Unfortunately I have been too deeply involved with these questions for too
long to be able to attempt a simple response. My last effort ran to about
50 pages, I think.
Dwight Atkinson's discussion is certainly one helpful way in to some of
these questions, and I may respond individually to a couple of the other
comments. Hasan is a careful writer but not easy to read; Jim Martin has
not usually given as extended discussions of these issues, but they are
implicit in his practice. Hasan and Martin see more principled differences
between their views than many others, concerned mainly with the practical
issues of discourse analysis, do. Halliday has tried to avoid taking sides
on some of the more delicate points of difference, but basically has much
more to say about register than about genre.
Martin makes his case most systematically I think in his book _English
Text_, pp. 501-508, but really it's the whole chapter that is relevant.
Hasan's "Conception of Context in Text" argues against Martin's overall
position, but does not address each specific argument. Hasan has a long new
paper that also deals with these issues, not yet published, "Speaking with
reference to context". My own long entry in this sweepstakes is a paper
that rather fell between the cracks and never got published, and was
probably too long in any case; a shorter version "Typology, topology,
topography: Genre semantics" is something I can make available on my
website (see under New Additions in a few days, maybe sooner).
Herewith just a few provocations --
1. Historically the discourse traditions that gave rise to the notions of
genre and of register are pretty much separate. The notions don't really
have much of anything directly to do with one another, but they do address
the same "phenomenon" in our folk-culture view of texts and language. While
the term "genre" has a long literary critical pedigree, the modern sense as
used in Martin or Hasan or most anybody mentioned in this discussion
derives from Propp's _Morphology of the Folktale_: it is a descriptive
model for the similarities among texts like Russian folktales extended to
the similarities among texts like experimental scientific research papers.
Bakhtin's use of the term translated "speech genres" is rather different,
though again addressing what we intuitively feel is the same phenomenon.
The notion of register is really more recent, more or less invented in its
modern form by Halliday and other British functionalists, but it has
multiple senses, which adds a bit to the confusion. Both Hasan and Martin
have set out to create unified theoretical discourses in which these
initially quite separate ideas would have a systematic relationship to one
another. The EFFORT has been very fruitful; whether the results are is not
so clear.
2. Also influenced by Halliday, Terry Threadgold (Monash University,
Melbourne, English) and Gunther Kress (U of London, Education) developed a
third theory, one that was more in the literary critical tradition and more
interested in the social and cultural phenomena that texts mediate. Each
group developed a genre theory for different purposes. Hasan's view of
genre originated in an attempt to systematize the linguistic similarities
and differences among children's narratives at a larger textual SCALE than
that to which register applies. Her view of register ultimately is part of
an effort to theorize the notion of context itself; with a full theory of
context, the relationship between genre and register can be explicated.
Martin, as noted here before, largely developed his theory of genre as an
aid to the pedagogy of writing, but also as one component of a larger
theory connecting culture-writ-in-language to text types.
3. Martin is the most explicit about the relationship of genre to register,
but while most of the time this relationship is glossed in rather
simple-sounding terms "realized by", the actual basis of the model is
Hjelmslev's very subtle notion of connotative semiotics (borrowed and
simplified by Roland Barthes in his _Elements of Semiology_ and parts of
_Mythologies_). In formal terms, Martin wants to use the notion of genre to
systematize and "explain" why certain grammatical and semantic features of
texts both co-occur and sequentially follow one another in recognizable and
repeated ways. Hasan believes the explanation of this phenomenon requires a
more adequate linguistic and semiotic theory of context. Threadgold and
Kress are more interested in where text types come from and how they
function in the social order, a bit more like Foucault's view, say.
4. Having warned against oversimplifications in this discussion, I will now
try to help out by making a few targeted and ad hoc simplifications for the
purpose of this discussion context.
The phenomenon that both register theory and genre theory are trying to
describe and make sense of is our learned perception that there are types
or kinds of very similar texts. What makes them seem similar to us? in what
ways are they similar? why do these similarities matter and others seem
mere accidental? I believe that we need a general theory of
_intertextuality_ to answer these questions, as much as any theory of
context. What kinds of connections do we make from text to text? including
here construing relations of similarity? what are the most culturally
important ways in which texts are similar for us?
Register theory approaches the phenomenon of recognizable "text types" from
the semantic analysis of the clause and its constituent or functional
units. In essence it says that in a particular situational context people
tend to make certain sorts of situation-appropriate (or
activity-appropriate) meanings, and to do so in part through language, and
so by drawing on the typical grammatical resources for making just those
kinds of meanings. Thus "register" in the Hallidayan model has a
situational sense (what kinds of situations or activities are there?), a
semantic sense (what kinds of meanings get made in these situations?), and
a lexico-grammatical sense (what words and grammatical forms get used in
making these meanings in these situations?). Above all, register is a
_paradigmatic_ approach to the problem: which of each of these as opposed
to what other situations, meanings, wordings? And the later versions of the
theory allow that there are merely greater _probabilities_ for certain
meanings and wordings in a situation type (and seek to specify the actual
numerical probabilities).
Genre theory on the other hand takes a more _syntagmatic_ approach. What
are the typical parts that go to make up each text-type as a whole? how do
these sequences of specialized parts serve to describe a text-type and
distinguish it from others? what are the biggest parts that recur from one
text of the type to another? what smaller-scale parts also recur? how can
these units be defined functionally in terms of how they contribute to the
overall meaning-effect of the text as a whole?
You could say that genre theory starts from the whole text and analyzes
down to the clause or whatever smallest unit tends to recur in texts of a
given type or "genre". Register theory on the other hand starts from the
clause (or below) and asks what is similar within all clauses of the text,
or what occurs distinctively frequently in many clauses of the text, and
how this is predictable from the context of situation in which the text
functions socially (context of production, context of use).
When genre theory gets down to delicately specifying the parts of a text of
a given genre, eventually it has to say something about the micro-register
of the part: i.e. what kinds of meanings are getting made in it, and by
what kinds of typical wordings. There is a register-like theory, called
phasal analysis, that empirically determines the units of a text by looking
at how the local micro-register shifts from sentence to sentence and
paragraph to paragraph. Small or no shifts within a unit; big shifts
between units, still bigger shifts between larger-scale units. Register
theory can predict many features of a text, but NOT the systematic
sequential changes in meaning from one part of the text to the next (what I
call the semantic heterogeneity of a text). Phasal analysis can describe
these but not predict them. Genre theory can describe them in a way that at
least seems to explain them. Genre theory is a syntagmatic rhetorical
theory: it describes sequential differences in linguistic form in terms of
sequential differences in rhetorical function. Register theory is a
paradigmatic semantic theory: it describes frequencies of typical meanings
in relation to the typicality of such meanings in particular situational or
activity contexts.
All this is about register and genre theories about TEXTS. There are also
of course theories about the text-types themselves: what kinds of
similarities and differences in text-types correspond to ditto in situation
or activity types? (register approach); what systematic relationships are
there between particular individual text-types? are some more structurally
alike than others? are there families of text-types related by their roles
in larger-scale activities and social networks? (genre approach). In more
sophisticated versions, one recognizes a dialectic mutual constitution
between text and context, rather than a determinative relationship from
context or situation or activity to text.
Finally, for those of you who are getting hungry, here is the analogy I use
with my undergraduate students. We have an American tradition, esp. for
holidays, of constructing elaborate colored and flavored gelatin
compositions, called "jello molds". Of course the "mold" is actually the
copper or aluminum structure that gives an overall shape to the
composition, and it can have various subsections or chambers into which the
still liquid jello is poured before being refrigerated to solidify. There
can also be multiple layers. The mold form is the genre structure. The
flavors of jello are the linguistic registers. Every drop of language has
its own flavor/color, and register theory is the semiotics of these
flavors/colors. The fancy mold forms arrange these flavors/colors into
recognizable larger-scale compositions, and by cultural convention certain
parts of the mold structure are assigned certain colors/flavors to produce
the overall effect of the whole.
Of course some flavors may permeate the whole composition, and not all
texts are produced with neatly prescribed molds. It is not at all clear
that novels or conversations have genre structure, though all texts can be
analyzed as to register(s). If genre as such, or register as such, is a
semiotic resource in its own right, then we can hybridize genres and
registers, parody them, ventriloquate them, subvert them, and generally
play with them. Threadgold and Kress take this view and propose that there
are also other such higher order semiotics built out of language, such as
typical cultural narratives or story-types. (This is close to Barthes' use
of the notion of connotative semiotics.)
I think that's enough. JAY.
---------------------------
JAY L. LEMKE
PROFESSOR OF EDUCATION
CITY UNIVERSITY OF NEW YORK
JLLBC who-is-at CUNYVM.CUNY.EDU
<http://academic.brooklyn.cuny.edu/education/jlemke/index.htm>
---------------------------