This investigation tests the validity of three theories of languages applied to data on lexical gender in the Medieval period of the Romance languages. By analyzing and comparing data from the Miracles of Our Lady, in combination with data from various other researchers, certain patterns of regularity and irregularity can be observed. The primary focus addresses the lexical gender assignment of nouns in the various Romance languages, in particular in the merger of the Latin neuter gender with the masculine and feminine genders. The overall changes and stasis of the history of lexical gender in the Romance languages, can be applied to various theories of language change in order to analyze and understand various phenomena. The phenomena examined in this investigation include the topics of lexical gender change and stasis, 'undecided' and ambiguous lexical gender assignment, and mass-gender elements. In addition, the theories of Lightfoot (1979, 1991), Keller (1994) and Croft (1996, 2000) are used to eludidate contemporary theoretical understanding of this topic. Lightfoot and Keller are chosen as representatives of Formalism and Functionalism, respectively, as they are often used in order to describe and discuss historical language change; the theory of Croft is a new theory which can benefit from further research. It is found that the theory of Croft (2000) best explains the research questions, but more work is recommended in order to fully understand the various phenomena of lexical gender in the history of the Romance languages, and in order to ensure the veracity of the theory, Croft's (2000) Evolutionary Theory must be tested further.