deutsch, ed. P. Schlobinski , p. 27–54. Opladen: Westdeutscher Verlag.10.1007/978-3-322-88924-9_2Search in Google Scholar
Schlachter, E. 2012. Syntax und informationsstruktur im althochdeutschen. Untersuchungen am Beispiel der Isidor-Gruppe. Heidelberg: Winter.Search in Google Scholar
Schwitalla, J. 2003. Gesprochenes Deutsch. Eine Einführung. 2nd ed. Berlin: Erich Schmidt.Search in Google Scholar
Sigurðardóttir, S. 2019. “Case(s) of V3 orders in icelandic with temporal adjuncts.” Working Papers in Scandinavian Syntax 102, 1–17.Search in Google Scholar
Speyer, A. and Weiß, H. 2018. “The prefield after the old high german period.” In: Clause Structure and Word Order in the History of German, eds. A. Jäger , G. Ferraresi and H. Weiß , p. 64–81. Oxford: Oxford University Press.10.1093/oso/9780198813545.003.0005Search in Google Scholar
Speyer, A. 2008. “Doppelte Vorfeldbesetzung im heutigen Deutsch und im Frühneuhochdeutschen.” Linguistische Berichte 216, 455–85.Search in Google Scholar
Szczepaniak, R. 2013. “Satztyp und sprachwandel.” In: Satztypen des Deutschen, eds. J. Meibauer , M. Steinbach and H. Altmann , p. 738–63. Berlin & New York: Mouton de Gruyter.10.1515/9783110224832.738Search in Google Scholar
te Velde, J. R. 2017. “Temporal adverbs in the kiezdeutsch left periphery: Combining late merge with deaccentuation for V3.” Studia Linguistica 71(3), 301–36.10.1111/stul.12055Search in Google Scholar
Urban, E. 1999. ‘Exactly’ stranding. M.A. thesis, University of California, Santa Cruz.Search in Google Scholar
Volodina, A. and Weiß, H. 2010. Wie einfach ist das deutsche Vorfeld? Paper presented at GGS 2010. FU Berlin: IDS Mannheim/Goethe-Universität Frankfurt a.M. http://www.annavolodina.de/dokumente/Vortrag_GGS_naemlich_7-05_lit.pdf (1 February, 2018).Search in Google Scholar
Walkden, G. 2015. “Verb-third in early west germanic: A comparative perspective.” In: Syntax over time: lexical, morphological, and information-structural interactions, eds. T. Biberauer and G. Walden , p. 236–48. Oxford: Oxford University Press.10.1093/acprof:oso/9780199687923.003.0014Search in Google Scholar
Wiese, H. 2010. Kiezdeutsch – grammatik, informationsstruktur und perzeption eines neuen dialekts. Ms. Universität Potsdam.Search in Google Scholar
Wiese, H. 2013. “What can new urban dialects tell us about internal language dynamics? The power of language diversity.” Linguistische Berichte Sonderheft 19, 207–45.Search in Google Scholar
Wiese, H. , Freywald, U. , Mayr, K. and Schalowski, S. 2012. “Das KiezDeutsch-Korpus. Spontansprachliche Daten Jugendlicher aus urbanen Wohngebieten.” Deutsche Sprache 40, 97–123.Search in Google Scholar
© 2021 Nicholas Catasso, published by De Gruyter
This work is licensed under the Creative Commons Attribution 4.0 International License.