The concept of linguistic register has been described by Trudgill (1983:101) as follows:
Linguistic varieties that are linked ... to occupations, professions or topics have been termed registers. The register of law, for example, is different from the register of medicine, which in turn is different from the language of engineering---and so on. Registers are usually characterized solely by vocabulary differences; either by the use of particular words, or by the use of words in a particular sense.Halliday, McIntosh and Strevens (1964) devote a long section to register in their 1964 work. They also refer to register as `distinguished by use.'
Registers are simply a rather special case of a particular kind of language being produced by the social situation.
I would suggest an improvement on Trudgill's definition by expanding the definition of register to include, in many cases, a preference (or even a dispreference) for particular syntactic patterns or rhetorical devices.
A close examination of many different kinds of registers shows that they tend to prefer or eschew
as well as having a preference for certain lexical devices (such as acronyms or blends) as well as certain more established lexical items and resources, such as Greco-Latin vocabulary (western European languages) or other classical languages, e.g. Sanskrit or Chinese.
When I say `registers prefer' etc. I mean, of course, that decision-makers who control the standards of the register prefer or disprefer, and may explicitly state these preferences in style-manuals for various journals, etc. Some researchers have noted that register is related to uses rather than users. Scherer and Giles (1979:51-3) devote two pages to a description of both differences in lexicon and the `complex, unusual semantic relations amongst perfectly commonplace words' found in certain registers.
Example: While traveling by air to another city recently, I overheard two people next to me discussing an issue in their discipline, which turned out to be high-energy physics. One man kept using the word 'quench' in a way I had never heard used before. In a lull in their conversation I interrupted and asked about this usage, explaining that I was a linguist. They explained that in their register, it meant rapidly decrease the temperature of a hot gas. My own understanding of this word was more like 'put out a fire; alleviate a person's thirst.'Let us tentatively propose the following definition of Register:
A set of specialized vocabulary and preferred (or dispreferred) syntactic and rhetorical devices/structures, used by specific socio-professional groups for special purposes. A register is a property or characteristic of a language, and not of an individual or a class of speakers.
Crucial for our discussion of register in the context of multilingualism and language policy is the fact that some languages lack certain registers: in western industrial societies they may lack ethno-scientific registers (folk taxonomies for classifying plants, animals or natural phenomena), or specialized poetic registers, specialized politeness systems, or registers for speaking in a trance.
This illustrates how even a numerically small and preliterate language like Toda may have three registers that are so different linguistically that they constitute separate and mutually-unintelligible codes, i.e., the existence of complex registers is not just a characteristic of post-industrial western languages.
In pre-industrial societies the languages lack legal, technical, scientific, and medical registers and subvarieties of these (for example, the register that airline pilots use to communicate with air traffic controllers). Such languages either function without such registers, which relegates them to a marginal status within a larger multilingual society (Stewart's `g' [group] function), or the members of such linguistic cultures acquire proficiency in these registers in other languages. In many postcolonial societies, of course, the registers they acquire proficiency in are registers of English or another ex-colonial language.
What this illustrates, of course, is that registers for a particular language may be di- or even tri-glossic: certain registers are in the domain of the H variety (religion, literature, ethno-history), some in the domain of the L-variety (conversation, jokes/stories, intimacy/courtship, auto-mechanical, building/construction trades etc.) and certain registers (high-tech, higher-education) may be in the domain of a totally different language.