[by the author] The notion of 'language' is not a primary concept. There do not appear to be many languages, possibly not even a single language, that has a word exclusively reserved to denote 'language.' The notion of 'language' always seems to be historically or synchronically derived from more basic senses. These earlier or basic senses metonymically relate to speech organs such as the tongue, aspects of linguistic action such as speaking and basic linguistic units such as the word. All these elements form part of a "language frame." This paper assumes that the metonymic vehicles reflect a folk model of language. Like the expert model of language, the folk model comprises different levels. It includes, in particular, the levels of (i) phonetics, focusing on articulation and speech organs, (ii) speech (i.e. parole), focusing on various aspects of linguistic action, and (iii) language (i.e. langue), focusing on selected linguistic units. These folk-linguistic levels interact in such a way that one can metonymically shift from the level of phonetics to that of speech and from the level of speech to that of language, or, by skipping the level of speech, immediately from the level of phonetics to that of language. These metonymic shifts form a motivated, unidirectional chain, i.e. the reverse direction is hardly possible. The paper investigates metonymies and metaphtonymies used for expressing notions of 'speaking' and 'language' in various, randomly chosen languages. The folk model of language which they reveal is systematic and fairly uniform cross-linguistically. It tends to associate articulators in the vocal tract with different aspects of language: for example, the throat and voice reflect sincere use of language, the tongue as the most salient speech organ tends to be associated with the production of language, and the lips as the upper articulators may distort meanings. The differences in meaning which different languages display in the metonymic shifts are due to implicatures. This study offers a new view of the function of metonymy. Traditionally, the function of metonymy has been seen in highlighting elements, i.e. metonymy is seen as being bound to a specific utterance. The function of conceptual metaphor, by contrast, has been seen as that of understanding abstract domains in terms of concrete domains, i.e. metaphor is seen in conceptual, language-independent terms. However, also metonymy appears to have as one of its functions that of systematically understanding abstract notions in terms of concrete, physical experiences. The only difference is that the concrete and abstract areas of experience systematically related in the metonymic shifts belong to the same domain of experience. (Günter Radden)