experimental conditions (e.g. Mills, 1978) (see also Szagun, 2004, for reports
of similar errors with der). With respect to gender, children are said to acquire
the rule system at around age three for masculine and feminine, whereas the
neuter category still causes problems at this age (Mills, 1985).
In terms of systematic learning of the system, older children (seven- to
nine-year-olds) seem to make use of phonological markers (e.g. based on word
endings in combination with syllable structure), suggesting productive com-
mand of the regularity of phonological features within the system (see, e.g.
MacWhinney, 1978). Similar findings are reported in Szagun et al. (2007)
from naturalistic data samples of children between 1;4 and 3;8 years of age.
Welsh
Our own studies of children’s acquisition of Welsh gender (e.g.
Gathercole & Thomas, 2005; Gathercole et al., 2001; Thomas, 2001; Thomas
& Gathercole, 2007) have tended to be experimental cross-sectional studies
looking at children who are developing bilingually with English. This type
of research is valuable in determining the extent to which children learning
specific languages rely on certain types of cues (semantic and/or syntactic)
to determine the gender of nouns. However, unlike with the German stud-
ies, there are no studies of children’s early emergence of gender in Welsh.
The studies that we do have, looking at older children’s competence, pro-
vide the following information regarding patterns of development among
Welsh-speaking children:
First, unlike the German data, gender is not learned until after age 9
years; second, children mark nouns for humans more often than nouns for
animals and inanimates; third, performance increases with increasing levels
of exposure to Welsh in the home (and to a lesser extent at school), suggest-
ing a crucial role for input when learning gender alongside English; fourth,
children often use the masculine as a default; and, finally, the system is
learned in a piecemeal, item-by-item fashion.
Summary: Gender
What is emerging out of the data from children acquiring gender in
German and in Welsh is that German children seem to acquire the system
relatively easily whereas children learning Welsh do not. What is not known
is whether learning two such systems simultaneously alters the course of
development relative to (near-) monolinguals. (All children learning Welsh
will hear English in the community and on the media. Even children who
are spoken to in Welsh at home are also exposed to English in this way.) Does
learning Welsh gender alongside a more transparent system bolster acquisition
of Welsh relative to (near-)monolinguals? Or does learning German gender
alongside a highly opaque system delay development in German, relative to
46 Part 2: Bilingual Language Development