As a “natural” semantic shift alternative (a) [human vs. non-human > human male vs. other] is much less motivated than alternative (b) [human male vs. other > human vs. non-human].
Standardized
It is more natural for a two-term gender distinction in the plural, human vs. non-human, to come about, through semantic reanalysis, from the two-term distinction human male vs. other than the other way round.
Keywords
gender, animacy, human, masculine, feminine
Domain
inflection, syntax, semantics
Type
target < source
Status
diachronic
Quality
statistical
Basis
impressionistic, based on Dravidian and reconstruction of Proto-Dravidian
The allegedly more natural reanalysis would consist in extending [human male] to [human] and restricting [non-male] to [non-human] – which is a simplification, feature-wise. The less natural reverse reanalysis would consist in restricting [human] to [human male] and extending [non-human] to [non-human, human female] – which is a complication. Also, the [human male] forms are the resolution forms for mixed groups also including human females in Dravidian, hence should alleviate the semantic reanalysis.
The allegedly more natural reanalysis would consist in extending [human male] to [human] and restricting [non-male] to [non-human] – which is a simplification, feature-wise. The less natural reverse reanalysis would consist in restricting [human] to [human male] and extending [non-human] to [non-human, human female] – which is a complication. Also, the [human male] forms are the resolution forms for mixed groups also including human females in Dravidian, hence should alleviate the semantic reanalysis.