raphy (2001) offers a helpful analysis of four autobiographical “I’s” (the histor-
ical-I of the real author, the narrating-I, the narrated-I, and the ideological-I)
but that the analysis leaves an unfortunate gap between the real author and
the narrating-I. Since Smith and Watson contend both that the real author is
ultimately unknowable and that the narrating-I may have multiple voices, I
suggest that their model should fill the gap with the implied authorial-I. This
implied authorial-I would be the agent responsible for choosing which of the
multiple voices of the narrating-I to employ at which points in the narration.
This implied authorial-I would also be able to communicate whether the nar-
rating-I was a reliable or unreliable spokesperson.
In the very impressive second edition of Reading Autobiography, Smith
and Watson respond to my suggestion by politely rejecting it. They contend
that the “narrating and narrated ‘I’s are temporally too interlinked to both be
effects of an implied author, and the project of self-narration is too involved
with its own process of reading and interpretation to sustain, or require, this
third term” (76). As long as the model of autobiographical telling empha-
sizes that both the narrating-I and the narrated-I are potentially mobile, the
concept of the implied author is unnecessary: “Phelan’s model of positing a
triangular situation (with narrating ‘I,’ narrated ‘I,’ and implied author) seems
to depend on a narrating ‘I’ fixed in one temporal plane. We would argue that
the dynamism of much autobiographical work, its ability to put the narrative
situation into play, makes such a category redundant” (76).
From my perspective, what Smith and Watson’s response demonstrates
most powerfully is something that any advocate for the concept of the implied
author ought to be aware of: among many narrative theorists, resistance to
the concept runs deep. The most telling evidence of Smith and Watson’s resis-
tance is that their response only obliquely addresses the specific argument I
make about the gap in their schema, that it leaves them with no clear way of
accounting for changes in the narrating-I. Rather than directly showing how
their model deals with that gap, Smith and Watson appeal to the complexity
of the interrelations of the narrating- and narrated-I’s and the potential dyna-
mism of both, and go from there to the conclusion that the concept of the
implied author is redundant.
But a little reflection on the example I use to extend my proposal for the
utility of the implied author, Frank McCourt’s Angela’s Ashes, calls that con-
clusion into question. In McCourt’s memoir, the narrating-I is very mobile,
as McCourt employs the vision and voice of his younger self at multiple ages
from three to nineteen.^1 Within Smith and Watson’s model, we have two
- In this connection, I find it puzzling that Smith and Watson claim that my model
depends on the narrating-I being fixed in one temporal plane, since my discussion of An g e l a’s
Ashes emphasizes the shifts in the vision and voice of McCourt’s narrating-I.
IMPLIED AUTHOR, DEFICIENT NARRATION, AND NONFICTION • 197