This paper proposes a review of the literature on the role of lexical frequency on the phonetic production of language acquisition data and spontaneous speach data. Language acquisition is vastly influenced by the frequencies of linguistic items learners experience in their contact with a certain language. This phenomenon is called ‘frequency effects’, and refers to the computation of the times the learner is exposed to language data in the input, and the effects such exposure has on his/her linguistic output (Gries 2008). Thus, language learning is argued to be an implicit phenomenon that is based on statistical learning triggered by frequency distributions of structural regularities within a language, as a consequence of the strength of associations between representations (Ellis 2006a). In phonetic-phonological learning, one may argue that regularities from the input are learned under the same principle of frequency effect, that being, the more a speech sound occurs in the input a learner is exposed to, the more this sound will be perceived and produced by the learner. In spontanous speech, however, it is noticed another type of influence of frequency distributions in the léxicon, which is the more frequent a word is the less accurate is its production (Ernestus, 2008). For instance, the word natuurlijk in Dutch is a highly frequent word in the lexicon, so it is very often produced by native speakers as “tuurlijk” or “tuur”, even when speakers have no conscience of such reductions. It is interesting to observe that the portion of the word which is preserved in production in such reductions is the prominent syllable of the word, presuming that the speaker has the listener “in mind” when he/she speaks. This phenomenon and a series of others will be approached in this review and one will try to account for what has been the answer to the question on the effect of lexical frequency on phonetic production by learners and native speakers of the languages.