Agreement is a recurrent device in language which helps to add coherence in it. More in particular, number agreement (which constitutes the main focus of this thesis) is one of the most useful tools in language, and some form of it is present in the majority of the languages of the world. The present thesis’ aim is that of analyzing which are the inner (mental) mechanisms that lead to the resolution of number agreement in language. To this purpose, some potential forces have been analyzed, syntax and semantics being the most important ones. The literature on this topic has never reached a solid conclusion as regards the power exerted by each of the previously mentioned forces on the final resolution of agreement. Syntactocentric theories, as those exposed by Fodor (1983) and Chomsky (1995, 1999, 2001) among others, considered agreement as an encapsulated “phase” inserted within a purely morphosyntactic process in which semantics is completely absent. On the contrary, for cognitive grammar, agreement is deeply based on semantics and therefore this source of information cannot be disregarded (Pollard & Sag, 1988; Barlow, 1999; Vigliocco et al., 1996; Thornton & MacDonald, 2003; Haskell & MacDonald, 2003). Therefore, this thesis tries to shed some light on this debate by presenting some experimental research (based on language errors) in which (a priori) semantic variables such as emotionality or concreteness have been manipulated in a series of experimental tests in order to ascertain to what extent were they responsible for the final resolution of agreement marks. In addition, these tests were carried out in two structurally different languages (Spanish and English). The ultimate aim of this comparison was that of observing how two different morphological systems lead to two opposing agreement systems in which different forces are responsible for the final resolution of agreement marks.