Embedded narrative means any story-representation, either explicitly represented in narrative, or implicit, produced in a character's mind, which then is represented in the reader's mind. These embedded narratives originate in either retrospective interpretations or projections of the future. Ryan bases her notion of fabula on Chatman: a fabula is made believed to exist independently of any communicative act.
Ryan describes a plot grammar in detail: a graph with three types of nodes: physical events, physical states, mental acts; six types of semantic relations: causation, enablement, reaction, motivation, fulfillment, and termination. These units are then used to construct higher level concepts. Ryan's model is what she calls an intent-driven model of narrative: The core components of the fabula are deliberate actions. The desires of the characters makes them formulate plans that determine the direction of the plot. In Ryan's view, a fabula is not a linear sequence, it is a network of embedded narratives - the possible outcomes as represented by the characters.
The tellability of a fabula can be predicted by its complexity of system of embedded narratives. A fabula with a linear structure (few branches) will not be as tellable as a fabula with many branches. Ryan exemplifies her model of embedded narratives using the story of Cinderella and Fowles' The French Lieutenant's Woman.
An actualized embedded narrative represents the direction the fabula takes (e.g., a character's plan that is successful). A virtual embedded narrative is a possible direction for the fabula, but one that it does not take (e.g., a character's plan that is unsuccessful). Actual embedded narratives often have no verbal manifestation, while virtual ones do. Ryan gives an interpretative guideline: if a plan is expressed in detail, it is likely to prove unsuccessful, and if a character gives a detailed interpretation of past action, it is likely to be wrong (except in the last chapter of a detective novel). This is a guideline for writers as well.
The reader, as well the characters, makes retrospections and projections, but not from within the narrative like the characters, but from without. This history is part of the reading history, not the history of the narrative universe. These embedded narratives of the reader and of the characters may overlap. These reader processes are the basis for narrative effects like suspense.
In her conclusion, Ryan encourages us to look not only at the direction the narrative is actually taking, but to pay attention to the alternative paths of narratives, that is, embedded narratives.