If a language is taken as a historical object, i.e. a way of communicating by language bound to the coordinates of a particular speech community (rather than as a self-contained system), its use varies along a set of dimensions that has been called the architecture of a language (Flydal 1951). The following set of four dimensions is proposed in Coseriu 1981.

Architecture of a language: dimensions of variation
dimensionexplanationexamples
diaphasicin different communicative settings, different levels of style/register are usedoral vs. written language, foreigner talk, vulgar style
diastraticin different social groups (according to age, sex, profession ...), different sociolects are usedyouth language, hunters' language
diatopicin different places and regions of the linguistic area, different dialects are spokenCockney English, Saxonian German
diachronicvariants and even historical stages follow each other on the diachronic axisextinct, obsolete, old-fashioned, current, fashionable expressions

Variants that characterize a particular generation characterize a particular social group and constitute, to that extent, diastratic variation. Only to the extent that such variants correspond to the diachrony of the language are they also diachronic variants.

The relationship between a variety and a language system is complicated. The set of properties that characterize a variety and distinguish it from the others does not exhaust a whole language system. This is clearest in the case of a professional special language, which may be characterized only by some additional vocabulary. Similarly, a dialect may share its syntax with the other dialects or with the standard while differing from them in phonological and lexical aspects.

Bibliographical references

Coseriu, Eugenio 1981, "Los conceptos de 'dialecto', 'nivel' y 'estilo de lengua' y el sentido propio de la dialectologia." Lingüística española actual 3:1-32.

Flydal, Leiv 1951, "Remarques sur certains rapports entre le style et l'état de langue." Norsk Tidsskrift for Sprogvidenskap 16:240-257.