Terho Itkonen
1. T h e a u t h o r deals with some of t h e developments in the sound system of his son (M.) in 1967—70, between the ages 1 ;6 a n d a b o u t 4 years. T h e material consists of written notes a n d transcribed tape recordings. development in general. As an example, the a u t h o r cites a stage (at 1 ;6) at which M . ' s established vocabulary con-sisted of 5 words. T h e vowel system of -nunciation of 'cow' is amma: does this b e a r witness to an opposition of two low tends to coincide with a; no u is recorded in t h e 2nd syllable, the only labial vowel being the o of auto, which tends to merge with a just as i does. Neither u n o r i occurs as the sonant of the 1st syllable;
instead of u, o appears in t h e p r o p e r n a m e
A glimpse of the vowel system of M . at 2;0 confirms these doubts. T h e r e seems to be a low vowel p h o n e m e with back a n d front variants (partly in comple-m e n t a r y distribution, p a r t l y in free antici-pation of a sound acquisition realized m u c h later). At 2 ; 0 , u a n d i a r e well-established phonemes in n u m e r o u s words;
in light of this, t h e u a n d i at 1 ;6 were rather anticipations of this later stage t h a n sounds with a clear p h o n e m i c status.
T h e position of o is most interesting. I t occurred d u r i n g the whole latter half of the 2nd year in auto (but in practically n o other words), a n d did so a t 2 ; 0 . This was a correct imitation of the a d u l t word a n d , taken together with t h e correct pronunciation of aiti at 1 ;6, it shows t h a t such »progressive idioms» are possible long before their elements have achieved a firm foothold in the phonemic system.
(Cf. Charles A. Ferguson a n d Carol B. Farwell, L g . 5 1 , 1975, p . 432.)
3. According to J a k o b s o n , »A diffe-rentiation of r o u n d e d vowels according to degree of aperture cannot arise in child language as long as t h e same opposition is lacking for u n r o u n d e d vo-wels. T h e pair u ~ o cannot, therefore, precede t h e pair { ~ e, a n d there are no children who have a n o-phoneme without having acquired a n *-phoneme»
(Child L a n g u a g e , Aphasia a n d P h o n o -logical Universals, p . 56). T h e acquisition of o a n d e in M . was a r e m a r k a b l y long process which does not seem to be fully compatible with this position. For a long time, M . was very selective in adopting adult words with o a n d e: u p to 2;1 h e used practically no e-words (and no correct pronunciation of e h a d been recorded) a n d , besides t h e frequent word auto, other o-words occurred only occasionally (usually with a correct p r o n u n -ciation). At 2;2 h e acquires several a d u l t e-words b u t almost always substitutes i for e. At the same time m a n y new a d u l t
304
T E R H O I T K O N E N
o-words occur, a n d now b o t h t h e old o-words (especially auto) a n d the new ones a r e p r o n o u n c e d either correctly (in a slight majority of instances) or with a instead of o. At 2;3 the acquisition of e-words continues; d u r i n g the first half of the m o n t h he continuously substitutes i for e, b u t in the latter half b o t h inter-mediate stages between e a n d i a n d a p u r e e emerge alongside i. T h e e-index (an average of all instances given the values e = 1, t = 0 a n d t h e intermediate stages = 0.5) is 0.08 for the first half a n d 0.44 for the second half of the m o n t h . At the same time, the percentage of o-like or p u r e o-pronunciation in o-words grows, t h e o-indexes being 0.68 a n d 0.89 of e only later. T h e complete acquisition of t h e two sounds, however, was contem-poraneous. T h e role of early extra-systemic pronunciations, w h i c h was not considered in the J a k o b s o n i a n model, to distinguish perceptually between i, u a n d e, o. At this stage, e a n d o were target phonemes which h e could not yet pronounce (with the exception of t h e progressive idiom auto a n d some other sporadic o-instances). Before t h e p h o n e m e b a r opened, he was compelled to adopt words also from t h e lexical stock behind the b a r , a n d so i temporarily replaced e.
F o r the sake of symmetry, one would expect a simultaneous substitution of u for o. T h i s occurred to a limited extent only. T h e fact t h a t the c o u p - d ' e t a t did not m e e t w i t h total success seems to be d u e in particular to the early pioneer of o, auto (and the other sporadic o-words):
they p a r t l y resisted substitution a n d began to attract new cases of o. At this point symmetry was instantiated in a n o t h e r a n d final way, t h r o u g h t h e g r a d u a l emergence of e as t h e front equivalent of o.
This development supports t h e ob-servation of Ferguson a n d Farwell (Lg.
51, p p . 433—4), according to which t h e avoidance of words with u n p r o n o u n -ceable sounds is c o m m o n in early speech development, a n d partially confirms their
thesis t h a t new p r o n u n c i a t i o n appears word by word. Note, however, t h a t auto was subjected to vacillation between o a n d u before the final victory of o. Since then, no pioneers of o (or e) can be discerned a n d t h e lexicon as a whole vacillates until i a n d u yield to e a n d o.
4. T h e presumed lexical diffusion of sound acquisition in children agrees, according to M a t t h e w Y. C h e n a n d William S.-Y. W a n g (Lg. 5 1 , p p . 263—4), with the view t h a t historical sound changes are g r a d u a l in the same sense.
T h e present writer would see the equiva-lent of sound acquisition in the adoption of phonological innovations, e.g., t h r o u g h t h e collision of two dialects (where the speakers h a v e a target pronunciation b e -fore their eyes) rather t h a n in sound change p r o p e r (where no target p r o -nunciation is present). A well-known example is the Finnish vocalization of syllable-final stops: this was a regular change in Western dialects, b u t in Eastern dialects, where the vocalized forms are borrowings from W . , their distribution varies from word to word. I n fact, it spread of stress alternation of the type abstract — abstract in English is due to proportion analogy rather t h a n to continuous sound c h a n g e and, in t h e u n -conditioned tonal split of t h e Chaozhou dialect in C h i n a , only the mysterious result of the split (in which about a half of t h e relevant items h a d changed tone) is to be seen while nothing is k n o w n a b o u t the assumed lexical graduality of this development. or Sudden Sound Changes ?», op. cit.
p p . 434—8) t h e present writer has given evidence for this view. If phonological acquisition in children^ c a n b e demon-strated to be g r a d u a l in the same sense, it would offer equally strong support for this position as does the phonetically g r a d u a l realization of a new p r o n u n c i a -tion a d o p t e d t h r o u g h interdialectal
con-T E R H O I con-T K O N E N
tacts. Notice t h a t in b o t h cases a certain target pronunciation is present: a priori one would expect this to prevent rather t h a n promote its adoption by way of intermediate phonetic stages.
5. T h e intermediate pronunciations observed in M . between i a n d e, u a n d o seem to indicate t h a t the acquisition of e a n d o was phonetically g r a d u a l rather t h a n a b r u p t . His acquisition of the inter-vocalic alveolar trill r gives more detailed evidence of a similar development. (Cf.
the u p p e r d i a g r a m on p . 294.)
M . avoided r-words r a t h e r consistently until 2;6. Between 2;6 a n d 2;9 he a d o p t e d m a n y such words, substituting zero for intervocalic r. At 2;9 a barely perceptable consonantal segment, a weak-ly articulated voiced alveolar fricative, appears instead of hiatus. This segment strengthens a n d the percentage of seg-mental pronunciation increases gradually until at 2 ; 11 the zero instances are in the minority, a n d the spectrum of the seg-mental pronunciation begins to extend from the alveolar fricative d to liquids ranging between a non-trilled or de-ficiently trilled r ( = d in F U tran-scription) and /. T h e borderline be-tween 3 and S, viz. d and / is not sharp, and there are also intermediate stages between I and d which consist of a lateral articulation with a M i k e co-articulation. I n the d i a g r a m on p . 294 such instances have been treated as »<5».
At 3;0 the zero instances disappear a n d t h e percentage of / diminishes re-m a r k a b l y ; a pronunciation which t h e a u t h o r has identified as a fully trilled r occurs sporadically at the same time, and after the disappearance of d at 3;2 the development manifests itself through a g r a d u a l increase of r at t h e cost of d (the distinction between these two is, in borderline cases, often as a r b i t r a r y as t h a t between <5 a n d d). F r o m 3;4 on the percentage of fully trilled r is constantly m o r e t h a n 9 0 ; trilled r can now be said to be in M.'s control. (From 3;1 on zero reappears, but its average percentage between 3;1 and 3;5 is less t h a n 5.)
I n t h e beginning t h e acquisition of r was influenced by a tension between this sound a n d t h e closely related sound, /. Although the first substitute for inter-vocalic I was likewise zero, a correctly p r o n o u n c e d I established its position at be-tween the two liquids could not b e fully avoided at a stage where non-trilled d as a reflex of r still caused difficulties. No dramatics of a b r u p t sound acquisition occurs in this history. T h e spectrum of pronunciations reflecting r was b r o a d until t h e final battle between fully a n d not fully trilled r was won. At every developmental stage the spectrum (or at least the greatest p a r t of it) constituted a phonetic c o n t i n u u m , a n d the acquisition of r manifested itself only as a step-by-step shift of the average p r o n u n c i a t i o n and, in the final stages, as a step-by-step decrease of dispersion within the spec-t r u m . This is a clear example of g r a d u a l sound acquisition t h r o u g h a phonetic c o n t i n u u m . I t is also a n example of solidarity in the advance of t h e lexical front as a w h o l e : no single word c a n be discerned in which r (or a n antecedent of r) would have a p p e a r e d earlier t h a n in the r e m a i n d e r of t h e vocabulary.
6. T h e existence of phonetically g r a d u a l acquisition of sounds does not exclude the possibility of a n a b r u p t acquisition or of various combinations of g r a d u a l a n d a b r u p t development. Both g r a d u a l a n d a b r u p t a d v a n c e m e n t occur in t h e p a t h along which M . learned his last conso-n a conso-n t , the voiced alveolar stop d. ( I t should be noted t h a t Finnish children often very acquire d very late. This seems to be d u e to the marginal position of this consonant in the S t a n d a r d Finnish p h o n e -mic system: it is the only voiced stop, with b a n d g occurring in foreign-learned words only; its frequency is very small (textually about 0.78 per cent of all p h o n e m e s ) ; a n d in most instances it is during the next months, however, is zero.
This corresponds exactly to the zero outcome of r at t h e same time, a n d its further development is also, for a long time, similar to t h a t of r: thus <5, / (pure or with 5-like coarticulation), 6 a n d r a p p e a r at the same time a n d with roughly same percentages as in the r p a t t e r n . T h e outcome r, however, never becomes as d o m i n a n t as in t h e p a t t e r n of r. This is chiefly d u e to t h e fact that
306
T E R H O I T K O N E N
t h e zero outcome, which seemed to have disappeared at 3 ; 0 — 3 ; 1 , begins to in-crease remarkably at 3 ; 2 ; u p to 3;5 its percentages r a n g e from 20 to 30. At 3;6—3;7 this zero wave has abated while in its place ^-instances increase to some extent. T h e correct d p r o n u n -ciation has been recorded intermittently as early as 3;2. At 3;6—3;8 it reoccurs sporadically, until the final acquisition at 3;8 takes place drastically in a few days (of which event M . is well aware).
After t h a t point t h e percentage of cor-rectly p r o n o u n c e d d remains consistently between 80 a n d 9 0 ; during a period of several weeks a r o u n d 4;0 it reaches t h e 100 p e r cent level (but only temporarily, for during the following months, as during preceding ones, trilled or non-trilled r a n d zero still occur sporadically as well).
(Cf. the lower d i a g r a m on p . 294.) T h e early shift from t t o zero in t h e acquisition history of d would seem enigmatic, if the succeeding stages h a d not b e e n recorded. I n fact, t h e spectrum to adopt t h e corresponding voiced stop (note t h a t the two other voiced stops b a n d g were lacking), a n d so the p a t t e r n of r was substituted for d. This, together with similar evidence from other Finnish children, seems to show t h a t the feature a n d its first indications are chiefly in-direct. T h e most i m p o r t a n t is the re-m a r k a b l e increase of zero outcore-mes at 3 ; 2 — 3 ; 5 . I t seems to b e d u e chiefly to analogy: in most zero instances the vo-calic element preceding d is long, a n d nijen = niden 'their'). (The zero reflexes of these presumable model words are not included in t h e above-mentioned per-centages of zero forms nor in the d i a g r a m on p . 294.) It should be emphasized that this situation differed clearly from the contemporaneous r p a t t e r n , where t h e zero outcome was very rare, thus showing
t h a t the actual target of M . in t h e d-words was at t h a t time d a n d not r.
T h e final acquisition of d was a b r u p t in three respects. Firstly, t h e frequency of r for d drops suddenly to a minority when d emerges. Secondly, d appears after r without intermediate phonetic stages (if t h e slight increase of d just before the emergence of d is not regarded as such a bridge between r a n d d). T h e h y p o t h e t i c a l phonetic continuum from r (via d a n d <5?) to d is perhaps not as n a t u r a l as t h e c o n t i n u u m from zero to r;
one difficulty is that consistent progress along this route would have d e m a n d e d a retreat from r to developmental stages which h a d been overcome with great difficulty during previous m o n t h s . T h i r d -ly, d appears simultaneously in all relevant items. T a k e n in conjunction with t h e acquisition history of r, this indicates t h a t a word-by-word acquisition of sounds cannot be a universal p h e n o -m e n o n .
7. I n 2 a n d 3 above, the correctly p r o n o u n c e d extrasystematic anticipations of o w e r e discussed. A similar anticipation of d was mentioned in 6. A n opposite case was t h e g r a d u a l disappearance of r for d at a stage w h e r e d was already in M.'s control. At first glance, such sur-vivals of former developmental stages d o not seem particularly p r o b l e m a t i c : they agree with general learning theories, according to which the frequency of in-correct performances decreases gradually after t h e first instance of correct per-formance. T h e explanation of such sur-vivals is, however, not always so straight-forward. This is illustrated by a special case not discussed a b o v e : t h e t r e a t m e n t this substitution continued in t h e l-clusters for a b o u t five months after M . h a d acquired intervocalic /. T h e develop-m e n t a l unity of/ a n d r develop-must be responsible for this lag. As mentioned, intervocalic / was opposed from the beginning (2;6)
to a zero reflecting the target p h o n e m e r.
This zero could not easily b e generalized
T E R H O I T K O N E N
into preconsonantal position, where it would have broken the prosodic p a t t e r n :
*koke 'high' instead of korke or korjke.
This difficulty probably g u a r d e d the nasal not only in r-clusters b u t also in /-clusters, which h a d hitherto been on a p a r with t h e former. / did not p u s h t h e nasal aside until a segment similar to r (i.e. <5 or S) h a d replaced the nasal in words with a r-cluster some days before 3;0.
T h e removal of nasals from /-clusters was established in a short time, a n d their removal from (--clusters too was total for several weeks, until a regression b e g a n a n d the nasals r e a p p e a r e d alongside the trill at 3 ; 1 . This regression persisted for ten months, with the percentage of nasal forms varying d u r i n g the first, second, third, fourth a n d seventh months from 12 to 32 per cent (for the remainder less t h a n 10 per cent was noted). T h e beginning (although not the end) of this regression coincides roughly with the weak new wave of t h e zero outcome in inter-vocalic r-items a n d the m u c h stronger new wave of zero in (/-items. T h e same explanation, however, cannot account for all three instances. Perhaps a slackening of control some weeks after the acquisition of trills or trill-like fricatives contributed to the regression in all three groups, b u t the a b u n d a n c e of zero in (/-items must be due chiefly to analogy, as explained above. Analogy alone, however, cannot explain either the relatively few instances of intervocalic zero or the n u m e r o u s instances of preconsonantal nasals in
r-words; another factor is involved here. As soon as M . noticed t h a t he h a d passed t h e earlier stage, h e dared to m a k e use of it consciously in his b a b b l i n g register. Con-comitant p h e n o m e n a , e.g. changes of intonation a n d stress in t h e u t t e r a n c e a n d irregular or regressive changes of p r o -nunciation in neighbouring words, w e r e often noted in connection with this kind of regression. W h e n zero stands for d, similar p h e n o m e n a do n o t occur, a n d this seems to indicate that the »survival» of a former p r o n u n c i a t i o n does n o t have the same stylistic function as in r-words.
O n the other h a n d , the possibility exists t h a t not all zero a n d nasal instances in r-words a r e genuine examples of a register shift. Clear criteria are not always avail-able, b u t sometimes it seems as if the stylistic effect of t h e nasal clusters h a d lost its strength so t h a t t h e nasal could occur in the basic register too. ( C o m p a r e with a similar feedback in adult speech:
a joke easily ceases to be a j o k e a n d is flattened to a n o r m a l expression.)
T o sum u p : a detailed investigation c a n reveal a kaleidoscopic variation in both the anticipations a n d survivals of the child's sound system, as it reveals in the whole of t h e sound acquisition p r o -logy in general», in the words of Ferguson a n d F a r w e l l (Lg. 5 1 , p . 438).