What are the differences between theoretical perspectives of the uses of the term "register"?

Question

I'd be interested in asking people about their understanding of the term register and what this signifies for them. This would be a discussion about a specialised term and I'm sure there are multiple interpretations out there, depending on your education.

I think that there are multiple definitions of register. Simplified definitions focus on "informal" and "formal" registers, however, these fail to formally identify what linguistic features realise the register variables. As an ESL teacher, I find many textbooks over-simplify this issue, resulting in students learning lists of "informal" and "formal" words e.g. get = receive, buy = purchase, and = furthermore etc. Although there may be some contexts where this is possible, this is simply not the case. Because of this, I'm trying to work out a better way to explain this and engage with people through shared understanding.

As a student of Halliday's systemic functional linguistics, the term "register" is further technicalised to include notions of what's going on (the field of discourse), who's involved (the tenor) and how these are textually linked together (the mode). However, even within SFL, there is still much discussion about what this actually means and there are different camps with different definitions. One camp believes there are registers for domains of use e.g instructional, regulatory, representational, etc. Another camp believes that register relates to the context of situation e.g. medical, legal, educational etc. The reason I bring this is up is that just last night I was discussing differences between Halliday and Martin's understanding of register (see http://functionallinguistics.springeropen.com/articles/10.1186/2196-419X-1-3)

Halliday sees register as a stage on the cline of instantiation, i.e. between the potential system and a given instance of text or language, similar to Saussure's notions of langue and parole. This means that the entire of the potential grammatical choices in a system will not occur in any particular instances of language in a given context. As such, the potential meaning (semantics) in any text will be mediated through register.

I have been schooled in Martin's view of register and understand it to be a complex interplay between the field, tenor and mode, each of which can be related to semantic systems, which in turn can be related to lexico-grammatical systems. E.g. Field can be related to whether the meanings are more common or specialised/technicalised (e.g. dog - canine), tenor can be related to whether the relationship between interactants is more close or distant (e.g friends and strangers) and whether the power relations are more equal or unequal (e.g. student - teacher), and mode can be related to whether the message is more spoken or written (remembering that written language can be spoken!). From this I understand "formal" to be a combination of more specialised/technical vocabulary (field), greater distance and power differential between interactants (tenor), and tends to be more written language (mode). "Informal" can be seen as a combination of more common/everyday vocabulary, closer distance and more equal power relations, and tends be more spoken. As such, jargon may be classified as a form of specialised language, which may in turn suggest a particular register, but the relations between the language and the context must be made explicit.

I understand this is a lot of linguistic theory but strongly believe this is relevant to the discussions about meaning and language that take place here.

Answer 1

I answered this in the other place this question was put, here, by referring to the work on "variationist" models, which described a dialect as a collection of speech styles called "lects". I have another answer, which I might as well put here, for variety.

In generative phonology, variation in pronunciation has usually been described by making certain phonological rules optional, meaning that as the system of phonological rules apply in order, one by one, to produce a pronunciation from an underlying form, a rule designated as optional needn't apply. Then for each applicable phonological rule which is optional, there are potentially two surface pronunciations produced, one in whose derivation the rule applies and one in whose derivation it does not apply (even though applicable).

This theory of variation in pronunciation was attacked as inadequate by Labov and C.-J. Bailey (sorry I have no references at hand), but I'm not concerned with those criticisms here. Instead, I have a suggestion to make about using the general model of a dialect as a set of lects to repair a conceptual defect in the account of variation given in generative phonology.

The problem with optional rules in generative phonology is that optionality is bound up with the assumption of linear ordering of phonological rules, proposed by Morris Halle, attacked by many (e.g., Stephen Anderson), and defended by no one other than Halle himself. I will assume here that the linear order of phonological rules is an incorrect assumption.

Now, without the assumption of linear order of phonological rules, here is the problem with rule optionality. With linear order, one can imagine constructing a phonological derivation in a kind of pseudo-time, where each optional rule has a time to apply, and then one has a choice whether to apply it or not. But without the linear order assumption, when is a rule's "time to apply"? The notion of optionality no longer has any clear sense.

The set-of-lects model of a dialect comes to the rescue. It provides a reasonable way of interpreting the notion of optionality that is independent of what assumptions are made about rule order.

Let us say that a phonological lect is a variety of speech with no variation induced by application of phonological rules, and a variable rule is a phonological rule which applies (obligatorily) in the phonological systems of some lects, but not others. A phonological dialect is a set of lects differing only by what variable rules apply in each lect.

In the special case in which a variable phonological rule applies in a lect independently of the application of other variable rules, we'll characterize the rule as "optional". It is an empirical issue as to whether all variable rules are optional, as conventional generative phonology assumes.

The case where all variable rules are optional is interesting, because this part of generative phonology works well, descriptively (in my opinion, of course), and the set-of-lects model makes available a formally simple way to characterize it. When the applicability of a variable rule in a dialect is independent of whether any other variable rule is applicable in a dialect, the phonologies of the lects correspond to the power set of the variable rules.

For the case where optional phonological rules simplify articulation, we can relate any two lects in complexity of articulation according to whether the set of variable rules of one is included in the set of variable rules of the other, and the dialect as a whole is partially ordered by this relation of complexity. That is to say, a dialect is a poset.

Although I have used variable phonological rules which simplify articulation to describe this approach to optionality, the account can easily be generalized to any variable language trait. Instead of the presence of a phonological rule in a lect, just substitute a word choice, a particular syntactic construction, or any other language trait of interest for phonological rule in the preceding.