II. PRELANGUAGE DEVELOPMENT
Sounds emitted by very young infants have recently been investigated by Ringel and Kluppel (1964), Bosma and Lind (1962), and Fisichelli et al., (1961), each using set of instruments. The difficulties and pitfalls of sound analysis of the prelanguage child have been discussed by Lane and Sheppard (1965) who have contributed a computer-based technique to this line of investigation. Although the purely naturalistic approach to the problem is clearly unsatisfactory, instrumental analysis still leaves much to be desired. A warning is particularly pertinent with regard to the uncritical use of the commercially available sound spectrograph (the “sonagraph” manufactured by Kay electric). This instrument does not perform well when the fundamental frequency of the speaker’s voice approaches 300 cps, which is the case in children under five years, particularly in the neonates. There are, nevertheless, certain generalizations that may be made safely from this kind of analysis.
近年來Ringel和Kluppel（1964）、Bosma和Lind（1962）和Fisichelli et al.（1961）各用不同的一套儀器設備來調查研究小小嬰兒發出的聲音。語言前期兒童語音分析的困難和缺失已經被Lane和Sheppard（1965）探討，他們也促成一套電腦輔助方法運用在這領域的研究。雖然對於這個問題完全自然的方法是確定不足的，但是工具格的分析仍然有很大的進步空間，關於商業用聲譜圖（由Kay電製造的聲譜圖）的不當使用有恰當的一個警告機制。當說話者聲音的基礎頻率接近300周/秒時這個儀器不能發揮良好的效用，而300周/秒的頻率正是五歲以下的小孩尤其是嬰兒的語音頻率，不過仍然有一些普遍的且可靠方法來分析兒童的語音。
Naturalistic observations as well as acoustic studies indicate that there are two distinct types of vocalization and that each has its own development history. The first type includes all sound related to crying. It is present at birth (and potentially present even before the end of normal gestation). It undergoes modifications during childhood and then persists throughout life. These sounds as well as other sounds more immediately related to vegetative functions seem to be quite divorced from the developmental history of the second type of vocalization, namely all of those sounds which eventually merge into the acoustic productions of speech.
This second type of sound emerges only after the sixth to eighth week. It begins with brief, little cooing sounds that fairly regularly follow the smiling response. It has the characteristics of a reflex that may be elicited by a specific stimulus, namely a nodding object resembling a face in the visual field of the baby (Spitz and Wolf, 1946), (Lenneberg, Rebelsky, and Nichols, 1965). Cooing sounds are obtained most easily and indiscriminately, however, between the tenth and thirteenth week; after this age, the visual and social stimuli become more and more differentiated. Soon it is necessary that the face be a familiar one in order to elicit smiling and cooing.
第二型的聲音要在六-八週後才會出現，這類型的聲音一開始是規律的出現在微笑反應後短小的咕咕聲，這個特性可能是其他特定刺激誘發出來的反應，像是一個寶寶看起來是臉的點頭物體（Spitz and Wolf, 1946）, （Lenneberg, Rebelsky, and Nichols, 1965）。咕咕聲是最容易且隨意能取得的聲音，然而再第十-十三週之後視覺和社會的刺激要越來越多變化。
The spectrograms of crying noises illustrate the absence of any articulation, apart from opening or closing the mouth. In essence, the infant simply blows his horn without operating the keys. What little modulation is present, is achieved primarily by laryngeal changes and variations in subglottal pressure. The control of voicing mechanisms is not yet well-developed in the two-week old, causing the fundamental frequency to waver irregularly over periods of 20 to 50 msec. This creates the variation in the density of the vertical stria, seen in Fig. 7.1.
The cooing sounds that begin to appear toward the end of the second month are acoustically fairly distinct from crying sounds. Their duration is characteristically about half a second, and energy is distributed over the frequency spectrum in a way that soon reminds us of vowel formants. This impression is reinforced because cooing contrasts with crying in that it shows resonance modulation almost at once in addition to fundamental frequency modulation. In other words, during cooing some articulatory organs are moving (mostly tongue), whereas during crying they tend to be held relatively still.
Although cooing sounds are “vowel-like”, we must guard against describing them in terms of specific speech sounds of English, for example. They are neither acoustically, nor motorically, nor functionally speech sounds. For instance, their acoustic onset differs from the more common vocalic onset in Germanic languages by having either no glottal stop at beginning or an overaspirated glottal stop such as never occurs in standard English. Early vocalizations seem to be different motorically from adult speech sounds, because the articulating organs move somewhat erratically and discoordinately.
By about six months the cooing sounds become more differentiated into vocalic and consonantal components. New articulatory modulations appear.
There are varying degrees of sound attenuation: they still occur somewhat randomly. This produces spectrographic patterns that show a general configuration not unlike that of connected speech; but it is important to stress once more that the analysis of these sounds makes it quite clear that they may not yet be regarded as isolated occurrences of speech sounds of a natural language. The child is articulating, but the sound patterns are very different from nonsense words that follow the pfonological rules of English. Figure 7.2 demonstrates the difference between the sound productions of the prelanguage child and adult sound production. A mother who is imitating the cooing sounds of her own child makes very different noises from her baby.
FIG. 7.2. (a) Spectrogram of a three-months-old boy cooing; (b) mother imitating her child after listening many times to a tape-loop on which the baby’s noises are recorded.
圖7.2. (a) 一個三個月大男孩的咕咕聲；(b) 母親在多次聆聽錄有嬰兒聲音的錄音帶後模仿她的小孩
Acoustic analysis of most vocalizations (except crying) throughout the first year continues to show evidence of poorly coordinated interplay among respiratory, laryngeal, and oropharyngeal mechanisms.
The articulatory signs of immaturity are, interestingly enough, not correlated with the onset of language proper. The first identifiable words occur at a time when articulation is still coarse. The attainment of good control over the motor acts necessary for fluent speech is not the milestone at which language has its fires beginnings.