This research aims to shed more light on the hypothesis of 'stress deafness' in French speakers in L2 Spanish and to determine the extent to which French speakers are 'stress deaf' to L2 Spanish stress violations in a listening comprehension task. Furthermore, this work aims to observe whether this stress deafness is 'persistent' or not in immersion learners (advanced level of Spanish) compared to a group of intermediate level French speakers in Spanish by means of an accentual or segmental error detection task. The second objective of this work is to see the impact of this accentual "deafness" on the production of lexical and morphological accentual pairs in Spanish by asking participants to pronounce correct sentences with lexical or morphological stress. Finally, the last objective of this work is to observe whether there is a correlation between the perceptual and productive skills of French speakers in L2. Our hypotheses predict that in perception, the advanced group will show a higher sensitivity to stress and segmental errors than the intermediate group. Our hypotheses also predict that in production, the advanced group will produce the proposed stimuli in a way that is close to native speakers and will perform better than the intermediate group. Finally, regarding both comprehension and production tasks, our hypothesis predicts the presence of a correlation between good perception and good production of the Spanish lexical accent by French speakers. The results of the analyses show, in accordance with our initial hypotheses, that in perception, the group of advanced French speakers shows sensitivity to errors close to that of native speakers and better than the intermediate group. Moreover, in production, the group of advanced French speakers performs better than the intermediate group regarding the different stress patterns proposed. Finally, in the comparison based on the regression slopes between the sensitivity of the participants in perception and the average of the judges who evaluated the participants' production, we observe the presence of a correlation between good perception and good production of morphological and lexical items. The results obtained are interpreted in line with existing psycholinguistic models.