In semiotics, the principle of indexicality refers to a token that refers to an object, not because it is visually similar to that object (as in iconicity) nor because it is analogous to that object (as in symbolism), but rather because the token is associated with the same general traits and connotations as its referent. A classic example of iconicity is the presence of smoke, which indicates fire. In this case, smoke is an index of fire.
Virtually any linguistic trait that is particular to a speech community can be indexical of that community. In her groundbreaking article Indexing Gender, Elinor Ochs systematized this sociolinguistic indexicality, which explains how certain aspects of speech (e.g., particular lexical items, syntactic complexity, hypercorrection and intonation) can be indices for the gender of the speaker. However, it is important to highlight Ochs’ caveat that indexicality is not static, rather it evolves across time and space. Thus, a trait that is indexical of a speech community at one time and place can be indexical of a different community at a different time and/or place.
Since Ochs’ work, it has been demonstrated that linguistic and paralinguistic features can be indices for far more than just gender. Alexandra Jaffe, for example, has done work demonstrating that a speaker’s nonstandard orthography can serve as an index for that person’s nonstandard speech. Beyond orthography, even the choice of writing system in itself can be indexical. To quote Robert King in his book Nehru and the Language Politics of India: “The Urdu script means Muslim, the Devanagari script means Hindu. The Urdu script as seen by an angry, inflamed Hindu mob summons up talismanic images from the present and the past.” Yet evidently the Urdu script does not stir up anger in everyone who reads it; the connections implicit in its iconicity depend on the culture, background and values of every particular interlocutor. To offer another example, Gothic script might indicate Germanness to some readers, Medievalness to others—and to others still, Nazism. As in speech pragmatics, the context in which a piece of writing is created (and read) is of vital importance to its meaning.
Still, it may be true that the majority of readers do not appreciate the differences between, for instance, Times and Garamond. As any typographer would assert, the use of one over the other plays a large role in the reader’s interpretation of the text, whether they notice it or not. In other words, indexicality may be subtle or even subconscious. Sometimes, though, we do confront salient paragraphemic features—such as the repeated used of italics—that we cannot help but notice. Nina Nørgaard examined a few examples of this phenomenon in novels and concluded that the use of marked typography (that is, when portions are clearly different from the majority of the text) can be indexical. For example, the inclusion of a handwritten excerpt in the middle of a novel functions as an index of the real-life existence of the character that wrote it.
Though this is a relatively understudied subject, sitting at the intersection of linguistics and information studies, it has seen renewed interest with respect to how writing systems are used in electronic communication. If you’re interested in the subject, I highly recommend checking out The Multilingual Internet, a brilliant book edited by Brenda Danet and Susan Herring.