These vocalisations were not related in any obvious way to the adult language spoken around the child. However, this was quickly superseded by attempts to produce adult words. In the case of Nigel, this seems to have started very abruptly during the course of just one day, when a whole host of adult-like utterances were recorded. It’s very hard to generalise about exactly when a child will start trying to produce the adult system, but a typical picture would be for the first words to appear any time between ten and fifteen months (if the child is learning more than one language, this onset may be later). Sometimes, there is a great deal of variation in the pronunciation of these early words, though on occasions the child may pronounce words very accurately. A famous case of the latter from the research literature is that of Hildegard Leopold, who was studied by her linguist father as she learnt English and German. Her first English word was pretty, pronounced more or less as in adult English.
Phonological processes in acquisition
After the child has acquired fifty or so words, a sudden change often
takes place. Children simplify their pronunciations and at the same time start acquiring a great many new words extremely quickly. Words which may have
been pronounced correctly at first suffer this simplification: Hildegard’s pretty is again an appropriate illustration. During this period, her near adult form gave way to [pɪtɪ] and then [bɪdɪ]. By the age of about four or five, however, children have mastered all but the trickiest articulations in their language (such as English [θ]).
What route children take towards this remarkably quick mastery and how they
navigate their route are interesting questions.
As already noted, it appears that children generally know more than they can say. Thus, one little boy, Amahl Smith, whose development between the ages of two and four was studied by his father, at one stage pronounced both mouth and mouse as [maʊs]. However, in perception he didn’t confuse the two words, as
indicated by the fact that he reliably identified pictures of a mouth or a mouse when asked to do so by his father. In fact, Amahl provided more subtle production evidence for this claim: at an earlier stage, he couldn’t pronounce [θ ð s z] and these came out as [t d]. Thus, he pronounced mouth and mouse as [maʊt]. At this stage, he was also learning how to pluralise nouns. Given his phonological
system, a word like cats was pronounced as [kæt] – the plural /s/ became [t] and the resulting sequence [tt] in [kætt] was simplified to [t]. However, his plural for mouse was [maʊtɪd], not [maʊt]. This is understandable if we assume that he knew that mouse really ended in /s/ and not /t/ and that words ending in /s/ normally form
98
sounds
their plural by the addition of [ɪz] (cf. bus/buses, kiss/kisses, etc.). Interestingly, this sort of example (which is far from unique) also shows that the child can’t have been just imitating plural forms: the child will not have heard a form mouses to imitate. We shall return to this mismatch between perception and production
below.
An influential theory about the way children learn articulation is based on the generative theory of phonology introduced in section 5. There we saw that
phonologists relate underlying representations (URs) to surface representations (SRs) by means of phonological rules, which are a way of referring to phonological processes. We can use this idea to account for aspects of child phonology by assuming that the child perceives and stores the adult forms of words more or less correctly (the evidence cited in the previous paragraph is consistent with this in the case of Amahl), and then imposes a set of phonological rules to simplify those pronunciations. The forms actually pronounced by the child are therefore equivalent to surface representations. This is an appealing model because a good many of the distortions introduced into children’s speech seem to be regular, and in many cases can be regarded as the consequence of phonological processes rather similar to those observed in adult languages (see below for illustration). As the child develops, the simplifying processes will be altered, to permit a greater variety of output forms, or lost altogether (so that the child’s form is the same as the adult’s).
For instance, to account for Hildegard Leopold’s form /pɪtɪ/ for pretty on this model, we can assume that she imposed a process of consonant cluster simplification onto adult forms, the effect of which is to transform the sequence /pr/ into /p/.
This process is shown schematically in (72), where C stands for any consonant: (72)
C C
→ C
The schematic picture emerging from this way of looking at things is represented in figure 29.
Other common types of phonological process for which children present
evidence are illustrated in the speech of Amahl Smith. At the age of two, he simplified almost all consonant clusters to a single consonant, e.g. stamp → [dap], drink → [gɪk], socks → [gɔk], scales → [geil], crumb → [gʌm], bring → [bɪŋ], Figure 29 Preliminary model of child phonology
Child phonology
99
spoon → [buːn]. The only clusters he produced were in words such as camera →
[gæmdə], bandage → [bændɪt], cheque-book → [gɛkbʊk], but this is easy to
understand once we recognise that the most complex type of syllable Amahl was able to pronounce was of the form consonant + vowel + consonant (CVC). If a
word (for him) had two syllables, such as bandage, then this would give rise to a cluster in the middle of the word, provided the two syllables were individually pronounceable, hence [gɛk] + [bʊk] to give [gɛk.bʊk] and so on.
Notice that in the examples we have cited here, at the beginning of a word the consonant is voiced, even if the adult word has a voiceless consonant or consonants in this position. Thus voiceless /s/ gives rise to voiced [g] in sock, voiceless /st/ becomes voiced [d] in stamp, etc. Voiceless sounds immediately followed by a vowel are very frequently voiced in early child speech, a phenomenon known as Prevocalic Voicing. It is not very common to find Prevocalic
Voicing in adult phonologies, though there is a rather similar phenomenon in a large number of languages in which a voiceless sound is voiced if it occurs
between two vowels.
Also illustrated in some of the above words is a very common process in child phonology, often know as Stopping, in which a fricative such as [f z ʃ] or an affricate [ʧ, ʤ] is simplified to the corresponding stop consonant, i.e. [p d t] or
[t d]. This kind of process could not be found in this form in adult phonologies, because it takes all fricatives and affricates and turns them into stops. Thus, if it were to occur in an adult phonology we would never know, because we would
never see any fricatives or affricates in the language in the first place. Prevocalic Voicing is a process which occurs in a specific environment or context, and such a process is called context-sensitive. Most of the phonological rules of adult phonologies (including that discussed in the previous section which determines whether a plosive is aspirated or not), are of this kind. On the other hand, Stopping is a process which occurs in all contexts or environments and therefore is called context-free.
Still another process apparent from Amahl’s forms is one in which a velar sound
[k g ŋ] at the end of a word appears to influence a coronal sound, such as [s t d ʧ], at the beginning of that word. Thus, the /d/ of drink becomes [g] in the context of the following /k/. Now, this phenomenon is rather reminiscent of phonological processes found in a variety of languages, and which are termed harmony
Читать дальше