32
EDY VENEZIANO
for /'ʒi'raf/, ‘girafe’) or, more often, nonadditional, word internal, vowels in initial
position (for example, /a'ʃe/ for /'ka'ʃe/, ‘to hide, hidden’) (for more details, see
Veneziano & Sinclair, 2000). The same analysis performed on the data of the boy G
has given the same result for the period between 1;9 and 2;1.
The premorphological interpretation of early PAEs is supported also by the good
fit existing between the proportional occurrence of the different types of PAEs pro-
duced by the child, and that of the vocalic sounds belonging to the grammatical mor-
phemes occurring in the position most adjacent to nouns, produced by the mother in
her CDS. This good correspondence doesn’t seem to have its source in the child trying
to reproduce the grammatical morphemes expected in the corresponding positions.
Indeed, considering PAEs as approximations of grammatical morphemes, and retain-
ing the set of grammatical morphemes allowed by the produced PAEs, errors in gram-
matical morpheme production turn out to be rather high (higher than 60%, see Veneziano
& Sinclair, 2000, for details). Another interpretation for the good correspondence sees
it as a result of the estimate of the proportional occurrence of the different types of
vocalic sounds expected, something very young children appear to be very good at.
5.2. The beginnings of differentiation between word categories and the emergence
of grammaticalization in the children’s verbal production
5.2.1. PAEs and grammatical morphemes
The initial differentiation between word categories, supported by the selective
production of PAEs and by the appearance of PMR variations only on verb-words, co-
occurs with other signs of grammaticalization in the children’s production.
Some of the PAEs appear as phonologically well-formed grammatical morphemes
that could occur in the positions where PAEs are produced. That is the case, for example,
of /lə'ba/, ‘the bathtub’, and of /la'paʒ/, ‘the page’. At 1;10.12, 27% of the prenominal
PAEs produced by C are of this kind, while only an average of 4.9% of the PAEs pro-
duced at earlier sessions were. At 2;2, this proportion goes up to 86% (see also Fig. 1).
For the boy, at 2;2 and 2;3, respectively 28% and 27% of the PAEs correspond to
phonologically well-formed grammatical morphemes, while at the previous sessions
the average was only 5.9%.
It should also be noted that at this time plurisyllabic targets are seldom reproduced
with a word-internal vowel in word-initial position. At 1;10.12, only 11% of the
plurisyllabic targets are reproduced in this way (vs. 55% found at the earlier session).
These words are now reproduced in a ‘CV’CV(C) pattern and may have PAEs in
initial position as well (for example, /'epapj'e/ for /'papj'e/, ‘paper’.
Concerning the grammatical function of PAEs, we find an interesting progression
for C. At 1;10.12, although errors of omission12 are still very high (69% of the
prenominal and 33% of the preverbal positions), the kinds of PAEs produced approxi-
12 An error of omission is counted when neither a PAE nor a phonologically well-formed grammatical
morpheme is found in those prenominal and preverbal positions where such an absence would not be
adequate under any possible interpretation of the child’s utterance.