Irregular V2 Relative Clauses in German: A Comparative Study, Lecture notes of Communication and Presentation Skills

Download Irregular V2 Relative Clauses in German: A Comparative Study and more Lecture notes Communication and Presentation Skills in PDF only on Docsity! On the theory of presentation STEFAN HUBER (LUND UNIVERSITY) Abstract The aim of the paper is to introduce a new information-structural category presentation and to argue for its relevance and independence from other information-structural layers. It is shown that a successful presentation consists of two steps, namely the import of a new discourse participant and its establishing (process) in the ongoing context. The employment of import and process is a coherence strategy that involves at least two sentences. Therefore, these presentational notions must be distinguished from merely intrasentential information structure concepts such as topic, focus, contrast, theme, rheme, etc. The paper also discusses some morphological and syntactic reflections of presentation. Notably, it will be claimed that both non-given (i.e., rhematic) and given (i.e., thematic) information can be (re-) presented. Moreover, the autonomy of the import-process layer from other information-structural concepts will be supported theoretically. 1 Information Structure: Syntax and Pragmatics Throughout the last decade, much attention has been paid to the investigation of information-structural features, such as topic, focus, theme, rheme, contrast, etc. Moreover, at least since Rizzi’s (1997) theoretical proposal, numerous attempts have been made to integrate such features into core syntax. The aim of this paper is to revisit and discuss some of these informational features, both from a syntactic and pragmatic view. In particular, it will be argued that there is a demand for a model that consists of at least four different layers of information structure, i.e., a new layer called presentation will be introduced. In order to facilitate the ensuing discussion, a short survey on both sides of information structure – syntax and pragmatics – will be given in this introductory section. The goal of the syntactic investigation of information structure has been to prove that categories such a focus, topic, contrast, etc. have a more or less severe impact on word order. These claims have been extended from discourse- configurational languages such as Hungarian (cf. Brody 1995, É. Kiss 1998), Finnish (Molnár & Järventausta 2003) or Basque (Arregi 2002) to Indo- European languages as well (e.g., Rizzi 1997 for Italian; É. Kiss 1998, Meinunger 1998, den Dikken 2005 for English; Frey 2004a, 2004b for German and many others). Traditionally, the main interest has been on the so-called left periphery. Especially for the features topic, focus and contrast, there have been attempts to create specific functional categories within the initial sentence domain. An item carrying one of these features has been claimed to raise to the relevant position to check its feature, according to (earlier) minimalist assumptions (Chomsky 1995). The evidence for such functional positions in Stefan Huber 40 discourse-configurational languages such as Hungarian and Basque is compelling, since a phrase can only be interpreted as a topic or (narrow) focus if it occurs in these designated positions (cf. Brody 1995, É. Kiss 1998). It has been more difficult to prove the existence of pragmatically motivated categories for Germanic languages. For instance, focus assignment in Germanic corresponds to accentuation and the accent placement in Germanic is rather free, i.e., a focus can also be interpreted in situ. As for the pragmatic evaluation of information structure, Molnár (1991, 1993) suggests that three distinct layers have to be assumed, according to the following pattern: (1) Message: → Topic – Comment Hearer: → Theme – Rheme Speaker: → Background – Focus Molnár differentiates between the layers of the message, the hearer and the speaker. Within each layer, there is a particular bifurcation: The message layer can be divided into a part that the utterance is about (topic; T)1 and a part that contains the actual information-structural predication (comment; C). Furthermore, the speaker organizes the utterance with regard to his/her assumptions as to hearer-old information (theme; Th) or new information (rheme; Rh). Finally, the speaker marks the more important (focus; F) and less important (background; B) information. Molnár points out that in the prototypical case the three layers coincide as in (2): (2) (I met a man.)2 He]T/Th/B [was a professor]C/Rh/F . However, Molnár discusses a wide array of other examples that demonstrate that the three layers need to be kept apart. In particular, she emphasizes that the concepts of topic and focus are not mutually exclusive, but rather apply on different levels. Two relevant examples for this claim are given in (3), which involves multiple foci, and (4), which presents an all-new sentence that still can be divided into a topic and a comment part: 1 Throughout the paper, the following abbreviations are used: AT = aboutness topic, B = background, C = comment, DT = discourse topic, F = focus, FT = frame topic, I = import, P = process, PT = pre-topic, Rh = rheme, T = topic, Th = theme. 2 Within the examples, parentheses are used to indicate context material, whereas brackets are meant to mark the relevant informational units. On the theory of presentation 43 they analyze these constructions as coordinative structures (cf. also Gärtner 2001), they define a pre-topic as some sort of anchor that links the second conjunct to the first one. The analysis of example (9a) in (9b) is a rough sketch of Gärtner’s (2001) and Endriss & Gärtner’s (2005) approach: (9) a. Apfeldorf hat viele Häuser, die stehen leer. Apfeldorf has many houses these stand empty ‘Apfeldorf has many houses that are abandoned.’ b. Apfeldorf hat [viele Häuser]PT (&) die stehen leer. Note that also the DP Apfeldorf must be considered an (aboutness) topic for the following proposition. Thus, it seems that there are two different topics in this sentence. As for the concept of focus there have been ambitions to differentiate this category as well. Not only have foci been classified according to their scope (minimal, intermediate, maximal foci, cf. Rosengren 1993 and many others), but more recently, a necessary distinction between focus and contrast has been repeatedly argued for (Molnár 2002, Frey 2004b), the claim being that a contrast always involves some sort of exclusion of individuals from a given set, which is not necessarily true for a (narrow) focus. Although information structure is closely tied to textual coherence, only few researchers have addressed concepts and strategies that include intersentential relationships. Note that all of the terms discussed thus far refer to single sentences as the limiting category. This circumstance is certainly due to the dominance of syntactic theories over information-structural investigations. Note, however, that some notions, such as thematic or given information, can only be analyzed if a proper co(n)text exists. It is the purpose of this paper to argue for the relevance of yet another information-structural strategy that reflects the coherence of two sentences by means of presenting an item. I will call this strategy presentation (in line with Lambrecht 1988, Brandt 1990 and others). By the term presentation I mean that an item is introduced into the ongoing discourse and that subsequently this newly presented element is established in the discourse. I will call the introducing part import and the establishing part process. The necessity of this additional layer of information structure will be supported by the fact that the import-process dichotomy does not coincide with other informational dichotomies such as background-focus or topic-comment. In this context it will be shown that yet another important distinction has to be made, namely the one between presentational and existential constructions although the two notions have commonly been used interchangeably. As will be demonstrated, both the meaning of the items involved and the structures surrounding them differ considerably. Stefan Huber 44 In section 2 the basic idea will be presented. In section 3 a survey over some relevant presentational constructions is given, which should corroborate the claim that presentation is reflected linguistically. In section 4 we will return to the theoretical framework and plead for the independence and importance of the presentational import-process layer. The investigation mainly deals with English and German data. 2 The Basic Idea: Presentation as a Two-Step Strategy It is often argued that the pragmatic function of existential sentences like (10) is to present or introduce an item into the ongoing discourse. Therefore they have been labeled “presentational” as well: (10) There are mice in the church. It is undoubtedly possible to introduce an (indefinite) phrase like mice into the context by means of a there is/are-construction in English. The reverse, however, is not necessarily true as the two different contexts for (10) in (10’a, 10’b) demonstrate: (10’) a. (As you can see from the trails on the floor,) there are mice in the church. (They seem to thrive in cool, dark places like St. Matthew’s.) b. (Can you believe it?) There are mice in the church! (It’s disgusting! No wonder nobody is Christian any longer …) In (10’a), the information unit that denotes mice re-occurs in the following text passage as an anaphoric pronoun (they). Hence, it has been properly introduced as discourse participant. In (10’b), on the other hand, there is no such thematic progression, i.e., no resumption of the referent mice. Nevertheless, in both sentences we find so-called “existential” constructions (i.e., a there are DP- construction). As one easily can understand, only (10’a) can be described as a true presentational construction. For this reason, it seems imperative to keep the notions “existential” and “presentational” apart. Presentation, thus, has to consist of at least two steps: first, an item is introduced into the context. As I aim to demonstrate in section 3, there are numerous morphosyntactic means to provide such an introduction. Second, the newly introduced item is being resumed and integrated into the discourse. Only if both conditions are met, an item can be said to be successfully presented. In the following, I will call the introducing part the importing (or short import, I) of an item. The second step will be called the processing of the item in the On the theory of presentation 45 discourse (or short process, P). The term import is meant to signal the destination of the item into the contextual flow, whereas the term process is chosen to indicate the transformation the information undergoes when being presented. If one looks at the concept of processing an introduced item, one can notice that there are at least two ways in which an item is established. First, the imported item can be described according to its properties. This strategy can be iterative, i.e., several properties can be mentioned in the subsequent sentences: (11) There are [mice]I in the church. [They seem to thrive in cool, dark places like St. Matthew’s.]Pa At any rate, [they have been living here for centuries.]Pb Another way of processing the presented element is to establish it as a new participant in the context. In this case, new elements might be presented within the process part so that the presentation becomes recursive: (12) There are [mice]I1 in the sanctuary St. Matthew’s. [They are building [a nest]I2 there.]P1 [It hosts their off-spring.]P2 Example (12) seems to be an instance of normal thematic progression (cf. Daneš 1974). However, although the imported items (mice and a nest) might correspond to rhemes in a standard theme-rheme analysis, a corresponding theme cannot be understood as identical to a process. As can be seen, the processes P1 and P2 in (12) are more complex, i.e., they contain propositions, whereas the themes (e.g., they, it) are single phrases. Thus, the concepts of theme-rheme and import-process need to be kept apart. I will return to this point in section 4.1. On a closer look, it seems that a prototypical import involves an indefinite DP. However, it will be shown that also definite DPs can re-enter the context, for instance if the speaker wants to re-establish a participant that is assumed not to be salient or present in the hearer’s consciousness at the point of the utterance. A typical case of such a re-presentation is what Lambrecht (1988) and Brandt (1990) call “continuative relative clauses:” (13) After many years, I finally met Peter Giles again, who immediately told me about Kevin Curby’s death. (Apparently Kevin had died already in 1998.) My suggestion would be that Peter Giles is re-imported into the context, whereas the extraposed relative clause represents its re-establishing/processing. Note that within the relative clause the DP Kevin Curby is re-imported as well Stefan Huber 48 An important question in this context is whether such a specific reading is dependent on the choice of the verb. Lambrecht (1988), who discusses “presentational relative constructions” and “presentational amalgam constructions” (see below, 3.2.2.), only mentions examples containing there is/are or have. Apparently, more predicates can initiate specific readings and/or presentations, e.g., verbs of perception (16) or encounter (17): (16) a. I saw a man. He was blind. b. I heard a lovely song. It reminded me of the old days. c. She felt a strange tingle in her right hand. It started on her fingertips and spread all over her arm to her upper body. (17) a. In Munich, Max met a very busy priest. He hardly had the time to take Max’s confession. b. I ran into a young man the other day. He said he was a former student of mine. Certainly, there are other verbal classes as well that allow for presentations. The question, then, is whether all these predicates have some sort of a common semantic feature. At this point, I doubt that this is the case, but I have to leave this question open. So far, a specific reading of an internal argument seems be compatible with any predicate in a proper context. On the other hand, it is not the case that any specific reading automatically triggers a presentation. For instance, in situations where the element pointed out is given in the surroundings of the utterance, a verbally explicit process part can be skipped. If one adheres to the rather strict definition given in section 2, no proper presentation takes place. Consider the following example, where two people in a discotheque look at a male person dancing in a very spastic way. The referent in question is denoted by the indefinite pronoun somebody: (18) Speaker A: Somebody forgot his valium this morning. Speaker B: No doubt. Although there is no real process part, it could be claimed that somebody acts as a deictic indicator of a specific person in the room. The specificity of somebody becomes evident through the choice of the anaphoric possessive his. With non- specific DPs, on the other hand, like the one in (19), no such agreement can be observed. (19) describes a situation where a person finds a couple of valium tablets in an unexpected place. The matching possessive is their, since it is not clear whether it is one person (and in this case: whether this person is male or female) or several persons who forgot the tablets: (19) Look at that: Somebody forgot their valium this morning. On the theory of presentation 49 Thus, a specific reading can occur without processing the specific item, i.e., presenting it. Let us turn now to the process part: In the examples discussed so far, an independent sentence is chosen to process the indefinite DP and establish it as a discourse participant. However, also a sentential coordination is acceptable: (20) I met interesting people in London and they showed me the city. Furthermore, a process part might be expressed by a relative clause. Lambrecht (1988) calls these constructions “continuative” or “presentational relative constructions”, depending on the matrix predicate: (21) I met interesting people in London, who showed me the city. (21) shows that the relative clause has a non-restrictive character and occurs in extraposition, i.e., after the locative adverbial in London. As for their counterparts in German, the label “weiterführende Relativsätze” has been used (cf. Brandt 1990, Reis 1997 and others). According to Brandt (1990) there has to be a balance between the matrix and the relative clause. At least on an information-structural level, the entire sentence ought to be considered displaying a coordinative rather than a subordinative structure. Brandt also makes a strong point about the independent focus-background structure of such relative clauses. We will return to a more thorough pragmatic analysis of these constructions in section 4.3.9 Note, however, that such a “balanced” account requires that presentations always involve some sort of bisententiality, i.e., that the process part has to be contained within a separate sentence. Apparently, in coordinative structures two syntactically independent sentences are present, each of which hosts one of the two presentational steps. The bisententiality requirement is also reflected by yet another strategy to process the imported item, namely the employment of parentheses. An example is given in (22): (22) I met interesting people in London – they showed me the city –, but I never made any real friends during my three months in the UK. 9 The syntactic difficulty of this approach is how to coordinate a subordinate and a matrix clause. This is especially problematic for German since there are two strongly diverging word orders for matrix clauses (V2) and dependent clauses (V-last). It has, however, been pointed out repeatedly that this combination is licit, at least in certain constructions, e.g., in conditional clauses (cf. Höhle 1989): (i) Wenn du nach Hause kommst und der Gerichtsvollzieher steht vor der Tür, … if you to home come and the bailiff stands in-front the door ‘If you come home and you meet the bailiff at the door …’ Stefan Huber 50 If one regards non-integrated parentheses10 as a pragmatic strategy of last-minute modification, the speaker of (22) uses the string they showed me the city to alter the status of interesting people into a specific DP. Thus, the parenthesis has to be regarded as a sentence with an independent focus-background structure and, hence, meets the criteria to serve as a proper process part of a presentation. 3.1.2 “Indefinite Demonstratives” It has been long known that English displays demonstratives in certain contexts that cannot be described as truly definite,11 cf. the investigations by Prince (1981), Gundel, Hedberg, & Zacharski (1993), Endriss & Gärtner (2005), and others. These DPs have been named “indefinite demonstratives”. An example with context is given in (23): (23) Yesterday, I met this American woman. (She was just breath-taking. We talked for hours. …) Endriss & Gärtner (2005) also present and discuss cases of German indefinite demonstratives: (24) a. Im Sommer gab es plötzlich diesen Moment, da klappte alles.12 in-the summer was there suddenly this moment then worked-out all ‘In the summer there was suddenly this moment when everything came together.’ b. Da waren noch diese Studenten, […]. Die haben ständig geschwätzt. there were also these students they have permanently chattered ‘There were also these students who chattered all the time.’ It appears that all these DPs are highly specific. By uttering indefinite demonstratives, the speaker clearly indicates that s/he has additional information about the denoted referents and that s/he wants to share this information.13 S/he does so in the subsequent sentence or clause. I regard the morphological marking of a DP as an indefinite demonstrative as an unambiguous strategy by the speaker to initiate a presentation. The utterance of such a specific item without any form of textual progression – for instance, 10 For the difference between integrated and non-integrated parentheses see Reis (1996). 11 Note that the notion of (in-) definiteness has been used both in a morphological and in a semantic sense (cf. Diesing 1992, Frey 2001). In this paper, I try to use these terms mainly as morphological categories. 12 The example in (24a) involves a so-called “V2 relative clause”, which will be discussed at length in subsection 3.1.3. 13 Dekker (2004) talks about “referential intentions” with indefinite demonstratives. On the theory of presentation 53 (29) a. * Kein Fallschirmspringeri beachtete ein Haus, das konnte eri schlecht sehen. b. Kein Fallschirmspringeri beachtete ein Haus, das eri schlecht sehen konnte. no skydiver noticed a house that he badly see could ‘No skydiver noticed a house that he couldn’t see clearly.’ (26b) and (27b) demonstrate the fact that V2 relative clauses cannot be moved along with their topicalized or passivized heads, whereas regular relative clauses can (26c, 27c). Co-reference between a pronoun and an R-expression seems to be licit with V2 relative clauses (28a), but not with regular relative clauses (28b). On the other hand, a variable within a V2 relative clause cannot be bound by a quantifier (29a), whereas this is not problem with regular relative clauses (29b). Gärtner (2001:103) points out another striking morphological fact: With locative adverbial heads, the regular relative pronoun wo is banned. Instead the weak demonstrative da must be used: (30) a. Ich war in einem Land, OKda/*wo kostet das Bier ein Vermögen. I was in a country there/where costs the beer a fortune ‘I was in a country where beer costs a fortune.’ b. Ich war in einem Land, *da/OKwo das Bier ein Vermögen kostet. Gärtner reaches the conclusion that there cannot be a regular relative dependency between the two clauses. Instead, he suggests that the link between the clauses is a phonetically empty coordinative head π°REL which projects a coordination phrase (πP): (31) πP CP1 π′ Ich war in einem Land π°REL CP2 (= V2 relative clause) da kostet das Bier ein Vermögen By introducing the relative conjunction π°REL Gärtner (2001:108ff.) underlines his ambition to corroborate the similarity between canonical restrictive relative clauses and V2 relative clauses. He employs a row of semantic tests to Stefan Huber 54 demonstrate the purported parallel behavior of V2 relative clauses and restrictive relative clauses. One important test is that a V2 relative clause does not seem to trigger implicatures as do regular coordinations:17 (32) a. Das Blatt hat eine Seite, die ist ganz schwarz (= Gärtner 2001:112) the sheet has a page it is completely black ‘The sheet has a page that is completely black.’ b. # Das Blatt hat eine Seite und die ist ganz schwarz. According to Gärtner, the reading (32b) implicatures that the sheet has only one page which is incompatible with world-knowledge (hence #). On the other hand, (32a), which in his theory is some sort of an amalgam of a coordination and a relative subordination, arguably does not trigger such a reading but rather the appropriate one “one page of the sheet is completely black”. Gärtner’s main problem, however, is that he cannot explain why the heads of V2 relative clauses cannot be marked as definite phrases. As a solution, he refers to the semantic-pragmatic approach by Heim (1988): The V2 structure of these clauses renders an assertive rather than a presuppositional interpretation. On the other hand, definite DPs containing restrictive relative clauses rather presuppose than assert the proposition embedded in the relative clause. Thus, according to Gärtner, the unacceptability of these constructions can be accounted for by referring to semantic-pragmatic incompatibilities. Endriss & Gärtner (2005) basically agree with Gärtner’s assumptions regarding the coordinative character of constructions involving V2 relative clauses and the incompatibility of V2 relative clauses with presuppositionality. Furthermore, Endriss & Gärtner (2005) claim that the head of a V2 relative clause is some sort of a topic (“Prä-Topik”, “pre-topic”) which requires a subsequent comment in the shape of a V2 clause, i.e., the V2 relative clause. Den Dikken (2005), investigating V2 relative clauses in Dutch, reaches a similar conclusion. 18 17 Another (morphological) test shows that it is possible to one-replace a V2 relative clause in conjunction. This is also possible with regular relative clauses (cf. Gärtner 2001:111): (i) Hans kennt einen Philosophen, der mag Achternbusch, und Maria kennt auch einen. Hans knows a philosopher he likes Achternbusch and Maria knows also one ‘Hans knows a philosopher who likes Achternbusch, and Maria knows one, too.’ 18 As to the syntactic form of these constructions, however, den Dikken (2005: 702) takes a somewhat different stand than Gärtner (2001) and Endriss & Gärtner (2005). He suggests that the matrix clause of these constructions moves to the specifier position of a topic phrase (TopP). The V2 relative clause on the other hand serves as the complement of the empty topic head Topº. According to den Dikken (2005), this is the designated comment position, cf. (den Dikken 2005:701): (i) a. Er waren twee jongens op het strand die hadden geen zwembroek aan. there were two boys on the beach they had no swimsuit on On the theory of presentation 55 Although I agree with Gärtner (2001) in that V2 relative clauses should not be considered relative clauses in the canonical sense19 and that these constructions involve some form of coordination, the introduction of a “relative conjunction” π°REL is rather problematic. To my knowledge, there is no independent evidence for such a coordinating element. The fact that coordinative conjunctions with more or less specified meanings exist (aber ‘but’, denn ‘because’, etc.) does not provide sufficient motivation for an empty conjunction with a “relative” meaning.20 Moreover, the stipulation of such an element does not solve one of Gärtner’s main problems, i.e., the impossibility of definite “heads” for V2 relative clauses. Lambrecht (1988) and Brandt (1990:49) suggest that V2 relative clauses need to be analyzed as presentational constructions since they introduce a new element into the discourse. I would like to adopt this view and claim that V2 are used to import the head of the V2 relative clause within the matrix clause and to process it by way of the following V2 relative clause. Moreover, since I will make a distinction between (aboutness) topics and imports, I will not assume den Dikken’s (2005), and Endriss & Gärtner ’s (2005) pragmatic analyses of these structures as (pre-)topic-comment structures. Given the assumption that imports carry a specificity feature which requires further specification, the “relative nature” (cf. Gärtner 2001:122) of the V2 relative clause constructions follows naturally: The imported item is looking for an additional specification and finds it in the extraposed V2 clause. The “restrictiveness effect”, then, is a mere epiphenomenon, since further descriptions of indefinite, but specific DPs can only be interpreted in that way. Thus, if one assumes that the DP eine Seite in (34b) is interpreted as specific – some lexical material such as gewiss or bestimmt (‘certain’) or besonders (‘particular’) might be helpful for this purpose –, an overt coordinative conjunction is fine:21 ‘There were two boys at the beach that had no swimsuits on.’ b. [TopP [S1 Er waren twee jongens op het strand] [Top′ Ø [S2 die hadden geen zwimbroek aan]]]. Apart from the fact that a syntactic topic phrase lacks independent evidence in Dutch, the interpretation of such a structure – if possible at all – would not be the appropriate one, i.e., the whole clause er waren twee jongens op het strand cannot bet he topic of the whole sentence since the obvious comment part corresponds only to the DP twee jongens and not the whole proposition in which it is embedded. 19 Nevertheless, I will adhere to the term “V2 relative clauses” throughout the remainder of the paper. 20 Gärtner (2001:106) entertains, though, the possibility of other semantically flavored empty conjunctions (e.g., ”π°CONCESS”), however without arguing for their existence. 21 In German the indefinite article and the numeral for 1 are identical (ein). English makes a clear distinction between these two items (a vs. one). Thus the sentence is bound to trigger Stefan Huber 58 (34) a. There’s a lot of people don’t know that. (= Lambrecht 1988:319) b. I have a friend in the Bay Area is a painter. The first impression that comes to mind when reading these sentences is that they involve truncated relative clauses, i.e., relative clauses whose relative pronoun or relative complementizer have been phonetically deleted: (34’) a. There’s a lot of people who/that don’t know that. b. I have a friend in the Bay Area who/that is a painter. Note that in all these examples it is the subject that has been deleted in the relative clause. With objects, the phonological deletion is not only possible but sometimes even required: 24 (35) a. There’s a lot of people (who/that) I don’t know. b. I have a friend in the bay area (who/that) I haven’t visited for years. Lambrecht (1988:335) reaches the conclusion that the amalgam character of these sentences has to be captured syntactically as follows: (36) a. There was a farmer had a dog. (= Lambrecht 1988:335) (i) a. Der er et tog går klokken seks. (Spoken Danish) (= Engdahl 1997:61) there is a train goes six ‘There is a train leaving at six o’clock.’ b. Jeg har to døtre bor i København. (North Sealandish) I have two daughters live in Copenhagen. ‘I have to daughters who live in Copenhagen.’ 24 One of the most convincing theoretical explanations of these facts is given by Platzack (1998), who argues for a subject position within the C-domain. Once the C-domain is visible (i.e., phonetically filled) at Spell Out by virtue of the overt subject there is no need for a complementizer or any other C-element any more. On the theory of presentation 59 b. SPAC SPres SRel NPLOC V NP VP V NP There was a farmer had a dog In accordance with Lakoff (1974), Lambrecht calls such structures “presentational,” which makes sense since an (indefinite) item is introduced as a new discourse participant by way of an existential clause or as the object of have. In a second step, i.e., in the relative clause (SREL), this item is established by being processed. The morphosyntactic strategy of amalgamating two sentences into one is very interesting, particularly since it underlines the fact that the import part and the process part belong tightly together. However, I concur with Engdahl (1997) in that these clauses must contain a phonologically deleted relative complementizer/pronoun, i.e., the amalgam structure emerges first at PF and is not the product of a syntactic derivation. At a minimum, a structure like (36b) would violate the theta-criterion and should thus render ungrammaticality. The PF analysis, on the other hand, is fairly plausible since it relieves from the burden of modifying standard theoretical assumptions. Thus, a more adequate tree structure would be (37): (37) S There was DP a farmer CP who C′ VP Vº DP had a dog Stefan Huber 60 The deletion of the relative pronoun might be motivated by the very presentational character of the construction, comparably to the one found in V2 relative clauses in German (see above, 3.1.3.). 3.2 Definites 3.2.1 Simple Definites Presenting an item by means of import and process can take place in at least two different ways: Either a completely new element is introduced into the ongoing context or an element is presented that the speaker assumes to be known by the hearer, although it might not be as salient as other given elements at the time of the utterance. In this case, the speaker can choose a re-import strategy. Note that a re-imported item needs to be processed in a proper way, too. Some examples of these re-presentations are given in (38). (38) a. After a month I met [the Indian grocer]I again. [He seemed to be ten years older.]P b. I often think of [my foster daughter]I. [She must be grown up by now.]P c. All of a sudden, [Peter Miller]I, – [he seemed very upset]P – entered the room. We had to interrupt our meeting. In Germanic languages, the morphological marking of given (noun) phrases is usually achieved by using the definite article (38a), another definite determiner (as e.g., the possessive in 38b), or inherently definite expressions such as proper names (38c). In the prototypical case, the definite article also indicates the specificity of the definite DP. Thus, a definite DP is compatible with the specificity requirement for imports, as stated above. As for the morphosyntactic make-up of the process part with definite imports, they basically follow the same pattern as indefinites, i.e., we find two-sentence structures (38a), continuative relative clauses (38b), and parentheses (38c). Note that – pace Gärtner (2001) and Endriss & Gärtner (2005) – I find an asyndetic string of two V2 clauses with a definite import (= constructions containing V2 relative clauses with a definite head) not generally bad, cf.: (39) (Was hast du gesehen?) what have you seen ‘What did you see?’ a. Ins Zimmer kam ein Mann, der war total \AUFgeregt. into-the room came a man he was totally upset ‘A man entered the room who was totally upset.’ On the theory of presentation 63 b. continuative relative clause: I met the old cockroach again, who now lived in a paper bag and who/OKhe was very poor. In a semantic-pragmatic respect the attributive relative clause presupposes and the continuative/ presentational asserts the embedded proposition. Lambrecht (1988) and Brandt (1990) notice that the continuative relative clause type forms a comment to a topic within the matrix clause, i.e., the head of the relative clause or the entire first clause. In this view, the whole sentence structure has to be analyzed as a coordinative rather than a subordinative structure. The division between restrictive and appositive relative clauses is legitimate as is the division between restrictive and continuative relative clauses. However, as will be shown in the upcoming section, the anchor of a continuative relative clause cannot be regarded as a topic in the strict sense but rather as an import within a presentational strategy. The continuative relative clause, then, forms the process part for the presented item. The notions of topic and import need to be kept apart, mainly because topic is an intrasentential category, whereas import refers to intersentential relations (see 4.2.). The taxonomy presented in Lambrecht (1988) and Brandt (1990) is also problematic from another perspective: It appears that there is no clear semantic and/or syntactic distinction between the appositive and the continuative relative clause type, but merely a topological one. Thus they could be of the same basic type: Both seem to display independent focus-background structures and both can be used as process parts with respect to the strategy of (re-) presentation. The only remaining difference is that continuative relative clauses obligatorily occur in extraposition whereas appositive relative clauses are free to be placed right-adjacent to their heads, cf.: (46) a. continuative relative clause: I invited [Max]I to my birthday party [, who immediately started talking about Paul’s accident]P. b. appositive relative clause: I invited [Max]I, [who immediately started talking about Paul’s accident,]P to my birthday party. Lambrecht’s coordination test shows that the relative pronoun within appositive relative clause can be replaced by a resumptive personal pronoun in the second conjunct, too: (47) I invited [Max]I, [who immediately started talking about Paul’s accident and who/OKhe doesn’t seem to be too sympathetic with him,]P to my birthday party. Stefan Huber 64 Finally, both continuative and appositive relative clauses equally assert rather than presuppose the embedded proposition and they also have both a parenthetic flavor to them.25 So far, continuative and appositive relative clauses seem to display the same formal and interpretative properties and there is no need to keep them as separate categories.26 Maybe there are some divergences on a textual level, but I will have to leave this question to further research. 4 Back to Theory: The Relevance of Import-Process The definition of presentation provided in the beginning of section 2 stated that presentation is a two-step measure, in which an item is (re-) introduced in the context and established in a second sentence. What has been called “import” throughout this paper is a category that seems comparable to the one of aboutness topic. Both den Dikken (2005) and Endriss & Gärtner (2005) argue that the presented material should be considered some sort of (pre-) topic as well. Also, the imported entity is arguably not quite as salient to the hearer as truly given information. Thus, one could assume that imported items necessarily need to be new information, even though they might be definite. Finally, it seems that an import must be part of the focus domain of the sentence in which it occurs, whereas the process part does not necessarily have to meet this requirement. The aim of this section is to address the theoretical justification of the introduced layer of presentation. In particular, I will show that the dichotomies of topic-comment, theme-rheme, and background-focus must be distinguished from the layer import-process. We will start with the comparison of theme-rheme and import-process. 4.1 Import-Process vs. Theme-Rheme In section 2 it was argued that a presentation can involve both indefinite and definite DPs. The prototypical case might be an indefinite and non-given, i.e., rhematic element, but also thematic elements occur: (48) Yesterday, I met [an/the old man]I. [He was very angry.]P 25 There might, however, be a difference between the continuative/appositive and the presentational type. It could be claimed that a sentence like (40d) above displays one single focus-background structure, i.e., that the relative clause is integrated in the focus domain of the matrix clause. See also 4.3. below. 26 This is mainly problematic for Lambrecht (1988), who offers strongly diverging syntactic analyses for appositive and continuative relative clause constructions. On the theory of presentation 65 The claim made above was that in both cases the imported DP was specific. The purpose of the specification is to establish the introduced DP as a new discourse participant. Apparently, the term “introduction of an item into an ongoing discourse” implies some sort of newness. However, as (48) shows, the category “import” does not have to contain (completely) new information which the speaker assumes not to be present in the hearer’s mind at all. There have been various attempts to differentiate thematic and rhematic material (amongst others Gundel, Hedberg, & Zacharski 1993, Birner 1994). For our purposes, a two layer model of (non-) givenness might suffice. Thus, one could argue that indefinite imports are all-new, whereas definite imports are given, but not activated (in the sense of Gundel, Hedberg, & Zacharski 1993). The following table shows the difference: (49) Given Non-given Activated A B Non- activated C D Given and activated material (A) can be considered truly thematic, whereas non- given and non-activated material (D) is truly rhematic. In this sense, D can be understood as the category of indefinite imports. Given, but non-activated material (C) is information that has been mentioned previously but cannot be assumed to be completely salient in the hearer’s mind at the very point of the discourse. Therefore, it needs to be re-introduced. Definite imports ought to belong to this category. The residual category “non-given, but activated” (B), might be information that is not evident from the context, but from the cotext, i.e., from the circumstances that surround the speech situation, but have not been verbally expressed. As for the process part, I suggest that the information involved has to be non- given. Although it might contain some elements previously mentioned, 27 the entireness of the proposition must be new. Otherwise it would be irrelevant in the sense of Grice (1975). The very notion of focus by definition demands some form of newness. 27 As Molnár (1993) shows, focal material can consist of purely given material: (i) Speaker A: Apologize to Sandra. You insulted her! Speaker B: No, [SHE insulted ME.]F None of the lexical material in the B-part is new (or “non-activated”), not even the verb insult. Still, the semantic roles and thus the perspective of the elements involved are different. Therefore, the utterance becomes relevant. Stefan Huber 68 (53) (Speaking of Paul, I have to tell you this:) In 1996, Paul met this woman. She totally bewitched him. (Later he even married her) It was suggested in 3.1.2. that indefinite demonstratives such as this woman in (53) obligatorily require the presentation of the denoted element. Hence, a sentence without a process part would be infelicitous (see 23’). However, in (53) there is not only a proper aboutness topic in the sentence (Paul), but also a frame topic (in 1996). It has been suggested that there might be several topics within one sentence (see e.g., Rizzi 1997) and they arguably must all appear within designated functional projections in the C-domain. A DP in an object position such as this woman can hardly be argued to occupy such a projection. Therefore, the imported item this woman cannot be a topic. I suggest the following information structure for (53): (53’) [In 1996]FT [Paul]AT1 [met [this woman]I.]C1 [[She]AT2 [totally bewitched him.]C2]P Note, too, that even a process itself can be divided into a topic-comment structure (AT2 – C2). This data provide strong evidence that the layers import- process and topic-comment must be kept apart. If one agrees with Gärtner (2001) in the assumption that sentences like (52) must be considered coordinative rather than relative constructions, there is yet another interesting fact that corroborates the necessity of distinguishing between these two layers of information structure. Topic and comment are notions that are tied to one single sentence. Import and process, on the other hand, pertain to more ore less independent sentences. Thus, the dichotomy import-process is rather text-linguistic in its nature whereas topic-comment mainly applies on an intrasentential level. Summing up: It seems necessary to make a clear distinction between the layer of topic-comment and the layer of import-process. Despite some intuitive similarities between these categories, there is an array of crucial differences as well. The most important fact is certainly that the two dichotomies can occur simultaneously in different positions of the same sentence(s). More generally speaking, it appears that topics must be realized in the beginning of a sentence whereas imports tend to occur sentence-finally.30 Also, topic and comment are 30 One question that I have to leave to further research is whether imports could ever occur in the topic position and whether the categories of topic and import could coincide. It is relatively easy to construct examples with an imported item in the first position, cf. the following V2 relative clause construction with a fronted head: (i) Sachen gibt’s, die gibt’s gar nicht (= Gärtner 2001:97) things are-there they are-there at all not On the theory of presentation 69 intrasentential categories, while import and process refer to intersentential relations. In order to avoid terminological confusion, I believe it is beneficial to dispense with the notion “pre-topic” (Endriss & Gärtner 2005) and to adhere to the new term “import” instead. 4.3 Import-Process vs. Background-Focus If there are intuitive parallels between the dimension of topic-comment and the dimension of import-process, things become slightly more complicated when it comes to the comparison of the layers import-process and background-focus. Note first that in the prototypical case an imported item seems to belong to the focus domain of the sentence in which it occurs. The process part, on the other hand, seems to be contained in a second sentence with an independent focus- background structure, cf. the focal structure of (53): (53’’) [Paul]B1 [met [this woman]I]F1. [[She]B2 [totally bewitched him.]F2]P This fact alone is evidence enough to keep the dichotomies background-focus and import-process separated because they refer to different informational domains. But there are more arguments for such a distinction: It is interesting to notice that some of the V2 relative clause constructions discussed in 3.1.3. can be uttered as single-focus structures, especially those that Lambrecht (1988) labeled “presentational relative clauses” (cf. ex. 40d above). A German example for such a construction would be (52), which we have already discussed above. The background-focus analysis of (52) is given in (52’): (52’) [Es gibt natürlich [Philosophen]I, [die kommen aus Grönland.]P]F Given the assumption that this structure involves the coordination of two syntactically independent V2 clauses we find that the maximum focus domain includes both clauses. However, within this single focus domain a presentation, i.e., a division into an import and a process part, can take place. ‘There are things that are just unbelievable.’ Likewise, one might argue that the specific subject DP some people in (ii) seems to be processed by the following sentence and hence properly presented: (ii) Some people can’t just be helped. They want to make careers, but they don’t want to work. Note, though, that even if one allows for a description of the fronted DPs in (i) and (ii) as topics and imports in one, the corresponding comments and processes differ from each other (intra- vs. intersentential realizations, see above). Stefan Huber 70 Thus, the categories of import and process can be found within one or two focus-background structures, which, in turn, demonstrates the importance of assuming separate information-structural levels. 5 Conclusion The theoretical claim put forward in this paper has been that an independent, intersentential layer of information structure must be assumed in order to capture the idea of introducing an item into the ongoing context. Thus, presentation was defined as a method of (i) mentioning the new or inactive discourse participant (import) and (ii) delivering enough additional specification to establish it as a constituent in the context (process). Not only were there a row of morphosyntactic patterns that confirmed the necessity of such an information- structural layer, but also theoretical considerations made the distinction of this layer from the traditional ones (theme-rheme, topic-comment, background- focus) inevitable. It was first and foremost due to the fact that import and process parts could be found independently from the other categories that their relevance became evident. As a summary I would like to offer the following extended version of Molnár’s (1991, 1993) model of information structure. Since I assume that presentation involves textual progression (i.e., intersententiality), I will call this layer context: (54) Sentence 1 Sentence 2 + (Sentential) Message Topic – Comment Hearer Theme – Rheme Speaker Background – Focus Context Import – Process Further differentiations might be necessary. These differentiations, as well as a more thorough investigation of the four layers’ complex interdependencies, will hopefully be provided by future research. 6 References Arregi, K. 2002. Focus on Basque movements. Doctoral Dissertation. Cambridge, MA: MIT. Averintseva, M. 2005. German right dislocation and discourse topic. Paper given at Sinn und Bedeutung 10. October 2005.