This note expands on certain aspects of Notes on Narrative, which should preferably be read first.

The narrative below, from the book Life Stories: The Creation of Coherence, by Charlotte Linde (Oxford, 1993, page 84), is relatively simple and undramatic, but its formal structure is clear, and it nicely illustrates what such narratives look like, what their parts are, and how they fit together. The first clause is a request from the interviewer, and the rest is the response of the inverviewee.

William Labov proposed a theory of narrative, which we call the Labov theory of narrative structure, although our version includes improvements due to Linde. The structural aspects of this theory can be formalized as a grammar, the instances of which correspond to the legal structures for narratives. The following uses an "extended BNF" notation,

The sort <Narr> is for narratives, and <Cls> is for narrative clauses, which we take as potentially including evaluative material, while <Eval> is for stand-alone evaluative clauses, <Open> is for the opening section, which may include an orientation and/or abstract, and <Coda> is for the closing section. It is a good exercise to parse the example narrative given above with respect to the above grammar, i.e., to give its parse tree, labelled by the sorts of the grammar.

It should be noted that the above BNF grammar (which is equivalent to a context free grammar) is not suffient for describing all aspects of narrative, which in some cases may be context sensitive, or restricted by certain axioms, or even not formally describable. For example, the above grammar fails to address the full range of ways in which evaluation can occur: some alternatives to explicit evaluative clauses include repitition of words or phrases (which serves to emphasize them), noticeably unusual lexical choices (which may serve to emphasize, de-emphasize, or otherwise spin something), and noticeably unusual syntactic choices (which also may emphasize or de-emphasize). An important use of this theory is to extract values from narratives. This approach can be used to determine the values of individuals, groups, and organizations, and it is a bit scary how effective it can be in practice.

Notice that all the main verbs in our example are in the narrative past tense, as is often the case; an alternative that is common with younger speakers is the narrative present tense, such as "Like, I go to the mall, see, and this guy, like, he come up to me, and ..." It is important to remember that the Labov theory rests on an empirical basis of naturally occurring oral narratives of personal experience; it is not intended for written forms like novels, or for human computer dialogues, though it can still yield insights into such forms, because oral narratives of personal experience are so basic.

The narrative presupposition says that the order in which the narrative clauses is given is the same as the order in which they events that they describe occur, unless explicitly indicated otherwise; the last phrase is the reason for call this a "presupposition' - it is a default convention, not an inevitable fact. The above grammar also does not address the ways in which narratives can be embedded in other narratives, e.g., via flashbacks or flashforwards.

The above structure for narratives can also be described as a semiotic system, in the sense of An Introduction to Algebraic Semiotics, with Applications to User Interface Design. The top level sort is of course <Narr>; the second level sorts include <Cls>, <Eval>, <Open>, and <Coda>, and <Abs>, while <Ornt> are third level sorts. It is another good exercise to write out further formal details of this sign system, including priorities and constructors that are left out here. (The webnote Semiotic Morphisms gives some basics of algebraic semiotics, while An Introduction to Algebraic Semiotics, with Applications to User Interface Design provides more detail.)