Pattern recognition in a constantly growing field of research. Identification of pattern in images, for instance, is a first step towards their interpretation. More generally, all formal systems handling strings of symbols involve parsing phases to recognize certain patterns. Regular expressions is one of the techniques to specify simple patterns [26]. It leads to practicable algorithms available under most operating systems or edition tons especially with Unix. String-matching is a particular case of pattern recognition. It consists in locating a word inside another word, called the text. Solutions to this problem can be divided into two families. In the first one the text is considered as fixed while the word is variable. This situation occurs when the text is a dictionary, for example. The basic solution of that sort is due to Weiner who introduced the notion of position trees [29]. It is a kind of index which as been improved in different ways (see [21], [5], [10]). For the second family of solutions to string-matching, it is the word that is fixed. The two most famous and efficient string-matching algorithms of this family have been designed by Knuth, Morris & Pratt [t8] and Boyer & Moore [7]. They have been subject to several studies, improvements or extensions (see [1], [11], [13-16], [22], [23], [25], [28]). A variation to the initial problem happens when approximate patterns are considered (see [20], [27]). Stringmatching is close to detection of repetitions in strings (see [3], [10], [17], [25]). In fact, the study of regularities in strings is a part of the analysis of string-matching algorithms. In this paper, two string-matching algorithms belonging to the second family are presented. They respectively obey to time and space constraints. Both algorithms start by a first phase during which the word alone is processed. Then, the search is done during a second phase which essentially supports the contraints.