The goal of this book is to explore the extent to which salient meanings – meanings that are decoded in our mental lexicon and foremost on our mind – affect our speech production and comprehension. “It aims to shed light, primarily empirically, on how, in addition to contextual information, salient meanings and senses of words and fixed expressions shape our linguistic behavior” (p. 9). According to Rachel Giora's graded salience hypothesis, salient meanings are more accessible than less salient ones. Of the multiple meanings of a given word or expression, the most salient one, which can be either literal or figurative, is always activated first regardless of context.

The book consists of nine chapters. The first three outline the theoretical background against which the empirical studies reported in the subsequent chapters were conducted. Chap. 1 offers some linguistic and nonlinguistic examples of how our salience-bound mind works. It then introduces three information-processing models that vary in respect to the temporal stage at which context affects language comprehension. According to the interactionist, direct access view, the contextual information interacts with lexical processes very early on and enables the appropriate interpretation to be tapped directly, without involving a contextually inappropriate stage initially. The modular view, in contrast, holds that linguistic information is initially processed independently of contextual information. When the output of the linguistic module does not cohere with the contextual information, it needs to go through a later stage of revision and adjustment to the contextual knowledge. The graded salience hypothesis posits that more salient meanings are always activated before less salient meanings in the linguistic module. They are accessed upon encounter regardless of contextual information. A highly predictive context, however, may speed up derivation of the appropriate meaning while it would not obstruct inappropriate salient meanings upon encounter of the lexical stimulus.

Chap. 2 defines the notions of salience and context and their relationship. Salient meanings are cognitively prominent owing to frequency, familiarity, conventionality, or prototypicality. The notion of salience is both graded and dynamic. That is, various meanings of a word or expression are ranked in salience, and their ranking is not fixed but is subject to reshuffling over time. Context may be predictive and affect the availability of meanings early on. However, contextual processes do not interact with lexical processes initially but run in parallel. Language comprehension comprises two phases: the initial phase of activation, and the subsequent phase of integration. In the initial phase, “contextually appropriate and salient meanings are activated – the latter automatically and independently of contextual information, the former as a result of a predictive context” (38). In the subsequent phase, “the activated meanings are either retained for further processes or suppressed as contextually disruptive” (38).

Chap. 3 addresses the issue of lexical access, on which the graded salience hypothesis differs from both the direct access view and the modular model. With selective access (the direct access view), the interaction between contextual and lexical processes early on selects and activates the contextually appropriate meanings exclusively, irrespective of their salience. With exhaustive access (the modular model), lexical access is autonomous, with all the meanings of a word or expression activated regardless of context or salience. Contextual processes interact only with the output of lexical processes, selecting the contextually compatible meaning. With ordered access (the graded salience hypothesis), lexical access, though autonomous, is sensitive to salience: More salient meanings are activated faster than less salient ones. Contextual knowledge retains or suppresses the salient meaning, and in the latter case selects a less salient but contextually compatible meaning. That is, context, though predictive of certain meanings, cannot block initial activation of salient meanings even though they are contextually incompatible.

Chaps. 4 through 7 study the comprehension of figurative language. Chap. 4 looks into how irony is processed and argues that experimental findings support the graded salience hypothesis, which makes different predictions regarding familiar and unfamiliar ironies with respect to the initial processes. With familiar ironies, their ironic interpretations, as well as their literal meanings, are salient and therefore are activated directly and automatically in the initial phase. With unfamiliar ironies, on the other hand, only their literal meanings are salient and accessed initially, and the contextual intervention in the subsequent processes will derive their ironic counterparts. In either case, however, irony retains its contextually incompatible salient (literal) meaning for contrast in irony interpretation. Only when familiar irony is used in a literally biasing context can its salient ironic interpretation be suppressed and discarded.

Chap. 5 examines how metaphors and idioms are processed. Experimental results show similar patterns in the activation of meanings for metaphors and idioms, as well as for irony. Both literal and figurative meanings of familiar metaphors and idioms are activated initially, irrespective of context, because they are both salient. The literal meaning, though contextually incompatible, is retained because it is conducive to the interpretation of the metaphor or idiom. For unfamiliar metaphors and idioms, the salient literal meaning is accessed initially and will give way to the less salient, figurative meaning in the subsequent phase of integration as dictated by contextual information. When familiar metaphors and idioms are embedded in a literally biasing context, however, their salient metaphoric or idiomatic meanings will be suppressed and discarded because they interfere with the comprehension of their literal counterparts.

Chap. 6 focuses on joke comprehension. Most jokes involve either polysemous interpretation of related meanings, or ambiguous interpretation of unrelated meanings, of a key word or expression. Under the graded salience hypothesis, jokes start with the salient meaning and end with a less salient meaning. That is, they tend to invite the processing of the more salient but eventually incompatible meaning first before leading to the activation of a less salient but congruent meaning for a reinterpretation. Although jokes and other tropes, such as irony and metaphor, share similar initial processes of activating salient meaning first, they differ in the subsequent integration processes. Whereas irony and metaphor comprehension retains salient but contextually incompatible meanings for interpretation, the comprehension of jokes needs to discard such meanings. This is because salient but contextually incompatible meanings do not play an instrumental role in the comprehension of jokes; sometimes they may even prohibit comprehension.

Chap. 7 discusses the role of salience in aesthetic innovation. According to the graded salience hypothesis, “novelty stands in some complementary relation to salience” (176); “it is not extreme novelty but ‘optimal’ innovation – novelty that allows for the recoverability of the familiar – that is most pleasurable” (176). In other words, pleasure is “a function of both salience and innovativeness,” and is “the gratification in discovering the familiar in the novel” (182). It is also argued that optimal innovation in the use of language is a mode of subversiveness, designed not only for aesthetic purposes but also for social and political change. Such change is both induced and represented by the deviation from the linguistic norm.

Chap. 8 reviews contemporary literature to see if its findings can be accounted for by the graded salience hypothesis. The evidence is only partly consistent with the standard pragmatic view or the direct access view, but almost entirely consistent with the graded salience hypothesis. Thus, the traditional view cannot account for the fact that context actually facilitates initial comprehension of figurative language. The direct access view can, however, because it does not require that the literal meaning of figurative language be processed first. The graded salience hypothesis can, too, since it predicts that salient meanings of familiar figurative utterances would be processed initially regardless of context. On the other hand, the direct access view cannot account for the fact that the literal meaning of figurative language is activated even though it is contextually incompatible. This fact, however, can be accounted for by both the traditional view, which mandates the initial activation of the literal meaning of figurative language before contextual intervention, and the graded salience hypothesis, which predicts initial activation of the salient literal meaning of unfamiliar figurative language regardless of context. It is concluded, therefore, that the graded salience hypothesis is superior in that it enables the reconciliation of conflicting views based on conflicting findings. The implication is that the distinction between the salient and less salient, rather than between the literal and figurative, is applicable to initial processes. As is shown, figurative and literal utterances involve different processes when diverging in salience, but similar processes when converging in salience.

Chap. 9 suggests the implications of the study and raises some questions for future research. It concludes with a note on autonomy of the mind as reflected in the superiority of salient information, which could affect our thinking and understanding in various ways irrespective of contextual information.

In this book, Giora provides a comprehensive psycholinguistic account of how figurative, as well as literal, language is comprehended, based on extensive review of literature and sensible analysis of experimental findings. It is a useful book for linguists and psychologists interested in figurative language, especially in the processes of language production and comprehension.