Abstract

This article presents novel evidence for cyclic wh-movement. Two experiments show that the question-sensitive particle denn in German is more readily accepted when occurring in an independent interrogative clause than in a dependent clause embedded in a wh-question. Importantly, however, acceptance of denn in a dependent clause increases significantly when a wh-phrase has been moved out of that clause. We argue that denn is locally licensed by an interrogative Force head. In dependent clauses, this licensing can be mediated by long wh-movement leaving a transient representation of interrogative force in Spec,CP. Without such a mediating trace in Spec,CP, the licensing of the particle fails. In conclusion, denn in a dependent clause indicates cyclic wh-movement.

1 Introduction

Wh-dependencies can span more than a single clause, as is apparent in long wh-movement. While it seems uncontroversial in transformational grammar that movement leaves a trace in the base position, despite arguments in favor of intermediate positions it remains a matter of debate whether intermediate positions play a role at all, and if so, which positions are targeted on the way out of an embedded clause. In this article, we argue that Q(uestion)-sensitive discourse particles (DiPs) provide novel evidence for cyclic movement via Spec,CP, where Spec,CP is seen as the place in which illocutionary force must be syntactically established if it is syntactically established at all. The data rest on acceptability judgments of German wh-questions containing the Q-sensitive DiP denn (lit. ‘then’). Given that DiPs lack any overt grammatical reflexes such as morphological agreement or relations to argument/event structure, their acceptability can only be based on pragmatic appropriateness and syntactic environment. In this sense, acceptability data for wh-questions systematically varying the presence and position of the particle reflect by far subtler judgments than data yielded by overt reflexes of displacement. On the other hand, the fact that Force must be rooted in the C domain provides direct evidence for the activation of an intermediate position in long wh-movement. Our data show that in wh-questions, the DiP denn can only occur in a clausal complement X if a long (i.e., transclausal) wh-dependency in the sense of cyclic movement connects X with the interrogative Force head of the root clause. If the particle remains in X without a link to the Force head of the root clause, it cannot be licensed. This is the case if wh-movement applies only in the root clause. The contrast is illustrated in (1).

(1)

  • a.

    Wen meintest du, dass wir denn zu der Tagung einladen sollten?

    who.ACC thought you that we DENN to the conference invite should

    ‘Who did you think that we should invite to the conference?’

  • b.

    *Wer meinte, dass wir denn einen Gastredner einladen sollten?

    who thought that we DENN a guest.speaker invite should

    ‘Who thought that we should invite a guest speaker?’

As our data show, this contrast is real. We conclude that the occurrence of a Q-sensitive particle in a dependent (nonroot) clause emerges as an independent diagnostic of long wh-movement targeting the C domain on the way out of the dependent clause.

The article is organized as follows. Section 2 gives an introduction to the role of DiPs in German and their licensing by illocutionary force. The emphasis is on wh-questions and Q-sensitive DiPs. Section 3 provides a survey of evidence for the involvement of the CP’s left edge in cyclic wh-movement. Section 4 presents two experiments, which provide empirical evidence for the contrast in (1). From the finding that denn is licit in a dependent clause only when whmovement has taken place out of this clause, we conclude that denn can be used as a diagnostic for cyclic wh-movement. Section 5 develops a formal syntactic analysis that rests on cyclic agreement between a Q-feature in the Force projection and a corresponding Q-feature of the particle. Section 6 concludes the article.

2 Discourse Particles in Wh-Questions

2.1 Properties of Discourse Particles

German is a language with the reputation of having many DiPs (also called “modal” particles). However, semantically comparable elements seem to occur in almost every language. DiPs are generally polyfunctional elements. The particle denn, for example, which is at the center of the present study, is also a subordinating conjunction, and in its original shape, dann, it was a temporal adverb. Historically, it derives from Old High German thanne ‘then’ (see Abraham 1991, Wauchope 1991, Wegener 2002). As far as the historical development of DiPs can be traced back, their status is the result of a grammaticalization process by which a lexical item is subject to simultaneous reduction of its phonological and semantic features and its syntactic freedom.1 DiPs cannot be coordinated or modified, they usually cannot bear phonological prominence, they lack proforms, and they are immobile.2 In a case like Er ist doch schon alt ‘He is, as you should know, already old’, topicalization of doch destroys the ‘as you should know’ reading. In Doch ist er schon alt, doch equals the adverb ‘however’: ‘Contrary to expectations, he is already old’. In other words, doch as a DiP cannot front whereas doch as an adverb can do so easily. The syntactic properties of DiPs are generally compatible with the view that particles are heads rather than phrases. We assume here without further discussion that at least some DiPs are heads and that denn belongs to this group.3 DiPs are sensitive to clause type. Some occur in declaratives, others in imperatives or in exclamatives, still others—like denn, which we examine here—in interrogatives. DiPs are prototypical root phenomena because the root clause is an utterance, and only utterances have interpretable illocutionary force.4 As we will show, this general impression requires certain qualifications.

2.2 Discourse Particles in Wh-Questions

Let us concentrate more closely on DiPs that arise in constituent questions. Consider the difference between (2) and (3).

(2)

  • Wo wohnst du?

  • where live you

  • ‘Where do you live?’

(3)

  • Wo wohnst du denn?

  • where live you DENN

  • ‘Where do you live? (I am wondering.)’

While (2) is a plain information-seeking question that does not reveal any attitude of the speaker, (3) signals that the speaker is in a particular way “concerned” about the proposition that the answer would yield. The particle denn, related to English then, makes reference to some common ground between speaker and hearer beyond the presupposition p, p = there is a place x such that you live in x. No existence of such an additional common ground of shared knowledge is assumed in (2). Denn is a quasi anaphoric expression, roughly translatable as ‘under some circumstances known to both of us’.5 This makes denn unfit for out-of-the-blue questions.6

Although we will concentrate here on the particle denn, there are other Q-sensitive particles in German that to some degree function along the lines of denn, such as nur (lit. ‘only’) and bloß (lit. ‘barely’).

(4)

  • Wo hast du nur / bloß meine Schlüssel hingelegt?

  • where have you NUR / BLOSS my keys put.down

  • ‘Where did you put my keys? (I have already looked everywhere.)’

In a wh-question like (4), the DiP signals that the speaker has already tried several times to find an answer. Bayer and Obenauer (2011) refer to such questions as “can’t-find-the-value-of-x questions” (CfvQs). Another particle is schon (lit. ‘already’). In wh-questions, schon yields a rhetorical question (RQ).7

(5)

  • Wer zahlt schon gerne Steuern?

  • who pays SCHON gladly taxes

  • ‘Who likes paying taxes? (Nobody!)’

Another one is wohl (lit. ‘well’). Denn and wohl are more general and can easily combine with nur, bloß, schon (see Thurmair 1989). When combined, these particles appear in a fixed order, as is familiar from Cinque’s (1999) work on adverbs and the proposed universal adverb hierarchy; for an application to German DiPs, see Coniglio 2011. Thus, cases of particle stacking as in (6) are naturally available in German.

(6)

  • a.

    Wo hast du denn wohl {nur / bloß} meine Schlüssel hingelegt?

    where have you DENNWOHLNUR / BLOSS my keys put.down

    ‘Where did you put my keys? (I have already looked everywhere.)’

  • b.

    Wer zahlt denn wohl schon gerne Steuern?

    who pays DENNWOHLSCHON gladly taxes

    ‘Who likes paying taxes? (Come on, nobody!)’

2.2.1 Discourse Particles and Illocutionary Force

Since DiPs are linked to illocutionary force, a syntactic account suggests that they need to access the Force projection in the sense of Rizzi’s (1997) proposal of a split CP. Force is present at the top of the root clause or of a root-like clause. A root-like clause is formally subordinate but nevertheless counts as quoted speech. Consider the following minimal pair from Bayer 2012:15 (see also Krifka 2014):

(7)

  • a.

    Christine fragte, warum Klaus denn so blass ist.

    Christine asked why Klaus DENN so pale is

    ‘Christine asked why Klaus is so pale.’

  • b.

    *Christine weiß, warum Klaus denn so blass ist.

    Christine knows why Klaus DENN so pale is

In (7a), the attitude of wondering/being concerned is ascribed to Christine and not to the speaker who utters the sentence. Thus, Christine is responsible for the use of denn. In (7b), however, because of the verb wissen ‘to know’, which does not set up a quotational context, only the speaker who utters (7b) would be available. But the speaker utters a declarative—a clause type that does not license the particle denn. The conclusion must be that the form of the embedded clause alone cannot be responsible for providing force. Instead, it is the interpretation conditioned by the matrix verb that renders the embedded clause as reported speech or not.8

2.2.2 Denn

in Dependent Clauses Despite their association with illocutionary force, Q-sensitive DiPs can also arise in dependent clauses, as in the following examples:

(8)

(9)

(10)

All of (8)–(10) are questions. As (11)–(13) show, the possibility of licensing denn as a DiP in noninterrogative contexts can be excluded.

(11)

  • Ich meine, dass er (*denn) ihrer Region besser helfen würde.

  • I think that he DENN your region better help would

  • ‘I think that he would help your region in a better way.’

(12)

  • Er denkt, dass es (*denn) irgendwie weitergehen wird.

  • he thinks that it DENN somehow go.on will

  • ‘He thinks that it will somehow go on.’

(13)

  • Du glaubst, dass er (*denn) ein falsches Bild von mir haben könnte.

  • you believe that he DENN a wrong picture of me have could

  • ‘You believe that he could have a wrong impression of me.’

The grammatical cases in (8)–(10) involve complement clauses from which a wh-phrase has been extracted. The question is how essential extraction from the complement clause is. Could it be the case that it doesn’t matter as long as denn is in a complement clause while the (immediately) dominating root clause has the right illocutionary force to license the particle? If this is so, the subject-wh clauses in (14)–(16) should be fine.

(14)

  • Welche Leute meinten, dass er denn ihrer Region besser helfen würde?

  • which people thought that he DENN your region better help would

  • ‘Which people thought that he would help your region in a better way?’

(15)

  • Wer denkt, dass es denn irgendwie weitergehen wird?

  • who thinks that it DENN somehow go.on will

  • ‘Who thinks that it will somehow go on?’

(16)

  • Welche Leute glauben, dass er denn ein falsches Bild von mir haben könnte?

  • which people believe that he DENN a wrong picture of me have could

  • ‘Which people believe that he could have a wrong impression of me?’

According to our intuitions, these sentences are degraded, but it is not clear how strong the effect really is. Section 4 presents two experiments showing that the contrast between short wh-movement and long wh-movement is real.

2.2.3 Apparent Counterexamples

Our search for relevant examples of the short wh-movement type also yielded sentences like (17) and (18), in which denn occurs in a dependent clause from which no wh-movement takes place.9

(17)

  • Wer sagt, dass Klamottenstil denn teuer sein muss?

  • who says that clothing.style DENN expensive be must

  • ‘Who says that clothing style must be expensive?’

  • http://www.team-ulm.de/Forum/1/145700/749 18 July 2013

(18)

All the examples we could find involve the subject wh-pronoun and the verb sagen, and all of them appear in contexts in which the question is interpreted as an RQ. Who says that p? amounts to the polar question Does anyone say that p? and yields the conversational implicature that hardly anyone or even no one says that p. This puts (17)–(18) on a par with (19). As a matter of fact, polar questions allow embedded denn under the condition that the embedding predicate is reduced to a marker of evidentiality.10

(19)

Predicates like mögen ‘to want’ and sich vorstellen ‘to imagine’ appear to refuse this interpretive option, although they show long wh-extraction with denn in the embedded CP.11 Examples (17)–(19) show that denn may occur in the dependent CP when the question under discussion (also called the “main point”; see Simons 2007) targets this dependent CP rather than the matrix clause. The matrix predicate is backgrounded to such an extent that it functions rather like an interrogative prefix to the embedded CP. As a result, denn is licit in the CP complement, as it is in corresponding simplex polar questions like Muss Klamottenstildennteuer sein? ‘Must clothing style be expensive?’ or Möchte dieser Manndennernsthaft eine Beziehung führen? ‘Does this man seriously want to be in a relationship?’12 Assume for the sake of the argument that for cases like (17) and (18) there is a reanalysis by which the wh-clause who says is interpreted as does anybody say (with the implicature nobody says). If so, denn is licensed as in (19). In this case, one could argue that there is no actual semantic embedding, and the question is functionally a simplex clause. Denn in the embedded CP with wh-extraction, as in (8)–(10), appears to be different, though, and can hardly escape a syntactic account. As Marga Reis (pers. comm.) points out, denn is licensed in the complement CP even if the embedding predicate is part of the speaker’s question. To see this, consider the following dialogue:

(20)

  • A:

    Mein Freund hat große Pläne für die Zukunft. Er glaubt, dass er bald befördert my friend has great plans for the future he believes that he soon promoted wird, und ins Ausland gehen kann.

    will.be and in.the abroad go can

  • B:

    Und wie1 glaubst DU, t′1 dass es denn dann mit euch t1 weitergehen soll? and how believe you that it DENN then with you.PL go.on should

    ‘And how do YOU believe that it will then go on with the two of you?’

The middle link t′1 of the chain headed by the wh-operator wie must have locally licensed the occurrence of denn although the matrix predicate is an essential part of B’s question; that is, the question cannot be semantically reduced to the question under discussion, ‘How will it go on with the two of you?’. Further evidence in this direction will be given in section 5.13 So whatever the ultimate analysis for (17)–(19) may be, the occurrence of denn in (8)–(10) seems to rely on the specific syntactic dependency that characterizes long wh-movement.

The difference between short and long wh-movement, which appears to be contradicted by examples like (17) and (18), can be further clarified if we shift our attention to the elements schon and nur, both of which may function as DiPs in wh-questions but—unlike denn—not in polar questions. Consider the interpretive difference between the pairs in (21) and (22).

(21)

  • a.

    An welche Regierung glaubst du, dass man schon gerne Steuern zahlt?

    to which government believe you that one SCHON gladly taxes pays

    ‘To which government do you believe one is fond of paying taxes? (To none at all!)’

  • b.

    Wer glaubt, dass man an unsere Regierung schon gerne Steuern zahlt?

    who believes that one to our government already gladly taxes pays

    ‘Who believes that one is already fond of paying taxes to our government?’

(22)

  • a.

    Von wem glaubst du, dass der Junge das nur abgeschrieben haben könnte?

    from who.DAT believe you that the boy this NUR copied have could

    ‘From who on earth do you believe the boy could have copied this?’

  • b.

    Wer glaubt, dass der Junge das nur abgeschrieben haben könnte?

    who believes that the boy this only copied have could

    ‘Who believes that the boy could have only copied this?’

While (21a) yields the RQ interpretation without any effort, that interpretation is very hard if not impossible to get in (21b). Even though schon appears in a wh-question, the preferred interpretation is one according to which schon is a regular adverb, and as such neutral about the type of its embedding sentence. The contrast in (22) may be even clearer. (22a) yields the CfvQ interpretation very naturally; the speaker signals that he or she has so far unsuccessfully tried to find an answer. However, the only interpretation of (22b) is that the speaker wants to know who believes that the boy could have done nothing else to a certain text but copy it. The only reading of nur here is the focus particle interpretation. In both of these cases, the interpretive distribution corroborates a syntactic account that attributes it to the fact that the distant DiP can be locally licensed under long wh-movement but not under short wh-movement.

It may be worth mentioning here that German allows another type of long-distance dependency that involves, not a C-headed complement, but a V+T-headed one. It is standardly understood as extraction from a V2 clause as in Wie1 [C′denkst du [CPt1soll es t1weitergehen mit euch]]? ‘How do you think it should go on with the two of you?’. Interestingly, Q-sensitive DiPs can freely occur in such complements, and no markedness effect obtains.

(23)

  • Wie1 [denkst du [t′ 1 soll es denn t1 weitergehen mit euch]]?

  • how think you should it DENN go.on with you.PL

  • ‘How do you think it should go on with the two of you?’

Reis (1995, 1996) proposes to reanalyze the purported matrix clause as a verb-first integrated parenthetical (VIP). According to Reis, denkst du in (23) is a VIP that may be inserted in a root wh-question. Her reasoning turns (23) into (24).

(24) Wie1 [denkst du] soll es denn t1 weitergehen mit euch?

If she is right, the occurrence of denn is unsurprising. The particle can access Force as in any simplex wh-question. Since wh-movement does not cross a CP boundary, denn is licensed without mediation by an intermediate trace of the wh-operator.14 The fact that VIPs are optional and can “float” in the clause, as in Wie1soll es [denkst du] denn t1weitergehen mit euch?, is compatible with the VIP’s potential role as an evidential marker. The wh-questions considered in this article are different. They are complex structures in which the wh-operator must have moved out of the dependent CP. Of course, this does not preclude the role of a semantic factor by which the matrix predicate may in fact attain properties of an evidential marker. The bridge construction denkst du in (9) and the VIP denkst du in (24) appear to play the same semantic role in the two cases, but the syntax is different. Licensing of denn in a VIP-modified question is as local as in any simplex wh-clause, whereas licensing of denn in the bridge construction must be achieved by an extra syntactic step, namely, creation of an intermediate trace in the left periphery of the embedded CP. Sentences like (17)–(19), (21b), and (22b) are superficially similar to (8)–(10), (21a), and (22a) but lack this extra step. As a result, Q-sensitive DiPs are unlicensed in their complement.15 A formal analysis will be given in section 5.

2.2.4 Intermediate Summary

In German, DiPs are distributed roughly like adverbs, they lack morphological agreement, and they do not interfere with argument/event structure. Thus, their role in the clause must be evaluated on the basis of nothing other than their pragmatic appropriateness and their syntactic environment. This distinguishes DiP licensing from gap licensing, in which Case, argument/event structure, and s-selection provide important cluesfor the computation of the dependency.16 The licensing of DiPs in wh-questions requires access to the C/Force domain. Exactly this constellation is provided by wh-movement through the embedded C/Force projection. If our reasoning is on the right track, the acceptability of Q-sensitive DiPs would constitute an important new diagnostic for cyclic movement across the CP phase. The particle’s dependency is superimposed on the familiar Ā-dependency and is thus crucially related to the intermediate landing site of wh (the C/Force projection). As such, denn as a diagnostic opens up a new line of research and testing.

3 Existing Evidence for Cyclic Wh-Movement

One of the most widely known pieces of evidence for successive-cyclic wh-movement is the fact that in Irish, the complementizer changes from the regular form go to a special form aL when a wh-operator has passed through its specifier. (Examples (25)–(27) are from McCloskey 2002: 185, 189.)

(25)

  • Creidim gu-r inis sé bréag.

  • believe.1SGGO-PAST tell he lie

  • ‘I believe that he told a lie.’

(26)

  • Céacu ceann a dhíol tú?

  • which one aL sold you

  • ‘Which one did you sell?’

(27)

  • an t-ainm a hinnseadh dúinn a bhí ar an áit

  • the name aL was.told to.us aL was on the place

  • ‘the name that we were told that it was on the place’

See McCloskey 2002 for discussion of the Irish facts in connection with general questions of successive-cyclic movement. Overt morphological effects of successive-cyclic movement have also been reported in Chamorro by Chung (1982) and her following work. Vestiges of cyclic movement have also been identified in Malay and in Bahasa Indonesia by Saddy (1991), Cole and Hermon (2000), and Sabel (2000), and for the Bantu language Kinande by Schneider-Zioga (2009). Widely discussed examples come from connectivity effects, as first investigated by Barss (1986). (28a) is ambiguous, as shown in (28b).

(28)

  • a.

    Which pictures of himself did John think (that) Fred liked?

  • b.

    [Which pictures of himselfi/j]1 did Johni think [t′ 1 (that) [Fredj liked t1]?

If the trace in the lower Spec,CP is a copy that contains [... himself ], the anaphor can be bound by the next accessible subject, which is John. Similar examples that provide an argument for an intermediate trace/copy could be given for German.

Various languages with overt cyclic wh-movement have been found to employ strategies by which the moved element or part of the moved element appears overtly on the clausal edges of the CPs involved. These constructions have become known as partial movement and copy movement. Partial movement has been found in German (see (29)), including Tyrolian (Alber 2008), colloquial Dutch (Barbiers 2014), Frisian (Hiemstra 1986), Hungarian (Horvath 1997, Lipták and Zimmermann 2007), Iraqi Arabic (Wahba 1992), Romani (McDaniel 1989), Hindi (Dayal 1994, Mahajan 1990), and Bangla (Bayer 1990, 1996); see also Lutz, Müller, and von Stechow 2000.

(29)

  • Was glaubst du, was Gerda meint, wem wir vertrauen können?

  • what believe you what Gerda thinks who.DAT we trust can

  • ‘Who do you believe Gerda thinks we can trust?’

In partial movement, the lower clause shows standard wh-movement while the scope of this wh-operator is recursively extended into the immediately dominating clause by the unmarked wh-pronoun was. Copy movement has been found at least in German (see (30)), including Tyrolian, Frisian, colloquial Dutch, Afrikaans, and Romani. For an overview and references, see Felser 2004; for a recent account, see Pankau 2013.

(30)

  • Wem glaubst du, wem Gerda meint, wem wir vertrauen können?

  • who.DAT believe you who.DAT Gerda thinks who.DAT we trust can

  • ‘Who do you believe Gerda thinks we can trust?’

Copy movement shares with partial movement at least the surface effect by which a wh-pronoun appears in each dominating clause and thus extends the scope of the wh-element of the lowest cycle. In the copy construction, it is, roughly speaking, the primary wh-element itself that is repeated in Spec,CP. As Felser (2004) and Pankau (2013) show, partial movement, as in (29), differs in various ways from overt movement and also from copy movement, as in (30), while the latter two have more properties in common. In section 4.2, we will extend our investigation to partial movement, and we will also discuss its relation to copy movement. For the time being, it suffices to note that in each of these cases, an overt reflex of the wh-dependency is spelled out in the CP’s edge position, which syntactic theory has traditionally claimed to host an intermediate Ā-trace in cyclic movement.17

In the processing literature, the assumption of traces in general and of intermediate traces in particular is controversial (Phillips and Parker 2014, Pickering and Barry 1991). Priming effects at purported in-situ trace positions, for instance, can be explained without reference to traces (e.g., as effects of simple lexical association). It is therefore desirable to search for ways of probing into long-distance dependencies that are independent of lexical influences. DiPs are independent of the clause’s argument/event structure and the lexical elements that control them. In fact, they belong to the “expressive dimension” (Potts 2005) and thus to a distinct vocabulary. Their acceptability in an embedded noninterrogative complement requires access to the locally available C/Force projection. No lexical cue can be made responsible for the long-distance dependency. This amounts to a new and independent way of exploring the presence of intermediate traces.

4 Experimental Investigations

In this section, we present an empirical investigation of speaker judgments about German wh-questions containing the DiP denn. The corpus examples discussed in section 2 suggest that denn can occur not only in root wh-questions but also in dependent clauses when licensed by wh-extraction out of a that-clause. However, we found only a small number of corpus examples— possibly because DiPs mainly occur in less formal spoken registers—so that firm conclusions were not possible. To clarify the facts, we designed two experiments examining the acceptability of denn in dependent clauses.

The experiments were conducted with the help of a questionnaire using magnitude estimation (ME), a method originally developed in psychophysics (see Stevens 1975) and later adapted for collecting linguistic judgments (see Bard, Robertson, and Sorace 1996, Cowart 1997, Sorace 1992). ME has been successfully applied to studying a wide range of syntactic phenomena, including extraction (Cowart 1997, Featherston 2005a,b, McDaniel and Cowart 1999). In an ME experiment, participants are first confronted with a so-called reference stimulus and then are asked to evaluate all further stimuli relative to this reference stimulus. They assign numerical values according to two constraints: (a) the numerical value can be arbitrarily small but must be greater than zero; (b) the ratio between the reference value and the item value should reflect the perceived acceptability ratio between the reference stimulus and the current stimulus. ME yields judgments on a continuous scale and is therefore suited to standard statistical procedures. Some of the underlying theoretical assumptions of ME have been questioned (e.g., Featherston 2008, Sprouse 2011, Weskott and Fanselow 2011), but this does not diminish its usefulness as a method for gathering fine-grained acceptability judgments (see Schütze and Sprouse 2014).

4.1 Experiment 1

Experiment 1 investigates the acceptability of denn in main and embedded clauses, depending on whether a sentence-initial wh-word originated in the main clause (“short wh-movement”) or in the embedded clause (“long wh-movement”). A sample sentence involving short wh-movement is provided in (31). Denn is licensed in the main but not in the embedded clause. Acceptability ratings should therefore be high in the former case and low in the latter.

(31)

  • Short wh-movement

  • Weri berichtete ti ihr (denn), dass die Einbrecher (denn) gefasst wurden?

  • who told her DENN that the burglars DENN caught were

  • ‘Who told her that the burglars were caught?’

Short wh-movement sentences contrast with long wh-movement sentences in which the sentence-initial wh-phrase relates to a variable in the embedded clause.

(32)

  • Long wh-movement

  • Weni vermutete er (denn), ti dass die Polizei (denn) ti festgenommen hat?

  • who.ACC suspected he DENN that the police DENN arrested has

  • ‘Who did he suspect that the police arrested?’

In (32), denn is licensed in the main as well as in the embedded clause. It should therefore be equally acceptable in both positions. Overall, however, sentences with long wh-movement should be less acceptable than sentences with short wh-movement, given the well-known problematic status of long extraction for (some) speakers of German (Kvam 1983, Paul 1919). To control for this effect, the experiment includes sentences with the clause-type-neutral temporal adverb damals ‘back then’ in place of denn.

4.1.1 Method

The participants, materials, and procedure for Experiment 1 were as follows.

4.1.1.1 Participants

Ninety-seven students from the University of Konstanz completed a questionnaire for course credit or payment. All participants were native speakers of German and naive with respect to the purpose of the experiment.

4.1.1.2 Materials

Experiment 1 had a three-factor design. The first factor was the between-items factor Movement (short vs. long wh-movement). One set of 16 sentences involved short wh-movement as in (31) and a second set of 16 different sentences involved long wh-movement as in (32). We decided to realize Movement as a between-items factor because this imposed less severe constraints on the lexical materials used in the sentences. Each of the total set of 32 sentences appeared in four versions resulting from crossing the two factors Particle (denn vs. damals) and Particle Position (main clause vs. embedded clause).18Table 1 gives a complete set of stimuli. All experimental sentences consisted of a matrix clause and an embedded clause. The matrix clause always began with a wh-phrase that originated either in the matrix clause (short wh-movement) or in the embedded clause (long wh-movement). Short wh-movement always involved the subject wh-pronoun wer ‘who’, whereas long wh-movement involved various wh-words, both arguments (wem ‘who.DAT’ and wen ‘who.ACC’) and adjuncts (wie ‘how’, wo ‘where’, woran ‘whereby’/‘at what’, and wohin ‘where to’). Each sentence contained either the clausetype-sensitive DiP denn or the clause-type-neutral adverbial damals ‘back then’, either in the main clause or in the embedded clause.

Table 1

Sample stimuli from Experiment 1

  • Short wh-movement, particle/adverb in main clause

  • Wer berichtete ihr denn/damals, dass die Einbrecher gefasst wurden?

  • who told her DENN/back.then that the burglars caught were

  • ’Who told her DENN/back then that the burglars were caught?’

 
  • Short wh-movement, particle/adverb in dependent clause

  • Wer berichtete ihr, dass die Einbrecher denn/damals gefasst wurden?

  • who told her that the burglars DENN/back.then caught were

  • ’Who told her that the burglars were caught DENN/back then?’

 
  • Long wh-movement, particle/adverb in main clause

  • Wen vermutete er denn/damals, dass die Polizei festgenommen hat?

  • who.ACC suspected he DENN/back.then that the police arrested has

  • ’Who did he suspect DENN/back then that the police arrested?’

 
  • Long wh-movement, particle/adverb in dependent clause

  • Wen vermutete er, dass die Polizei denn/damals festgenommen hat?

  • who.ACC suspected he that the police DENN/back.then arrested has

  • ’Who did he suspect that the police arrested DENN/back then?’

 
  • Short wh-movement, particle/adverb in main clause

  • Wer berichtete ihr denn/damals, dass die Einbrecher gefasst wurden?

  • who told her DENN/back.then that the burglars caught were

  • ’Who told her DENN/back then that the burglars were caught?’

 
  • Short wh-movement, particle/adverb in dependent clause

  • Wer berichtete ihr, dass die Einbrecher denn/damals gefasst wurden?

  • who told her that the burglars DENN/back.then caught were

  • ’Who told her that the burglars were caught DENN/back then?’

 
  • Long wh-movement, particle/adverb in main clause

  • Wen vermutete er denn/damals, dass die Polizei festgenommen hat?

  • who.ACC suspected he DENN/back.then that the police arrested has

  • ’Who did he suspect DENN/back then that the police arrested?’

 
  • Long wh-movement, particle/adverb in dependent clause

  • Wen vermutete er, dass die Polizei denn/damals festgenommen hat?

  • who.ACC suspected he that the police DENN/back.then arrested has

  • ’Who did he suspect that the police arrested DENN/back then?’

 

The experimental sentences were distributed over four lists according to a Latin square design. Each list contained each sentence in only one of its four versions and contained an equal number of sentences in each condition. Finally, three versions of each list were constructed by randomizing the order in a list. The resulting twelve lists were then transformed into questionnaires. In total, each questionnaire contained 108 sentences: 32 experimental items and 76 filler items. About half of the filler items served as experimental items in an unrelated experiment investigating argument structure alternations. Filler items were structurally related to the experimental items insofar as they also contained embedded clauses. Some of them were wh-questions (short and long wh-movement) while others were declaratives. Filler items were both grammatical and ungrammatical sentences. Ungrammatical fillers involved various kinds of violations: for example, island violations, Case violations, agreement errors, lexical deviations, and a few morphological errors. The strength of the violations was expected to range from very subtle to very strong.

4.1.1.3 Procedure

The experiment tested participants using a paper questionnaire. To familiarize the participants with the procedure, the questionnaire began with a training in which they had to judge the length of lines relative to a reference line. First, an instruction was given together with an example. Next, seven lines were given, each with a box to its left. The first line served as the reference item and was assigned the value 50. The participants’ task was to estimate the length of the subsequent lines relative to the reference line and write down a value in the box. The next page of the questionnaire contained an instruction for estimating sentence acceptability together with an example. The actual experiment began on the third page. At the top of this page and all subsequent pages, the reference sentence in (33) appeared together with the reference value 50.

(33)

  • Ich glaube, dass den Bericht der Chef in seinem Büro gelesen hat.

  • I think that the.ACC report the.NOM boss in his office read has

  • ‘I think that the boss read the report in his office.’

The reference sentence in (33) is adapted from the reference sentence that Keller (2000) used for his Experiment 2. The sentence is grammatical but exhibits scrambled (i.e., noncanonical) word order. In the absence of any discourse licensing of the object-before-subject order, sentences like (33) are usually perceived as somewhat degraded (see Keller 2000, Pechmann et al. 1994). The experimental sentences occurred below the reference sentence, each with a box to its left in which the participants were instructed to write their numerical judgment. The first eleven sentences were filler items.

4.1.2 Results

All analyses were run using R (R Core Team 2012). Before analysis, the ratings obtained in Experiment 1 were normalized by applying the z-transformation. Each participant’s mean rating was subtracted from each individual rating, and the resulting value was divided by the participant’s standard deviation. Positive values indicate ratings above the average value; negative values indicate ratings below the average. The averaged normalized ratings obtained in this way are shown in figure 1. They were analyzed by means of linear mixed-effect modeling using the R-package lme4 (Bates, Maechler, and Bolker 2012). The experimental factors were coded in such a way that all contrasts reported below test whether differences between means are significant. Following the advice given in Barr et al. 2013, we first computed a model containing the full factorial design in the random slopes. Since this model did not converge, we simplified the random effect specification by removing the three-way interaction from the random item factor. Table 2 shows the estimated contrasts, their standard error, and the resulting t-value. Since an exact p-value cannot be computed, we consider contrasts with an absolute t-value greater than 2 as significant (see Barr et al. 2013).

Figure 1

Acceptability ratings in Experiment 1 (z-scores)

Figure 1

Acceptability ratings in Experiment 1 (z-scores)

Table 2

Summary of fixed effects in the mixed-effects model for Experiment 1

EstimateSEt-value
(Intercept) .01 .045 0.19 
Movement (short vs. long) .43 .090 4.82 
Particle (damals vs. denn−.14 .046 − 3.15 
ParticlePosition (main vs. embedded) .16 .038 4.14 
Movement∗Particle −.44 .090 − 4.89 
Movement∗ParticlePosition .56 .069 8.09 
Particle∗ParticlePosition .47 .066 7.03 
Movement∗Particle∗ParticlePosition −.29 .125 2.29 
EstimateSEt-value
(Intercept) .01 .045 0.19 
Movement (short vs. long) .43 .090 4.82 
Particle (damals vs. denn−.14 .046 − 3.15 
ParticlePosition (main vs. embedded) .16 .038 4.14 
Movement∗Particle −.44 .090 − 4.89 
Movement∗ParticlePosition .56 .069 8.09 
Particle∗ParticlePosition .47 .066 7.03 
Movement∗Particle∗ParticlePosition −.29 .125 2.29 

Table 2 shows that all main effects and all interactions are significant. To explore the data in more detail, we first computed separate statistical models for sentences with short and sentences with long wh-movement. The results of the two-factor mixed-effect models are shown in table 3; the exact acceptability scores and t-values for pairwise comparisons are given in table 4.

Table 3

Summary of the fixed effects in the mixed-effects models for Experiment 1, separately for sentences involving short wh-movement and sentences involving long wh-movement

Short wh-movementLong wh-movement
EstimateSEt-valueEstimateSEt-value
(Intercept) .23 .067 3.35 −.21 .061 −3.40 
Particle −.37 .084 −4.35 .08 .036 2.14 
ParticlePosition .43 .057 4.67 −.12 .045 −2.67 
Particle*ParticlePosition .61 .101 6.07 .33 .066 4.95 
Short wh-movementLong wh-movement
EstimateSEt-valueEstimateSEt-value
(Intercept) .23 .067 3.35 −.21 .061 −3.40 
Particle −.37 .084 −4.35 .08 .036 2.14 
ParticlePosition .43 .057 4.67 −.12 .045 −2.67 
Particle*ParticlePosition .61 .101 6.07 .33 .066 4.95 
Table 4

z-scores in Experiment 1 and t-values in mixed-effects models for each contrast (t-values in italics)

damalsdennt-value
Short wh-movement 
damals/denn in main clause .46 .42 .80 
damals/denn in embedded clause .35 −.33 5.74 
t-value 1.80 9.25  
Long wh-movement 
damals/denn in main clause −.39 −.15 4.58 
damals/denn in embedded clause −.11 −.19 1.54 
t-value 5.92 .55  
damalsdennt-value
Short wh-movement 
damals/denn in main clause .46 .42 .80 
damals/denn in embedded clause .35 −.33 5.74 
t-value 1.80 9.25  
Long wh-movement 
damals/denn in main clause −.39 −.15 4.58 
damals/denn in embedded clause −.11 −.19 1.54 
t-value 5.92 .55  

Figure 1shows an interaction between Particle and Particle Position for both types of movement. This interaction is significant in both cases (see table 3). The source of the interaction, however, is very different for the two types of movement. For sentences with short wh-movement, pairwise comparisons (see table 4) show that the acceptability of sentences with damals does not depend on the position of the adverb (|t| < 2). For sentences with denn, in contrast, the position of the particle has a strong effect (|t| > 2). When denn is contained in the main clause proper, it is as acceptable as damals. When denn is part of the embedded clause, however, acceptability is much reduced, not only compared with denn in the main clause but also compared with damals in the embedded clause.

For sentences with long wh-movement, a rather different picture emerges. For these sentences, the position of the particle has no significant effect on denn but does have a significant effect on damals. When damals is contained in the main clause, acceptability is reduced. The drop in acceptability is, however, not as strong as for denn in the embedded clause of sentences with short wh-movement. When contained in the embedded clause, denn and damals do not differ, whereas acceptability of damals is reduced in the main clause.

Turning finally to the two types of movement, we see that long wh-movement sentences are judged as less acceptable than short wh-movement sentences, with one exception: when they contain denn in the embedded clause, short wh-movement sentences are less acceptable than long wh-movement sentences.

To summarize, Experiment 1 shows a general penalty for long wh-movement and, more importantly, interactions of Movement, Particle, and Particle Position. The temporal adverb damals is disliked in the main clause of sentences with long wh-movement, while the DiP denn is disliked in the embedded clause of sentences with short wh-movement.

4.1.3 Discussion

Let us first explain the result that the temporal adverb damals in sentences with long wh-movement is less acceptable in the main clause than in the embedded clause. We think there are two reasons for this unexpected effect. First, the presence of the temporal adverb affects the bridge property of the attitude verbs that were used in this condition. Long wh-movement is maximally unmarked in the context of pure bridge verbs (see Erteschik-Shir 1973). The temporal adverb reduces the bridge property of the matrix predicate—but so does the DiP denn. This brings us to the second reason. Obviously, it is semantically sound but in the absence of an appropriate context nevertheless poorly motivated to modify an attitude predicate with damals; why should someone think or believe back then (instead of right now) that p? Although merger of denn affects the bridge property in a comparable way, it does not have a comparable semantic effect.19 In the short wh-movement condition, the matrix verbs were mainly verbs of communication, that is, eventive verbs. Neither merger of the adverb damals nor merger of the particle denn is expected to cause any awkwardness. The reduced acceptance of the adverb damals in the main clause of sentences with an epistemic matrix verb and long wh-movement must therefore be seen as a confound due to the way in which the materials for Experiment 1 were constructed.

The most important finding of Experiment 1 is the asymmetry observed for denn. In root clauses, denn is as acceptable as damals. This is expected because root clauses have a layer of structure that encodes illocutionary force. In dependent nonroot clauses, denn received substantially lower ratings, and clearly so in clauses that lacked interrogative force.20 Crucially, however, denn is more acceptable in dependent clauses from which wh-movement has taken place. Our claim is that the licensing of denn is achieved by the intermediate trace/copy of wh in Spec,CP. If this reasoning is on the right track, the acceptability of denn provides independent evidence for the creation of intermediate traces/copies in cyclic movement.21

Before accepting this conclusion, we must address a potential complication of Experiment 1. This complication has to do with the fact that long wh-movement across the complementizer dass is not a preferred option in many varieties of German. As already noted by Paul (1919: 321ff.), who speaks of Satzverschlingungen ‘sentence intertwinings’, there are regional differences with respect to the acceptability of extraction from dass-complements (see Kvam 1983, Lühr 1988). The facts are still not totally clear. Prescriptive grammar bans it, while in practice it appears here and there. It seems to be more current in southern German dialects and regiolects than in varieties spoken in northern Germany. In a sentence completion study, Fanselow and Weskott (2010) found that speakers from southern regions in fact produced more long wh-extractions than speakers from northern regions. However, it is doubtful that this difference rests on a parameter.22

Participants in Experiment 1 had a bias against wh-extraction out of dass-clauses. Acceptance of long wh-extraction was relatively low. We might therefore wonder whether the lack of an additional penalty for denn in the embedded clause of long wh-extraction sentences is just a floor effect. Two arguments speak against this possibility. First, in an ME experiment it is always possible to assign a value below the lowest values assigned so far. Second, lower values were assigned in several cases. The mean ratings in two conditions are lower than the mean ratings for long wh-extractions with denn in the embedded clause: long wh-extractions with damals in the main clause and short wh-extractions with denn in the embedded clause. Furthermore, many of the ungrammatical fillers received z-scores lower than the mean for sentences with long whmovement and denn in the embedded clause. Hence, we can exclude that the lack of a position effect for denn in sentences with long wh-movement is a floor effect.

Fortunately, German has an alternative construction that yields interpretive effects very similar to (albeit not the same as) overt extraction but does not involve long wh-extraction: partial movement, as illustrated in (34) (see Fanselow to appear, Lutz, Müller, and von Stechow 2000, and section 5.3 below).

(34)

  • Was meint Gerda, wem wir vertrauen können?

  • what thinks Gerda who.DAT we trust can

  • ‘Who does Gerda think that we can trust?’

Partial movement is “incomplete” in the sense that the wh-phrase does not reach the actual scope position, which for s-selectional reasons must be outside the scope of the attitude verb meinen ‘to think’. In (34), wem ‘who.DAT’ moves, but it moves only to the clause-initial position of the embedded clause. However, it is interpreted with scope over the matrix clause—that is, as if occurring in Spec,CP of the matrix clause. Nevertheless, DiPs appear to be licit in the dependent clause of partial movement constructions.

(35)

  • Was meint Gerda, wem wir {denn, nur, schon} vertrauen können?

  • what thinks Gerda who.DAT we DENNNURSCHON trust can

The wh-phrases in (34) and (35) are associated with each other. Was in the upper clause marks the scope of the wh-phrase in the lower clause. The exact nature of the dependency is a matter of debate. The debate about this construction is dominated by different approaches, to which we will turn in section 5.3. What matters at this point is the partial similarity with overt movement. Q-sensitive DiPs seem to be licensed in CPs whose actual interrogative force—as signaled by wh-scope—is absorbed by the root clause. Importantly, this is achieved without overt wh-movement.

4.2 Experiment 2

Experiment 2 replaces long wh-extraction from Experiment 1 by partial movement, as shown in (36). This avoids the complications of long wh-movement that could have jeopardized Experiment 1. Apart from this, the experimental design and materials are identical with the ones in Experiment 1.

(36)

  • a.

    Was vermutete er denn, weni die Polizei ti festgenommen hat?

    what suspected he DENN who.ACC the police arrested has

    ‘Who did he suspect that the police arrested?’

  • b.

    Was vermutete er, weni die Polizei denn ti festgenommen hat?

    what suspected he who.ACC the police DENN arrested has

    ‘Who did he suspect the police arrested?’

4.2.1 Method

The participants, materials, and procedure for Experiment 2 were as follows.

4.2.1.1 Participants

One hundred and two students at the University of Konstanz completed the questionnaire for course credit or payment. All participants were native speakers of German and naive with respect to the purpose of the experiment.

4.2.1.2 Materials

The sentence material for Experiment 2 was derived from the sentence material of Experiment 1. Experiment 2 had the same three-factor design as Experiment 1. All 16 sentences from Experiment 1 involving short wh-movement were imported into Experiment 2 without any change. In the other 16 sentences, long wh-movement was replaced by partial movement. As in Experiment 1, each of the 32 sentences occurred in four versions according to the two within-item factors Particle and Particle Position. A complete stimulus sample is given in table 5.23

Table 5

Sample stimuli from Experiment 2

  • Short wh-movement, particle/adverb in main clause

  • Wer berichtete ihr damals/denn, dass die Einbrecher gefasst wurden?

  • who told her back.then/DENN that the burglars caught were

  • ’Who told her back then/DENN that the burglars were caught?’

 
  • Short wh-movement, particle/adverb in embedded clause

  • Wer berichtete ihr, dass die Einbrecher damals/denn gefasst wurden?

  • who told her that the burglars back.then/DENN caught were

  • ’Who told her that the burglars were caught back then/DENN?’

 
  • Partial wh-movement, particle/adverb in main clause

  • Was vermutete er damals/denn, wen die Polizei festgenommen hat?

  • what suspected he back.then/DENN who.ACC the police arrested has

  • ’Who did he suspect back then/DENN that the police arrested?’

 
Partial wh-movement, particle/adverb in embedded clause Was vermutete er, wen die Polizei damals/denn festgenommen hat? what suspected he who.ACC the police back.then/DENN arrested has ’Who did he suspect that the police arrested back then/DENN?’ 
  • Short wh-movement, particle/adverb in main clause

  • Wer berichtete ihr damals/denn, dass die Einbrecher gefasst wurden?

  • who told her back.then/DENN that the burglars caught were

  • ’Who told her back then/DENN that the burglars were caught?’

 
  • Short wh-movement, particle/adverb in embedded clause

  • Wer berichtete ihr, dass die Einbrecher damals/denn gefasst wurden?

  • who told her that the burglars back.then/DENN caught were

  • ’Who told her that the burglars were caught back then/DENN?’

 
  • Partial wh-movement, particle/adverb in main clause

  • Was vermutete er damals/denn, wen die Polizei festgenommen hat?

  • what suspected he back.then/DENN who.ACC the police arrested has

  • ’Who did he suspect back then/DENN that the police arrested?’

 
Partial wh-movement, particle/adverb in embedded clause Was vermutete er, wen die Polizei damals/denn festgenommen hat? what suspected he who.ACC the police back.then/DENN arrested has ’Who did he suspect that the police arrested back then/DENN?’ 

The 32 sentence quadruples were distributed over four lists according to a Latin square design. Each list was then randomized twice, and the resulting eight lists were interspersed in longer lists containing items from an unrelated experiment on argument structure alternations and the same 40 genuine filler items used in Experiment 1. Overall, each questionnaire contained 112 sentences.

4.2.1.3 Procedure

The same procedure was used as in Experiment 1. Acceptability judgments were again obtained via a questionnaire, using the ME method. The reference sentence was the same as in Experiment 1, but this time it was assigned a value of 10, which was also the value for the reference line in the preceding training. The choice of the reference value (10, rather than 50 in Experiment 1) did not affect the results.24

4.2.2 Results

As for Experiment 1, we used linear mixed-effects modeling for further statistical analyses. Again, we first computed a model containing the full-factorial design in the random slopes. Since this model failed convergence, we simplified the random effect specification stepwise by removing interactions from the random factors. The maximal converging model contains the addition of the three experimental factors in both the random subject factor and the random item factor. The model summary in table 6 shows main effects of Particle and Particle Position but no general effect of Movement (short vs. partial). The factor Movement interacts, however, with both Particle and Particle Position. The interaction of Particle and Particle Position is also significant. Unlike in Experiment 1, the interaction involving all three factors (Movement, Particle, and Particle Position) failed to reach significance in Experiment 2.

TABLE 6

Summary of fixed effects in the mixed-effects model for Experiment 2

EstimateSEt-value
(Intercept) .23 .040 5.72 
Movement (short vs. partial) −.04 .076 −.47 
Particle (damals vs. denn−.21 .039 − 5.36 
ParticlePosition (main vs. embedded) .30 .039 7.77 
Movement∗Particle −.33 .070 − 4.64 
Movement∗ParticlePosition .45 .070 6.32 
Particle∗ParticlePosition .77 .049 15.79 
Movement∗Particle∗ParticlePosition .11 .098 1.12 
EstimateSEt-value
(Intercept) .23 .040 5.72 
Movement (short vs. partial) −.04 .076 −.47 
Particle (damals vs. denn−.21 .039 − 5.36 
ParticlePosition (main vs. embedded) .30 .039 7.77 
Movement∗Particle −.33 .070 − 4.64 
Movement∗ParticlePosition .45 .070 6.32 
Particle∗ParticlePosition .77 .049 15.79 
Movement∗Particle∗ParticlePosition .11 .098 1.12 

Figure 2 reveals that something different is going on in sentences with short wh-movement and sentences with partial movement. To explore this difference, we computed separate models for sentences with these two types of movement (see table 7).

Figure 2

Acceptability ratings in Experiment 2 (z-scores)

Figure 2

Acceptability ratings in Experiment 2 (z-scores)

For sentences with short wh-movement, the model (left part in table 7) indicates main effects of Particle and Particle Position as well as a significant interaction of the two factors. As in Experiment 1, the main effects have to be qualified by the interaction. Pairwise comparisons within a mixed-effects model (see table 8) show that the main effect of Particle (higher acceptability scores for damals) is driven by sentences in which damals/denn occurs in the embedded clause. The main effect of Position is driven by sentences containing the DiP denn. Only in this case is acceptability higher when the DiP occurs in the main clause than in the embedded clause. Taken together, acceptability is high in three conditions: damals in both the main and embedded clauses, and denn in the main clause. These three conditions do not differ significantly from each other. When denn occurs in the embedded clause, however, we see a striking decrease in acceptability.

TABLE 7

Summary of the fixed effects in the mixed-effects models for Experiment 2, separately for sentences involving short wh-movement and sentences involving partial wh-movement

Short wh-movementPartial wh-movement
EstimateSEt-valueEstimateSEt-value
(Intercept) .21 .066 3.24 −.25 .044 5.69 
Particle −.37 .054 − 6.79 −.05 .049 −.95 
ParticlePosition .53 .065 8.05 .08 .042 1.87 
Particle∗ParticlePosition .83 .110 7.57 .72 .090 7.98 
Short wh-movementPartial wh-movement
EstimateSEt-valueEstimateSEt-value
(Intercept) .21 .066 3.24 −.25 .044 5.69 
Particle −.37 .054 − 6.79 −.05 .049 −.95 
ParticlePosition .53 .065 8.05 .08 .042 1.87 
Particle∗ParticlePosition .83 .110 7.57 .72 .090 7.98 
TABLE 8

z-scores in Experiment 2 and t-values in mixed-effects models for each contrast (t-values in italics)

damalsdennt-value
Short wh-movement 
damals/denn in main clause .45 .42 .88 
damals/denn in embedded clause .33 −.44 8.10 
t-value 1.10 14.24  
Partial wh-movement 
damals/denn in main clause .13 .44 4.30 
damals/denn in embedded clause .40 .00 6.25 
t-value 5.50 _6.03  
damalsdennt-value
Short wh-movement 
damals/denn in main clause .45 .42 .88 
damals/denn in embedded clause .33 −.44 8.10 
t-value 1.10 14.24  
Partial wh-movement 
damals/denn in main clause .13 .44 4.30 
damals/denn in embedded clause .40 .00 6.25 
t-value 5.50 _6.03  

Sentences with partial movement exhibit a different pattern. The corresponding model (right part in table 7) attests a significant interaction of Particle and Particle Position but no main effects. The acceptability of damals is higher in embedded clauses than in the main clause. In contrast, the acceptability of denn is higher when denn is a proper part of the main clause. In the main clause, the acceptability of denn is higher than the acceptability of damals, whereas the reverse is true when the DiP or the adverb, respectively, is contained in the embedded clause.

In summary, the results of Experiment 2 confirm the results of Experiment 1. Overall, denn is less acceptable in embedded clauses, but acceptability increases strongly when the left periphery of the embedded clause contains a wh-element whose scope is linked to the root clause.

4.3 Discussion

Experiment 2 overcomes a potential confound in Experiment 1. Substituting partial movement for long wh-movement, which some speakers of German dislike, increased acceptability in general. Overall, sentences with partial movement are as acceptable as sentences with short wh-movement. The crucial difference between the two clause types concerns the acceptability of denn in embedded clauses. Sentences with short wh-movement received a strong penalty when denn occurred in the embedded clause, whereas the penalty was reduced in sentences with partial movement. Hence, Experiment 2 attests the same asymmetry as the one found in the long wh-movement condition of Experiment 1. Although partial movement, in contrast to long wh-movement in Experiment 1, does not remove the penalty for denn in a dependent clause, it clearly reduces the penalty.

As in Experiment 1, the acceptability of damals is degraded when damals occurs in the main clause of a sentence with short wh-movement. Unlike what is observed for denn, the penalty for damals is quite constant across Experiments 1 and 2—that is, comparable for long wh-movement and partial movement. We attribute this difference to the nature of the violation. While the licensing of denn depends on the syntactic environment, the acceptability of damals depends on the semantic environment. We can generalize that denn is sensitive to the syntactic difference between long wh-movement and partial movement whereas damals is not.

Taken together, Experiments 1 and 2 show that the DiP denn results in degraded acceptability when occurring in a dependent clause. Acceptability increases when a wh-phrase has been extracted from that clause (long wh-movement in Experiment 1) or achieves matrix scope via association with the scope-marking wh-pronoun was (partial movement in Experiment 2). In section 5, we propose how to capture this finding in syntactic theory.

5 Licensing Discourse Particles in Wh-Questions

5.1 Local Licensing

Fundamental to our syntactic explanation of the facts considered so far is the assumption of feature sharing between a clause-type-specific Force head and a matching DiP. A Force head with the interpretable feature iQ can probe in its local domain for a DiP with a corresponding unvalued and uninterpretable feature uQ. According to Bayer (2012) and Bayer and Obenauer (2011), denn is a functional element that heads a Particle Phrase (PrtP) above vP. The relevant phrase structure is shown in (37). Prt is often preceded by topics (the aboutness topic as well as discourse topics), but—for reasons not well understood so far—the space between Force and Prt may also remain empty.

(37) [ForceP Force [... [PrtP Prt [vP . . . ]]]]

It is the Force feature that is interpretable, not the corresponding feature on Prt. This is one of the reasons why we adopt a feature-sharing theory rather than the standard Minimalist checking theory. In standard checking theory, the uninterpretable feature is always in the probe, while the goal bears an interpretable feature that values and ultimately deletes the probe’s uninterpretable feature. This sort of directionality is abandoned in the feature-sharing version of Agree proposed by Pesetsky and Torrego (2007).25 The novelty of their system is that an interpretable (i) feature can be unvalued, signaled by empty square brackets [ ], and an uninterpretable (u) feature can be valued, signaled by some arbitrary number in square brackets (e.g., [4]). With this in mind, an uninterpretable Q-feature on Prt will be able to value an unvalued interpretable Q-feature on Force.

(38)

graphic

The unification account requires featural agreement according to which the same feature is present in both locations. In (38), this is expressed by an arbitrary shared value, here 4. It is important to see that the particle remains exactly where it is merged. It does not change its scope in the course of the derivation as it would do if it underwent LF movement or feature movement. Instead, the effect of agreement is that the particle gains access to Force at a distance and therefore can give rise to the fine-tuning of Force without becoming a phrase-structural subconstituent of Force. As Bayer and Obenauer (2011) show, wh-questions can be turned into special questions such as RQs, surprise-disapproval questions, and CfvQs through merger of different Q-sensitive DiPs in interrogative clauses. Force is prima facie neutral with respect to these shades of meaning. Given that questions can be interpreted in the absence of DiPs, interrogative force must be represented in some way. The observed semantic enrichment is due to a featural link, the Q-feature in (38), which establishes compositionality between Force and the particle. We assume with Rizzi (1997, 2004) and following work that Force is part of the architecture of the C domain and that it is the topmost layer of the clause. Assuming with Chomsky (2000) that CP and vP are phases and that the Phase Impenetrability Condition holds, probing from Force to Prt will not induce any locality violation.26

5.2 Local Licensing in Embedded Clauses

The challenging question is how to account for DiPs in dependent clauses. The experiments reported in section 4 have shown that in wh-questions in which wh and the particle occur in different clauses, the uninterpretable Q-feature on the particle cannot value its counterpart on the unvalued interpretable Force head unless wh-movement has taken place from the complement clause. The same is trivially true for any type of island; see footnote 15. Obviously, long wh-movement via Spec,CP leaves a representation of Q-Force in the intermediate Spec,CP that can be valued by the corresponding Q-feature on the particle without any violation of the Phase Impenetrability Condition. Sentence (9), repeated in (39) for convenience, is then analyzed as in (40).

(39)

  • Wie denkst du, dass es denn weitergehen soll mit euch?

  • how think you that it DENN go.on should with you.PL

  • ‘How do you think it should go on with the two of you?’

(40)

graphic

The Q-Force feature on denn can be probed by Q-Force thanks to the intermediate representation of Q-Force that has been established by cyclic wh-movement. The question is how Q-Force can be established in the intermediate Spec,CP given that (a) the matrix verb is a propositional attitude verb that disallows interrogative interpretation of its complement, and (b) the embedded CP seems to fail as a representative of the relevant speech act altogether.

Pesetsky and Torrego (2007) suggest for long wh-movement that an uninterpretable feature uQ[ ] on C is valued by an uninterpretable [+interrogative] wh-element. Thus, C and therefore its projection will be “[+wh]” but without any interpretive consequence. Otherwise, the CP could not remain in the scope of the verb denken ‘to think’. If this is correct, it is easy to see how the particle in the dependent clause can locally agree by virtue of its uninterpretable Q-feature. We propose that the Q-feature on Prt values an uninterpretable Force head. Of course, this head could in principle also be interpretable. This would be the case under merger of a verb like fragen ‘to ask’ as in (41), where CP counts as an embedded interrogative speech act.

(41)

  • Jemand fragte [CP wie [C [es [PrtP denn wie weitergehen soll mit euch]]]].

  • someone asked how it DENN go.on should with you.PL

  • ‘Someone asked how it should go on with the two of you.’

Since there is no lookahead device, there can be no a priori determination of the interpretability of Q-Force that goes beyond what is in the relevant numeration. This is exactly what the liberalized framework of probe-goal agreement guarantees. Since interpretability and valuation are separated, there can be agreement between two uninterpretable features; that is, an unvalued uninterpretable feature of the probe can be valued by an uninterpretable feature of the goal.27

As the controller of the DiP, Bayer (2012) and Bayer and Obenauer (2011) suggest a single Force head that involves a specification of clause type and a related feature for illocutionary force. Here, we follow Coniglio and Zegrean’s (2012) proposal of splitting up Force into Clause Type (CT) and Speech Act (SA).28 With this atomization of Force in mind, consider the derivation of (9)/(39) in which the embedded clause has CT but lacks SA.29 The particle denn has the interrogative CT-feature Q and the SA-feature ILL(ocution). Both are uninterpretable, their interpretable correspondents being located in the root clause’s left periphery. The complementizer heads a CP; the moved finite verb heads a FinP. For reasons of readability, we omit derivational steps that do not relate to the issue under discussion.

(42)

  • a.

    [vP wie [vPwie weitergehen soll mit euch]] →→ Merge denn

  • b.

    [PrtP dennuQ[ ], uILL[ ] [vP wie [vPwie weitergehen soll mit euch]]] → Move wh

  • c.

    [CTP CTuQ[ ], uILL[ ] [CP wie dass [TP es [PrtP dennuQ[ ], uILL[ ] [vPwie[vPwie. . . weitergehen . . . ]]]]]] → Agree →

  • d.

    [CTP CTuQ[1], uILL[2] [CP wie dass [TP es [PrtP dennuQ[1], uILL[2] [vPwie [vPwie. . . weitergehen . . . ]]]]]] → . . . → Move wh

  • e.

    [SAP SAuQ[ ], iILL[ ] [CTP CTiQ[ ], uILL[ ] [FinP wie [Fin′ denkst [TP du [CTP CTuQ[1], uILL[2] [CPwie dass [TP es [PrtP dennuQ[1], uILL[2] [vPwie [vPwie. . . weitergehen . . . ]]]]]]]]]]] → Agree →

  • f.

    [SAP SAuQ[1], iILL[2] [CTP CTiQ[1], uILL[2] [FinP wie [Fin′ denkst [TP du [CTP CTuQ[1], uILL[2] [CPwie dass [TP es [PrtP dennuQ[1], uILL[2] [vPwie [vPwie . . . weitergehen . . . ]]]]]]]]]]]

The important points are these: In (42d), the particle denn is locally licensed by agreement for CT as well as for SA. Nevertheless, both CT and SA are uninterpretable because the numeration contains the verb denken ‘to think’, and denken bans a semantically interpretable wh-complement in its scope. On the other hand, denn is locally linked and therefore stays available during further derivational steps by which wh moves higher up. In (42e) (i.e., at the root level), CT is interpretable for Q; SA, which shares the feature Q with CT, provides the features that are relevant for interpreting (41) as an utterance. The link between interpretable Force (i.e., SA) and the particle is local in the sense of phase theory and strict cyclicity. Cyclic agreement with the particle’s CT/ILLfeatures comes about as a consequence of cyclic wh-movement and its feature for interrogativity, be it interpretable at a certain stage of the derivation or not (yet). In (42f), the uninterpretable CT-feature on SA is valued by CT’s interpretable Q-feature.30

Den Dikken (2009) argues that wh-movement to Spec,CP is terminal and can never be transient. If we keep to the split-CP theory according to which Force is represented in the top left clausal periphery, however, we can hardly avoid the conclusion that Force is established via Spec,CP and that the wh-phrase, having passed through Spec,CP, is the overt sign of Q-Force. Q-Force, the composition of Q and ILL in (42), is the locally available probe that agrees with the Q/ILL-feature on the particle denn. Assuming that vP is a phase, Den Dikken’s proposal is that the wh-phrase passes through the vP phase but not through the CP phase; however, vP does not qualify as the locus of force. Given the structure in (37) with the particle projection above vP, Force would not even c-command its goal. Therefore, we see no reason to deviate from the standard assumption of a CP-related intermediate landing site.31

As the experimental results from section 4 have established, licensing of the particle in a dependent clause by means of long wh-movement contrasts strikingly with ungrammatical examples like (43).

(43)

  • *Wer berichtete wer ihr [CP dass die Einbrecher [PrtP denn gefasst wurden]]?

  • who told who her that the burglars DENN caught were

Probe-goal agreement provides a straightforward explanation. Wh-movement in these cases is confined to local movement within the matrix clause, while the Q-sensitive particle denn is in the embedded clause. It does not matter whether the embedded CP has a Force projection or not. If it lacks such a projection altogether, the uninterpretable Q-feature on denn cannot be valued, and the derivation crashes. But even if the embedded CP did have a Force projection, it would be unsuitable for valuing the Q-feature on denn, because the CP would at best have declarative force.32 Nevertheless, as the derivation of (42) shows, the embedded CP must have seen an uninterpretable interrogative CT-feature in the CP layer. This feature is obviously responsible both for wh-movement to Spec,CP and for probing the Q-feature on the particle denn.

5.3 Local Licensing in Partial Movement

The established term partial movement subsumes a number of rather different proposals.33

The direct dependency approach (DDA) (see, e.g., Manetta 2011, McDaniel 1989, Müller 1997, Van Riemsdijk 1983) assumes a syntactic link between a wh-scope marker and the locally moved wh of the dependent clause. The most common version of the DDA assumes that the scope marker is an expletive that is overwritten at LF by the wh-phrase of the subordinate clause.

The indirect dependency approach (IDA) (see Dayal 1994 and much following work) assumes that the wh-scope marker is actually the direct object of the matrix clause. What is coindexed with the wh-scope marker is not the locally moved wh of the dependent clause but the whole wh-CP. In principle, examples like (36) could be simply two sequential wh-clauses: What did he suspect? Who did the police arrest? As indicated in German by V-final word order and binding facts, however, the second clause must be subordinate. On the basis of properties of Hungarian that are not at issue here, Horvath (1997) suggests that the scope marker is the direct wh-object of the matrix verb and that the dependent wh-clause undergoes LF pied-piping and adjoins to the wh-object. Wide scope results from wh-agreement.34Dayal (1994) and various linguists who follow her lead likewise assume that the wh-scope marker is the direct object of the matrix clause, but they propose a purely semantic account of the IDA. The idea is that the wh-object, an existential quantifier ranging over propositions, is semantically restricted by the embedded wh-clause. The latter, being an “open” proposition, is of a matching type, but it is not a question. Consider (34), Was meint Gerda, wem wir vertrauen können? (34) consists not of two questions but of one. A natural answer would be Karl. It translates into the following Hamblin/Karttunen-style logical representation: ppq[qx[q=ˆwe cantrust x]& p=ˆthinks (Gerda, q)]. By virtue of asking what Gerda thinks (p), the speaker asks about the set of persons x for whom the proposition (q)—namely, that we can trust x—is true.

Of the two approaches, the DDA is immediately compatible with our finding about DiPs in partial movement constructions. Since the DDA is more or less an LF variant of overt wh-movement, an Ā chain is built between the purported expletive was and the locally moved wh-phrase. The CT of the dependent clause is interrogative and probes the Q-sensitive DiP denn. Being selected by an attitude verb, neither CT nor ILL can be interpretable. Both are interpretable in the root clause (see (42e–f)). It is therefore predicted that the DiP enters the expanded semantic composition of Force in the same way as it does under overt wh-movement.

Within the IDA, nonsubordination can be excluded, not only because of the subordination-indicating V-final word order of the lower CP and other facts but also because Experiment 2 has demonstrated that the DiP in the second CP is, to a certain extent, degraded. Such an asymmetry would be unexpected if the second clause were simply another root wh-question. Assuming syntactic dependency, the IDA must express the relation between interpretable Q-Force in the root clause and the DiP in the dependent CP. According to our proposal, the dependent CP has a CT head with uninterpretable features for CT and ILL. These are responsible for the local valuation of the corresponding features on the DiP. In partial movement constructions, the dependent CP is overtly of the proper CT, so there can be no doubt about the local licenser of the DiP. The next issue is how to reach an interpretable licenser. Horvath-style CP-adjunction to was could do the job, but the operation may not be available under restrictive Minimalist assumptions. Within the semantic account of the IDA, the embedded CP cannot be a question. Given that in our account neither the CP’s CT nor its ILL is interpretable, the presence of their uninterpretable counterparts cannot pose a semantic obstacle. If this is correct, the logical form of a version of (34) with denn in the embedded CP, Was meint Gerda, wem wirdennvertrauen können?,is blind to the fact that the DiP ultimately requires an interrogative context. Assume now that because of its wh-feature, the wh-CP may be probed by was before it moves to its ultimate destination. Full interpretation is, however, only achieved when CT/ILL become interpretable in the SAP.35 If we are right, our explanation for the occurrence of DiPs in embedded clauses carries over to accounts of partial movement in terms of the IDA.36

Recall that partial movement as investigated in Experiment 2 is but one alternative to overt movement. As mentioned in section 3, German has yet another scoping strategy: copy movement. Copy movement, illustrated in (30), differs in various respects from partial movement (see Felser 2004, Pankau 2013). The copies are something like the featural spell-out of the locally Ā moved wh-phrase. The semantic mechanics of the IDA are not applicable. Nevertheless, denn in the dependent clause of (44) appears to us as unproblematic as it is in partial movement.37

(44)

  • Wen vermutete er, wen die Polizei denn festgenommen hat?

  • who.ACC suspected he who.ACC the police DENN arrested has

  • ‘Who did he suspect the police arrested?’

Uninterpretable CT in the lower CP probes the DiP. Interpretable CT and SA of the root clause agree with the relevant features of the embedded CP. Thus, it should be obvious that the DiP connects to the root question and joins the composition of illocutionary force. Since copy movement was not investigated in the present study, we leave the issue here.38

5.4 Predicting More Data

Since long wh-movement is recursively iterable depending on how much structure is built, it is expected that DiPs that depend on interrogative force can show up in any pre-vP position along the chain that spans the space between the matrix wh and the lowest copy. An example is given in (45a), with the simplified analysis in (45b).

(45)

  • a.

    Wie denkst du, dass Max denn meint, dass es weitergehen soll mit euch?

    how think you that Max DENN thinks that it go.on should with you.PL

    ‘How do you think that Max thinks that it should go on with the two of you?’

  • b.

    Wie denkst du [CPwie dass Max [PrtP denn meint [CPwie dass es wie weitergehen soll mit euch]]]?

According to our intuitions, examples of this type are fully grammatical. Under the current account, this is expected since the Q-sensitive particle can be probed from the left periphery of the medial CP.39

Coniglio (2011) and Thurmair (1989) have shown in detail how German DiPs can be combined, and that under combination they usually occur in a fixed order. In combinations of denn with other Q-sensitive particles such as schon, nur, bloß, and wohl, the particle denn always takes scope over the others.

(46)

  • a.

    Wo wird er sich denn schon versteckt haben?

    where will he REFLDENNSCHON hidden have

    ‘Where after all will he be hiding? (There is only one possible place.)’

  • b.

    Wem könnte man denn dieses Auto nur verkaufen?

    who.DAT could one DENN this car NUR sell

    ‘Who on earth could one sell this car to!? (I have no answer so far.)’

Any order in which denn is preceded by another Q-sensitive DiP is excluded: *schon > denn, *nur > denn,*bloß > denn,*wohl > denn. The interesting point in the present connection is that particles can occur in different CPs and that the order is retained. Consider (47a), with the analysis in (47b).

(47)

  • a.

    Wo glaubst du, dass Fritz denn meint, dass man hier schon Benzin kriegt?

    where believe you that Fritz DENN thinks that one here SCHON gasoline gets

    ‘Where do you believe that Fritz thinks after all that one can get gasoline here? (Nowhere/Hardly anywhere!)’

  • b.

    Wo glaubst du [CP wo dass Fritz [PrtP denn meint [CP wo dass man hier [PrtP schon wo Benzin kriegt]]]]?

As (47b) shows, the wh-operator wo moves through the CP cycles, targeting in each cycle a Q-Force head by which a DiP can be licensed. Complex examples in which the particles retain their property as Q-sensitive DiPs are hard to construct. To the extent that constructing such examples is possible, it seems that ordering violations lead to ungrammaticality.

(48)

  • a.

    Wo glaubst du, dass Fritz denn meint, dass man hier nur ein Hotelzimmer

    where believe you that Fritz DENN thinks that one here NUR a hotel.room finden kann?

    find can

    ‘Where do you believe Fritz thinks one can get a hotel room here? (I have so far not found an answer.)’

  • b.

    *Wo glaubst du, dass Fritz nur meint, dass man hier denn ein Hotelzimmer finden kann?

While (48a) allows the CfvQ interpretation, and while (48b) is grammatical without nur, the ordering *nur > denn leads to a crash. The reading of (47a) with schon as the trigger of the RQ interpretation of the question and the reading of (48a) with nur as the trigger of the CfvQ interpretation show that the lower DiP is probed via successive-cyclic wh-movement.

5.5 Scope

DiPs appear under strict scope requirements. The theory that has been developed so far integrates this semantic fact straightforwardly. One classical insight is that DiPs do not move. If they are part of the functional skeleton of the clause, as is the case in our account, this property is captured without stipulations. How do they access what we continue to abbreviate as the “Force layer” (i.e., what is actually decomposed into CT and SA)? We have argued for the adoption of a version of probe-goal agreement. Probe-goal agreement differs from LF movement in important respects. It enables Force to communicate with the Force-sensitive particle without movement. This is what distinguishes agreement from movement. The dependence of DiPs on illocutionary force is uncontroversial. The technical implementation of this dependency is not. For instance, an account in terms of LF movement has been suggested by Zimmermann (2004, 2008). In Zimmermann’s (2008) account of wohl, a DiP that can appear in declaratives as well as in interrogatives, the idea of LF movement rests on the intuition that the strength of the propositional commitment is encoded by ForceP and that the strength can be modified by a suitable DiP. It is then natural to conclude that this modification must be brought about by LF movement. LF movement could be a viable solution if one assumes, as Zimmermann does, that the particle is an adverb (i.e., an XP that can undergo movement) and if the modification of Force is always clause-bound. According to the present account, however, the DiPs under investigation quite consistently display properties of functional heads. Like other functional heads, they are immobile. Bayer (2012) shows that denn naturally turns into a clitic, that is, an X0 element. Furthermore, our data show that Q-sensitive DiPs do arise in dependent clauses, and if they do, they must associate with interpretable Force in the root clause. This leads to a conflict with movement. LF movement across the CP boundary is generally not attested. Thus, accounts in terms of LF movement seem to be in trouble.

Whatever the merits of LF movement may be, we do not even need to go as far as theorizing about its pluses and minuses. We simply need to inspect semantic intuitions. Consider the following minimal pair of RQs, first discussed in Bayer 2011:5.

(49)

  • a.

    Wo glaubst du, dass man hier nachts um 3 Uhr schon Benzin kriegt?

    where believe you that one here at.night at 3 o’clock SCHON gasoline gets

    ‘Where do you believe one can get gasoline here at 3 o’clock in the morning? (Nowhere/Hardly anywhere.)’

  • b.

    Wo glaubst du schon, dass man hier nachts um 3 Uhr Benzin kriegt?

    where believe you SCHON that one here at.night at 3 o’clock gasoline gets

In (49a), the speaker asks about the places x such that the addressee believes that there is a plausibility ranking of places x according to which one can get gasoline in x at 3 o’clock in the morning.40 In (49b), however, the speaker asks about the places x such that there is a plausibility ranking of the addressee’s believing that one can get gasoline in x at 3 o’clock in the morning. Both questions are semantically well-formed, but (49b) is awkward. The reason must be that a plausibility ranking of places with respect to the believe-predicate that is headed by an attitude verb is pragmatically strange. LF movement of X reorders X’s scope. Thus, if the interpretation of the particle schon in (49a) rested on schon taking scope in the root, the interpretations of (49a) and (49b) should be indistinguishable, and (49a) should be as awkward as (49b). But none of this is true. (49a) and (49b) are easily distinguishable. The judgments are clear enough and corroborate the account we have developed above. This demonstrates that the DiP does not move. It is interpreted exactly where we perceive it in PF. Given the ongoing discussion of the contribution of DiPs to the full interpretation of utterances (real speech acts), this is an important clarification. It justifies the claim that the DiP has certain more or less concrete lexical properties that are semantically activated after merger. Further interpretations that have to do with its dependence on Force appear to be largely independent of these lexical properties. The relation to the high left clausal periphery is a matter of formal syntactic licensing by which Force enters a telescoping relation with the particle. As we hope to have shown here, this relation is parasitic on cyclic wh-movement.

6 Summary

Wh-questions constitute the prototypical example for syntactic dependencies that are seemingly unbounded. So far, they have been investigated mainly on the basis of intuitions about displaced and therefore locally missing constituents as definable by argument and event structure. DiPs in German wh-questions offer a fresh look at the topic of transclausal dependency. We have argued that DiPs provide a new kind of evidence for cyclic movement that is only indirectly dependent on displacement. We have shown that seemingly “unbounded” dependencies also control the licensing of DiPs by a syntactic exponent of illocutionary force. Cyclic wh-movement appears to license Q-sensitive DiPs across CP phases. This finding, first reported by Bayer (2012) and Bayer and Obenauer (2011), rested previously on rather infrequent examples in corpora and on the authors’ individual judgments. This article has clarified the theoretically relevant part of these data with the help of two experiments. These experiments showed that German speakers are sensitive to the syntactic environment in which the Q-sensitive particle denn can reshape a question’s illocutionary meaning. Experiment 1 demonstrated that long wh-movement has the effect of connecting the particle across the CP phase to the Force projection of the root clause, while short wh-movement in the root clause fails to do so. The interpretation of Experiment 1 was complicated by the fact that many speakers feel uncomfortable with long wh-movement. Therefore, Experiment 2 explored the same question by replacing overt movement with partial movement. The results replicated the results of Experiment 1 while removing the problem of the bias against long wh-movement. The results of both experiments are explained by a theory in which an interrogative Force feature probes a Q-sensitive subfeature of the DiP denn, and in which this probing can be performed cyclically. The theoretical assumptions underlying this proposal have been made explicit. The Force projection has been argued to be split into projections of Clause Type and Speech Act. Wh-movement to Spec,CP enables the Force features to act as probes of uninterpretable clause type and illocutionary features on the DiP. Although the experiments were limited to the particle denn, it is obvious that other Q-sensitive particles like schon behave accordingly. Thus, the licit appearance of Q-sensitive DiPs in clausal complements must be seen as a novel diagnostic of phase-bound movement. The study shows in addition that parts of “expressive meaning” (see Potts 2005), which appear to be idiosyncratic to some extent, are nevertheless deeply rooted in core grammar and appear to be hardwired in syntax in hitherto unexpected ways.41

Notes

Our research has partly been supported by the Deutsche Forschungsgemeinschaft, grant Ba 1178/8-1. Many thanks to Hans-Georg Obenauer, Marga Reis, Maribel Romero, Marco Coniglio, Hans-Martin Gärtner, Ellen Brandner, Andreas Trotzke, and Yvonne Viesel, as well as to three anonymous reviewers. None of them must be held responsible for any errors or misconceptions. Thanks to Simon Hopp, Uwe Braun, and Marc Meisezahl for technical assistance.

1Denn must be seen as a weakened version of dann with vowel reduction and further reduction to dn or ’n. Semantically, it has been reduced from a temporal adverb to a general deictic element that makes reference to the common ground that the speaker shares or purports to share with the addressee.

2 An apparent exception are examples in which discourse particles form a constituent with a wh-phrase and travel along with it, as in (i).

3 See Meibauer 1994 for discussion of their X-bar status. Cardinaletti (2011) and Coniglio (2011) argue that discourse particles are not heads but “weak adverbs,” a distinction that will not play a role here; see Bayer and Obenauer 2011 and, for different particle constructions, Cable 2010.

4 Opinions about the syntactic representation of illocutionary force differ to some extent, but there is no doubt about its representation in the root clause. See among many others Thurmair 1989 and more recently Zimmermann 2008, where it is proposed that ForceP provides an epistemic reference point for the interlocutors. Krifka (2014) argues for the possibility of embedded speech acts.

5 An artifact of its presence is that questions of type (3) are often perceived as “more intimate” or “friendlier” than those of type (2), but this is really no more than a consequence of the speaker’s attitude. Questions with denn can certainly cater to all kinds of emotions and registers.

6 For examples and discussion, see Bayer 2012, König 1977, and Wegener 2002. One reviewer objects that a stranger who enters one’s private garden may be asked Was machen Sie denn da? ‘What are you doing here?’. However, this question is not contextless. The speaker assumes that the hearer knows that he or she is on someone’s private territory. This common ground suffices to license denn. The particle seems to make reference exactly to this circumstance.

7Schon has other applications as well. For a recent study, see Egg 2012.

8 Analogous examples can be found with the particles schon, nur, and so on, in embedded clauses and with the same restrictions. For details about embedded root phenomena, see Heycock 2005. Neither in these cases nor in those to be considered shortly can one find “double access” to the speaker or to the issuer of the embedded proposition as discussed for German by Meinunger (2004). Meinunger’s focus is on embedded verb-second (V2) clauses. V2 complements play no role in the present study.

9 The rest of this section has benefited from critical questions by one of the reviewers.

10 Examples like (19) are invariably in 2nd person. We could not find any in 3rd person. If this reflects a constraint on polar questions, it must be absent in long wh-movement. The materials used in the experiments reported in section 4 involve 3rd person matrix predicates.

11 Attested examples with wh-extraction are the following:

In an informal classroom judgment study, 132 speakers rated questions with long wh-extraction from complements under the volitional predicate mögen/möchten ‘to want’ (see Meinunger 2006, 2007 for discussion). The particle denn was either in the matrix or in the embedded clause.

  • (iii) An wen1 möchtest du (denn), dass ich deinen Brief (denn) t1 weiterleite? to who.ACC want you DENN that I your letter DENN pass.on ‘Who do you want me to pass your letter on to?’

Comparison was made with short wh-extraction, as in (iv).

  • (iv) Wer1 möchte (denn) t1, dass ich deinen Brief (denn) weiterleite? who wants DENN that I your letter DENN pass.on ‘Who wants me to pass your letter on?’

(iv) yielded a sharp difference between denn in the matrix clause and denn in the embedded clause, whereas the contrast was almost canceled in (iii). This interaction suggests that the difference between long extraction in (8)–(10) and short extraction in (16)–(18) also holds for predicates that cannot be understood as evidential markers.

12Denn in the embedded CP is nevertheless far less frequent than denn in the root CP.

13Simons (2007) discusses comparable cases of the variability of the “main point” of a question.

14 For more recent discussion of VIPs, see Bayer and Salzmann 2013, Pankau, Thiersch, and Würzner 2010, and Viesel 2011, and for experimental explorations, see Kiziak 2010.

15 They are, of course, also unlicensed in arbitrary islands. (i) is an example of an infinitival adjunct; there is no way for denn to access the illocutionary force of the root clause.

(i) *Warum ist Hans [ohne denn einen Führerschein zu haben] Auto gefahren? why is Hans without DENN a driver.license to have car driven ‘Why did Hans drive a car without having a driver’s license?’

16 While it is uncontroversial that a wh-filler must be associated with a trace position in order to satisfy restrictions of Case, argument structure, and so on, it has remained controversial which positions elements cycle through on their way out. There was an early agreement that Spec,CP must be involved (Chomsky 1973). In the Barriers framework (Chomsky 1986), wh-movement also had to pass through a position adjoined to VP. In the Minimalist Program (Chomsky 1995, 2001), this has led to the suggestion that vP is a phase in addition to CP.

17 See Den Dikken 2009 for arguments against this standard view. See also section 5.2 below.

18 For ease of reference, we subsume the particle denn and the adverb damals under the label particle, although the two must not be confused in theory.

19 To see that merger of denn blocks the purely evidential interpretation of the attitude verb, consider slifting and other VIPs (see Reis 2000 and the discussion at the end of section 2.2). Compare (i) with the deviant (ii).

  • (i) Wen hat die Polizei (meinst du) festgenommen (meinst du)?

    who.ACC has the police think you caught ‘Who did the police catch, do you think?’

  • (ii) Wen hat die Polizei (*meinst du denn) festgenommen (*meinst du denn)?

Notice that long wh-movement in cases like (32), Wen vermutete erdenn, dass die Polizei festgenommen hat?, is grammatical. This shows that the matrix predicate survives as a bridge even if it cannot be reduced to an evidential marker; see Simons 2007 for pertinent discussion of English.

20 Recall that formally subordinate clauses—as signaled, for instance, by subjunctive mood—are nevertheless able to have root properties. Reconsider the contrast in (7), where we argued that the matrix predicate provides for a quotational context that reports speech.

21 Notice that the wh-element must move overtly. LF movement, as assumed in previous theories, is insufficient.

  • (i) Wer sagt, dass Peter (*denn) WEN getroffen hat?

    who says that Peter DENN

    who.ACC met has ‘Who says that Peter met who?’

By all criteria, at no stage of the derivation has the embedded CP had an interrogative Force head. The same is true for simplex clauses with wh-in-situ. Although they may be used as some sort of interrogatives (echo or exam questions), they can never license the particle denn. By formal criteria, they seem to lack interrogative force altogether.

  • (ii) Peter hat (*denn) WEN getroffen?

    Peter has DENN who.ACC met

    ‘Peter met who?’

According to Reis (1991) and Trissler (1999), such “questions” fail to be questions in any syntactic sense.

22Wh-extraction can be found in literary texts that are not of southern German origin. For example:

  • (i) Welchen wollet ihr, dass ich euch losgebe? Barrabam oder Jesum, von dem gesagt wird, er sei which.one want you that I you.PL free.give Barrabas or Jesus of whom said is he be Christus?

    Christ

    ‘Which one do you want me to release? Barrabas or Jesus, of whom they say he is Christ?’

    (J. S. Bach, St. Matthew Passion, BWV 244)

23 Because of an error in the materials, two sentences had to be excluded from the analysis.

24 In fact, the choice was made on purpose in order to test the independence of the method from the reference value. Comparisons of the two experiments show that the reference value did not affect the results. The z-scores (for identical items) are comparable across the two experiments. The mean z-scores for the 40 genuine filler items are virtually identical (r = .99). Thus, as expected given that ME requires relative judgments, it did not matter whether participants worked with a reference value of 10 or of 50.

25Zeijlstra (2012) defends a restrictive system in which the interpretable agreeing feature, the goal, is always the upper one. In other words, he proposes the reversal of the probe-goal dependency: probing goes upward. In our account, however, the Q-sensitive particle’s Q-feature is not interpretable and therefore cannot probe an uninterpretable Q-feature in the C domain.

26 This is not to say that there are no problems; see Richards 2012 for discussion. All we want to say here is that, given its fixed pre-vP position, probing the discourse particle from Force would not require deviating from this assumption.

27 Languages with nominal agreement such as Latin and German suggest that this is required in any event.

28Coniglio and Zegrean’s (2012) account is inspired by Haegeman’s (2002, 2006) work and adopts an improved version of the agreement proposal originally suggested by Bayer (2012) and Bayer and Obenauer (2011). There are, of course, alternatives. According to Lohnstein (2007), for example, there is no motivation for a uniform Force projection, as he takes what he calls “sentential force” to be determined by different components of the grammatical system.

29 As already suggested, the embedded CP may also have an SA layer; see the discussion of (7) and (41). In this regard, German offers challenges that are beyond the scope of this article. As (i) shows, wh-interrogatives, in which denn can occur felicitously, can in addition host particles like ja and doch even though the latter are incompatible with interrogative contexts.

  • (i) Wo hast du [DP diesen [AP ja unwahrscheinlich begabten] Pianisten] denn gehört? where have you this JA incredibly gifted pianist DENN heard ‘Where did you hear this indeed incredibly gifted pianist? (I’m wondering.)’

As Struckmeier (2014) and Viesel (to appear) point out, the DP contains an AP with a clause-like structure. This clause obviously follows the default of an assertive clause type and therefore can license a CT-feature on ja. On the other hand, the speaker who takes responsibility for the adequacy of ja is identical with the speaker who takes responsibility for the adequacy of denn. The proposition [incredibly gifted (x)] is presupposed. Given that ja requires an SA, this SA must be linked to the SA of the root clause. For further discussion of DiPs in adnominal and adverbial clauses, see Hinterhölzl and Krifka 2013.

30 One reviewer argues that one could dispense with the CT-feature because every embedded wh-clause has an uninterpretable ILL-feature that can be valued under agreement with the matrix predicate or an interpretable ILL-feature in the root SA. While this may be possible in the core cases, there are many cases in which ILL requires syntactic information from CT. Since German DiPs are formally CT-dependent but at the same time in need of illocutionary interpretation, the independence of CT and ILL seems to be well-motivated. For discussion of their role in German, see Coniglio and Zegrean 2012.

31 Using novel data from the Nilotic language Dinka, Van Urk and Richards (2015) explore which phases are involved in long-distance extraction. They show that both the vP and the CP phase are visibly involved in successive-cyclic movement.

32 Declarative/Assertive appears to be the unmarked value created by the absence of any of the other mood types. As example (i) in footnote 29 shows, the DiP ja can appear in APs that express a predication and can therefore be considered declarative by default.

33 This section has profited from critical questions by two reviewers.

34 See Fanselow and Mahajan 2000 for a general account along these lines that integrates German and Hindi.

35 As it happens, this may explain a long-standing problem: why, in multiple questions, the in-situ was cannot license partial movement (see Bayer 1996:228, Fanselow and Mahajan 2000:206). The reason could be that in its underlying position, was lacks the relevant interpretable features.

36 The use of a DiP does not raise the embedded wh-CP to a higher logical type. Therefore, the extensions of the IDA suggested in Lipták and Zimmermann 2007 and Sternefeld 2001, 2002, suggested by one reviewer, do not seem to be relevant here.

37 Consider the following example from the Internet:

38 In fact, we left copy movement out of the empirical studies because it does not appear to be equally available for all speakers.

39 Example (45) is in fact another example that cannot be accounted for by a nonsyntactic theory according to which the question under discussion is Wie soll es weitergehen mit euch? If it were, the slifting paraphrase in (i) should be acceptable, contrary to fact.

  • (i) *Wie soll es weitergehen mit euch, denkst du, dass Max denn meint? how should it go.on with you.PL think you that Max DENN thinks ‘How do you think that Max thinks that it should go on with the two of you?’

40 The implicature is that the space of the scale that renders “one can get gasoline in x early in the morning” true is very small or even nonexistent.

41 For German, this is particularly noticeable because traditional and prescriptive grammar have long viewed DiPs as “optional” and more or less “meaningless” parts of speech.

References

References
Abraham,
Werner
.
1991
.
The grammaticalization of German modal particles
. In
Approaches to grammaticalization II
, ed. by
Elizabeth Closs
Traugott
and
Bernd
Heine
,
331
380
.
Amsterdam
:
John Benjamins
.
Alber,
Birgit
.
2008
.
Tyrolean A-bar movement: Doubling and resumptive pronoun structures
. In
Microvariation in syntactic doubling
, ed. by
Sjef
Barbiers
,
Olaf
Koeneman
,
Marika
Lekakou
, and
Margreet van der
Ham
,
141
170
.
Bingley
:
Emerald Group
.
Barbiers,
Sjef
.
2014
.
Syntactic doubling and deletion as a source of variation
. In
Linguistic variation in the Minimalist framework
, ed. by
M. Carme
Picallo
,
197
223
.
Oxford
:
Oxford University Press
.
Bard,
Ellen Gurman
,
Dan
Robertson
, and
Antonella
Sorace
.
1996
.
Magnitude estimation of linguistic acceptability
.
Language
72
:
32
68
.
Barr,
Dale J
.,
Roger
Levy
,
Christoph
Scheepers
, and
Harry J.
Tily
.
2013
.
Random effects structure for confirmatory hypothesis testing: Keep it maximal
.
Journal of Memory and Language
68
:
255
278
.
Barss,
Andrew
.
1986
.
Chains and anaphoric dependence: On reconstruction and its implications
.
Doctoral dissertation, MIT, Cambridge, MA
.
Bates,
Douglas M
.,
Martin
Maechler
, and
Ben
Bolker
.
2012
.
lme4: Linear mixed-effects models using S4 classes
.
R package version 0.999999-0
.
Bayer,
Josef
.
1990
.
Directionality of government and Logical Form: A study of focusing particles and wh-scope
.
Habilitation thesis, University of Konstanz
.
Bayer,
Josef
.
1996
.
Directionality and Logical Form: On the scope of focusing particles and wh-in-situ
.
Dordrecht
:
Kluwer
.
Bayer,
Josef
.
2011
.
Discourse particles in questions: Syntactic constituency, scope and decomposition
.
Paper presented at the workshop Structures of Meaning – in Honor of Arnim von Stechow, University of Konstanz, 10–11 November 2011. Available at http://ling.uni-konstanz.de/pages/home/bayer/pdf/Stechow-Colloq-2011.pdf
.
Bayer,
Josef
.
2012
.
From modal particle to interrogative marker: A study of German denn
. In
Functional heads: The cartography of syntactic structures, vol. 7
, ed. by
Laura
Brugé
,
Anna
Cardinaletti
,
Giuliana
Giusti
,
Nicola
Munaro
, and
Cecilia
Poletto
,
13
28
.
Oxford
:
Oxford University Press
.
Bayer,
Josef
, and
Hans-Georg
Obenauer
.
2011
.
Discourse particles, clause structure, and question types
.
The Linguistic Review
28
:
449
491
.
Bayer,
Josef
, and
Martin
Salzmann
.
2013
.
That-trace effects and resumption: How improper movement can be repaired
. In
Repairs: The added value of being wrong
, ed. by
Patrick
Brandt
and
Eric
Fuß
,
275
333
.
Berlin
:
de Gruyter
.
Bayer,
Josef
, and
Andreas
Trotzke
.
2015
.
The derivation and interpretation of left peripheral discourse particles
. In
Discourse-oriented syntax
, ed. by
Josef
Bayer
,
Roland
Hinterhölzl
, and
Andreas
Trotzke
,
13
40
.
Amsterdam
:
John Benjamins
.
Cable,
Seth
.
2010
.
The grammar of Q: Q-particles, wh-movement, and pied-piping
.
Oxford
:
Oxford University Press
.
Cardinaletti,
Anna
.
2011
.
German and Italian modal particles and clause structure
.
The Linguistic Review
28
:
493
531
.
Chomsky,
Noam
.
1973
.
Conditions on transformations
. In
A festschrift for Morris Halle
, ed. by
Stephen R.
Anderson
and
Paul
Kiparsky
,
232
286
.
New York
:
Holt, Rinehart and Winston
.
Chomsky,
Noam
.
1986
.
Barriers
.
Cambridge, MA
:
MIT Press
.
Chomsky,
Noam
.
1995
.
The Minimalist Program
.
Cambridge, MA
:
MIT Press
.
Chomsky,
Noam
.
2000
.
Minimalist inquiries: The framework
. In
Step by step
, ed. by
Roger
Martin
,
David
Michaels
, and
Juan
Uriagereka
,
89
155
.
Cambridge, MA
:
MIT Press
.
Chomsky,
Noam
.
2001
.
Derivation by phase
. In
Ken Hale: A life in language
, ed. by
Michael
Kenstowicz
,
1
52
.
Cambridge, MA
:
MIT Press
.
Chung,
Sandra
.
1982
.
Unbounded dependencies in Chamorro grammar
.
Linguistic Inquiry
13
:
39
77
.
Cinque,
Guglielmo
.
1999
.
Adverbs and functional heads: A cross-linguistic perspective
.
New York
:
Oxford University Press
.
Cole,
Peter
, and
Gabriella
Hermon
.
2000
.
Partial wh-movement: Evidence from Malay
. In
Wh-scope marking
, ed. by
Uli
Lutz
,
Gereon
Müller
, and
Arnim von
Stechow
,
101
130
.
Amsterdam
:
John Benjamins
.
Coniglio,
Marco
.
2011
.
Die Syntax der deutschen Modalpartikeln: Ihre Distribution und Lizensierung in Haupt-und Nebensätzen
.
Berlin
:
Akademie Verlag
.
Coniglio,
Marco
, and
Iulia
Zegrean
.
2012
.
Splitting up Force: Evidence from discourse particles
. In
Main clause phenomena: New horizons
, ed. by
Lobke
Aelbrecht
,
Liliane
Haegeman
, and
Rachel
Nye
,
229
255
.
Amsterdam
:
John Benjamins
.
Cowart,
Wayne
.
1997
.
Experimental syntax: Applying objective methods to sentence judgments
.
Thousand Oaks, CA
:
Sage Publications
.
Dayal,
Veneeta
.
1994
.
Scope marking as indirect wh dependency
.
Natural Language Semantics
2
:
137
170
.
Dikken,
Marcel den
.
2009
.
Arguments for successive-cyclic movement through SpecCP: A critical review
.
Linguistic Variation Yearbook
9
:
89
126
.
Egg,
Markus
.
2012
.
Discourse particles at the semantics-pragmatics interface
. In
Modality and theory of mind elements across languages
, ed. by
Werner
Abraham
and
Elisabeth
Leiss
,
297
333
.
Berlin
:
de Gruyter
.
Erteschik-Shir,
Nomi
.
1973
.
On the nature of island constraints
.
Doctoral dissertation, MIT, Cambridge, MA
.
Fanselow,
Gisbert
.
To appear
.
Partial wh-movement
. In
The Blackwell companion to syntax
, [new edition] ed. by
Martin
Everaert
and
Henk
van Riemsdijk
.
Oxford
:
Blackwell
.
Fanselow,
Gisbert
, and
Anoop
Mahajan
.
2000
.
Towards a Minimalist theory of wh-expletives, wh-copying, and successive cyclicity
. In
Wh-scope marking
, ed. by
Uli
Lutz
,
Gereon
Müller
, and
Arnim von
Stechow
,
195
230
.
Amsterdam
:
John Benjamins
.
Fanselow,
Gisbert
, and
Thomas
Weskott
.
2010
.
A short note on long movement in German
.
Linguistische Berichte
222
:
129
140
.
Featherston,
Sam
.
2005a
.
That-trace in German
.
Lingua
115
:
1277
1302
.
Featherston,
Sam
.
2005b
.
Universals and grammaticality: Wh-constraints in German and English
.
Linguistics
43
:
667
711
.
Featherston,
Sam
.
2008
.
Thermometer judgements as linguistic evidence
. In
Was ist linguistische Evidenz?
, ed. by
Claudia Maria
Riehl
and
Astrid
Rothe
,
69
89
.
Aachen
:
Shaker Verlag
.
Felser,
Claudia
.
2004
.
Wh-copying, phases, and successive cyclicity
.
Lingua
114
:
543
574
.
Haegeman,
Liliane
.
2002
.
Anchoring to speaker: Adverbial clauses and the structure of CP
. In
Georgetown University working papers in theoretical linguistics 2
, ed. by
Simon
Mauck
and
Jennifer
Mittelstaedt
,
117
180
.
Washington, DC
:
Georgetown University, Department of Linguistics
.
Haegeman,
Liliane
.
2006
.
Conditionals, factives and the left periphery
.
Lingua
116
:
1651
1669
.
Heycock,
Caroline
.
2005
.
Embedded root phenomena
. In
The Blackwell companion to syntax, vol. 2
, ed. by
Henk van
Riemsdijk
and
Martin
Everaert
,
174
209
.
Oxford
:
Blackwell
.
Hiemstra,
Inge
.
1986
.
Some aspects of wh-questions in Frisian
.
Nowele
8
:
97
110
.
Hinterhölzl,
Roland
, and
Manfred
Krifka
.
2013
.
Modal particles in adverbial and adnominal clauses
.
Ms., Università di Venezia, Ca’Foscari and Humboldt-Universität zu Berlin
.
Horvath,
Julia
.
1997
.
The status of “wh-expletives” and the partial wh-movement construction of Hungarian
.
Natural Language and Linguistic Theory
15
:
509
572
.
Keller,
Frank
.
2000
.
Gradience in grammar: Experimental and computational aspects of degrees of grammaticality
.
Doctoral dissertation, University of Edinburgh
.
Kiziak,
Tanja
.
2010
.
Extraction asymmetries: Experimental evidence from German
.
Amsterdam
:
John Benjamins
.
König,
Ekkehard
.
1977
.
Modalpartikeln in Fragesätzen
. In
Aspekte der Modalpartikeln
, ed. by
Harald
Weydt
,
115
130
.
Tübingen
:
Niemeyer
.
Krifka,
Manfred
.
2014
.
Embedding illocutionary acts
. In
Recursion: Complexity in cognition
, ed. by
Tom
Roeper
and
Margaret
Speas
,
59
87
.
Berlin
:
Springer
.
Kvam,
Sigmund
.
1983
.
Linksverschachtelung im Deutschen und Norwegischen: Eine kontrastive Untersuchung zur Satzverschränkung und Infinitivverschränkung in der deutschen und norwegischeón Gegenwartssprache
.
Tübingen
:
Niemeyer
.
Lipták,
Anikó
, and
Malte
Zimmermann
.
2007
.
Indirect scope marking again: A case for generalized question formation
.
Natural Language and Linguistic Theory
25
:
103
155
.
Lohnstein,
Horst
.
2007
.
On clause types and sentential force
.
Linguistische Berichte
209
:
63
86
.
Lühr,
Rosemarie
.
1988
.
Zur Satzverschränkung im heutigen Deutsch
.
Groninger Arbeiten zur germanistischen Linguistik
29
:
74
87
.
Lutz,
Uli
,
Gereon
Müller
, and
Arnim
von Stechow
, eds.
2000
.
Wh-scope marking
.
Amsterdam
:
John Benjamins
.
Mahajan,
Anoop
.
1990
.
The A/A-bar distinction and movement theory
.
Doctoral dissertation, MIT, Cambridge, MA
.
Manetta,
Emily
.
2011
.
Peripheries in Kashmiri and Hindi-Urdu: The syntax of discourse-driven movement
.
Amsterdam
:
John Benjamins
.
McCloskey,
James
.
2002
.
Resumption, successive cyclicity, and the locality of operations
. In
Derivations in the Minimalist Program
, ed. by
Samuel David
Epstein
and
T. Daniel
Seely
,
184
226
.
Oxford
:
Blackwell
.
McDaniel,
Dana
.
1989
.
Partial and multiple wh-movement
.
Natural Language and Linguistic Theory
7
:
565
604
.
McDaniel,
Dana
, and
Wayne
Cowart
.
1999
.
Experimental evidence of a Minimalist account of English resumptive pronouns
.
Cognition
70
:
B15
B24
.
Meibauer,
Jörg
.
1994
.
Modaler Kontrast und konzeptuelle Verschiebung: Studien zur Syntax und Semantik deutscher Modalpartikeln
.
Tübingen
:
Niemeyer
.
Meinunger,
André
.
2004
.
Verb position, verbal mood and the anchoring (potential) of sentences
. In
Syntax and semantics of the left periphery
, ed. by
Horst
Lohnstein
and
Susanne
Trissler
,
313
341
.
Berlin
:
de Gruyter
.
Meinunger,
André
.
2006
.
The discourse status of subordinate sentences and some implications for syntax and pragmatics
. In
The architecture of focus
, ed. by
Valéria
Molnár
and
Susanne
Winkler
,
459
487
.
Berlin
:
de Gruyter
.
Meinunger,
André
.
2007
.
In the mood of desire and hope: Remarks on the German subjunctive, the verb second phenomenon, the nature of volitional predicates and speculations on illocution
. In
Tense, mood and aspect
, ed. by
Louis de
Saussure
,
Jacques
Moeschler
, and
Genoveva
Puskas
,
155
176
.
Amsterdam
:
Rodopi
.
Müller,
Gereon
.
1997
.
Partial wh-movement and Optimality Theory
.
The Linguistic Review
14
:
249
306
.
Pankau,
Andreas
.
2013
.
Replacing copies: The syntax of wh-copying in German
.
Doctoral dissertation, Utrecht University
.
Pankau,
Andreas
,
Craig
Thiersch
, and
Kay-Michael
Würzner
.
2010
.
Spurious ambiguities and the parentheticals debate
. In
Language and logos: Studies in theoretical and computational linguistics
, ed. by
Thomas
Hanneforth
,
Gisbert
Fanselow
, and
Peter
Staudacher
,
129
146
.
Berlin
:
Akademie Verlag
.
Paul,
Hermann
.
1919
.
Deutsche Grammatik
. Bd. 3, Teil 4:
Syntax
.
Halle an der Saale
:
Niemeyer
.
Pechmann,
Thomas
,
Hans
Uszkoreit
,
Johannes
Engelkamp
, and
Dieter
Zerbst
.
1994
.
Word order in the German middle field: Linguistic theory and psycholinguistic evidence
.
Computational Linguistics at the University of the Saarland, Report No. 43
.
Pesetsky,
David
, and
Esther
Torrego
.
2007
.
The syntax of valuation and the interpretability of features
. In
Phrasal and clausal architecture: Syntactic derivation and interpretation
, ed. by
Simin
Karimi
,
Vida
Samiian
, and
Wendy K.
Wilkins
,
262
294
.
Amsterdam
:
John Benjamins
.
Phillips,
Colin
, and
Dan
Parker
.
2014
.
The psycholinguistics of ellipsis
.
Lingua
151
:
78
95
.
Pickering,
Martin
, and
Guy
Barry
.
1991
.
Sentence processing without empty categories
.
Language and Cognitive Processes
6
:
229
259
.
Potts,
Christopher
.
2005
.
The logic of conventional implicatures
.
Oxford
:
Oxford University Press
.
R Core Team
.
2012
.
R: A language and environment for statistical computing
.
R Foundation for Statistical Computing, Vienna
.
Reis,
Marga
.
1991
.
Echo-w-Sätze und Echo-w-Fragen
. In
Fragesätze und Fragen
, ed. by
Marga
Reis
and
Inger
Rosengren
,
49
76
.
Tübingen
:
Niemeyer
.
Reis,
Marga
.
1995
.
Wer glaubst du hat recht? On so-called extractions from verb-second clauses and verb-first parenthetical constructions in German
.
Sprache und Pragmatik
36
:
27
83
.
Reis,
Marga
.
1996
.
Extractions from verb-second clauses in German?
In
On extraction and extraposition in German
, ed. by
Uli
Lutz
and
Jürgen
Pafel
,
45
88
.
Amsterdam
:
John Benjamins
.
Reis,
Marga
.
2000
.
On the parenthetical features of German was ... w constructions and how to account for them
. In
Wh-scope marking
, ed. by
Uli
Lutz
,
Gereon
Müller
, and
Arnim
von Stechow
,
359
407
.
Amsterdam
:
John Benjamins
.
Richards,
Marc
.
2012
.
Probing the past: On reconciling long-distance agreement with the Phase Impenetrability Condition
. In
Local modelling of non-local dependencies in syntax
, ed. by
Artemis
Alexiadou
,
Tibor
Kiss
, and
Gereon
Müller
,
135
154
.
Berlin
:
de Gruyter
.
Riemsdijk,
Henk van
.
1983
.
Correspondence effects and the Empty Category Principle
. In
Studies in generative grammar and language acquisition
, ed. by
Yukio
Otsu
,
Henk van
Riemsdijk
,
Kazuko
Inoue
,
Akio
Kamio
, and
Noriko
Kawasaki
,
5
16
.
Tokyo
:
International Christian University
.
Rizzi,
Luigi
.
1997
.
The fine structure of the left periphery
. In
Elements of grammar
, ed. by
Liliane
Haegeman
,
281
337
.
Dordrecht
:
Kluwer
.
Rizzi,
Luigi
.
2004
.
Locality and left periphery
. In
Structures and beyond: The cartography of syntactic structures, vol. 3
, ed. by
Adriana
Belletti
,
223
251
.
Oxford
:
Oxford University Press
.
Sabel,
Joachim
.
2000
.
Partial wh-movement and the typology of wh-questions
. In
Wh-scope marking
, ed. by
Uli
Lutz
,
Gereon
Müller
, and
Arnim von
Stechow
,
409
446
.
Amsterdam
:
John Benjamins
.
Saddy,
Doug
.
1991
.
Wh scope mechanisms in Bahasa Indonesia
. In
More papers on wh-movement
, ed. by
Lisa Lai-Shen
Cheng
and
Hamida
Demirdache
,
183
218
.
MIT Working Papers in Linguistics 15. Cambridge, MA
:
MIT, MIT Working Papers in Linguistics
.
Schneider-Zioga,
Patricia
.
2009
.
Wh-agreement and bounded unbounded movement
. In
Merging features: Computation, interpretation, and acquisition
, ed. by
José M.
Brucart
,
Anna
Gavarró
, and
Jaume
Solà
,
46
59
.
Oxford
:
Oxford University Press
.
Schütze,
Carson T
., and
Jon
Sprouse
.
2014
.
Judgment data
. In
Research methods in linguistics
, ed. by
Robert J.
Podesva
and
Devyani
Sharma
,
27
50
.
Cambridge
:
Cambridge University Press
.
Simons,
Mandy
.
2007
.
Observations on embedding verbs, evidentiality, and presupposition
.
Lingua
117
:
1034
1056
.
Sorace,
Antonella
.
1992
.
Lexical conditions on syntactic knowledge: Auxiliary selection in native and nonnative grammars of Italian
.
Doctoral dissertation, University of Edinburgh
.
Sprouse,
Jon
.
2011
.
A test of the cognitive assumptions of magnitude estimation: Commutativity does not hold for acceptability judgments
.
Language
87
:
274
288
.
Sternefeld,
Wolfgang
.
2001
.
Partial movement constructions, pied piping, and higher order choice functions
. In
Audiatur vox sapientiae: A festschrift for Arnim von Stechow
, ed. by
Caroline
Féry
and
Wolfgang
Sternefeld
,
473
486
.
Berlin
:
Akademie Verlag
.
Sternefeld,
Wolfgang
.
2002
.
Wh-expletives and partial wh-movement: Two nonexisting concepts
? In
Issues in formal German(ic) typology
, ed. by
Werner
Abraham
and
C. Jan-Wouter
Zwart
,
285
306
.
Amsterdam
:
John Benjamins
.
Stevens,
Stanley S
.
1975
.
Psychophysics: Introduction to its perceptual, neural, and social prospects
.
New York
:
John Wiley
.
Struckmeier,
Volker
.
2014
.
‘Ja doch wohl’ C? Modal particles in German as C-related elements
.
Studia Linguistica
68
:
16
48
.
Thurmair,
Maria
.
1989
.
Modalpartikeln und ihre Kombinationen
.
Tübingen
:
Niemeyer
.
Trissler,
Susanne
.
1999
.
Syntaktische Bedingungen für w-Merkmale: Zur Bildung interrogativer w-Phrasen
.
Doctoral dissertation, UniversitätTübingen
.
van Urk,
Coppe
, and
Norvin
Richards
.
2015
.
Two components of long-distance extraction: Successive cyclicity in Dinka
.
Linguistic Inquiry
46
:
113
155
.
Viesel,
Yvonne
.
2011
.
Glaubt er, glaub ich, glaub: Integrierte V1-Parenthesen, Extraktion aus V2-Komplementen, Grammatikalisierung
.
Linguistische Berichte
181
:
1
42
.
Viesel,
Yvonne
.
To appear
.
Discourse particles ‘embedded’: German ja in APs
. In
Discourse particles: Formal approaches to their syntax and semantics
, ed. by
Josef
Bayer
and
Volker
Struckmeier
.
Berlin
:
de Gruyter
.
Wahba,
Wafaa Abdel-Faheem Batran
.
1992
.
LF movement in Iraqi Arabic
. In
Logical structure and linguistic structure
, ed. by
C.-T. James
Huang
and
Robert
May
,
253
276
.
Dordrecht
:
Kluwer
.
Wauchope,
Mary Michele
.
1991
.
The grammar of the Old High German modal particles thoh, ia, and thanne
.
Frankfurt am Main
:
Peter Lang
.
Wegener,
Heide
.
2002
.
The evolution of the German modal particle ‘denn’
. In
New reflections on grammaticalization
, ed. by
Ilse
Wischer
and
Gabriele
Diewald
,
379
393
.
Amsterdam
:
John Benjamins
.
Weskott,
Thomas
, and
Gisbert
Fanselow
.
2011
.
On the informativity of different measures of linguistic acceptability
.
Language
87
:
249
273
.
Zeijlstra,
Hedde
.
2012
.
There is only one way to agree
.
The Linguistic Review
29
:
491
539
.
Zimmermann,
Malte
.
2004
.
Zum wohl: Diskurspartikeln als Satzmodifikatoren
.
Linguistische Berichte
199
:
253
286
.
Zimmermann,
Malte
.
2008
.
Discourse particles in the left periphery
. In
Dislocated elements in discourse
, ed. by
Benjamin
Shaer
,
Philippa
Cook
,
Werner
Frey
, and
Claudia
Maienborn
,
200
231
.
New York
:
Routledge
.