Indexicality

The concept of indexicality has been greatly elaborated in the literature of linguistic anthropology since its introduction by Silverstein, but Silverstein himself adopted the term from the theory of sign phenomena, or semiotics, of Charles Sanders Peirce. As an implication of his general metaphysical theory of the three universal categories, Peirce proposed a model of the sign as a triadic relationship: a sign is "something which stands to somebody for something in some respect or capacity."[7] Thus, more technically, a sign consists of

• A sign-vehicle or representamen, the perceptible phenomenon which does the representing, whether audibly, visibly or in some other sensory modality;[8]^{: "Representamen"}
• An object, the entity of whatever kind, with whatever modal status (experienceable, potential, imaginary, law-like, etc.), which is represented by the sign;[8]^: "Object" and
• An interpretant, the "idea in the mind" of the perceiving individual, which interprets the sign-vehicle as representing the object.[8]^{: "Interpretant"}

Peirce further proposed to classify sign phenomena along three different dimensions by means of three trichotomies, the second of which classifies signs into three categories according to the nature of the relationship between the sign-vehicle and the object it represents. As captioned by Silverstein, these are:

• Icon: a sign in which "the perceivable properties of the sign vehicle itself have isomorphism to (up to identity with) those of the entity signaled. That is, the entities are 'likenesses' in some sense."[5]^: 27
• Index: a sign in which "the occurrence of a sign vehicle token bears a connection of understood spatio-temporal contiguity to the occurrence of the entity signaled. That is, the presence of some entity is perceived to be signaled in the context of communication incorporating the sign vehicle."[5]^: 27
• Symbol: the residual class, a sign which is not related to its object by virtue of bearing some qualitative likeness to it, nor by virtue of co-occurring with it in some contextual framework. These "form the class of 'arbitrary' signs traditionally spoken of as the fundamental kind of linguistic entity. Sign vehicle and entity signaled are related through the bond of a semantico-referential meaning"[5]^: 27 which permits them to be used to refer to any member of a whole class or category of entities.

Silverstein observes that multiple signs may share the same sign-vehicle. For instance, as mentioned, linguistic signs as traditionally understood are symbols, and analyzed in terms of their contribution to reference and predication, since they arbitrarily denote a whole class of possible objects of reference by virtue of their semantic meanings. But in a trivial sense each linguistic sign token (word or expression spoken in an actual context of use) also functions iconically, since it is an icon of its type in the code (grammar) of the language. It also functions indexically, by indexing its symbol type, since its use in context presupposes that such a type exists in the semantico-referential grammar in use in the communicative situation (grammar is thus understood as an element of the context of communication).[5]^: 27–28

So icon, index and symbol are not mutually exclusive categories—indeed, Silverstein argues, they are to be understood as distinct modes of semiotic function,[5]^: 29 which may be overlaid on a single sign-vehicle. This entails that one sign-vehicle may function in multiple semiotic modes simultaneously. This observation is the key to understanding deixis, traditionally a difficult problem for semantic theory.

In linguistic anthropology, deixis is defined as referential indexicality—that is, morphemes or strings of morphemes, generally organized into closed paradigmatic sets, which function to "individuate or single out objects of reference or address in terms of their relation to the current interactive context in which the utterance occurs.".[9]^: 46–47 Deictic expressions are thus distinguished, on the one hand, from standard denotational categories such as common nouns, which potentially refer to any member of a whole class or category of entities: these display purely semantico-referential meaning, and in the Peircean terminology are known as symbols. On the other hand, deixis is distinguished as a particular subclass of indexicality in general, which may be nonreferential or altogether nonlinguistic.

In the older terminology of Otto Jespersen and Roman Jakobson, these forms were called shifters.[10][11] Silverstein, by introducing the terminology of Peirce, was able to define them more specifically as referential indexicals.[5]

Non-referential indices or "pure" indices do not contribute to the semantico-referential value of a speech event yet "signal some particular value of one or more contextual variables."[5] Non-referential indices encode certain metapragmatic elements of a speech event's context through linguistic variations. The degree of variation in non-referential indices is considerable and serves to infuse the speech event with, at times, multiple levels of pragmatic "meaning."[12] Of particular note are: sex/gender indices, deference indices (including the affinal taboo index), affect indices, as well as the phenomena of phonological hypercorrection and social identity indexicality.

Examples of non-referential forms of indexicality include sex/gender, affect, deference, social class, and social identity indices. Many scholars, notably Silverstein, argue that occurrences of non-referential indexicality entail not only the context-dependent variability of the speech event, but also increasingly subtle forms of indexical meaning (first, second, and higher-orders) as well.[12]

One common system of non-referential indexicality is sex/gender indices. These indices index the gender or "female/male" social status of the interlocutor. There are a multitude of linguistic variants that act to index sex and gender such as:

• word-final or sentence-final particles:many languages employ the suffixation of word-final particles to index the gender of the speaker. These particles vary from phonological alterations such as the one explored by William Labov in his work on postvocalic /r/ employment in words that had no word final "r" (which is claimed, among other things, to index the "female" social sex status by virtue of the statistical fact that women tend to hypercorrect their speech more often than men);[13] suffixation of single phonemes, such as /-s/ in Muskogean languages of the southeastern United States;[5] or particle suffixation (such as the Japanese sentence-final use of -wa with rising intonation to indicate increasing affect and, via second-order indexicality, the gender of the speaker (in this case, female))[13]
• morphological and phonological mechanisms: such as in Yana, a language where one form of all major words are spoken by sociological male to sociological male, and another form (which is constructed around phonological changes in word forms) is used for all other combination of interlocutors; or the Japanese prefix-affixation of o- to indicate politeness and, consequently, feminine social identity.[14]

Many instances of sex/gender indices incorporate multiple levels of indexicality (also referred to as indexical order).[12] In fact, some, such as the prefix-affixation of o- in Japanese, demonstrate complex higher-order indexical forms. In this example, the first order indexes politeness and the second order indexes affiliation with a certain gender class. It is argued that there is an even higher level of indexical order evidenced by the fact that many jobs use the o- prefix to attract female applicants.[14] This notion of higher-order indexicality is similar to Silverstein's discussion of "wine talk" in that it indexes "an identity-by-visible-consumption[12] [here, employment]" that is an inherent of a certain social register (i.e. social gender indexicality).

Affective meaning is seen as "the encoding, or indexing of speakers emotions into speech events."[15] The interlocutor of the event "decodes" these verbal messages of affect by giving "precedence to intentionality";[15] that is, by assuming that the affective form intentionally indexes emotional meaning.

Some examples of affective forms are: diminutives (for example, diminutive affixes in Indo-European and Amerindian languages indicate sympathy, endearment, emotional closeness, or antipathy, condescension, and emotional distance); ideophones and onomatopoeias; expletives, exclamations, interjections, curses, insults, and imprecations (said to be "dramatizations of actions or states"); intonation change (common in tone languages such as Japanese); address terms, kinship terms, and pronouns which often display clear affective dimensions (ranging from the complex address-form systems found languages such a Javanese to inversions of vocative kin terms found in Rural Italy);[15] lexical processes such as synecdoche and metonymy involved in effect meaning manipulation; certain categories of meaning like evidentiality; reduplication, quantifiers, and comparative structures; as well as inflectional morphology.

Affective forms are a means by which a speaker indexes emotional states through different linguistic mechanisms. These indices become important when applied to other forms of non-referential indexicality, such as sex indices and social identity indices, because of the innate relationship between first-order indexicality and subsequent second-order (or higher) indexical forms. (See multiple indices section for Japanese example).

Deference indices encode deference from one interlocutor to another (usually representing inequalities of status, rank, age, sex, etc.).[5] Some examples of deference indices are:

The T/V deference entitlement system of European languages was famously detailed by linguists Brown and Gilman.[16] T/V deference entitlement is a system by which a speaker/addressee speech event is determined by perceived disparities of 'power' and 'solidarity' between interlocutors. Brown and Gilman organized the possible relationships between the speaker and the addressee into six categories:

1. Superior and solidary
2. Superior and not solidary
3. Equal and solidary
4. Equal and not solidary
5. Inferior and solidary
6. Inferior and not solidary

The 'power semantic' indicates that the speaker in a superior position uses T and the speaker in an inferior position uses V. The 'solidarity semantic' indicates that speakers use T for close relationships and V for more formal relationships. These two principles conflict in categories 2 and 5, allowing either T or V in those cases:

1. Superior and solidary: T
2. Superior and not solidary: T/V
3. Equal and solidary: T
4. Equal and not solidary: V
5. Inferior and solidary: T/V
6. Inferior and not solidary: V

Brown and Gilman observed that as the solidarity semantic becomes more important than the power semantic in various cultures, the proportion of T to V use in the two ambiguous categories changes accordingly.

Silverstein comments that while exhibiting a basic level of first-order indexicality, the T/V system also employs second-order indexicality vis-à-vis 'enregistered honorification'.[12] He cites that the V form can also function as an index of valued "public" register and the standards of good behavior that are entailed by use of V forms over T forms in public contexts. Therefore, people will use T/V deference entailment in 1) a first-order indexical sense that distinguishes between speaker/addressee interpersonal values of 'power' and 'solidarity' and 2) a second-order indexical sense that indexes an interlocutor's inherent "honor" or social merit in employing V forms over T forms in public contexts.

Japanese provides an excellent case study of honorifics. Honorifics in Japanese can be divided into two categories: addressee honorifics, which index deference to the addressee of the utterance; and referent honorifics, which index deference to the referent of the utterance. Cynthia Dunn claims that "almost every utterance in Japanese requires a choice between direct and distal forms of the predicate."[17] The direct form indexes intimacy and "spontaneous self-expression" in contexts involving family and close friends. Contrarily, distal form index social contexts of a more formal, public nature such as distant acquaintances, business settings, or other formal settings.

Japanese also contains a set of humble forms (Japanese kenjōgo 謙譲語) which are employed by the speaker to index their deference to someone else. There are also suppletive forms that can be used in lieu of regular honorific endings (for example, the subject honorific form of taberu (食べる, to eat): meshiagaru 召し上がる). Verbs that involve human subjects must choose between distal or direct forms (towards the addressee) as well as a distinguish between either no use of referent honorifics, use of subject honorific (for others), or use of humble form (for self). The Japanese model for non-referential indexicality demonstrates a very subtle and complicated system that encodes social context into almost every utterance.

Dyirbal, a language of the Cairns rain forest in Northern Queensland, employs a system known as the affinal taboo index. Speakers of the language maintain two sets of lexical items: 1) an "everyday" or common interaction set of lexical items and 2) a "mother-in-law" set that is employed when the speaker is in the very distinct context of interaction with their mother-in-law. In this particular system of deference indices, speakers have developed an entirely separate lexicon (there are roughly four "everyday" lexical entries for every one "mother-in-law" lexical entry; 4:1) to index deference in contexts inclusive of the mother-in-law.

Hypercorrection is defined by Wolfram as "the use of speech form on the basis of false analogy."[18] DeCamp defines hypercorrection in a more precise fashion claiming that "hypercorrection is an incorrect analogy with a form in a prestige dialect which the speaker has imperfectly mastered."[19] Many scholars argue that hypercorrection provides both an index of "social class" and an "Index of Linguistic insecurity". The latter index can be defined as a speaker's attempts at self-correction in areas of perceived linguistic insufficiencies which denote their lower social standing and minimal social mobility.[20]

Donald Winford conducted a study that measured the phonological hypercorrection in creolization of English speakers in Trinidad. He claims that the ability to use prestigious norms goes "hand-in-hand" with knowledge of stigmatization afforded to use of "lesser" phonological variants.[20] He concluded that sociologically "lesser" individuals would try to increase the frequency of certain vowels that were frequent in the high prestige dialect, but they ended up using those vowels even more than their target dialect. This hypercorrection of vowels is an example of non-referential indexicality that indexes, by virtue of innate urges forcing lower class civilians to hypercorrect phonological variants, the actual social class of the speaker. As Silverstein claims, this also conveys an "Index of Linguistic insecurity" in which a speaker not only indexes their actual social class (via first-order indexicality) but also the insecurities about class constraints and subsequent linguistic effects that encourage hypercorrection in the first place (an incidence of second-order indexicality).[12]

William Labov and many others have also studied how hypercorrection in African American Vernacular English demonstrates similar social class non-referential indexicality.

Multiple non-referential indices can be employed to index the social identity of a speaker. An example of how multiple indexes can constitute social identity is exemplified by Ochs discussion of copula deletion: "That Bad" in American English can index a speaker to be a child, foreigner, medical patient, or elderly person. Use of multiple non-referential indices at once (for example copula deletion and raising intonation), helps further index the social identity of the speaker as that of a child.[21]

Linguistic and non-linguistic indices are also an important ways of indexing social identity. For example, the Japanese utterance -wa in conjunction with raising intonation (indexical of increasing affect) by one person who "looks like a woman" and another who looks "like a man" may index different affective dispositions which, in turn, can index gender difference.[13] Ochs and Schieffilen also claim that facial features, gestures, as well as other non-linguistic indices may actually help specify the general information provided by the linguistic features and augment the pragmatic meaning of the utterance.[22]

In much of the research currently conducted upon various phenomena of non-referential indexicality, there is an increased interest in not only what is called first-order indexicality, but subsequent second-order as well as "higher-order" levels of indexical meaning. First-order indexicality can be defined as the first level of pragmatic meaning that is drawn from an utterance. For example, instances of deference indexicality, such as the variation between informal tu and formal vous in French, indicate a speaker/addressee communicative relationship built upon the values of power and solidarity possessed by the interlocutors.[16] When a speaker addresses somebody using the V form instead of the T form, they index (via first-order indexicality) their understanding of the need for deference to the addressee. In other words, they perceive or recognize an incongruence between their levels of power and/or solidarity and employ a more formal way of addressing that person to suit the contextual constraints of the speech event.

Second-Order Indexicality is concerned with the connection between linguistic variables and the metapragmatic meanings that they encode. For example, a woman is walking down the street in Manhattan and she stops to ask somebody where a McDonald's is. He responds to her talking in a heavy "Brooklyn" accent. She notices this accent and considers a set of possible personal characteristics that might be indexed by it (such as the man's intelligence, economic situation, and other non-linguistic aspects of his life). The power of language to encode these preconceived "stereotypes" based solely on accent is an example of second-order indexicality (representative of a more complex and subtle system of indexical form than that of first-order indexicality).

For demonstrations of higher (or rarefied) indexical orders, Michael Silverstein discusses the particularities of "life-style emblematization" or "convention-dependent-indexical iconicity" which, as he claims, is prototypical of a phenomenon he dubs "wine talk." Professional wine critics use a certain "technical vocabulary" that are "metaphorical of prestige realms of traditional English gentlemanly horticulture."[12] Thus, a certain "lingo" is created for wine that indexically entails certain notions of prestigious social classes or genres. When "yuppies" use the lingo for wine flavors created by these critics in the actual context of drinking wine, Silverstein argues that they become the "well-bred, interesting (subtle, balanced, intriguing, winning, etc.) person" that is iconic of the metaphorical "fashion of speaking" employed by people of higher social registers, demanding notoriety as a result of this high level of connoisseurship.[12] In other words, the wine drinker becomes a refined, gentlemanly critic and, in doing so, adopts a similar level of connoisseurship and social refinement. Silverstein defines this as an example of higher-order indexical "authorization" in which the indexical order of this "wine talk" exists in a "complex, interlocking set of institutionally formed macro-sociological interests."[12] A speaker of English metaphorically transfers him- or herself into the social structure of the "wine world" that is encoded by the oinoglossia of elite critics using a very particular "technical" terminology.

The use of "wine talk" or similar "fine-cheeses talk", "perfume talk", "Hegelian-dialectics talk", "particle-physics talk", "DNA-sequencing talk", "semiotics talk" etc. confers upon an individual an identity-by-visible-consumption indexical of a certain macro-sociological elite identity[12] and is, as such, an instance of higher-order indexicality.