The following is much less that a complete account of phonetic representations. The discussion focusses on the role of a phonetic representation in linguistic description.
The phonetic properties of some piece of text comprise the articulatory and auditory properties of its significans. These have, by definition, physically measurable correlates. In this, the phonetic properties contrast with the phonological properties of a piece of text, which are those properties of its significans which are distinctive in the language system.
A phonetic representation of some piece of text is a representation of the phonetic properties of its significans. A major distinction applies by the criterion of the methodological status of the piece of text in question (Lehmann 2004, §3.3):
- It is a piece of primary data if it is an utterance occurred as a specific speech event with its spatio-temporal coordinates; i.e. it is an object with a historical identity.
- It is a piece of secondary data if it is a syntagma instantiating a linguistic system. Such objects lack spatio-temporal coordinates and, therefore, a historical identity.
This distinction yields two kinds of phonetic representation:
- If the object represented is a piece of primary data, its phonetic representation is a phonetic transcription. A phonetic transcription is faithful to what was actually said in the event in question and may go to any degree of detail such that a skilled phonetician could, by reading it out, imitate the recording that is represented.
- If the object represented is a piece of secondary data, its phonetic representation is a systematic phonetic representation. A systematic phonetic representation represents the phones and suprasegmental features that the phonology of the language generates for the significans in question. By reading it out, one may produce a valid utterance of the language, though it may not sound quite idiomatic.
Both types of phonetic representation use the International Phonetic Alphabet. Systematic phonetic representations are also done in other notation systems, with feature matrices among them. The following considerations are limited to the linear IPA notation. Both types of phonetic representations are used in linguistic descriptions:
- A phonetic transcription is used if a particular recording is to be represented in the description. This applies to texts for which there are raw data, i.e. an audio or visual recording. The purpose is to faithfully represent a particular piece of data.
- A systematic phonetic representation is used if the text represented is to instantiate the linguistic system. The representation only specifies the auditory form of the significans of the text according to the rules of this system. This kind of representation applies either to secondary data or to primary data for which an audio recording does not exist or is of no relevance. The purpose is to provide a representation of the significans of the text which indicates its auditory form.
A linguistic description comprises both a section that documents the object language and solitary examples in the running metalinguistic text.
- A text edition may represent a particular recorded speech event. It may thus be presented in phonetic transcription. However, it may also represent a text that originated in writing. For such a text, only a systematic phonetic representation is possible.
- For solitary examples, phonetic transcriptions may occasionally be useful in the phonetic section of the language description. In other sections of the description, a systematic phonetic representation is more appropriate, provided a phonetic representation is useful at all.
In sections of the language system above the phonological level, i.e. in the grammar, the semantics and discourse constitution, phonetics plays at best a subordinate role and normally no role. Here, phonological representations, including importantly morphophonemic representations, generally take their place. However, even phonological representations are generally not needed above the morphological level. In sections of the linguistic description devoted to syntax, semantics and discourse structure, generally orthographic representations of examples suffice. In earlier phases of the linguistic discipline, sometimes phonetic or phonological representations were required at these descriptive levels, too. However, such requirements were made in situations were there was no established writing system for the language in question.
References
Lehmann, Christian 2004, “Data in linguistics.” The Linguistic Review 21(3/4):275-310. 
].