The central aim of this dissertation is to describe the behaviour of past participle agreement (PPA) within compound tenses in Majorcan Catalan, especially in transitive constructions, in two main contexts: a) with object clitics (3rd person accusative clitics, 1st and 2nd person accusative clitics, and the partitive clitic), as well as in constructions with wh-movement (relative clauses, wh- interrogative sentences, and wh- and degree exclamative sentences) or with some other type of object fronting/preposing (focal, resumptive or stylistic); and b) when the object remains in situ, in its postverbal canonical position — a possibility which nowadays is possible just for speakers from some villages. The starting point for the description has been, mainly, the development and the computer analysis of four corpora with Majorcan data: the Rondaies mallorquines by ANTONI M. ALCOVER, the Cançoner popular de Mallorca by RAFEL GINARD, the interviews by JOAN MIRALLES with elder people from Montuïri (gathered in the books Un poble, un temps and Veus del passat), and the interviews by JAUME CORBERA with elder people from several Majorcan villages, which have been transcribed in the DVD Antologia from the Arxiu Audiovisual dels Dialectes Catalans de les Illes Balears. Furthermore, starting from a critical review of previous theoretical accounts, and always within a comparative approach, I provide a formal analysis for PPA, within the Minimalist Program of Generative Grammar. One important contribution of this thesis is the ascertainment that PPA is not a unified phenomenon, but an epiphenomenon, which can be regarded as the by-product of several syntactic mechanisms. Chapter I is dedicated to the introduction of the topic, and to present the goals and the interest of the research. I also expose the methodology used for data collection and data classification, through the compilation, tagging, and computer processing of four corpora with Majorcan data regarding auxiliary selection and PPA. Moreover, I present a preliminary statistical analysis of the data. Chapter II is devoted to PPA with clitics, wh-movement, or some other kind of object fronting/preposing. Moreover, I also study PPA with periphrases and the polemic topic of PPA with the perception verbs veure (‘see’) and sentir (‘hear’). I also critically analyse LOPORCARO’s (1998) idea according to which preposed objects would be a spurious class. Finally, I compare the requirements for PPA with wh-movement with respect to French. In chapter III, I study PPA with the object in situ (PPAOIS). I highlight that, in Majorcan Catalan, at least for those speakers who admit this kind of PPA, two requirements are needed in order to license it: regarding information structure, the object must be a given topic or an aboutness topic (this requirement applies to Friulian Rhaeto-Romance too), and, besides this, regarding event structure and aspect, the internal argument must be completely affected by a dynamic event with bounded or resultative aspect. Chapter IV concentrates on some phenomena that, although they do not properly refer to PPA, they are indirectly related to it, such as supercompound tenses, the degree of integration between the auxiliary and the participle, and the «exceptional» selection of ESSER (‘BE’) within transitive or unergative constructions.