ABSTRACT This paper seeks to explore the nature of certain types of interaction between the source and the target language in the process of translation basing on the notion of language interference in the sense of any violation of the target language form or norm under the influence of the source language form or norm. It is suggested that apart from 'overt' manifestations of interference, that can be easily traced back to the source language, there also exists another type of interference, or 'cryptic' interference. Its mechanism consists in switching between the source and the target language, which affects the mental processing and results in producing instances of interference which can be traced back to the source language, however, not to the source language phrases, formulations, and lexical items found in the actual source language text; these phenomena are due to the process of re-analysis (or reformulation) of the source language message occurring before the actual translation is performed. 1. Introduction The aim of this paper is to present the findings of an empirical study of the process of sight translation, based on an experiment during which eight translators with varying professional experience were asked to perform the same translational task. The goal of the experiment was to inspect the issues connected with language interference during interpreting. (1) Naturally, language interference can be a significantly variable factor and it is reasonable to assume that both its scope and amount change relative to interpreting conditions, e.g. whether a translation task is performed from or into one's native language. The case in point here was to provide some insights into the mechanism of Foreign Language (FL) interference and the influence it exerts over the Native Language (NL) interpreting performance, as well as into the interpreters' shortcomings in their mother tongue. After inspection of the results of the experiment it turned out that a number of translation errors (2) could not be ascribed to language interference as it is understood in foreign language teaching. (3) The subjects had certain problems also with the use of their native language, a phenomenon of large significance, since all of the experiment participants had Polish as their mother tongue. Most importantly, however, the experiment revealed that regardless of their professional experience, all translators produced instances of interference of a particular character, given the working name cryptic interference. This paper is devoted to the description of its mechanism and the implications it may have for the mental aspects of the process of translation. 2. The experiment The subjects of the experiment were two groups of four. The first group featured active interpreters with professional experience varying from 18 to 3 years in the trade. The second group featured individuals with little (or no) professional experience in interpreting, but with two years' training in interpreting. It was hoped that inclusion of interpreters with varying professional experience would broaden the research spectrum and thus help render more comprehensive results. The experiment was made up of two sight-translation exercises: two attempts at interpreting one text, the first one after a careful perusal of the Second Language (SL) text, the second one after studying a text used as a prompt for the second performance--a model translation of the original SL text. For this purpose an experienced professional translator (c. 20 years in the trade) had been asked to produce a model translation. The text used in the experiment was carefully selected as it had to meet a number of requirements. It was to contain a selection of idiomatic phrases, stylistically specific for English and thus assumed to be difficult to render in a foreign language. The text selected was an authentic text, a news report from The Guardian Weekly of 25 April 1999. …