Linguistik
Refine
Year of publication
Document Type
- Part of a Book (591) (remove)
Language
- English (591) (remove)
Has Fulltext
- yes (591)
Is part of the Bibliography
- no (591) (remove)
Keywords
- Syntax (79)
- Spracherwerb (63)
- Deutsch (56)
- Phonologie (46)
- Semantik (42)
- Englisch (40)
- Sprachtest (33)
- Thema-Rhema-Gliederung (32)
- Intonation <Linguistik> (25)
- Morphologie (24)
Institute
- Extern (11)
- Sprachwissenschaften (1)
In this paper I put forward and justify a syntactic configuration that I call Complex Small Clause-structure. I show that this single syntactic structure can explain both the semantic value and the syntactic behavior of a range of constructions that up to now have been explored separately and, hence, proposed divergent analyses among them.
Even if we can generate a logical form, principles of use may limit the ways in which we can use it. In this paper, I motivate one such principle of use, and explore its effects. Much of the discussion involves kinds of sentences that have received attention in the literature on "individual-level predicates".
The role of migration and language contact in the development of the Sino-Tibetan language family
(2001)
The effects of different forms of predication have been insightfully (and almost exclusively) studied for 'simple' cases of predication, of which the 'presentational sentence' is maybe the paradigm instantiation. It is the aim of this paper to show that thc same kind of effects as well as in fact the same kind of structures are present at embedded levels in thematically and otherwise more complex structures. Beyond presentational sentences, 'unaccusative' experiencing constructions involving a dative subject, 'double object constructions' and - to a lesser extent - spraylload constructions are discussed. For all of these, it is argued that they comprise a predication encoding the ascription of a transient temporal property to a location. On this basis, a proposal is made as to how the scope asymmetry between the two arguments involved in the colistructions can be explained. Furthermore, a proposal is made as to how what has been called 'argument shift' is motivated.
This paper compares secondary predication constructions (including small clause complements, resultatives, and/or depictives) in English and Korean and argues that these two typologically different languages employ different modes of satisfying the Case Filter (Chomsky 1981) with regard to the Case of the subjects of secondary predication constructions. More specifically, we argue that the subject of the secondary predicate in English is Accusative Case-marked by the higher governing verb, while that in Korean is Nominative Case-marked by default. Evidence for default Nominative Case will be provided from Korean and other languages.
In this paper, I argue that this informally given list of characteristics covers only a certain subclass of specific indefinites. […] In particular, I dispute the definition of specific indefinites as "the speaker has the referent in mind" as rather confusing if one is working with a semantic theory. Furthermore, I discuss "relative specificity", it. cases in which the specific indefinite does not exhibit wide, but intermediate or narrow scope behavior. Based on such data, I argue that specificity expresses a referential dependency between introduced discourse items. Informally speaking, the specificity of the indefinite expression something [...] expresses that the reference of the expression depends on the reference of another expression, here, on the expression a monk, not the speaker.
This paper pursues the question what the implications of the Anti-Locality Hypothesis could be for the syntax of secondary predication. Focus of the discussion will be an investigation of what their internal structure of small clause complements must look like, how these small clause complements connect to their matrix environments, and what the relevance could be for the formulation of anti-locality presented here. Anti-locality is defined over a tripartite clause structure (split into three Prolific Domains) and a PF-condition on the computation (the Condition on Domain-Exclusivity). The investigation revolves around two leading questions: (i) does the syntax of small clauses involve more structure than simply [SC DP XP] and (ii) do small clauses constitute their own Prolific Domain (or maybe even more)? The results, affirmative answers to both questions, are also relevant for other types of secondary predication.
Since the advent of nonlinear phonology many linguists have either assumed or argued explicitly that many languages have words in which one or more segment does not belong structurally to the syllable. Three commonly employed adjectives used to describe such consonants are 'extrasyllabic', 'extrametrical' or 'stray'. Other authors refer to such segments as belonging to the 'appendix'. [...] Various non-linear representations have been proposed to express the 'extrasyllabicity' of segments [...]. The ones I am concerned with in the present article analyze [...] consonants [...] structurally as being outside of the syllable [...]. For transparency I ignore here both subsyllabic constituency as well as higher level prosodic constituents to which the stray consonants are sometimes assumed to attach. For reasons to be made clear below I refer to syllables [...] in which the stray consonant is situated outside of the syllable, as abstract syllables.
Internals from elaboration
(2001)
Elaboration or Narration, as so-called discourse relations (or rhetorical relations), are modelled in Segmented Discourse Structure Theory (SDRT) as relations between discourse constituents (or constituents for short). These are either propositions that come into being by interpretation of sentences occurring in a text; the propositions then have the status of DRSes. Or, constituents are compounds of such DRSes, constructed from DRSes (or compounds of them) by discourse relations. Elaboration and Narration in that sense, rather than referring to text types, provide links between constituents that allow them to combine in ways that, for a recipient, a resulting text is coherent and has (some) elaborative or narrative properties.
The ultimate goal of this paper is to find a representation of modality compatible with some basic conditions on the syntax-semantic interface. Such conditions are anchored, for instance, in Chomsky's (1995) principle of full interpretation (FI). Abstract interpretation of modality is, however - be it "only" in semantic terms - already a hard nut to crack, way too vast to be dealt with in any comprehensive way here. What is pursued instead is a case-study-centered analysis. The case in point are the English modals (EM) viewed in their development through time - a locus classicus for a number of linguistic theories and frameworks. The idea will be to start out from two lines of research - continuous grammaticalization vs. cataclysmic change - and to explain some of their incongruities. The first non-trivial point here consists in deriving more fundamental questions from this research. The second, possibly even less trivial one consists in answering them. Specifically, I will argue that regardless of the actual numerical rate of change, there is an underlying and more structured way to account for the notions of change and continuity within the modal system, respectively.
Dynamic semantic accounts of presupposition have proven to quite successful improvements over earlier theories. One great advance has been to link presupposition and anaphora together (van der Sandt 92, Geurts 95), an approach that extends to integrate bridging and other discourse phenomena (Asher and Lascarides 1998a,b). In this extended anaphoric account, presuppositions attach, like assertions, to the discourse context via certain rhetorical relations. These discourse attachments constrain accommodation and help avoid some infelicitous predictions of standard accounts of presupposition. Further, they have interesting and complex interactions with underspecified conditions that are an important feature of the contributions of most presupposition triggers.
Deictic uses of definites, on the other hand, seem at first glance to fall outside the purview of an anaphoric theory of presupposition. There seems to be little that a discourse based theory would have to say. I will argue, however, that a discourse based account can capture how these definites function in conversation. In particular such accounts can clarify the interaction between the uses of such deictic definites and various conversational moves. At least some deictic uses of definites generate presuppositions that are bound to the context via a rhetorical function that I'll call unchoring, which if successful entails a type of knowing how. If this anchoring function is accepted, then the acceptors know how to locate the referent of the definite in the present context. I'll concentrate here just on definites that refer to spatial locations, where the intuitions about anchoring are quite clear. But I think that this view extends to other deictic uses of definites and has ramifications for an analysis of de re attitudes as well.
We will see how it is reasonable to speak of a minimum distance that an element must cross in order to enter into a well-formed movement dependency. In the course of the discussion of this notion of anti-localiry, a theoretical framework unfolds which is compatible with recent thoughts on syntactic computation regarding local economy and phrase structure, as well as the view that certain pronouns are grammatical formatives, rather than fully lexical expressions. The upshot will be that if an element does not move a certain distance, the derivation crashes at PF, unless the lower copy is spelled out as a pronominal element. The framework presented has a number of implications for the study of clause-typing, of which some will be discussed towards the end.
Recent research has adduced growing evidence for a distinct stratum of cultural practices that underlies various "tribal" traditions in the Himalayan region and that also seems to be characteristic of various local versions of the Bon tradition. Bon literature is not uncommonly embedded in cultural patterns that are more specifically Himalayan than belonging to the greater South Asian heritage. Two aspects of this that have received attention in Ramble's (1997) study of a Bon guide to the sacred Kong-po mountain (rKong-po bon- ri) are the symbolism of wild boar hunting involved in marriage rituals and poison cults with their corresponding beliefs about poisoning. Another pattern of cultoral organization that may help better understand the Bon tradition against its Himalayan background is spatial conceptualization.
In this paper, I discuss four different verb forms in Ndebele (a Nguni Bantu language spoken mainly in Zimbabwe) - the imperative, reduplicated, future and participial. I show that while all four are subject to minimality restrictions, minimality is satisfied differently in each of these morphological contexts. To account for this, I argue that in Ndebele (as in other Bantu languages) Word and RED are not the only constituents which must satisfy minimality: the Stem is also subject to minimality conditions in some morphological contexts. This paper, then, provides additional arguments for the proposal that Phonological Word is not the only sub-lexical morpho-prosodic constituent. Further, I argue that, although Word, RED and Stern are all subject to the same minimality constraint – they must all be minimally bisyllabic - this does not follow from a single 'generalized' constraint. Instead, I argue, contra recent work within Generalized Template Theory (see, e.g., McCarthy & Prince 1994, 1995a, 1999; Urbanezyk 1995, 1996; and Walker 2000; etc.) that a distinct minimality constraint must be formalized for each of these morpho-prosodic constituents.
Identical topic (IT henceforth) was previously known as copying topic (Xu & Liu (1998:141-157). It is fully or partially identical to a corresponding element (CE henceforth) occurring in the following part of the clause. Broadly speaking, IT is semantically empty. Being an unusual type of adding, it properly falls into the central concern of this volume.
It seems IT can be attested in all Chinese dialects, though the phenomena in question have been poorly documented and have scarcely been studied under a unified category. IT seems to be a better candidate to characterise topic prominent languages than many other topic types including the non-gap topic, which has long been called "Chinese style topic" since Chafe (1976) and has been viewed as a major characteristic of topic prominent languages (e.g., Li & Thompson, 1976, Xu & Langendoen 1985, Gasde 1999). I believe the study of IT structure is necessary to obtain a clearer and more complete picture of topic structure in general. As far as I know, Wu dialects of Chinese, including Shanghainese, are the ones which have the richest IT types and the greatest text frequency of IT. Therefore, this study will be based on both Mandarin and Shanghainese data.
The present investigation is concerned with German participles II (past participles) as lexical heads of adjuncts.
Within a minimalist framework of sound-meaning correlation, the analysis presupposes a lexicalist conception of morphology and the differentiation of Semantic Form and Conceptual Structure. It is argued that participles II have the same argument structure as the underlying verbs and can undergo passivization, perfectivization and conversion to adjectives. As for the potential of participles to function as modifiers, it is shown that attributive and adverbial participle constructions involve further operations of conversion. Participle constructions are considered as reduced sentences. They do not have a syntactic position for the subject, for an operator (comparable to the relative pronoun in relative clauses) or for an adverbial relator (as in adverbial clauses). The pertinent components are present only in the semantic structure.
Two templates serve the composition of modifiers - including participle constructions - with the modificandum. It is necessary to differentiate between modification which unifies two predicates relating to participants or to situations and frame setting modification where the modifier is given the status of a propositional operator.
The proposed analysis shows that the high degree of semantic underspecification and interpretative flexibility of German participle II constructions resides in the indeterminacy of participles II with respect to voice and perfect, in the absence of certain constituents in the syntactic structure and in the presence of corresponding parameters in the Semantic Form of the participle phrases.
This paper deals with restitutive and repetitive 'wieder'. Proceeding from the assumption that adverbial adjuncts have base positions which reflect their semantic relations to the rest of the sentence, it is shown that repetitive 'wieder' belongs to the class of event adverbs minimally c-commanding the base positions of all arguments whereas restitutive 'wieder' has many properties in common with process adjuncts, minimally c-commanding the final verb.
The argument-modifier distinction is less clear in NPs than in VPs; nouns do not typically take arguments. The clearest cases of arguments in NPs are in certain kinds of nominalizations which retain some "verbal" properties (Grimshaw 1990). The status of apparent arguments of non-deverbal relational nouns like sister is more controversial.
Genitive constructions like 'John's teacher', 'team of John's' offer a challenging testing ground for the argument-modifier distinction in NPs, both in English and cross-linguistically. On the analyses of Partee (1983/97) and Barker (1995), the DP in a genitive phrase (i.e. 'John' in 'John's') is always an argument of some relation, but the relation does not always come from the head noun. On those "ambiguity" analyses, some genitives are argument-like and some are modifier-like. Recent proposals by Jensen and Vikner and by Borschev and Partee analyze all genitives as argument-like, a conclusion we are no longer sure of.
In this paper we explore a range of possible analyses: argument-only, modifier-only, and ambiguity analyses, and consider the kinds of semantic evidence that suggest that different analyses may be correct for different genitive or possessive constructions in different languages.
The present paper offers evidence that there are two variants of adverbial modification that differ with respect to the way in which a modifier is linked to the verb's eventuality argument. So-called external modifiers relate to the full eventuality, whereas internal modifiers relate to some integral part of it. The choice between external and internal modification is shown to be dependent on the modifier's syntactic base position. External modifiers are base-generated at the VP periphery, whereas internal modifiers are base generated at the V periphery. These observations are accounted for by a refined version of the standard Davidsonian approach to adverbial modification according to which modification is mediated by a free variable. In the case of external modification, the grammar takes responsibility for identifying the free variable with the verb's eventuality argument, whereas in the case of internal modification, a value for the free variable is determined by the conceptual system on the basis of contextually salient world knowledge.
The paper investigates a recent proposal to resultativity by G. Jäger and R. Blutner (J&B). J&B say that the representation of result states of accomplishments by means of CAUSE and BECOME is not correct and should not be done in the syntax in terms of decomposition. They develop an axiomatic approach where each accomplishment/achievement is related to its result by a particular axiom. Modification of the result by "again" makes use of these axioms and the restitutive/resultative ambiguity is a matter of lexical ambiguity or polysemy. They argue that the classical decomposition theory cannot treat the restitutive reading of "A Delaware settled in New Jersey again" (there had been Delawares in New Jersey but not this particular one; and those earlier Delawares never moved to New Jersey but were borne there). I discuss (and dispute) these data and compare the two theories. J&B's contains an OT-part dealing with the disambiguating role of stress. While the decomposition theory cannot deal with the data mentioned, it can integrate the OT-part of J&B's theory.
Rethinking the adjunct
(2000)
The purpose of the present paper is twofold: first, to show that, when defining the adjunct, it is necessary to distinguish in a strict modular way between the syntactic level and the lexico-semantic level. Thus, the adjunct is a syntactic category on a par with the specifier and the complement, whereas the argument belongs to the same set as does (among others) the modifier. The consequence of this distinction is that there is no direct one-to-one opposition between adjuncts and arguments. Nor is there any direct one-to one relation between adjuncts and modifiers.
The second and main purpose of the paper is to account for the well-known difference between the position of a specific set of modifiers (cause, time, place etc.) in, on the one hand, English and Swedish, on the other, German. In English and Swedish the default position of these modifiers is postverbal, whereas in German it is preverbal. Further, in English and Swedish, these modifiers occur in a mirror order compared with their German counterparts, an order which, from a semantic point of view, is not the expected one. I shall demonstrate that this difference is due to the different settings of the verbal head parameter, the former languages being VO-languages and the latter being OV -languages. I shall further argue that in English and Swedish these modifiers are base generated as adjuncts to an empty VP, which is a complement of the main verb of what I shall call the minimal VP (MVP), whereas in German they are adjuncts on top of the MVP. Finally, I shall argue that the postverbal modifiers move at the latest at LF to the top of the MVP, in order to take scope over it, the restriction being 'Shortest move'. The movement results in the correct scope order of the postverbal modifiers.
The proposed structure also accounts for the binding data, in particular for the binding of a specific Swedish possessive anaphor 'sin'. This pronoun, which may occur within the MVP, must not occur within the postverbal modifiers in the empty VP. This supports the assumption that there is a strict borderline between the MVP and the assumed empty VP. The account is also in accordance with the focus data, the specific set of modifiers being potential focus exponents in a wide focus reading in English and Swedish, but not in German.
In this study, I investigate the positions and interpretations available to 'manner' adverbs in English. My central claim, contra Wyner (1994, 1998), is that an association does exist between 'manner' adverb positions and interpretations, which is best characterized in terms of Peterson's (1997) distinction between 'restrictive' and 'non-restrictive' modification. I also claim, however, that the association in question is not as general as commonly claimed; and, in particular, does not apply directly to 'manner' adverbs in 'fronted' and 'parenthetical' positions, which require special syntactic description.
This paper proposes that we can predict which adverbs cannot adjoin to the right in headinitial languages by means of a particular semantic property, that of being a "subjective" adverb, one which maps an event or proposition onto a scale with the high degree of indeterminacy and context-dependence. Such adverbs, such as 'probably' or 'luckily', cannot adjoin to the right with non-manner readings, while other adverbs (like 'politically', 'often', or 'deliberately') may. This supports the view that the distribution of adverbs depends heavily, and subtly, on their lexicosemantic properties.
It is argued that there is a surprising gap in the distribution of adverbial modifiers, namely that there are (practically) no adverbs that modify exclusively stative verbs. Given the general range of selectional restrictions associated with adverb/verb modification, this comes as a surprise. It is argued that this gap cannot be the result of standard selectional restrictions. An independently motivated account of the state-event verb contrast, in which state verbs are proposed to lack Davidsonian arguments is presented and argued to account for this stative adverb gap. Some apparent and real problems with the analysis are discussed.
In this paper I would like to show that the principles which have been proposed so far to account for the relationship between the informational level and the syntactic level in a Chinese utterance are unable to predict some interesting and regular facts of that language.
To my mind, the form and the position of the question operator in an interrogative utterance provide two distributional tests which univocally indicate where the new information lies. Hence, the pairing of affirmative and interrogative sentences might be a better approach to locate where the new information lies in a Chinese utterance.
The present study offers the analysis of the role of adverbials in the semantic structure of a sentence. To clarify this role new notions "Adverbials with floating and fixed semantic scope" are proposed. This classification also can clarify the role of adverbials from the point of view of the division into arguments vs. adjuncts.
This paper deals with a series of semantic contrasts between the copula "be" and the preposition "as", two functional elements that both head elementary predication structures. It will be argued that the meaning of "as" is a type lowering device shifting the meaning of its complement NP from generalized quantifier type to property type (where properties are conceived as relations between individuals and situations), while the copula "be" induces a type coercion from (partial) situations to (total) possible worlds. Paired with van der Sandt's 1992 theory of presupposition accommodation, these assumptions will account for the observed contrasts between "as" and "be".
This paper is concerned with the fact that a number of adverbal modifications involve a systematic reinterpretation of at least one of the expressions connected by the operation in question. It offers an approach in which such transfers of meaning turn out to be a result of contextually controlled enrichments of an underspecified as well as a strictly compositionally structured semantic representation. The approach proposed is general for three reasons: First, it takes into account not only reinterpretations in temporal but also such in non-temporal modification. Second, it allows considering so-called secondary predications as a particular kind of adverbal modification. Third, it explains the respective reinterpretations within a uniform formal framework of meaning variation.
This paper draws a link between the typological phenomenon of the paradigmatically supported evidentiality evoked by perfect and/or perfectivity and the equally epistemic system of modal verbs in German. The assumption is that, if perfect(ivity) is at the bottom of evidentiality in a wide number of unrelated languages, then it will not be an arbitrary fact that systematic epistemic readings occur also for the modal verbs in German, which were preterite presents originally. It will be demonstrated, for one, how exactly modal verbs in Modem German still betray sensitivity to perfect and perfective contexts, and, second, how perfect(ivity) is prone to evincing epistemic meaning. Although the expectation cannot be satisfied due to a lack of respective data from the older stages of German, a research path is sketched narrowing down the linguistic questions to be asked and dating results to be reached.
Editorial preface
(2000)
The present issue grew out of two sources. The main one was the workshop on Adding and Omitting (A & 0) held during the DGfS Conference organized in Konstanz at the beginning of 1999 by our ZAS project on Syntax der Fokusbildung. The purpose of the workshop was to bring together people working on topicalization (addition of expressions, in a sense) and ellipsis (omission, i.e. deletion of linguistic material) and their relations and interaction. Since the workshop was very successful and met with a great deal of interest on the part of both participants and outsiders, we decided to collect and publish the papers that were presented. Towards the end of 1999, a follow-up workshop on Ellipsis and Information Structure was organized by Kerstin Schwabe and Susanne Winkler (Tübingen). The papers given at this second meeting were supposed to be an integral part of the publication as well. More and more people got involved, further developing our common understanding of the topic phenomenon, so that there was too much material for a single volume. We therefore decided to split the enterprise into two volumes. The ellipsis papers are to be published by 'Benjamins' this year in Interpreting Omitted Structures.
This paper deals with the emergence of verb morphology in one German child up to the time mini-paradigms occur in the data. I will focus on the role of protomorphology as a transitional stage between rote learning and the productive use of morphological distinctions.
The study presents a first investigation of two different processes in the L1-acquisition of German: The acquisition of definite pronominal forms and the occurence of finite verbs. The aim of the study is to find out if there are inherent relations between both processes. Inherent relations are understood as developmental relations based on the structural properties which demand a correlated emergence of the finite verb and definite pronominal forms.
In these conclusions we can deal only with some of the tentative comparative results of the workshop papers on the early development of verb morphology. The main focus is on criteria of how the child detects morphology and how this emerging morphological competence develops in its earliest phases. In view of the purpose and tentative character of these conclusions, all references will be limited to the papers of the workshop and to earlier studies by workshop participants within the "Crosslinguistic Project on Pre- and Protomorphology in Language Acquisition". Much more will be given in the projected final publication.
Introduction
(2000)
Dulong/Rawang is a Tibeto-Burman language spoken on both sides of the China/Myanmar (Burma) border just south and cast of Tibet. [...] In this chapter, I will be using data of the Mvtwang (Mvt River) dialect, which is considered the most central of those dialects in Myanmar and so has become something of a standard for writing and intergroup communication, though most of the phenomena we will be discussing are general to dialects in both China and Myanrnar. I will use the short form 'Rawang' in referring to this dialect.
The distribution of trimoraic syllables in German and English as evidence for the phonological word
(2000)
In the present article I discuss the distribution of trimoraic syllables in German and English. The reason I have chosen to analyze these two languages together is that the data in both languages are strikingly similar. However, although the basic generalization in (1) holds for both German and English, we will see below that trimoraic syllabIes do not have an identical distribution in both languages.
In the present study I make the following theoretical claims. First, I argue that the three environments in (1) have a property in common: they all describe the right edge of a phonological word (or prosodic word; henceforth pword). From a formal point of view, I argue that a constraint I dub the THIRD MORA RESTRICTION (henceforth TMR), which ensures that trimoraic syllables surface at the end of a pword, is active in German and English. According to my proposal trimoraic syllables cannot occur morpheme-internally because monomorphemic grammatical words like garden are parsed as single pwords. Second, I argue that the TMR refers crucially to moraic structure. In particular, underlined strings like the ones in (1) will be shown to be trimoraic; neither skeletal positions nor the subsyllabic constituent rhyme are necessary. Third, the TMR will be shown to be violated in certain (predictable) pword-internal cases, as in Monde and chamber; I account for such facts in an OptimalityTheoretic analysis (henceforth OT; Prince & Smolensky 1993) by ranking various markedness constraints among themselves or by ranking them ahead of the TMR. Fourth, I hold that the TMR describes a concrete level of grammar, which I refer to below as the 'surface' representation. In this respect, my treatment differs significantly from the one proposed for English by Borowsky (1986, 1989), in which the English facts are captured in a Lexical Phonology model by ordering the relevant constraint at level 1 in the lexicon.
Identity effects in phonology are deviations from regular phonological form (i.e. canonical patterns) which are due to the relatedness between words. More specifically, identity effects are those deviations which have the function to enhance similarity in the surface phonological form of morphologically related words. In rule-based generative phonology the effects in question are described by means of the cycle. For example, the stress on the second syllable in cond[ɛ]nsation as opposed to the stresslessness of the second syllable in comp[ǝ]nsation is described by applying the stress rules initially to the sterns thereby yielding condénse and cómpensàte. Subsequently the stress rules are reapplied to the affixed words with the initial stress assignment (i.e. stress on the second syllable in condense, but not in compensate) leaving its mark in the output form (cf. Chomsky and Halle 1968). A second example are words like lie[p]los 'unloving' in German, which shows the effects of neutralization in coda position (i.e. only voiceless obstruents may occur in coda position) even though the obstruent should 'regularly' be syllabified in head position (i.e. bl is a wellformed syllable head in German). Here the stern is syllabified on an initial cycle, obstruent devoicing applies (i.e. lie[p]) and this structure is left intact when affixation applies (i.e. lie[p ]Ios ) (cf. Hall 1992). As a result the stern of lie[p]los is identical to the base lie[p].
I argue in this study that consonantal strength shifts can be explained through positional bans on features, expressed over positions marked as weak at a given level of prosodic structure, usually the metrical foo!. This approach might be characterized as "templatic" in the sense it seeks to explain positional restrictions and distributional patterns relative to independently motivated, fixed prosodic elements. In this sense, it follows Dresher & Lahiri's (1991) idea of metrical coherence in phonological systems, namely, "[T]hat grammars adhere to syllabic templates and metrical patterns of limited types, and that these patterns persist across derivations and are available to a number of different processes ... " (251). [...] The study is structured as follows: section 1 presents a typology of distributional asymmetries based on data from unrelated languages, demonstrating that the stress foot of each of these languages determines the contexts of neutralization and weakening of stops. Section 2 elaborates the notion of a template, exploring some of its formal properties, while section 3 presents templatic analyses of data from English and German. Section 4 explores the properties of weak positions, especially weak onsets, in more detail, including discussion of templates in phonological acquisition. Section 5 summarizes and concludes the study.
The aim of this paper is to show what role prosodic constituents, especially the foot and the prosodic word play in Polish phonology. The focus is placed on their function in the representation of extrasyllabic consonants in word-initial, word-medial, and word-final positions.
The paper is organized as follows. In the first section, I show that the foot and the prosodic word are well-motivated prosodic constituents in Polish prosody. In the second part, I discuss consonant clusters in Polish focussing on segments that are not parsed into a syllable due to violations of the Sonority Sequencing Generalisation, i.e. extrasyllabic segments. Finally, I analyze possible representations of the extrasyllabic consonants and conclude that both the foot and the prosodic word play a crucial role in terms of licensing. My proposal differs from the ones by Rubach and Booij (1990b) and Rubach (1997) in that I argue that the word-initial sonorants traditionally called extrasyllabic are licenced by the foot and not by the prosodic word (cf. Rubach and Booij (1990b)) or the syllable (cf. Rubach (1997)). For my analysis I adopt the framework of Optimality Theory, cf. McCarthy and Prince (1993), Prince and Smolensky (1993), in which derivational levels are abandoned and only surface representations are evaluated by means of universal constraints.
In this work, I examine a set of languages which appear to require resyllabification postlexically; in less derivational terms, a word's syllabification in isolation differs from its syllabification in a phrase-internal context. Although many people, myself included, have been looking at such cases in isolation over the years, I bring together several examples here to see what features they share and how an Optimality Theory analysis improves upon rule-based derivational approaches.
The purpose of this paper is to provide a unified (i.e. independent of lexical categories) account of Persian stress. I show that by differentiating word- and phrase-level stress rules, one can account for the superficial differences exemplified in (1) above and many of the stipulations suggested by previous scholars. The paper is organized as follows. In section 1, I look at nouns and adjectives and propose a rule that would account for their stress pattern. In section 2, I extend the stress rule to verbs and show the problem this category poses to our generalization. The main proposal of this paper is discussed in section 3. I introduce the phrasal stress rule in Persian and show that by differentiating word-level and phrase-level stress rules, one can come to a unified account of Persian stress. Section 4 deals with some problematic eases for the proposed generalization and discusses some tentative solutions and their theoretical consequences. Section 5 concludes the paper.
The paper proposes structural constraints for different adjunct classes in German and English. Approaches in which syntax has only the task to provide adjunct positions and in which principles of scope are supposed to explain the distribution of adjuncts are rejected as incomplete. The syntactic requirements are not as rigid as other approaches require, such that there is just one possible position for a given adjunct. Rather the syntactic constraints may be fulfilled in different positions.
In this paper the first results concerning the development of early verb morphology in an L1-English speaking child are presented. Adopting the framework of morphological development of Dressler (Dressler, this volume) the data of a girl from the CHILDES database, Nina of the Suppes corpus, is analysed with regard to the emergence of early verbal categories (e.g. number and person) and their appearance in a first mini-paradigm. In the sessions analysed so far the child Nina has reached an age of 2;2 when the first mini-paradigm emerges.
This paper deals with early verb development (e.g., person, tense) until the emergence of verb-paradigms in two French-speaking children.
I will show the parallelism between the two children in the gradual building of paradigms, despite considerable differences in the rate of development. Individual differences on the other hand will bring me to reconsider the broad category of premorphological rote-learnt forms which already displays some patterning in one of the children's data.
This paper studies the acquisition process of Spanish verbal morphology in a monolingual child. The study focuses on the period of the first 50 verb lemmas. This covers the period from age 1;7 till 1;10.
The data shows that the verb acquisition process of this Spanish child follows three main stages:
1. A lexical stage in which verbs are only acquired as a lexical element.
2. A syntactic stage in which the verb, still contemplated as a non-split word, becomes the main element in the development of thematic and semantic relations.
3. A morphological stage in which verb suffixes begin to be analysed separately. At this stage, the relationship between form and meaning starts and the functional categories linked to the verb (tense, aspect, agreement, mood... ) begin to be acquired. Just at this moment, the first miniparadigms appear, which suggests that the acquisition process of verb morphology has started.
The first two stages are premorphological and cover in our child the period till 1;9. In the last stage, which begins at 1;10, the child enters the protomorphological stage.
The source of the data used in this paper are recordings of conversations with a Lithuanian girl, Rūta. Rūta lives in Vilnius and is the only child in the family. Both parents speak standard Lithuanian without dialectal influences. The recordings were taken on a free basis without a fixed schedule, then transcribed by the mother of the child, double-checked and coded in accordance with CHILDES by the author of the paper. At the moment of writing this contribution the data taken between 1;7-2;5 have been fully processed. Over this period about 34.5 hours of recordings were collected.
The Present Perfect in Portuguese has the curious property of forcing iteration of the eventuality described. This paper proposes an account of the iterativity in terms of selectional restrictions of the Present Tense and independent properties of the Perfect and argues against the account of Giorgi and Pianesi 1998 in which the Portuguese Present Perfect is treated as containing two main verbs.
An adjunct-DP in the free instrumental case occurs in a number of surface positions where the DP is syntactically optional. does not depend on any element in the sentence, and has a number of different interpretations. We introduce Bailyn's proposal which postulates a uniform syntactic environment for all the uses of instr. This calls for a uniform semantics of these DPs which can nevertheless accomodate the different interpretations. Starting with the hypothesis of Roman Jakobson about the semantics of the instrumental case we formulate a semantic interpretation theory based on abduction. We give a uniform semantics for three different adjunct uses of instr in this framework. In the concluding part of the paper we discuss some possible alternatives and ramifications as well as questions and objections raised with respect to the treatment proposed in this paper.
This article aims to recast the properties of topic-prominent languages and their differences from subject-prominent languages as documented in the functionalist literature into the framework of the Principle-and-Parameter approach. It provides a configurational definition of the topic construction called Topic Phrase (TP), with the topic marker as its head. The availablity of TP enables topic prominent languages to develop various topic structures with properties such as morphological marking; cross-categorial realization of topics and comments; and mutiple application of topicalization. The article elaborates the notion of topic prominence. A topic prominent language is characterized as one that tends to activate the TP and to make full use of the configuration. Typically, it has a larger number and variety of highly grammaticalized topic markers in the Lexicon and permits a variety of syntactic categories to occur in the specifier position and the complement position of TP.
The complexity of human languages has always inspired research for some human faculty that makes language learning possible. The system that generates the complexity of human languages, ideally, is simple and effective. Recent developments of the generative grammatical theory explore deeper into the issue of simplicity or economy. The Minimalist Program developed in Chomsky (1991, 1993, 1995) tries to provide contents to such notions. What does it mean to be more economic or least effort? An important instantiation of such notions is the proposal that movement is the last resort assuming that movement is more costly than non-movement. Processes occur only because they are necessary. The definition of necessity generally is cast in morphological terms. Moreover, the notion of "economy" or "least effort" is deterministic of the appropriate derivations for sentences: a shorter derivation is better than a longer one. In this work, we show that the notion of "least effort," - do minimally if possible - is manifested not only in derivations but also in other aspects of the grammar. We take Chinese as an example and show that this language exhibits the properties manifesting some "least effort" guidelines in the area of movement and reconstruction, and in the projection of syntactic positions: when there is a choice, non-application of moyement/reconstruction and non-projection of a position are adopted. These phenomena essentially are attested in topic structures. The question arises as to why topic structures exhibit such minimal effort effects. We suggest that this is due to the fact that topic structures can be derived by movement or base-generation. When there are morpho-syntactic clues that reconstruction is necessary, the structure is a movement structure. Otherwise, the less costly non-movement structure is assumed. Moreover, because of the possibility of assuming a topic NP to be base-generated, bearing a predication (or aboutness) relation with the comment clause, the argument position which otherwise would be related to the topic (conveniently termed the trace position) is not projected when there is a choice of projecting or not projecting it.
This paper shows the early development of the first approximately 50 verbs found in the recorded speech production of one Croatian girl. The aim is to analyse and interpret the child's verb development in terms of the distinction of a pre- and a protomorphological phase before modularised morphology in language acquisition (Dressler & Karpf 1995). Furthermore, focus will be laid on the emergence of first verb paradigms.
The left periphery has enjoyed extensive study over the past years, especially drawn against the framework of Rizzi (1997). It is argued that in this part of the clause, relations are licensed that have direct impact on discourse interpretation and information structure, such as topic, focus, clause type, and the like. I take this line of research up and argue in favour of a split CP on the basis of strictly left-peripheral phenomena across languages. But I also want to link the relation of articulated clause structure, syntactic derivations, and information structure. In particular, I outline the basics of a model of syntactic derivation that makes explicit reference to the interpretive interfaces in a cyclic, dynamic manner.
I suggest a return to older stages of generative grammar, at least in spirit, by proposing that clausal derivation stretches over three important areas which I call prolific domains: the part of the clause which licenses argument/thematic relations (V- or θ-domain), the part that licenses agreement/grammatica1 relations (T- or ϕ-domain), and the part that licenses discourse/information-relevant relations (C- or ω-domain). It is thus a rather broad and conceptual notion of "adding" and "omitting" that I am concerned with here, namely licensing of material to relate to information structure, and the desire to find an answer to the question which elements might be added or omitted across languages to establish such links.
This paper is a preliminary comparative study of the relation between word order and information structure in three Null Subject Languages ((NSLs) Spanish, Italian and Greek). The aim is twofold: first I seek to examine the differences and the similarities among these languages in this domain of their syntax. Secon, I investigate the possible derivations of the various patterns and attempt to localize the differences among these languages in different underlying syntactic structures.
The ultimate goal of the study is to examine the acquisition of intensifiers in English and German. In this paper an overview of the first results regarding four L1 English-speaking children will be given. Contrary to previous claims in the literature (e.g. Thomas 1990), it will be argued that intensifiers are used by children in early phases of language acquisition. Intensifiers play an important role in early phases of language acquisition since they can be used to express the wish either to be included or excluded in a certain action and thus contribute to structuring a central aspect of the child's discourse.
This paper presents an analysis of secondary predicates as aspectual modifiers and secondary predication as a summing operation which sums the denotation of the matrix verb and the secondary predicate. I argue that, as opposed to the summing peration involved in simple conjunction, there is a constraint on secondary predication; in the 0 case of depictives, the event introduced by the matrix verb must be PART-OF the event introduced by the secondary predicate, where e1 is PART-OF e2 if the running time of e1 is contained in the running time of e2 and if e1 and e2 share a grammatical argument. I argue resultative predication differs from depictive predication in that the PART-OF constraint holds in resultative constructions between the event which is the culmination of e1 and e2: formally, while depictive predication introduces the statement PART-OF(e1,e2), resultative predication introduces the statement PART-OF(cul(e1),e2). I show that this is all that is necessary to explain the well-known properties of resultative predication.
The distinction between COMPLEMENTS and ADJUNCTS has a long tradition in grammatical theory, and it is also included in some way or other in most current formal linguistic theories. But it is a highly vexed distinction, for several reasons, one of which is that no diagnostic criteria have emerged that will reliably distinguish adjuncts from complements in all cases – too many examples seem to "fall into the crack" between the two categories, no matter how theorists wrestle with them.
In this paper, I will argue that this empirical diagnostic "problem" is, in fact, precisely what we should expect to find in natural language, when a proper understanding of the adjunct/complement distinction is achieved: the key hypothesis is that a complete grammar should provide a DUAL ANALYSIS of every complement as an adjunct, and potentially, an analysis of any adjunct as a complement. What this means and why it is motivated by linguistic evidence will be discussed in detail.
The unusual development of the PDE [present-day English] s-genitive can be historically motivated, if the 's form is supposed to be not a mere leftover of the Old English (henceforth OE) casemarking, but the outcome of the merging of two patterns: the inflectional genitive ending (levelled to -s) and the construction "John his book" (henceforth 'possessive-linked genitive') during the Middle and the Early Modem English phases.
As my corpus analysis will show, the semantic and syntactic constraints ruling the occurrence of the 's pattern in the time interval of the rise of the 's-pattern (1400 - 1650) are the same ones as those ruling the occurrence of the possessive-linked genitive.
This hypothesis is further confirmed by cross-language comparison (with the other West Germanic languages, especially Afrikaans).
The present study is concerned with Single Clitics, as weil as with Clitic Doubling and Clitic Left Dislocation constructions and will test the Uniformity Hypothesis (Sportiehe 1992), according to which all three constructions involve the same underlying structure. It will be shown that:
- acquisition data pose a problem for the Uniformity Hypothesis (Sportiche 1992) and support rather the idea that Single Clitic, Clitic Doubling and Clitic Left Dislocation constructions do not involve the same underlying structure,
- omission of definite articles in Clitic Doubling and Clitic Left Dislocation constructions parallels omission of definite articles in simple DPs,
- selective omission of some types of Determiners, i.e. definite articles and use of another type of Determiners, i.e. clitic pronouns, can be explained in terms of the different feature specification of words belonging to the category D and the different status of clitics vs. definite articles.
The article deals with the analysis of the development of aspectuality at the early stages of the acquisition of Russian. Data from seven children are investigated for this purpose. It is claimed that the category of aspectuality, being the property of the whole utterance, can be expressed at the early stages of language acquisition even before the verb itself occurs. During this period some children mark the basic aspectual opposition "process-result" by the linguistic devices at their disposal, namely by various uses of sound imitations or onomatopoetics. Onomatopoetics, when used once, can be said to be the predecessors of perfective verbs, while reduplicative use of onomatopoetics seems to correspond to the imperfective aspect. The paper presents an analysis of the early verb lexicons of six children. Among their 24 earliest verbs both aspects are represented. As revealed by the analysis, aspect (and Aktionsart) clusters with tense in a specific way: imperfective verbs are mainly used in the present while perfectives are used mostly in the past.
The paper investigates the issue whether the stage-level/individual level contrast introduced by Carlson 1977 requires the assumption of two homonymous copulas depending on the categorization of the predicative. We argue that instead of a uniform stage-level/individual level distinction we have to distinguish several similar but independent contrasts, none of which crucially depend on the semantics of the copula. In the second part of the paper, we concentrate on one group of phenomena-the distribution of weak subjects-and propose an explanation in terms of an interaction between topic/comment structure and aspectual properties of the predicate.
In this paper I investigate the properties of the copula-like verb 'ficar' in Brazilian Portuguese using Pustejovsky's generative lexicon (GL). The verb 'ficar' can be translated as 'stay' or 'become', depending on its complement. With locatives, only the STAY reading is possible. With adjectival complements, both BECOME and STAY readings are possible. I propose that 'ficar' takes an eventuality as its complement and I argue that there is no need to create multiple lexical entries for it, since the readings are the result of the possible combinations between the transition denoted by 'ficar' and the properties of the stative complements.
I argue that the BECOME reading with adjectival predicates is the result of combining part of the qualia of the adjectival predicate with the TRANSITION of 'ficar'. The STAY readings of 'ficar'+adjective are the result of shadowing the transition. In the case of 'ficar'+locative, the BECOME reading is unavailable. Departing from the hypothesis that subevents have to be linked to arguments in order to be able to be modified by certain types of modifiers or be selected by certain types of heads, I argue that the transition, in the case of locative complements, is not associated to any argument because nothing in the qualia of the locative complement is compatible with a transition, given that there is not motion component in either 'ficar' or the locative. Unlinked to any argument, the TRANSITION can only be part of the 'constant' meaning of the verb, which explains why it is not available for modification.
This paper investigates syntactic properties of verbless constructions in Chinese. Verbless constructions differ from constructions with overt verbs in three major respects. First, there is a VP-internal nominal raising in Chinese, which is optional if an overt verb shows up, and obligatory if there is no overt verb. Second, while an overt verb can select various kinds of argument, the internal argument of a verbless construction cannot be indefinite. Third, there are two types of object depictive secondary predication constructions, and only one of them allows for a null verb.
Simplicity as a methodological orientation applies to linguistic theory just as to any other field of research: ‘Occam’s razor’ is the label for the basic heuristic maxim according to which an adequate analysis must ultimately be reduced to indispensible specifications. In this sense, conceptual economy has been a strict and stimulating guideline in the development of Generative Grammar from the very beginning. Halle’s (1959) argument discarding the level of taxonomic phonemics in order to unify two otherwise separate phonological processes is an early characteristic example; a more general notion is that of an evaluation metric introduced in Chomsky (1957, 1975), which relates the relative simplicity of alternative linguistic descriptions systematically to the quest for explanatory adequacy of the theory underlying the descriptions to be evaluated. Further proposals along these lines include the theory of markedness developed in Chomsky and Halle (1968), Kean (1975, 1981), and others, the notion of underspecification proposed e.g. in Archangeli (1984), Farkas (1990), the concept of default values and related notions. An important step promoting this general orientation was the idea of Principles and Parameters developed in Chomsky (1981, 1986), which reduced the notion of language particular rule systems to universal principles, subject merely to parametrization with restricted options, largely related to properties of particular lexical items. On this account, the notion of a simplicity metric is to be dispensed with, as competing analyses of relevant data are now supposed to be essentially excluded by the restrictive system of principles.
It is widely believed that existential quantifiers can bring about the semantic effects of a scope which is wider than their actual syntactic scope (See Fodor & Sag (1982), Cresti (1995), Kratzer (1995), Reinhart (1995) and Winter (1995), among many others.) On the other hand, it is assumed that the syntactic scope of universal quantifiers can be determined unequivocally by the semantics. This paper shows that this second assumption is wrong; universal quantifiers can also bring about scope illusions, though in a very specific environment. In particular, we argue that in the environment of generic tense, universal quantifiers can show the semantic effects of a scope which is wider than the one that is actually realized at LF. Our argument has four steps. First, we show that in generic contexts, universal quantifiers escape standard “scope-islands” (Section 1). Second, we show how the effects of wide scope in generic contexts can be achieved without syntactic wide scope (Section 2.1). Third, we show that this result is actually forced on us, once we take seriously certain independent issues concerning the interpretation of generic tense (Sections 2.2 - 2.4). Finally, the semantics of generic tense and, in particular, its interaction with focus, will yield some intricate new predictions, which, as we show, are borne out (Sections 3 - 5).
Part of linguistic competence involves the ability to construct and interpret words. The average high school student knows about 60,000 words whose form and meaning are not derived from those of other words. Such words including read, language, on, cold, and if, to name but a few - must be learned and stored as separate items in the lexicon (or mental dictionary). However, countless other words can be constructed and comprehended by the application of quite general rules to more basic words. For example, any speaker of English who knows the meaning of the noun fax - and the verb derived from it - could form and interpret words such as faxable (for things that can be faxed) and fax machine (for the device that sends and receives faxes). The system of categories and rules involved in word formation and interpretation is called morphology. This chapter presents an introduction to the study of morphology, beginning with the inventory of notions relevant to the analysis of word structure.
Coreference in dialogue
(1997)
Since the early days of discourse analysis coreference has always been considered a major factor in the formation of texts and dialogues. The repetition of nominal elements and the anaphoric use of pronouns in successive sentences is a fundamental cohesive pattern which ties sentences together and contributes to the coherence of sequences. "La coherence transphrastique trouve dans la pronominalisation un des procedes les plus efficaces" (Stati 1990, 160). The basic structural pattern on which linguists focused their interest in the early 1970s is captured by the following examples: (1) A man entered the house. After closing the door, the man sat down. He was tired. (2) Peter The man entered the house. He was tired. He ...
Eins der signifikanten Probleme in der maschinellen Übersetzung japanische in deutsche Sprache ist die fehlende Information und Definitheit im japanischen Analyse-Output. Eine effiziente Lösung dieses Problems ist es, die Suche nach der relevanten Information in den Transfer zu integrieren. Transferregeln werden mit Präferenzregeln und Default-Regeln kombiniert. Dadurch wird Information über lexikalische Restriktionen der Zielsprache, über die Domäne und über den Diskurs zugänglich.
In LaPolla 1990, I presented arguments to show that Chinese is a language in which there has been no grammaticalizalion of the syntactic relations "subject" and "object". This being the case, then syntactic relations cannot be what determines word order in Chinese. In this paper I will argue that, aside from a semantic rule that the actor of a verb, if expressed, must precede that verb, it is pragmatic relations (information structure) that are the main determinants of word order in Chinese.
This paper presents the first results of a comprehensive project on comparative Tibeto-Burman (TB) morpho-syntax. Data on morphological forms and typological patterns were collected from one hundred fifty-one languages and dialects in the TB family. For this paper the data were surveyed for nominal 'ergative' or agentive case marking (postpositions), in an attempt to determine if it would be possible to reconstruct an ergative case marker to Proto-Tibeto-Burman (PTB), and in so doing learn more about the nature of grammatical organization in PTB. Ablative, instrumental, genitive, locative, and other case forms were also surveyed for possible cognacy with ergative forms, as suggested in DeLancey 1984.
Speakers of various Southern german dialects may be heard to use two syntactic variants of subordinate clauses which are represented by the following Swabian examples: (1) daß er den net will komme lasse (2) daß er den net komme lasse will Of these two variants of the three-element verbal complex, only the non-dialect counterpart of (2) is accepted as standard modern written German: (3) daß er ihn nicht kommen lassen will In earlier periods of the German language, however, both variants were used by authors of written texts.
When in 1934, Robert BLEICHSTEINER published the Caucasian language specimina contained in the "travel book" of the 17th century Turkish writer Evliya Çelebi , he was struck by the amount of reliability he found in Evliya’s notations: "(Die Sprachproben) sind, von einzelnen Mißverständnissen abgesehen, und wenn man die falschen Punktierungen und Irrtümer der Kopisten abrechnet, außerordentlich gut, ja zuweilen mit einem gewissen phonetischen Geschick wiedergegeben, was der Auffassungsgabe und dem Eifer Evliyas ein hohes Zeugnis ausstellt. Man muß bedenken, wie schwer das arabische Alphabet, ohne weitere Unterscheidungszeichen, wie sie die islamischen Kaukasusvölker anwenden, die verwickelten, oft über 70 verschiedene Phoneme umfassenden Lautsysteme wiederzugeben imstande ist. Wenn trotzdem die Entzifferung der Sprachproben zum größten Teil geglückt ist, so muß man der ungewöhnlichen Begabung des türkischen Reisenden und Gelehrten schrankenlose Bewunderung zollen" (85). ...
The study of what makes utterances difficult or easy to understand is one of the central topics of research in comprehension. It is both theoretically attractive and useful in practice. The more we know about difficulties in understanding the more we know about understanding. And the better we grasp typical problems of understanding in certain types of discourse and for certain recipients the better we can overcome these problems and the better we can advise people whose job it is to overcome such problems. It is therefore not surprising that comprehensibility has been the object of much reflection as far back as the days of classical rhetoric and that it is a center of lively interest in several present-day scientific disciplines, ranging from artificial intelligence and educational psychology to linguistics.
The semantics of gradation
(1989)
The term 'gradation' is meant to cover a range of phenomena which for the time being I shall call quantitative evaluations regarding dimensions or features. I shall actually be looking into the principles governing the way gradation is expressed in language. The quantitative aspect of the adjectives of dimension occupies a key position which can be systematically explained and this aspect will be the crucial point of the discussion. I shall focus on the various grammatical forms of comparison: comparative, equative, superlative and some related constructions, and indications of measurement and adverbial indications of degree.
It is by now a weIl-known topic in semantics that there are striking similarities between the meanings of nominal and verbal expressions, insofar as the mass:count distinction in the nominal domain is reflected in the atelic:telic distinction in the verbal domain (cf. Leisi 1953, Taylor 1977, Bach 1986, to cite just a few authors). However, these supposed similarities have not be made explicit in formal representations.
This paper is the second in a series arguing for a discourse·based analysis of grammatical relations in Chinese in which there is a direct mapping between semantic role and grammatical function, and there are no relation-changing lexical rules such as passivization that can change that mapping. The correct assignment of semantic roles to the constituents of a discourse is done by the listener purely on the basis of the discourse structure and pragmatics (real world knowledge). Though grammatical analyses of certain constructions can be done on the sentence level, the sentence is generally not the central unit for understanding anaphora and grammatical relations in Chinese. Two related arguments are presented here: the question of 'subject' and the structure of discourse developed from an analysis of the nature of discourse referent tracking.