Discourse Adjectives

Gina Taranto

• Publisher: Routledge
• Year: 2013
• ISBN:
  9781135500559
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

First Published in 2006. This volume introduces and provides a semantic analysis of Discourse Adjectives, a natural class of adjectives that the author argues includes apparent, clear, evident, and obvious among its prototypical members. With a main claim that Discourse Adjectives do not provide information about the facts of the world. Rather, they are used by interlocutors to negotiate the status of propositions in a discourse.

Chapter One

Introduction

In this dissertation I introduce and provide a thorough semantic analysis of Discourse Adjectives, a natural class of adjectives that I argue includes apparent, clear, evident, and obvious among its prototypical members. My main claim is that Discourse Adjectives do not provide information about the facts of the world. Rather, they are used by interlocutors to negotiate the status of propositions in a discourse. Consider the data in (1).

(1) a. Briscoe is a detective.

b. It is clear that Briscoe is a detective.

Sentence (1a) is an informative sentence of English; it provides new descriptive information about the state of the world. In particular, when a speaker utters (1a), her addressee is provided with new information about “Briscoe, ” namely, that he is a detective. In contrast, I claim that (1b) does not provide new descriptive information—(lb) adds no new information about “Briscoe” per se. However, the sentence is informative. It provides information about the conversation under discussion, namely, that in the current discourse, evidence to support the conclusion that Briscoe is a detective is readily perceptible, at least to a minimum vague standard of clarity. Discourse Adjectives are a predicate type whose semantics allows people to talk about their conversation in addition to talking about their world.

Discourse Adjectives are presented as an example of a way of meaning that is not typical of expressions in natural language. I show that Discourse Adjectives do not add new information to the Common Ground in the conventional manner (Stalnaker 1979:325; van der Sandt 1992: 367, etc.); they provide an example of an expression type whose semantics conventionally carries no descriptive content.

I make a second argument that is closely intertwined with the first. This has to do with the interpretation of the conceptually necessary experiencer argument that is often not syntactically expressed in utterances with Discourse Adjectives. Consider the data in (2).

(2) a. It is clear that Briscoe is a detective.

b. It is clear to me that Briscoe is a detective.

c. It is clear to everyone that Briscoe is a detective.

d. It is clear to us that Briscoe is a detective.

For a proposition to be “clear” there needs to be an interpreter, or judge, of clarity. While the experiencer may be implicit, that is to say not syntactically expressed, as in (1a), it may also be overt as in (1b)–(1d). I argue that when the experiencer of a Discourse Adjective is left implicit, the default is for it to be interpreted as the participants in the discourse (1d), rather than the speaker alone (1b), or people in general as in (1c). This is noteworthy, since with other adjectives, an implicit experiencer can only be interpreted as either the speaker or people in general (the so-called ‘arbitrary’ reading; see, for example Rizzi 1986), rather than both the speaker and the addressee. To illustrate, note the contrast between (2a) and (3).

(3) It is absurd that Mary is a doctor.

The difference between (2a) and (3) is that after a discourse model is updated with a sentence containing a Discourse Adjective such as (2a), the interlocutors are justified in behaving, at least for the purposes of their current conversation, as if the proposition expressed by Briscoe is a detective is believed to be clear by all discourse participants. With an utterance containing what will be identified below as an Attitude Adjective, such as absurd in (3), the proposition expressed by Briscoe is a detective is guaranteed to be believed absurd by the speaker only. That is, in contrast with Discourse Adjectives, the discourse participants are not justified in behaving as if all discourse participants are in agreement regarding the absurdity of Briscoe’s detectivehood.

There is ample previous research on the interpretation of implicit arguments. It has been shown that the anchoring of implicit arguments, or the resolution of their referents, may occur at both syntactic and discourse-representational levels (Partee 1989, Condoravdi and Gawron 1996, Bhatt and Izvorski 1998). Additionally, it has been shown that explicit pronominal arguments, that is syntactically overt pronouns, may find an anchor in a discourse context (Hankamer and Sag 1976). To my knowledge, the anchoring of implicit arguments in discourse contexts, aside from the implicit non-speaker content of the pronoun we, has not been extensively discussed in the literature.

I show that an explanation of the facts about Discourse Adjectives demand a reconsideration of the structure of discourse models, and the nature of the organization of the Common Ground in discourse. In order to talk about the effect of an utterance with a Discourse Adjective, I adopt Gunlogson’s (2001) interpretation of the standard Stalnakerian model (Stalnaker 1979, 1998). This move enables me to isolate the propositions that belong to what will be identified as the public commitments of individual discourse participants. Gunlogson’s framework is presented in Chapter Three.

After introducing Gunlogson’s framework, I modify it slightly to allow for the treatment of vagueness (Fine 1975, Williamson 1994, Kyburg and Morreau 2002, Barker 2002). After presenting the formal model, I show how the meaning of Discourse Adjectives crucially involves identification of degrees of probability, rather than degrees of truth, or even degrees of clarity. The analysis I propose accounts for the interpretation of the implicit experiencers of Discourse Adjectives, and underscores the value of uttering sentences with no descriptive content. Both of these aspects of meaning are inexplicable in other theories of semantics.

I begin Chapter Two by presenting empirical facts about the distribution of Discourse Adjectives as compared to Raising and Attitude Adjectives. I show that these three classes have partially overlapping but distinct distributions, and conclude that Discourse Adjectives should be recognized as a natural syntactic class. I go on to argue that despite their appearance as factives, the syntactic behavior of Discourse Adjectives cannot be attributed to their Factivity (Kiparsky and Kiparsky 1970), precisely because they are not true factives. While in (4a) (=lb) clear has the ability to seem factive, sentences (4b) and (4c) confirm that the presupposition of the complement (that Briscoe is a detective) is not entailed to be true when clarity is negated or questioned.

(4) a. It is clear that Briscoe is a detective.

b. It isn’t clear that Briscoe is a detective.

c. Is it clear that Briscoe is a detective?

Chapter Two provides preliminary semantic descriptions for all three classes of proposition modifying adjective, and shows that the three identifiable syntactic classes correspond to three distinct semantic classes as well. I suggest a correlation between the observed semantics and syntax. A generalization that emerges (consistent with findings of, among others, Torrego 1996) is that experiencers interact with raising in the following way: raising is prohibited in the presence of a syntactically realized experiencer noun phrase. In light of this generalization I discuss an apparent counterexample found in the class of raising verbs, a class for which raising is not blocked by the overt expression of an experiencer as shown in (5), with the predicate seem.

(5) a. It seems to me that Briscoe is a detective.

b. Briscoe seems to me to be a detective.

Following Rothstein (1983) I analyze the experiencer of seem as an adjunct, and suggest that the generalization about the interaction of experiencers in raising is that experiencer arguments disallow raising, but experiencer adjuncts do not.

Chapters Three, Four and Five constitute the heart of this dissertation. In the first part of Chapter Three, I provide background information on context-update approaches to modeling the Common Ground of a discourse. I discuss the basic idea of Stalnaker (1979) and the early implementation of this proposed by Heim (1982). The version of the Stalnakerian framework I subsequently adopt is the one proposed by Gunlogson (2001), which differs from other Stalnakerian models by separating out the distinct commitments of the participants in a discourse.

I also use Chapter Three to introduce the Paradox of Asserting Clarity (Barker and Taranto 2002), a paradox that characterizes Discourse Adjectives, but not Raising or Attitude Adjectives. The introduction of the paradox in concert with the introduction to the framework illustrates the issues that any semantic analysis of Discourse Adjectives must address, and underscores the benefits of adopting Gunlogson’s innovation to the standard Stalnakerian model.

The introduction of the paradox sets the stage for Chapter Four, in which I propose a modification to Gunlogson’s model that is necessary for a thorough semantic analysis of clear, the member I take as the prototypical representative for the entire class of Discourse Adjectives. The proposed modification involves representing degrees of likelihood, after Barker (2002). This move allows for an accurate representation of the fact that Discourse Adjectives are vague predicates that involve a scale of probability.

Following the presentation of the semantics of clear, Chapter Five discusses the class of Discourse Adjectives as a whole, and shows that these adjectives interact with implicational hierarchies involving strength of evidence in addition to degrees of probability. By understanding the interaction of Discourse Adjectives with both implicational hierarchies, it is possible to accurately characterize empirical facts about the conditions in which some speakers might select the use of one Discourse Adjective over another.

Finally, Gunlogson originally developed her model in order to provide an analysis of specific intonational phenomena. More specifically, she separates the distinct contributions that individual discourse participants make to the Common Ground of a discourse. This formalization of the Common Ground underlies her compositional analysis of the contribution that rising and falling intonation make to the interpretation of sentences with declarative form. Examples are shown in (6), in which sentence final ‘.’ and ‘?’ indicate rising and falling intonation.

(6) a. Briscoe is a detective.

b. Briscoe is a detective?

Her observation is that with falling intonation, an utterance of (6a) commits the speaker to the content of the proposition expressed by Briscoe is a detective, while an utterance of (6b) commits the addressee. While there does not seem to be an intonational contour that allows a speaker to commit all the participants in a discourse to the content of an expressed proposition, this dissertation shows that Discourse Adjectives fill this role in the grammar of English. I show that the use of a Discourse Adjective with a declarative sentence and falling intonation is a strategy a speaker can adopt to commit both speaker and addressee to the content ...