A paraphrase of a linguistic expression E is another linguistic expression P which is synonymous or near-synonymous with E and whose function it is to elucidate the meaning of E. This presupposes that the meaning of P is known. ¹ Typically, both E and P are sentences. The methodological situation is generally such that a certain meaning component of E is at stake, while the presence of the same meaning component in P is taken for granted. For this to be possible, P is generally more explicit – in the relevant respect – than E. To give an example: E = this is a stallion, P = this is a male horse. Both P and E contain the two semantic components at stake here, ‘male’ and ‘horse’, but P is more explicit than E because each of these semantic components is represented by a separate word in P, while they are just lexical features of one word in E.

The paraphrase method may be useful in syntactic analysis if the syntactic construction to be analyzed may be paired with a standard paraphrasis. Then E and P may be considered to form a syntactic paradigm. This is often a useful approach to word-formation processes. Consider noun incorporation in Yucatec Maya in and .

Here a clause containing an incorporative verb is paraphrased by a clause containing the corresponding base verb and a syntagma containing the erstwhile incorporated item. What is of special interest here is the semantic role of the incorporated item, brought out in the paraphrase.

Similarly, the function of the applicative derivation may be elucidated by paraphrasing relevant expressions with constructions based on the base verb, as in the following example from German:

Here a clause containing an applicative verb is paraphrased by a clause containing the corresponding base verb and a nominal dependent corresponding to the direct object of the applicative verb. The method shows that in all cases, there is a participant which is coded as the direct object of the applicative verb, but as a more peripheral dependent of the base verb. It thus contributes to discovering the function of the applicative.

Sometimes E allows of alternative paraphrases each of which makes a sense of E explicit. Then in a given context E may be ambiguous, but if it is replaced by P, the ambiguity vanishes: If P is appropriate in the given context, then that is the sense of E, and otherwise its sense corresponds to some other paraphrase. Consider information structure as an example:

Paraphrase test of contrastive topic status of X:

Given a sentence E containing a constituent X, then X is a contrastive topic in E (while the rest of E expresses an assertion on X) iff E may, in its context, be properly paraphrased by ‘As for X, [assertion].’

Paraphrase test of contrastive focus status of X:

Given a sentence E containing a constituent X, then X is a contrastive focus in E (while the rest of E expresses a presupposition) iff E may, in its context, be properly paraphrased by ‘It is X that [presupposition].’

Now let E = the answer in E1 and P = the answer in E1':

The result of this paraphrase test is, thus, that John and Paul are contrastive topics in E1, while Linda and Sue are contrastive foci in E1 – something that may not have been obvious from E1 alone.

There is a more elaborate version and a more elaborate German counterpart to this test.

¹ Ultimately, this is the problem of the infinite regress in definitions.