Printer Friendly

Focus as perspectivation (1).

Abstract

This paper aims at focus/background structures (FBSs) from a production point of view. Based on an analysis of retellings, we demonstrate that syntax-oriented conceptions of focus are applicable to these spoken data, resulting in minimal or maximal foci (Uhmann 1991). These foci do not always come up to the predictions of the textlinguistic quaestio model (Stutterheim 1997) with respect to focus realization. According to the quaestio model we would expect that the linguistic units referring to the respective events are focused. The minimal and maximal foei have different pragmatic functions that will be characterized as perspectivation process (cf. Glatz and Stutterheim, this issue) during conceptualization.

I. Introduction

This article aims at focus/background structures (FBSs) from a production point of view. We investigate the factors that determine why a speaker declares certain information as being focused and other as belonging to the (conversational) background. To determine these factors, it is obviously not very useful to confine the analysis to isolated sentences or, as it is common in many formal approaches to focus, to question-answer pairs. Rather, we must clarify how the focus/background structure of the respective utterance units unfolds and how it is linguistically realized in the overall text.

In developing a model of the production of utterances with FBSs, the insights of many focus theories cannot be taken over directly, because these theories investigate FBSs primarily as a parsing and interpretation problem (see, e.g., Bosch 1998; Buring 1997; Hajicova et al. 1998; Jackendoff 1972; Jacobs 1991; Rooth 1985, 1992; Sgall and Hajicova 1977). Issues associated with the relation between conceptualization and focus determination do playa central role in every comprehensive production model, but they have not been accounted for by these theories. However, this does not mean that our model of the production of utterances with focus/background structures must be developed from scratch. There are several theories that concern the linguistic realization of focus and that may directly be applied to a production model. There are theories dealing with the relation between syntax and phonology (e.g. Gussenhoven 1984; Rochemont 1986; Selkirk 1984), phonological analyses of focus expressions (e.g. Fery 1993; Uhmann 1991), and semantic approaches that tend to explain FBSs in terms of given and new information (e.g. Halliday 1967; Jackendoff 1972; Lambrecht 1994).

The nonapplicability holds substantially true for approaches to the meaning of focus that investigate the truth-conditions of sentences with FBSs. From a production point of view, the meaning of focus should be considered as a pragmatic one because cooperative speakers tailor their utterances to their beliefs about the addressee's interests and information states. (2) The emergence of FBSs is sensitive to these factors. We will show that a production-oriented, pragmatic account of the meaning of focus can be given in terms of perspectivation. Perspectivation, in tuna, is a ubiquitous process during conceptualization that can be traced back to choice from sets of options. Hence, a production-oriented approach to the pragmatics of focus can be given on the basis of the speaker's choice of informational units under certain constraints.

Our central assumption is that the notion of perspectivation leads to a deeper understanding of the cognitive mechanisms underlying the emergence of focus. Given the data, we can determine the bearers of focal accents and we are able to determine for each utterance unit syntactic constituents that express a focus by means of well-established projection rules. But a formal characterization of focus in terms of set-theoretic operations gives only some blurry impressions of the constraints that are at work during the speaker's determination of foci. As we will show, only two kinds of foci occur in the retellings, each with a specific function with respect to perspectivity. Minimal foci express contrast and require the speaker's monitoring of what has been said by her. Contrary to minimal focus domains, maximal focus domains express newness and are constrained by the communicative demand. These functions, however, are not specially marked by prosodic features.

1.1. Data used

As data, we used a corpus of offline retellings of the silent trick film Quest (3) to determine the aforementioned constraints. The movie lasts seven minutes. It shows a figure made of clay that, searching for water, passes through five inhospitable environments--a sand world, a paper world, a stone world, a metal world, and a world of machinery. Therefore, the movie consists of five clearly distinguishable episodes, each showing the events that happen to the clay figure in the respective environment. During its search, the figure is leaving the respective environment (except the last, lethal, one) by falling through some hole into a new landscape.

The film was first shown to the subjects at full length. After that, the film was shown again, and it was then paused after the paper and the stone episode. The subjects were asked to relay what they perceived from the respective scene in such a way that listeners of their retellings, unfamiliar to the film, would be able to comprehend the overall events. In any case, the subjects had no reason to believe that the listeners knew the movie; hence the speakers should have based their utterances on a minimized common ground.

The responses were taped and transcribed. (4) We analyzed these texts with respect to word order, accent positions (nuclear and prenuclear), and pitch contours. This was done auditorily and by means of the program Praat (see Boersma and Weenink 1996). (5) For the purpose of this paper, we then selected spoken retellings of ten subjects. All the subjects were native speakers of German.

We evaluated the data with respect to two focus-related approaches, the textlinguistic quaestio model, and syntactic projection approaches. We will show in Section 4 that the hypotheses of the quaestio model with respect to focus assignment cannot be confirmed by our data. However, as we will demonstrate in Section 3, projection models are applicable to the spoken retellings. Based on these findings, we will finally show the different pragmatic functions of foci in our data that are describable as perspectivation phenomena.

At this point, a word on the status of spontaneous speech may be in order. We are fully aware of the methodological problems using spontaneous speech. There is no complete control of the relevant variables, but data that have been collected by the reading of texts would not exhibit the desired effects. We aim at processes of conceptual planning, and reading a text aloud does not provide valid data in this respect. However, one has to be aware that our procedure has several repercussions, especially that "in spontaneous speech (...) one cannot expect to be able to identify all intonation phrase boundaries with a similar degree of certainty" (Grabe 1998: 41).

1.2. Terminology

Before we present the empirical analysis, we have to define the basic notions related to focus/background structure.

We are using the term focus for a semantic/pragmatic notion referring to that part of a proposition that is put into the foreground. The remaining complementary part of the proposition serves as the background. That is, focus and background are propositional attributes. The foci are mapped onto syntactic constituents, the so-called focus domains. We use the term "focus domain" for elements of form, not content, that is, a focus domain is a manifestation of focus in a linguistic structure.

A focus domain can comprise of different amounts of terminal elements. If only one [X.sup.0] element constitutes a focus domain, we will call it a minimal focus domain. (6) A maximal projection can also serve as a focus domain. If the maximal projection is a CP, IP, or VP, it is commonly called a maximal focus domain. That means, the term "focus domain" is underspecified regarding which maximal projection it exactly comprises. In our data, we find mainly VP focus domains. Each focus domain contains a focus exponent--the bearer of the focus accent of the respective focus domain, the prosodically most salient element of this focus domain. (7) That means, the focus exponent is the [X.sup.0] element providing the nuclear syllable, carrying the nuclear accent. This may be illustrated by the following example (syllables carrying nuclear accents are marked by capitals)

(1) Roberto gewinnt morgen den Giro d'ITAlia Roberto wins tomorrow the Giro d'Italia

While it is easy to specify in this simple example that Italia serves as focus exponent (for the sake of simplicity we equate the focal accent with the nuclues accent), it is not that easy to determine the focus domain and the focus intended by the speaker. Approaches to the interpretation of sentences with FBSs use question-answer pairs as a methodological dodge for this. FBSs are context-related phenomena. This context is simulated by a question and the information requested in the question is focused in the response (see, e.g., Hartmann 2001; Rochemont 1998; Rochemont and Culicover 1990). With this method, it is possible to demonstrate that example (1) can have more than one focus domain. The focus domain can be made visible by means of adding negative contrastive adjuncts:

(2) Warum bist du so frohlich? why are you that cheerful

a. Roberto gewinnt morgen den [[Giro d'ITAlia].sub.FD] und nicht den Berlin-Marathon 'and not the Berlin-Marathon'

b. Roberto [gewinnt morgen den Giro d'ITAlia]FD und nimmt kein uppiges Mahl zu sich 'and will not have a lavish meal'

c. [Roberto gewinnt morgen den Giro d'ITAlia]FD und es ist nicht der Fall, dass ich morgen in den Urlaub fahre 'and it is not the case that tomorrow I'll go on vacation'

Once the focus domain is determined, the focus can--from the perspective of language comprehension--be derived from it. For example, in alternative semantics (Rooth 1985, 1992) this is achieved by set abstraction in a compositional way.

From a production point of view, the problem with these analyses starts with the determination of focus domains. (8) The simple relation between wh phrase and the respective focus domain ignores the relevance of the speaker's and listener's different information states for focus assignment. For example, it is possible that sentence (1) answers the question "what's new?" with minimal focus domain although this question typically asks for a maximal focus domain.

(3) Q: Was gibt's Neues? (what's new?)

A: Roberto gewinnt morgen den [[Giro d'ITAlia].sub.FD] und nicht den Berlin-Marathon

The reason for this is that it is solely the speaker who declares what is new information. The status of being "new" is not given per se; it is ascribed by the speaker. The determination of what is new and what given information is rooted in the speaker's estimation of the listener's beliefs. Example (3) is a well-formed question-answer pair if speaker A believes that she and the questioner share the information that Roberto is winning an important sports contest tomorrow, but the questioner does not know which one Roberto will probably win. In other words, whether a sentence with focus/background structure functions as answer to a context-simulating question is context-dependent by itself. (9) Hence, in language production we cannot use the well-established response test to determine the foci. (10) Instead the task for a model of focus planning is to describe how the speaker's beliefs of the listener's information state influence focusing.

It is argued that minimal foci (sometimes also called narrow foci) should be identified with contrastive foci. Roughly speaking, the reason is as follows: contrastiveness is typically characterized as the explicit rejection of alternative informational units from some closed set of alternatives (cf. Bartels and Kingston 1994). By means of this characterization, contrast becomes a gradual notion: every focus is more or less contrastive, but the smaller the focus, the more contrastive it becomes because the set of rejected alternatives becomes smaller and, therefore, more "salient." However, example (3) shows that minimal foci do not need to be contrastive in the sense given above. Whether a minimal focus is contrastive or not does not depend on the amount of terminal elements in the corresponding focus domain but on its function to reject or correct information. As we will show below, some utterances in the retellings have a contrastive function. Most of them are--due to the monological nature of the data--a kind of self-correction. Typically, these contrastive foci are indeed minimal ones, but this is due to the fact that the speakers corrected concepts that are associated with single words.

Our final terminological clarification concerns the notion of perspectivity. Perspective or point of view are quite familiar terms in linguistics. They are used in a variety of ways, referring to the perceptual or conceptual position in terms of which verbalized situations and events are presented (cf. Prince 1987: 73). (11) However, even though it seems intuitively plausible that perspectivation is a pervading phenomenon, it is very tricky to pin down what is actually the common core of "perspective" in all its different aspects.

In a spatial domain it may seem to be obvious what it means to take a perspective. When a speaker aims at describing the relative location of objects, taking a perspective is comparable to the respective camera angle taken. Accordingly, the location of one and the same object can be described in different manners, depending on which vantage point the speaker takes, if and how salient objects are accessible within the visual and/or communicative context, and which inherent spatial properties the respective object has. In spatial descriptions, speakers have the option to choose between a limited number of origines of a restricted number of axial systems in order to locate objects relative to each other. Furthermore, in principle, the speaker has the option to choose between all objects to be mentioned as possible reference objects. Whether she uses the phone is to the right of the pencil or the pencil is in front of the phone in order to describe the location of both objects, depends on the chosen origo (pencil vs. speaker) and the choice of reference and primary object (pencil vs. phone), respectively. What spatial descriptions show is that perspective is parametrisable: the speaker has to choose the communicatively most relevant statement from a restricted set of options. For spatial descriptions, this involves the choice of the ongo, the reference and primary objects, the linearization strategy, and so on. The interplay of the chosen options results in the perspectivized spatial description. (12)

For focus/background structures, the relevant choice bases are already inherent in the semantic approaches to focus based on the notion of alternative sets, but we will try to put this into more concrete terms. The choice base is the discourse protocol that keeps track of what has been said by the interlocutors, or it is the sequence of events that the speaker perceived.

2. Focus planning in language production

Most psycholinguistic models of language production assume a division of the production process into three major stages, which are called, following Levelt (1989, 1999), conceptualization, formulation, and articulation. Conceptualization for language production comprises the selection, preparation, and lineanzation of prelinguistic information. These processes lead to a conceptual representation in a propositional format (the so-called preverbal message), which serves as the input to the formulator. Formulation involves the transformation of the output of the conceptual level into linguistic structures. This entails the selection of appropriate lexical items, and the retrieval of syntactic structures and prosodic information. The formulation process results in an articulatory plan. The articulation can be described as the realization of that phonetic plan and its execution on the motoric level.

The conceptual task of determining the content of an utterance can roughly be divided into two steps, which Levelt dubs macro- and microplanning respectively: The decision of which information is going to be verbalized, and the further tailoring of the propositional content selected during macroplanning, in order to ensure that this propositional content will be realizable by the linguistic means that the respective language provides.

This rough outline of necessary conceptual tasks for speaking is widely accepted in the psycholinguistic literature, and there is not much disagreement about the two-step process in conceptualization. However, it is currently anything but clear exactly which processes belong to macro-and microplanning and how they shape conceptual information for its linguistic realization. The same holds true for the various working models of conceptual preparations proposed in the natural language processing community.

2.1. Focus from a production point of view

Foci, focus domains, and focus exponents are realized on different layers in Levelt's production model. Foci are planned during conceptualization and provided in the preverbal message for their linguistic realization. The mapping of foci to focus domains takes place during syntactic encoding, and determining the focus exponent involves also phonological encoding.

However, the general model leaves the questions open why a speaker is planning foci and how this is done. While Levelt's model is principally able to answer the second question, it cannot be brought in to answer the first question, because it ignores the origins of the speaker's intention. (13) Levelt's model does not account for factors that enforce or drive speaking and how these factors influence the form of the linguistic utterance. Since focus/background structures are listener-oriented means of information structuring, the role of the speaker's beliefs of the listener's information state and its influence on the linguistic realization is not sufficiently describable. A more suitable approach is the textlinguistic quaestio model (Stutterheim and Klein 1989; Stutterheim 1997). The quaestio model takes the perspective of language production. It describes conceptualization processes and therefore also fundamental processes of focus/background structuring.

On the other hand, the quaestio model does not account for the relation between focused information and the position of the focus exponent that signals the focus. This relation is typically described by the percolation of an abstract focus feature [F] within the syntactic structure of the respective utterances. Therefore, in the next section we will first investigate whether these syntactic conceptions of focus result in focus domains that are also motivated by the quaestio model.

3. Percolation rules

It is widely agreed upon that the prosodic structure of utterances consists of a hierarchy of well-defined prosodic domain types as prosodic words ([omega]), phonological phrases ([phi]), intonational phrases ([iota]), and utterances. Within this prosodic hierarchy the level relevant for focus assignment is the intonation phrase. For reasons of simplicity we will assume that in case of maximal focus, a CP, IP, or VP is mapped onto t (see Hartmann 2001). It is also a quite common assumption that t is more tied up with semantic factors than the other categories of the prosodic hierarchy, even though there are several syntactic factors involved in the building up of intonation phrases. An intonation phrase has to have some minimal prosodic structure, including at least one phrase accent that serves as the nuclear accent and carries the nuclear tone, and a boundary tone (see, e.g., Fery 1993; Grabe 1998; Ladd 1996; Pierrehumbert 1980; Pierrehumbert and Beckman 1988; Uhmann 1991). However, a boundary tone is not a conditio sine qua non of the intonational phrase. Especially in spontaneous speech one will find many instances where boundaries tones are lacking. Therefore, intonational boundaries may also be marked by pauses or may only be determinable relying on syntactic or semantic criteria.

In accordance with Grabe (1998), we used as pitch-accent inventory for labeling the [F.sub.0] trace the bitonal H * + L and L * + H, and for the boundary specifications the boundary tones H% and L%. If the trailing tone is realized later than the postaccentual syllable, this delay is marked by the diacritic ">"; downstepped tones are marked by "!". Depending on the segmental material, the rones are object of several realizational effects, as truncation (as in [20]) or compression. For details, see Ladd (1996: 132ff.) and Grabe (1998: 149ff.). It has to be kept in mind that [F.sub.0] traces do not have phonological status. They are acoustic manifestations of phonological entities, as formants of vowels are, and they are not only manifestations of intonation, but also of several other factors as intrinsic pitch, etc.

In syntactic conceptions of focus, it is a common assumption that a syntactic feature [F] can be "freely" assigned to certain syntactic structures, which can be projected (or percolated, see Rochemont 1986: 33) under certain constraints, specifying a focus domain. That means, focus is taken to be a feature in the syntactic representation, which calls for a semantic and pragmatic interpretation besides a phonological and phonetic one (see Rooth 1996: 271).

The rules of percolation determine the class of admissable subtrees by which [F] and the accented terminal element serving as the focus exponent are related. We assume bottom-up percolation because it supports the incremental character of the formulation process and can be realized by an HPSG in a straightforward manner (for details, see Klabunde et al., in press).

As for German, [F]-percolation in HPSG-based syntactic formulation is guided by the following standard projection rules (see Gunther [1999] for a comprehensive overview): if the focus domain is not a verb phrase, the complement becomes the focus exponent. If the complement is itself a phrasal sign, the focus exponent is the deepest embedded complement. This means, during the generation process we have to look recursively for the most deeply embedded lemma, which becomes the complement of a head.

If the focus domain is a verb phrase, the focus exponent is located in the verb-adjacent complement. Again, it is the most deeply embedded complement. However, if more than one complement exists, then the nonadjacent complement receives a prenuclear accent and the adjacent one becomes the focus exponent. This simple pattern changes if the verb phrase contains an adjacent adjunct. In this case the verb becomes the focus exponent and the complement within the adjunct receives a prenuclear accent.

In the following, we will have a look at an example from our retellings to check if these rules of [F]-projection are applicable. The example is the last part of the paper episode. The speaker is a woman. We will see that one characteristic of our data is that the number of syntactic patterns with more than one complement or adjunct to the verb are comparatively low. (14)

(4) [TEXT NOT REPRODUCIBLE IN ASCII].

First, in example (4) (see Figure 1) we have two [iota]s, clearly marked by pauses. Both comprise structures with a prenominal modifying or quantifying item, carrying accents. The thematic function of these accents is clear: marking the identity of the intention of the protagonist with respect to the previously seen and mentioned "paper level" (namely, to search for water), and the statement that identity holds with respect to this very level, using the indexical DIEser (in contrast to the former sand level). In both cases percolation rules predict a minimal focus domain.

According to the percolation rules, the respective focus-domain options should be das [[gleiche].sub.FD] spiel. This phrase is of relatively high idiomaticity. The noun spiel has a somehow idiomatic reading, and it seems awkward to use the phrase with an indefinite article

(5) ??ein gleiches Spiel/??a same game

because of the inherent definiteness of the restrictive adjective gleich/ same. The semantics of gleich/same implies an antecedens, i.e., the modificatum is in some sense necessarily given, even though is does not have to be explicitly introduced. The consequence is that an accentuation like

(6) ??das gleiche SPIEL

seems quite odd (except in case of metalinguistic accentuation). The only other possibility, namely to mark the complete NP as focus domain, seems to be an accentuation pattern like

(7) das GLEIche SPIEL

Therefore, we may take the adjective as the focus domain only.

One problem with the phrase auf [[dieser].sub.FD] ebene nochmal is the scope marking of nochmal. In case of "association with focus" (*nochmal auf dieser Ebene vs. nochmal auf DIEser ebene) we will have a minimal focus domain anyway. But from a semantic point of view, nochmal seems to be related to the following phrase, videlicet we have nochmal die Suche nach Wasser. However, the nochmal seems to be incorporated into the preceding prosodic domain. (15)

Second, we have the head-complement structure [die suche [nach [[wasser].sub.FD]FD]FD], which also aUows the highest node to carry [F]. The accent on wasser is auditorily very prominent, the F0 trace shows a downstepped !H* + L. (16)

Regarding the context, it does not make much sense to assume that there is any real competition between wasser and any alternative. It makes more sense to assume that this phrase has a presentational function, implying that the complete NP serves as focus domain. This is in accordance with Fery (1993:158f.), who shows that downstep, signalling equal prominence of accents, enables two accents to be integrated into a common focus domain. In our example, this is a thetic NP. According to the quaestio model, (4) would be classified as a side structure (for this, see the next section), but it is obvious that it serves as a headlining function, resuming the retelling after the break.

In example (8) (see Figure 2 for the pitch contour),

(8) [TEXT NOT REPRODUCIBLE IN ASCII.]

we have two [phi]s, which might be the manifestation of a kind of self-correction (note the cut-off of the final t in such), and a [iota] ending with a nuclear accent on stelle, which seems to be a reflex of phonological constraints (namely that every IP [...][sub.[iota]] has to contain a nuclear accent) without any information structural value. The syntactic structure breaks off and this accent serves as the nuclear accent, obligatorily contained in any intonational phrase. A similar phenomenon is commonly found at the left border of parenthesis niches. Actually, we have something similar here, if we take example (9) (see Figure 3) as a kind of parenthetical element (compare the flattening of the [F.sub.0] range). Characteristically, it attaches to the preceding element without a gap.

(9) so quasi

(10) [TEXT NOT REPRODUCIBLE IN ASCII.]

In example (10) we assume the VP as the focus domain, with aufgeweicht serving as the focus exponent. According to the percolation rules, there is the option that the whole sentence serves as the focus domain. However, from a pragmatic point of view that does not make very much sense, because, as already mentioned, this specimen is the end part of the description of the fate of the clay figure in the paper world. Therefore, paper is a well-established referent. The prenuclear H*+L accent on papier may be a topic accent. (17) Assuming also that the topic expression cannot be included in the focus domain, this rules out the possibility that the whole sentence functions as a focus domain.

In example (11) (see Figure 4), the VP is also a possible focus domain. The complement contains the focus exponent.

(11) [TEXT NOT REPRODUCIBLE IN ASCII.]

The same holds true for the directional complement in example (12), which also shows a prenuclear accent on blatter, prosodically clearly separated from (11).

(12) [TEXT NOT REPRODUCIBLE IN ASCII.]

him (DAT) fly leaves really in-the face

The side structure (11) is coherently integrated in the discourse, because it serves as an evaluation of (12).

In example (13) (see Figure 5), which relates back to (8) and (10), we have four [iota]s, the boundaries mostly marked by pauses.

(13) [TEXT NOT REPRODUCIBLE IN ASCII.]

The accented aufgeweicht in the third t again might be a manifestation of the necessary nuclear-accent constraint. Evidence is given by the lacking morphological agreement, if the content of the fourth t is to be interpreted as a self-correction, and if originally a predicative construction was targeted, which does not show morphological agreement in German. The accent on findet allows only a minimal focus domain. Remembering the fact that in (10) the soaked paper was already mentioned, we can assume that the intended string would have been ... [[FINdet].sup.FD] er dann so aufgeweichtes papier. The optional focus domain [aufgeweichtes [paPIER].sub.FD] might again be the outcome of the distortion of the originally planned string. On aufgeweichtes we have a prenuclear accent as well.

The examples (14) and (15)-(16) (see Figure 6) do not present any problems.

(14) [TEXT NOT REPRODUCIBLE IN ASCII.]

In this case greifen and patschen serve as the pair of alternatives.

(15) [TEXT NOT REPRODUCIBLE IN ASCII.]

(16) [TEXT NOT REPRODUCIBLE IN ASCII.]

The prosodic structuring of examples (17) and (18) is quite clear (see Figure 7). We have three prosodic phrases. In the first one, the accented dem, forcing a deictic reading, is only compatible with a minimal focus domain. The second [iota] shows a nucleus accent on wechselt, which, being ah unaccusative verb, may allow a maximal focus domain; the subject carries a certain amount of prosodic prominence. The third prosodic phrase is once more a product of the necessary nucleus accent constraint.

(17) [TEXT NOT REPRODUCIBLE IN ASCII.]

(18) [TEXT NOT REPRODUCIBLE IN ASCII.]

Example (20) (see Figure 8) provides no problem. In example (19), the verb-adjacent argument does not seem to contain the focus exponent, but the VP is nevertheless able to serve as focus domain. The reason is clearly that the prefix functions as an incorporated argument (see Stiebels 1996). Therefore, the utterance satisfies the relevant constraint. (Patscht and papier carry prenuclear accents, PApier showing a stress shift to the left to avoid stress clash.)

(19) [TEXT NOT REPRODUCIBLE IN ASCII.]

(20) [TEXT NOT REPRODUCIBLE IN ASCII.]

In examples (21) and (22) (see Figure 9) we have a clear case of a minimal focus domain with contrastive function. The boundary of t is marked by an H%.

(21) [TEXT NOT REPRODUCIBLE IN ASCII.]

(22) [TEXT NOT REPRODUCIBLE IN ASCII.]

The next utterance (see Figures 10 and 11) again provides us with an accent position that would license the VP to serve as focus domain.

(23) [TEXT NOT REPRODUCIBLE IN ASCII.]

(24) [TEXT NOT REPRODUCIBLE IN ASCII.]

To sum up, applying the percolation rules results in minimal or maximal focus domains but no constituent in-between, as the aforementioned utterances show. For the sake of clarity we present them as the complete original text, marking only the nucleus syllables:

(25) und das [[GLEIche].sub.FD] [spiel].sub.[iota]] * auf [[DIEser].sub.FD] ebene [nochmal].sub.[iota]] * [die suche nach [WASser].sub.FD].sub.[iota] und ah er [ [SUCH].sub.FD].sub.[phi]] und [[FINdet].sub.FD] [dann].sub.[phi]] an einer [STELle].sub.[iota]] so quasi das papier [[ist AUFgeweicht].sub.FD].sub.[iota]] [vorher].sub.[phi]] [merkt man noch ein bi[e]sschen ne aggressive [UMwelt].sub.[phi]][iota]]FD] ihm fliegen BLATter richtig ins [geSICHT].sub.[iota] und ah ner [STELle].sub.[iota]] [[FINdet].sub.FD] er [dann].sub.[iota]] so [AUFgeweicht].sub.[iota]], so so [aufgeweichtes [paPIER].sub.[iota]]FD] sozusagen [WASser].sub.[iota]]FD] [[GREIFT].sub.FD] [hin].sub.[iota]] [[PATSCHT].sub.FD] [hin].sub.[iota]] und in [[DEM].sub.FD] [moment].sub.[iota]][[WECHselt].sub.[phi]] (*) wieder die [Ebene].sub.[phi]]FD] und er wird [WIEder].sub.[phi]] [PATSCHT dieses papier [DURCH].sub.[iota]]FD] es [BRICHT].sub.FD][iota]] also [naja].sub.[phi]] papier [[BRICHT].sub.FD] [nicht].sub.[iota]] es [[REISST].sub.[iota]]FD] es ** [[ist AUFgeweicht].sub.[iota].sub.FD]] und er FALLT wieder in ne NEUe [[ebene].sub.[iota]]

These focus domains have been derived by a syntax-driven approach. If we assume that focus domains and foci are linked in an unambiguous way (that is, each focus domain expresses one focus, and each focus can only be realized in one focus domain), the question comes up what the pragmatic functions of the corresponding minimal and maximal focus domains are. To determine these functions, we will introduce the textlinguistic quaestio model.

4. Focus/background structures in the quaestio approach

The quaestio approach views a text as a complex answer to an implicit question, the so-called quaestio. This quaestio represents the speaker's intention and provides constraints for the instantiation of conceptual domains (time, location, etc.) as well as for the realization of foci. The approach does not only make some predictions with respect to the referential instantiation of these domains but also with respect to the sequencing of the respective domains, i.e., with respect to the unfolding of these domains in the course of the text. This referential movement states whether discourse referents will be introduced as new ones, or will be maintained or moved referentially (e.g., in the temporal domain verbalized by and then ... and then ... and then ...).

The quaestio sets global constraints for textual development. Narratives, for example, answer the quaestio "what happened to protagonist P at time t?" or "what did P do at t?" In order to satisfy the informational demand, predicates will be used that refer to eventualities or actions. The temporal domain will be structured by specific temporal intervals. The order of the intervals is determined by the chronological order of the respective events. Modality is confined to factivity, the protagonist is instantiated as global topic. Other domains, such as locality, are not addressed by this quaestio.

However, every text does not only consist of parts answering directly to the quaestio (the so-called main structures), but it contains other parts as well, the side structures, which are nevertheless integrated in a coherent way. Side structures, for example, give the speaker's evaluations or additional information. Side structures have to be integrated in the text in a very specific way, which can also have an impact on prosodic structuring. Our data, the retellings, are based on the same quaestio but their text structure differs from that of narratives. While narratives refer to one reference world only, retellings refer to two worlds: the speaker's perception of the movie and the events shown in the movie. Therefore, retellings are blends of narratives with descriptions (cf. Dietrich 1992; Stutterheim 1997; Weinrich 1964). Consequently, every utterance that does not function as a proper part of a narrative is classified as a side structure. Although the quaestio approach seems to fail in this respect, this does not cause a serious problem for our analysis of focus/background structures. We are after the information demanded by the quaestio because this should be focused.

The quaestio induces a set of candidates from which one member will be chosen by the corresponding answer. The aforementioned quaestio induces a (very large) set of possible situations from which the answer will be a member. This set of possible candidates is called "topic" in the quaestio approach. The selected candidate is called "focus." 18 Note that this topic-focus distinction corresponds to the notions of alternative sets and focus in alternative semantics. Since the quaestio approach aims primarily at the structure of texts, its predictions with respect to focus is not confined to single sentences. The quaestio sets conditions for these alternative sets and foci.

It is important to state that there is a difference between fixed global instantiations of a referential domain (e.g. the protagonist or the temporal frame) and constraints on the possible relations between utterances (e.g. constraints on referential movement). By these means it is possible to specify constraints regarding possible focus/background structures for each utterance of the main structure of the text (contrary to side structures). Therefore, these constraints provide some necessary conditions for the selection of the means to verbalize the respective focus/ background pattern.

Global constraints can have a direct impact on single utterances. For example, the fixed instantiation of the protagonist as a global topic results in its being the primary candidate for the subject of the respective utterances (cf. Murcia Serra 2001).

In summary, the quaestio approach predicts for retellings that they function as an answer to the question "what did protagonist P do at time t?," which requires to determine a protagonist and to order the respective events chronologically. Since these events are what is asked for, they belong to the focus of the utterances.

Since in German the focus/background structure is coded primarily by means of accent position and word order, we would expect that these formal criteria in the respective utterances allow us to derive the informational distribution of the respective conceptual domains. But as our data show, this turns out to be problematic.

4.1. Foci in the quaestio model

Do the predictions made by the quaestio model match the foci we find in our data? In order to answer this question, we first have to determine in our data the focal accents. In a second step, we have to derive the focus domain from the accent position. The focus domain can unambiguously be mapped onto the underlying focus.

If the quaestio is interpreted in analogy to an explicit question, there must be a syntactic correspondence between this question and finally the single utterance serving as a partial answer to this question. This would match the approaches we have mentioned already, which use a question to simulate the context, rating the contextual aptness of a sentence with focus/background structure serving as an answer to this question.

However, such a naive quaestio to text relationship cannot be derived from the data.

Our starting assumption would be that the linguistic forms of the respective utterances in our video retellings allow the complete VP to serve as focus domain, if they are able to serve as an apt answer to the quaestio verbalized as "What happened to the protagonist?" However, in some cases our data do not allow this possibility. It is either the case that if word order would license a focus domain that enables the utterance to serve as a proper answer to the question, the accent position blocks the complete VP as focus domain, or it is the case that the accent position allows this interpretation, but word order blocks it. Sometimes both conditions occur.

To discuss this problem in more detail, let us have a closer look at the beginning of a further video retelling. This text covers the first episode of the movie. (The speaker again is a woman; we give only a selection of pitch accents.)

(26) in einer grossen [[SANDwuste].sub.[phi]1] * sind * FELSteile [zu sehen].sub.[phi]2]]t

in a big sand desert are rock parts to see

(27) die sich nach ner weile [erHEben].sub.t]

which REFL after a while raise

(28) [TEXT NOT REPRODUCIBLE IN ASCII.]

(29) das [AUFsteht].sub.t] * ausm [SAND].sub.t]

who gets up from=the sand

(30) [TEXT NOT REPRODUCIBLE IN ASCII.]

(31) [TEXT NOT REPRODUCIBLE IN ASCII.]

(32) nichts [DRIN].sub.t]

nothing in it

(33) und fangt dann an zu [BUDdeln].sub.t] * ] *

and starts then to dig

(34) [TEXT NOT REPRODUCIBLE IN ASCII.]

(35) [TEXT NOT REPRODUCIBLE IN ASCII.]

Example (26), the opening utterance of this text, contains two [theta]s, jointly marking one focus domain, which shows a kind of stacked structure, [[theta].sub.1], containing a PP, carries a phrasal accent on its complement. Therefore it is able to mark the complete PP as a focus domain by itself. As (26)-(28) are to be classified as a presentational structure, (26) gives the general setting, which has to be given first, before the speaker focuses on one specific part of that scene. The position of the nucleus accent on felsteile is in accordance to the general percolation rules. If we stayed with the quaestio for narratives given above ("what happens to P at t?" or "what did P do at t?") (26)-(28) would have to be classified as side structures. However, this classification would not do justice to the function of these utterances for two reasons. First, they are the very opening of the text. Second, they give essential information for the understanding of the text closely linked to the following main structures by the relative pronoun das in (29). This is one manifestation of the fact that retellings of movies instantiate a text structure that is a hybrid between a narrative and a description, because they contain two embedded reference worlds, the ficticious one and the real one in which the speaker perceives the course of events. According to the corresponding quaestio, all utterances (26)-(28) should display a maximal focus domain. A maximal focus domain is in accordance with word order and position of the nucleus accent in (27), which allow focus projection. However, (28) does not provide this possibility. The nucleus accent is clearly on mannchen (see Figure 12). A maximal focus domain is unlikely because in the given context the little man stands in contrast to the mentioned rock parts. The speaker perceived some rock parts, which turned out to be the little man after some time. Therefore, we assume a minimal focus domain with contrastive function.

[FIGURE 12 OMITTED]

Example (29) provides two ts, both marking focus domains by themselves, the second one marking an exbraciation. It is a very common technique to exbraciate a phrase, if the speaker wants to present it as an autonomous piece of information without distorting the information structure of the main clause. The quaestio would induce a focus mapped onto a focus domain like das [ausm sand A UFsteht], especially because "sand" was already introduced into the discourse in (26). That the speaker chose to exbraciate the PP ausm sand might be motivated by the transition from a descriptive to a narrative text part, which makes it necessary to gap the conceptual distance between sandwuste in (26) and sand in (29), induced by the text-structural border.

Example (30), carrying the nucleus accent on flasche (see Figure 13), does not allow for the VP to be taken as the focus domain, but only the DP eine flasche. However, the quaestio licenses only a maximal focus domain here. A maximal domain is provided by the subordinated clause in (31).

[FIGURE 13 OMITTED]

In (32), a side structure utterance provides essential information. The negation serves as a focus particle. The next utterance, (33), again presents the complete VP as focus domain, in accordance to the quaestio. Due to the position of the nucleus accent on the head noun of the object NP, utterance (34) allows only a focus domain [ein [tiefes [loch]]] (see Figure 14). The utterance provides important additional information to (33), which explains the disappearence of the clay figure, reported by (35), again with a maximal focus domain. The fact that (35) shows a downstepped !H* +L on the nucleus syllable is probably associated with its location in a subordinated clause. (19)

[FIGURE 14 OMITTED]

We have seen that the syntactic percolation rules are applicable to our data in principle, and that they allow us to determine for each utterance unit a maximal or minimal focus domain. The maximal focus domains correspond to the foci that are derivable from the underlying quaestio. In other words, the minimal focus domains seem to have a specific status with respect to their informativity.

5. Preliminary conclusions

Our previous analysis of the video retellings has shown three things: first, focus accents are not standardized as specific contours. Rather there is only a tendency not to realize focal accents by a rising contour. This contrasts sharply with the assumption that foci are realized phonologically in a standardized way).(20) Cohan (2000, 2001) shows convincingly that a standardized phonological realization does not hold for English as well. Second, taking the nucleus accents as focus accents, possible focus domains in our data of spoken German are determinable by syntax-driven pereolation rules. Third, if we assume a one-to-one mapping from focus domains onto foci, the resulting foci are semistandardized: they are minimal of maximal focus domain.

These results are completely formulator-related. They show that if conceptual material in the preverbal message is marked as being in focus, there are language-specific mechanisms that determine the focus exponent during syntactic and phonological encoding. The nonstandardized pitch contour associated with focus shows the existente of additional planning mechanisms that influence the phonological realization of stress.

However, we claimed that our analysis allows us to derive statements about perspectivation that takes place during conceptualization. So what do the previous analyses tell us about perspectivation?

6. Focus and perspectivation

Focus determination, as one aspect of perspective taking, is highly context sensitive, as is, for example, the choice of a spatial perspective (Grabowski and Miller 2000). Going back to the analysis of our data given in the previous sections, we see that the respective focus/background structures have two--very well known--functions that can be associated with two different aspects of conceptual preparation:

a. Marking the information that is new for the listener. This decision clearly belongs to the macroplanning tasks where a speaker decides which information to communicate and in which order. If a speaker selects certain information for verbalization she must also compare this informational unit with her beliefs of the addressee's information state in order to be communicatively relevant. An informational unit is in focus if the speaker believes that this information is not known by the addressee.

b. Contrasting information with already verbalized information. This includes the self-corrections in our data. Contrary to the previous function of focus, contrastiveness involves monitoring and is primarily the result of checking the current utterance with the discourse protocol.

These two functions of focus are localized at different stages in conceptualization. However, we have seen that these different conceptualization demands are not marked prosodically in different ways. Rather we find a strong correlation between the maximal and minimal focus domains we have determined by means of the percolation rules and the demands of the quaestio, and these two functions: minimal foci express contrast and require monitoring, maximal foci express newness and are constrained by the quaestio.

The focus-related linkage between perspectivation and choice is as follows: the choice base is the mental representation of the perceived events shown in the movie. The data clearly show that the speakers choose maximal or minimal foci, but nothing in between. In case of maximal focus, they choose that information from the perceived events that was new to the listener. Since the protagonist of the movie, the clay figure, has a highly salient status throughout the whole movie, the corresponding referent belongs to the background. Minimal foci only occur in corrections and self-repairs.

During the situation of retelling the movie, the speakers have the possibility to choose any event they have perceived, but that choice is confined by the communicative demand of the quaestio which induces a temporal sequentialization as linearization strategy. Therefore, for each event the speakers choose, these informational units--which serve as foci that express newness--are compared with the previously mentioned information. Actually, the speakers do not have a real choice with respect to focus because there are no alternatives to the new information. Everything previously mentioned is given, and the forthcoming events should not be mentioned due to the linearization principle of temporal order, which is derived from the quaestio.

7. Conclusion

To sum up, the analysis of pitch contours, syntax, and pragmatics of the retellings shows that focal accents have two main functions: contrast with the speaker's previously given information (a kind of self-repair) and signaling newness.

There was no correlation between these two functions and the pitch contours. In other words, the function of contrasting was not specially marked by prosodic features. Whether the speakers correct themselves of whether they provide new information is not marked by a special contrasting accent (Dietrich 1990). This statement probably does not hold for contrasting accents in general. We assume that the missing contrasting accent is due to the monological nature of the retellings where only the speaker provides information to a silent listener. Real conversation does not occur.

In most utterance units, feature percolation describes the focus exponent. This is somewhat unexpected since the spoken retellings do not consist completely of syntactically complete and well-formed sentences.

Finally, we have defined perspectivation as choice between options. In this sense, FBSs are linguistic means to express perspectivation: the choice between what the speaker declares as new or correcting information and what is "given."

Notes

(1.) The research reported in this paper has been carried out while both authors worked at the University of Heidelberg at the Center for Computational Linguistics. The work was funded by the Deutsche Forschungsgemeinschaft in the priority program "Language Production" under grant He 1467/3-2. We would like to thank two anonymous reviewers for their very helpful comments and suggestions and Anne Marie Xavier for polishing up our English. Any errors are the full responsibility of the authors. Correspondence address: Ralf Klabunde, Sprachwissenschaftliches Institut, Ruhr-Universitat Bochum, 44780 Bochum, Germany. E-mail: klabunde@linguistics.ruhr-uni-bochum.de.

(2.) Tailoring utterances to the addressee is by no means a trivial matter. See, e.g., Speck (1995); Brown and Dell (1987); Del] and Brown (1991).

(3.) The movie was produced by Thomas Stellmach in 1997. It got an Oscar award.

(4.) This was done by Christiane ron Stutterheim and her research group. We are very grateful to them for having been allowed to use these data.

(5.) Pitch was analyzed in time steps of 0.01 s with a minimum pitch of 75 and a maximum pitch of 600 Hz. If necessary, octave jumps were killed, and pitch was smoothed with a bandwidth of 10 Hz. The time given in the diagrams refer to the total running time of the respective retellings.

(6.) Selkirk (1984) uses the term narrow focus. The two terms are very often identified with contrastive focus.

(7.) Often the term focus is used in the sense of "focus exponent" (e.g. Buring 1997: 28; Steedman 2000: 656) or as a term for the focus domain (e.g. Rochemont 1998: 337; cf. also Coban 2000: 49); but we will keep strictly to our terminological differentiation, because it puts emphasis on the essential differences between the conceptual and semantic category focus and focus expression as its linguistic realization. Therefore, we will speak of VP focus domains, IP focus domains, etc., not of VP focus, etc.

(8.) Actually, the problem starts with the determination of the focal accent already. Assuming that the focus exponent is identical with the nucleus accent contradicts the facts. Furthermore, the focus exponent is not realized by a kind of standardized pitch contour as many semanticians assume. If this would be the case, the analysis of the data would be quite simple: one has to look for such specific pitch contours and to check whether it signals a focal accent. As Cohan (2000) shows for English, nucleus accents are not necessarily focal accents, and they are realized in different ways. This corresponds completely with our analysis of the retellings. The retellings show only that there is a tendency not to realize focal accents by a rising contour.

(9.) As Tancredi (1992) states, "the test [which he dubs the 'natural response test'] can only be used to determine the potential semantic foci compatible with any given phonological accent assignment in a sentence, and cannot (except in extreme cases) be used to determine what the actual semantic focus of a sentence is in a given instance" (Tancredi 1992: 64).

(10.) An additional drawback of this test is the taking-over of structural features of the question by the answer. See, e.g., Bock (1986).

(11.) There are also some other termas used, such as viewpoint or subjectivity (cf. Smith 200O).

(12.)If the main notion in perspectivation is choice, then the permeating nature of perspective taking is reflected in a variety of linguistic forms where choice between options becomes relevant. This includes, for example, the choice of communication, consciousness, and evaluative verbs to describe a situation, but also aspect and prosodic means.

(13.) "Where intentions come from is not a concern of this book" (Levelt 1989: 59).

(14.) The asterisk "*" marks a pause; "!" marks a downstepped tone.

(15.) According to the rules given by Nespor and Vogel (1986).

(16.) For reasons of space we will not go into detail regarding downsteps. See Fery (1993: 155ff.): Ladd (1996: 74ff.); Grabe (1998: 185ff.).

(17.) More on topic accents, see Fery (1993).

(18.) Cf. Stutterheim: "Wir nennen diejenigen Elemente, die die Menge ron Kandidaten, aus der einer zu spezifizieren ist, festlegen, die Topik und die Spezifikation selbst den Fokus einer AuBerung (vgl. Rooth 1985)" (Stutterheim 1997: 36).

(19.) See Fery (1993: 159).

(20.) For example, Buring claims that "in English and German, F [focus] is usually realized by a falling tone (H* L or H* L% [...])" (Buring 1998: 3).

References

Bartels, Christine; and Kingston, John (1994). Salient pitch cues in the perception of contrastive focus. In Focus & Natural Language Processing. Vol. 1: Intonation and Syntax, Peter Bosch and Rob van der Sandt (eds.), Working Paper 6, 1-10. Heidelberg: IBM Deutschland Informationssysteme GmbH, Scientific Centre, Institute for Logic and Linguistics.

Bock, Kathryn (1986). Syntactic persistence in language production. Cognition 31, 163-186.

Boersma, Paul; and Weenink, David (1996). Praat, a System for doing Phonetics by Computer, version 3.4. University of Amsterdam, Institute of Phonetic Science. Report 132. See also www.praat.org.

Bosch, Peter (ed.) (1998). Focus and Natural Language Processing, Linguistic, Cognitive, and Computational Perspectives. Cambridge: Cambridge University Press.

Brown, Paula M.; and Dell, Gary S. (1987). Adapting production to comprehension: the explicit mention of instruments. Cognitive Psychology 19, 441-472.

Buring, Daniel (1997). The Meaning of Topic and Focus. The 59th Street Bridge Accent. Routledge Studies in German Linguistics 3. London: Routledge.

Buring, Daniel (1998). Focus and topic in a complex model of discourse. Unpublished manuscript, Cologne University.

Cohan, Jocelyn Ballantyne (2000). The realization of focus in spoken English. Unpublished doctoral dissertation. University of Texas at Austin.

--(2001). Distinguishing pitch accent from focus. Unpublished manuscript, to appear in Chicago Linguistic Society 37.

Dell, Gary S. and Brown, Paula M. (1991). Mechanisms for listener-adaptation in language production: limiting the role of the "model of the listener." In Bridges Between Psychology and Linguistics: A Swarthmore Festschrift for Lila Gleitman, Donna Jo Napoli and Judy Anne Kegl (eds.), 105-129. Hillsdale, NJ: Erlbaum.

Dietrich, Rainer (1990). Zu Forro und Bedeutung der Kontrastintonation im Deutschen. Linguistische Berichte 129, 415-430.

--(1992). Modalitat im Deutschen. Zur Theorie der relativen Modalitat. Opladen: Westdeutscher Verlag.

Fery, Caroline (1993). German Intonational Pattern. Linguistische Arbeiten 285. Tubingen: Narr.

Grabe, Esther (1998). Comparative Intonational Phonology: English and German. MPI Series in Psycholinguistics 7. Nijmegen: Max Planck Institute for Psycholinguistics.

Grabowski, Joachim; and Miller, George A. (2000). Factors effecting the use of dimensional prepositions in German and American English. Object orientation, social context, and prepositional pattern. Journal of Psycholinguistic Research 29, 517-554.

Gunther, Carsten (1999). Prosodie und Sprachproduktion. Tubingen: Niemeyer.

Gussenhoven, Carlos (1984). On the Grammar and Semantics of Sentence Accents. Dordrecht: Foris.

Hajicova, Era; Partee, Barbara; and Sgall, Petr (1998). Topic--Focus Articulation, Tripartite Structures, and Semantic Content. Studies in Linguistics and Philosophy 71. Dordrecht: Kluwer.

Halliday, M. A. K. (1967). Notes on transitivity and theme in English. Journal of Linguistics 3, 199-244.

Hartmann, Katharina (2001). Right Node Raising and Gapping: Interface Conditions of Prosodic Deletion. Amsterdam: Benjamins.

Jackendoff, Ray (1972). Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT Press.

Jacobs, Joaclftm (1991). Focus Ambiguities. Journal of Semantics 8, 1-36.

Klabunde, Ralf; Glatz, Daniel; and Ebert, Christian (i.p.). Zur Formulierung von Fokus/ Hintergrund-Gliederungen. In Sprachproduktion, Christopher Habel and Thomas Pechmann (eds.). Wiesbaden: Deutscher Universitatsverlag.

Ladd, D. Robert (1996). Intonational Phonology. Cambridge Studies in Linguistics 79. Cambridge: Cambridge University Press.

Lambrecht, Knud (1994). Information Structure and Sentence Form. Cambridge: Cambridge University Press.

Levelt, W. J. M. (1989). Speaking. From Intention to Articulation. Cambridge, MA: MIT Press.

--(1999). Producing spoken language: a blueprint of the speaker. In The Neurocognition of Language, Colin Brown and Peter Hagoort (eds.), 83-122. Oxford: Oxford University Press.

Murcia Serra, Jorge (2001). Grammatische Relationen in Deutschen und Spanisehen. Eine empirische Untersuchung zur Rolle der einzelsprachlichen Form bei der Konzeptualisierung ron Ausserungen im Text. Europaische Hochschulschriften, Reihe 21, Linguistik; 236. Frankfurt am Main: Peter Lang.

Nespor, Marina; and Vogel, Irene (1986). Prosodic Phonology. Studies in Generative Grammar 28. Dordrecht: Foris.

Pierrehumbert, Janet B. (1980). The Phonology and Phonetics of English Intonation. Cambridge, MA: MIT Press.

--; and Beckman, Mary E. (1988). Japanese Tone Structure. Cambridge, MA: MIT Press. Prince, Gerald (1987). Dictionary of Narratology. Lincoln: University of Nebraska Press.

Rochemont, Michael S. (1986). Focus in Generative Grammar. Studies in Generative Linguistic Analysis 4. Amsterdam: Benjamins.

--(1998). Phonological focus and structural focus. In The Limits of Syntax, Peter W. Culicover and Louise McNally (eds.), 337-363. San Diego, CA: Academic Press.

--; and Culicover, Peter W. (1990). English Focus Constructions and the Theory of Grammar. Cambridge Studies in Linguistics 52. Cambridge: Cambridge University Press.

Rooth, Mats (1985). Association with Focus. Unpublished doctoral dissertation, University of Massachusetts at Amherst.

--(1992). A theory of focus interpretation. Natural Language Semantics 1, 75-116.

--(1996). Focus. In The Handbook of Contempory Semantic Theory, Shalom Lappin (ed.), 271-297. Oxford: Blackwell.

Selkirk, Elisabeth O. (1984). Phonology and Syntax: The Relation between Sound and Structure. Current Studies in Linguistics Series 10. Cambridge, MA: MIT Press.

Sgall, Petr; and Hajicova, Era (1977). Focus on focus. Prague Bulletin of Mathematical Linguistics 28, 5-51.

Smith, Carlota S. (2000). Accounting for Point of View and Subjectivity. University of Oslo, Germanistisk Institutt. SPRIK report 4.

Speck, Agnes (1995). Textproduktion im Dialog. Zum Einfluss des Redepartners auf die Textorganisatian. Opladen: Westdeutscher Verlag.

Steedman, Mark (2000). Information structure and the syntax-phonology interface. Linguistic Inquiry 31,649-689.

Stiebels, Barbara (1996). Lexikalische Argumente und Adjunkte. Zum semantischen Beitrag von verbalen Prafixen und Partikeln Studia Grammatica 39. Berlin: Akademie Verlag.

Stutterheim, Christiane von (1997). Einige Prinzipien des Textaufbaus. Tubingen: Niemeyer.

--; and Klein, Wolfgang (1989). Textstructure and referential movement. In Language Processing in Social Context, Rainer Dietrich and Carl F. Graumann (eds.), 39-76. Amsterdam: North-Holland.

Tancredi, Chfistopher (1992). Deletion, Deaccenting and Presupposition. MIT Working Papers in Linguistics. Cambridge, MA: MIT Press.

Uhmann, Susanne (1991). Fokusphonologie. Eine Analyse deutscher Intonationskonturen im Rahmen der nicht-linearen Phonologie. Linguistische Arbeiten 252. Tubingen: Niemeyer.

Weinrich, H. (1964). Tempus. Besprochene underzahlte Welt. Stuttgart: Kohlhammer.

Received 8 February 2002

Revised edition received 5 June 2002

Institut fur Deutsche Sprache, Mannheim

Ruhr-Universitat Bochum
COPYRIGHT 2003 Walter de Gruyter GmbH & Co. KG
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Glatz, Daniel; Klabunde, Ralf
Publication:Linguistics: an interdisciplinary journal of the language sciences
Date:Sep 1, 2003
Words:9654
Previous Article:Verbs and nouns convey different types of motion in event descriptions *.
Next Article:Publications received between 2 June 2002 and 1 May 2003.
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters