Beyond the involvement of mutation, there are additional features that
make the gender system even less transparent (see Gathercole et al., 2001;
Thomas & Gathercole, 2007): there is no one-to-one correspondence between
form and function (i.e. SM does not ‘mean’ any type of gender; other trigger-
ing environments – prepositions, degree markers, continuous aspect, etc. –
also trigger SM); there are contradictory triggers (e.g. feminine nouns trigger
SM in local constructs, whereas it is the masculine possessive adjective ei
‘his/its’ that triggers SM in distant constructs); there is a great deal of vari-
ability in adherence to gender marking and gender agreement in adult speech;
and there are cases in which there is no overt marker of gender – namely
words with non-mutatable word-initial consonants or vowels. Together,
these render the Welsh gender system a highly complex structure to master.
Given the relative complexity of the two systems, what do we know about
children’s attempts at acquiring these systems? The next section outlines what
is known about children’s acquisition of gender in German and in Welsh.
Children’s Acquisition of Gender in German
and in Welsh
German
Much of the acquisition data on gender in German come from longitudi-
nal case-studies involving a combination of bilingual pairings, including for
example German-English bilinguals (Mills, 1986), German-French bilinguals
(Müller, 1990; Meisel, 1994), German-Spanish bilinguals (Kuchenbrandt,
2008), and German-Italian bilinguals (Cantone, 1999; Müller et al., 2002), as
well as studies looking specifically at German monolinguals (e.g. Clahsen,
1984; Szagun, 2004). This type of research not only provides invaluable
cross-linguistic information about the rate and timing of acquisition, but also
allows for analysis of errors in production that can reveal interesting facts
about the acquisition process. Together, these studies provide the following
information regarding patterns of development among German-speaking
children:
Adjectives are used before determiners, and are largely marked correctly
from the outset (Mills, 1985). Full determiner forms can also appear early
(sometimes reported as early as 1;5 – Szagun et al., 2007 – but usually between
2;0 and 2;10 years – Mills, 1985). Data from early diary studies reveal that the
early use of the definite article appears largely error-free (see e.g. Mills, 1985,
for a review) relative to the errors found on indefinite forms, although chil-
dren do also become proficient with indefinite forms by age 3. Yet other stud-
ies have found no differences in errors across the determiner types. In terms
of the errors produced, many studies have observed that children overgener-
alise the use of die in naturalistic productions at these ages, as well as under
Cross-linguistic Influence and Patterns of Acquisition 45