This book presents a new theory of grammatical categories - the Universal Spine Hypothesis - and reinforces generative notions of Universal Grammar while accommodating insights from linguistic typology.
EDITORS: François Recanati, Isidora Stojanovic and Neftalí Villanueva TITLE: Context-Dependence, Perspective and Relativity SERIES TITLE: Mouton Series in Pragmatics 6 PUBLISHER: De Gruyter Mouton YEAR: 2010
Janneke Huitink, Institute of Linguistics, Goethe-University Frankfurt
This book brings together philosophical and linguistic papers about the pervasiveness of context-dependence. The collection originated at the conference ''Context Dependence, Perspective and Relativity in Language and Thought'' that was organized in Paris at the Ecole Normale Supérieure, November 9-11, 2007.
The editor's introduction provides a useful overview of the debates concerning the role of context in linguistic understanding. One important debate concerns the boundary between semantics and pragmatics. For instance, if Mary says ''Most people were drunk'', it is more or less agreed upon that all that is required for what she said to be true is that most people in the contextually relevant domain were drunk. But there is disagreement about the underlying mechanism. It could be the case that the sentence contains a free variable and can therefore only be interpreted if the context assigns this variable a value, or it could be the case that we turn to the context for purely pragmatic reasons, to make sense of what the speaker said.
Another debate that is of importance in the present volume is that between contextualism and relativism. It is commonly assumed that context plays two roles: (i) it helps to determine which proposition is expressed and (ii) it provides the indices relative to which we evaluate the proposition. Take the claim ''Tokyo is the capital of Japan''. While this is currently true, it was false in the past, as Kyoto used to be the capital of Japan. Contextualism maintains that the proposition expressed by this claim varies with the time of utterance. Relativism, on the other hand, holds that the same proposition is expressed, but receives different truth-values relative to different contextually determined times. It has often been argued that relativism is needed to account for so-called faultless disagreements. Suppose that A says ''This is delicious'' and B responds ''No, it isn't''. Intuitively, A and B are disagreeing, but contextualism fails to grasp this. According to this theory, A expresses the proposition that this is delicious to A, while B expresses that this is not delicious to B. These propositions are not contradictory. In contrast, the relativist would say that there is one proposition (that this is delicious) to which A and B assign opposite truth-values. Indices of evaluation thus do not just consist of a world and a time, but also include a person (the ''judge'').
The book is divided into five parts. Part 1: ''Implicit content: philosophical perspectives'' asks how far our thoughts and beliefs are context-dependent. It has been argued that beliefs may contain unarticulated constituents. A famous case in point, going back to Perry (1986) is the belief that it is raining, believed by a Z-lander. Z-landers do not travel to, or communicate with, residents of other places, and they have no name for Z-land. Therefore, the belief that it is raining doesn't represent the location, Z-land. In ''Unarticulated tension'', Lenny Clapp argues that this idea is incompatible with the Crimmins & Perry (1989) analysis of belief-reports. According to this analysis, developed to solve Frege's doxastic puzzle cases, belief reports do not only concern the content of the belief, but also refer to the particular mode of presentation the subject employs in the belief. Now consider the report: ''This Z-lander believes that it is raining''. Clapp argues that the weather location must be a part of reported belief, in order for the report to be accurate. That is, what I said was really something like ''This Z-lander believes that it is raining there'' (but I left the last word unpronounced). Following Crimmins & Perry, this is true only if there is a mode of presentation of the weather location in the relevant Z-lander's belief (with which a belief that it is raining there is associated). The problem is that, by assumption, such a mode of presentation does not exist.
In ''What is said'' Kepa Korta and John Perry consider some puzzling saying-reports. Suppose that Ivan is unaware of the Basque custom to use ''Donostia'' to refer to the city San Sebastian, while Tom and his friends do know about this, and in fact prefer to employ ''Donostia''. Scenario 1: Tom and his friends are looking for the bus to San Sebastian/Donostia and Tom overhears Ivan musing to himself ''This bus is not going to Donostia; it is going to San Sebastian''. Tom may now report this to his friends as ''Ivan says that this bus is going to Donostia''. Scenario 2: The bus passes a sign ''Free drinks at Noam's Bar in Donostia''. Even though Ivan likes free drinks, he doesn't show any enthusiasm. He mutters ''But this bus isn't going to Donostia''. Tom may now explain his lack of enthusiasm by saying ''Ivan says that this bus isn't going to Donostia''. To account for these reports, Korta & Perry propose that saying-reports refer to the particular words that the subject used (analogous to the Crimmins & Perry account of belief-reports). In scenario 1, the context determines that Ivan's use of ''San Sebastian'' is relevant. Because Ivan used this name in an utterance with the content that Tom would express by ''This bus is going to Donostia'', the report is true. In scenario 2, however, Ivan's use of ''Donostia'' counts. Because Ivan used ''Donostia'' in an utterance the content of which was that the bus isn't going there, the report is true.
Augustín Vicente argues in ''Context dependency in thought'' that the vehicle of thought is fully explicit. There are no unarticulated constituents in our thinking. Consider again our Z-lander's belief that it is raining. Perry (1986) claims that this belief concerns Z-land without representing it. Hence, the Z-lander's thought is not a proposition, but a propositional function. Vicente points out that this only follows if we insist that the Z-lander's concept of rain is like ours in that it picks out a dyadic property (a relation between times and places). Alternatively, we could assume that Z-landers have a different rain-concept, picking out a monadic property (of times). If the Z-landers would become nomads and would start having weather-thoughts about other places, we would simply have to say that their ''rain''-predicate has changed its meaning.
In Part 2, ''Implicit content (2): linguistic perspectives'' concerns context-sensitivity in language rather than thought. First, Urtzi Etxeberria and Anastasia Giannakidou argue in ''Contextual domain restriction and the definite determiner'' that quantifier domain restriction is explicitly encoded in the grammar, at least in Greek, Basque, and Salish. In Salish, for example, quantifiers in argument position must appear with a definite determiner modifying their nominal argument. Thus, in this language, a construction like (i) would be ungrammatical because ''all'' combines with the bare nominal ''children'' rather than with the modified nominal ''the children'':
(i) *I gave all children candy. (ii) I gave all the children candy.
The authors propose that (strong) quantifiers can only combine with a nominal argument after it has been contextually restricted. In Salish, the definite determiner takes care of this job, i.e. in (ii) ''the children'' yields a contextually salient set of children. If this story is on the right track, domain restriction is not a free pragmatic process, but is mandated by linguistic material.
''The model theory for words with context-sensitive implicit arguments'' by Brendan S. Gillon concerns verbs that have both an intransitive and a transitive use. A case in point is ''leave'': ''Bill left'' vs. ''Bill left Cairo''. Gillon argues that it is not necessary to postulate lexical ambiguity for ''leave''. Roughly, he proposes that ''Bill left'' contains an implicit argument, the semantic value of which is contextually determined. This is the value that would be assigned to a suitable pronoun, had the direct object slot been filled with a pronoun.
Finally, in ''Three types of ellipsis'' Jason Merchant defends a semantic slot-filling analysis of ellipsis. Consider the following example:
(i) Sanjay and Silvia are loading up a van. Silvia is looking for a missing table leg. Sanjay says, ''On the stoop.''
According to Merchant, it is best not to assume that Sanjay actually uttered the syntactic sentence ''It is on the stoop'' and left the first two words unpronounced. Rather, Sanjay simply uttered a predicate in isolation. However, Sanjay did manage to produce a semantic sentence, i.e. something of type t. After all, we have the intuition that if the table leg is not on the stoop, Sanjay has said something false. To account for this, Merchant proposes that the semantic value of ''On the stoop'' contains a free-variable: ''on-the-stoop(x)''. This variable can either be bound by a lambda-operator (which happens in syntactic sentences like ''The table is on the stoop''), but it may also be left free, which is the case in (i). Then the contextual assignment function must assign a semantic value to x. In (i), this function will map x onto the missing table leg. Merchant concludes that no extraordinary appeal to pragmatics is needed to cover ellipsis.
Part 3, ''Relativism, knowledge ascriptions, and predicates of personal taste'' deals with relativity to standards of taste, certainty, etc. In ''Relativism, disagreement and predicates of personal taste'' Barry C. Smith points out that, somewhat paradoxically perhaps, relativism makes the very problem it is supposed to solve disappear. Relativism is usually said to account for the fact that two people, A and B, can be genuinely disagreeing and can be right at the same time. However, it follows from relativism that one can only say that someone is right relative to a perspective. And from each individual perspective (even that of a third party), only one of A and B can be right, and the other must be wrong (though neither one can be said to be absolutely right (wrong)). It is therefore misleading to present relativism as the claim that A and B may issue incompatible claims and may both be right. Instead, relativism is simply the claim that truth is always truth-relative-to-a-perspective.
In ''Knowledge attributions and relevant epistemic standards'' Dan Zeman discusses knowledge attributions like ''I know that the bank will be open on Saturday''. Zeman's goal is to account for the data that Stanley (2005) reports in the introduction to his book ''Knowledge and Practical Interests''. Suppose that on Friday Hannah wants to deposit her paycheck the next day. If nothing hinges on her depositing her paycheck by Saturday, she may reason as follows: ''I was at the bank two weeks ago on a Saturday, so I know that the bank will be open tomorrow''. If, on the other hand, she has an impending bill coming due and it is very important that she make the deposit by Saturday, then someone pointing out that banks do change their opening hours may distort her reasoning. In this case, Hannah might truthfully utter: ''I do not really know that the bank will be open tomorrow''. In this same scenario consider Jill, for whom nothing is at stake, who reasons as follows: ''I saw Hannah at the bank two weeks ago on a Saturday, so Hannah knows that the bank will be open tomorrow''. Zeman reports that her statement is intuitively false. To account for this, he proposes a preference for the highest epistemic standards (the ''Highest Standard Principle''). Thus, once we have accepted the possibility that the bank has changed its opening hours as a live option, we have raised the bar for the application of ''know'', and this bar continues to be high while the story progresses. Jill's utterance ''Hannah knows that the bank will be open tomorrow'' will therefore be evaluated with respect to this high epistemic standard (instead of her own, lower one), and will come out false. Zeman presents his account as an extension of relativism, but as far as I can see, the main idea is compatible with contextualism as well.
Part 4, ''Perspectival shifts'' concerns the interpretation of indirect discourse. In ''On words and thoughts about oneself'' James Higginbotham writes about first-personal thought as reflected in language. Notoriously, (i) and (ii) do not mean the same thing, even if ''he'' refers to John:
(i) John expects (PRO) to win. (ii) John expects that he will win.
Suppose that John expects that the person with ticket 47 will win, and that John actually holds ticket 47, but that he is unaware of this (say that he has not looked at his ticket number yet). In this scenario, the attitude report in (ii) is true, but the one in (i) is false. This indicates that (i) must be interpreted as a de se expectation: i.e. an expectation of John that is necessarily from the first person perspective (he must think to himself: ''I will win''). It has been argued in the literature that this is best explained by an appeal to ''monstrous'' context shifts, but Higginbotham argues in favor of an account in terms of anaphora. Simplifying things considerably, his theory may be stated as follows. The grammatical subject of the embedded clause in (i), PRO, receives its reference through a definite description, namely ''the experiencer of the reported attitude''. This description is special, because one cannot conceive of a scenario in which John misidentifies the reference of this description. Therefore, (i) is necessarily de se. In contrast, ''he'' in (ii) gets its reference in the normal way, via its antecedent, the higher subject ''John'', which leaves room for misidentification.
In ''Indirect discourse, relativism, and contexts that point to other contexts'' Christopher Gauker observes that some words require adjustment to the speaker's context in indirect discourse, while for others adjustment is only optional. Imagine that Silas is at Jonas' birthday party. He is talking to Emma on the phone, who is at a different party. Silas tells Emma ''There is a clown here''. If Emma wants to report this, she must substitute Silas' word ''here'' for ''there'':
(i) Silas says that there is a clown there. (ii) *Silas says that there is a clown here.
Next Silas says ''Everybody is wearing a funny hat''. When Emma wants to report this utterance, she may add a more precise classification of the domain of “everybody”, but she doesn't have to. The following are both accurate:
(iii) Silas says that everybody at Jonas' party is wearing a funny hat. (iv) Silas says that everybody is wearing a funny hat.
Gauker proposes that a context c is an entity that assigns other contexts to relevant people. For instance, c(S) is the context determined for the person S by context c. He then proposes that ''S says that p'' is true in a context c if and only if there is a sentence q such that S uttered q in c(S), and p is an ''elevation'' of q to c from c(S). Emma's utterance of (i) is true in c, for Silas did utter ''There is a clown here'', and the sentence embedded in the speech report ''There is a clown there'' is an elevation of this sentence, because the place that ''there'' refers to in c is the same place that ''here'' refers to in the context that Emma's context assigns to Silas. Similarly, both (iii) and (iv) are true, because both ''Everybody at Jonas' party is wearing a funny hat'' and ''Everybody is wearing a funny hat'' count as elevations of Silas' utterance ''Everybody is wearing a funny hat''. In the latter case, Emma is simply quoting Silas' words, leaving it unclear what the relevant domain is.
Finally, Part 5 ''Perspectival facts'' addresses some metaphysical questions pertaining to relativism. In ''Is relativity a requirement for mind-dependence?'' Eyja M. Brynjarsdóttir asks how subjective properties like ''being funny'' differ from objective properties like ''being 1 meter tall''. One might be tempted to answer that subjective properties give rise to faultless disagreements, while objective properties do not. But Brynjarsdóttir argues that this is mistaken. For example, when Emma says ''Grover is funny'', a contextualist would say this means as much as ''Grover is funny-for-Emma''. Crucially, what makes this true is Emma's state of mind: if Grover triggers an amusement response in Emma's mind, then this may cause her to utter ''Grover is funny'', by which she assigns Grover the property of being funny-to-Emma, which in turn makes it the case that he has the property. In contrast, being 1 meter tall is not a property that Grover has in virtue of someone's mental response. Thus, Brynjarsdóttir proposes that a property is subjective just in case a mental state is responsible for it being instantiated.
In ''Perspectival truth and perspectival realism'' Giuliano Torrengo argues that, from a metaphysical point of view, it is not clear that there is a substantive distinction between relativism and contextualism. He first argues that relativists must embrace what he calls ''perspectival realism'': perspectives are ''realities'' akin to possible worlds. Furthermore, relativists must take a pluralistic stance, i.e. they must allow multiple perspectives to enjoy the same status: there is no such thing as the ''right'' fact. But pluralistic perspectival realism is not without problems. First, one may question the intelligibility of there being pair-wise incompatible perspectives of which the actual world is composed. Second, given this metaphysics we can no longer adjudicate between relativism and contextualism. Recall that relativism is supposed to account for faultless disagreements. However, even on this theory A and B do not only express contradictory propositions, but they also disagree on which perspective is the relevant one for evaluating their claims. But then relativism suffers from the same problem as contextualism: it is hard to see where the intuition that there is a genuine disagreement comes from.
The collection of papers does a good job illustrating the scope of the two context debates that were presented in the introduction. The book makes it clear that context dependence is at the heart of a wide range of phenomena, but also that the jury is still out on the limits of what context may achieve, in language, and in thought. Precisely because context dependence is so pervasive, the topics discussed in the papers are somewhat disparate, and for this reason it is not entirely clear that there is enough of a connection between the five parts that merits putting them together in a single volume. This is not to deny that each paper is worth reading on its own.
My main criticism concerns the division between linguistics and philosophy. First, in this volume the contextualism-relativism debate is dominated by philosophical arguments. This might create the impression that linguistics has little to bring to the table, which would be incorrect. For example, Sæbø (2009) offers new linguistic arguments that favor contextualism over relativism, while presenting diagnostic tools for subjective properties that go beyond the faultless disagreement test. Second, though the semantics-pragmatics interface debate is illustrated by both philosophical and linguistic work, there appears to be little to no dialogue between researchers from these disciplines. For instance, linguistic work on mixed quotation (Potts 2007, Geurts & Maier 2005) seems highly relevant to the papers on indirect speech reports, but is not discussed. At the same time, the linguistic papers in the book (with the exception of Etxeberria & Giannakidou's contribution) do not really address the philosophical question whether a free pragmatic account is a feasible alternative to slot-filling.
In conclusion, though the volume cannot be called a truly interdisciplinary endeavor, it certainly contains papers that will be of interest to both philosophers and linguists who are interested in context dependence.
Crimmins, Mark & John Perry. 1989. The prince and the phone booth: reporting puzzling beliefs. Journal of Philosophy 86. 685-711.
Geurts, Bart & Emar Maier. 2005. Quotation in context. Belgian Journal of Linguistics 17(1). 109-128.
Perry, John. 1986. Thought without representation. Supplementary Proceedings of the Aristotelian Society 60. 263-283.
Potts, Chris. 2007. The dimensions of quotation. In C. Barker and P. Jacobson (Eds.). Direct Compositionality. Oxford: Oxford University Press. 405-431.
Stanley, Jason. 2005. Knowledge and practical interests. Oxford: Clarendon Press.
ABOUT THE REVIEWER
ABOUT THE REVIEWER:
Janneke Huitink obtained a Ph.D. in Philosophy from the Radboud University
Nijmegen, the Netherlands. She is currently researching philosophical,
linguistic and psychological aspects of conditional sentences. Together
with Cécile Meier she hosted a workshop on subjective meaning at the
conference of the Deutsche Gesellschaft für Sprachwissenschaft (DGfS), the
goal of which was to find alternatives for relativism.