In April 2007 Andreas Schinner made a statistical study of the document in 'The Voynich Manuscript: Evidence of the Hoax Hypothesis. Cryptologia (Wikipedia page ) Volume 31 issue 2: 95-107 (2007).'
Schinner showed that the statistical properties of the manuscript's text were more consistent with meaningless gibberish produced using a quasi-stochastic method such as the one described by Gordon Rugg, than with Latin and medieval German texts.
However, in 2013 an article by Amancio et al. published online in PlosOne  (Diego R. Amancio, Eduardo G. Altmann, Diego Rybski, Osvaldo N. Oliveira Jr, Luciano da F. Costa (July 2013). "Probing the statistical properties of unknown texts: application to the Voynich Manuscript". PlosOne. doi:10.1080/01611190601133539. Retrieved 2014-02-17 .) argued precisely the opposite, namely that the Voynich manuscript "is mostly compatible with natural languages and incompatible with random texts" (Abstract).
However the system used only applies to certain types of languages: the case against would involve the cost of the materials (parchments and ink), the smoothness of the writing, and the time needed to create the manuscript. (German Wikipedia Voynich Manuscript at  refers.)