Journal of Phonetics (198 2) 10, 43 9-451
A dichotomy in the English sound system Henry Petersen Institute of Phonetics, University of Aarhus, Aarhus, Denmark Received 8th July 1982
Abstract:
fu:f versus non-diphthongization of these vowels gives rise to a dichotomy in the English sound system with profound effects throughout that system. These effects are demonstrated in the domains of absolute and relative duration, relative speed; the decelerating/accelerating effects of consonants on changes of rate of delivery and the effects of changes in the time/speed and physiological dimensions on the duration of consonants. The articulation of /i:/, fu:f is analyzed , and it is hypothesized that diphthongization of these vowels may start a fresh evolutionary cycle as (perhaps) it did in the late Middle Ages. It is shown that diphthongization of the high vowels /i:/,
Introduction
Torben Kisbye (1980) states that ME [i:] in !if through the transitional stages [ij] and [ei] ended up as [ai] in PE life. Similarly, ME [u:] in hous underwent a parallel development resulting in PE [au] in house. And Kisbye goes on: "What took place was probably this. When the two highest vowels shifted to diphthongs, the pair below moved up to fill the gap leaving room for the pair below to move, etc. (but this is only one explanation!). The transition was gradual and phonemic distance always maintained so as to avoid homophony". The above-mentioned gaps were filled by, for instance, [c:] in ME see, PE [i:] in sea and [o:] in MEfol, fool, PE [u :] in fool. The hypothesis offered by Kisbye seems to me a very plausible one, the more so as the said operative force, the diphthongization of /i:/ and /u:/, is still with us, so that its effects can be studied : all manuals of English phonetics state that /i: , u:/ may be realized as either conservative sounds with little or no glide or as advanced sounds with various glides. Especially /u:/ would seem to have many allophones, among them [u], which may be said to be a mutation (Gimson, 1962). Furthermore, my limited personal experience has taught me that when a person has a conservative /i:/, his /u :/ will also be realized without a glide. Similarly, a person who has an advanced /i:/ will also have an advanced /u:/ . If it could be demonstrated that beyond its mere auditory effects diphthongization also gave rise to other measurable dislocations in the English sound system, this would in my opinion lend support to the above-mentioned hypothesis and perhaps transfer it from the realm of hypotheses to that of theories. Furthermore, one might expect that PE diphthongization might be the starting point of a fresh evolutionary cycle as perhaps the corresponding phenomenon was in the years between Chaucer and Shakespeare. 0095-44 70/82/040439 + 13 $03.00/0
© 1982 Academic Press Inc. (London) Ltd.
440
H. Petersen
The purpose of this article is to show the profound effects of diphthongization on the PE sound system and the dichotomy caused by it throughout that system, so that it is proper not only to speak about conservative and advanced high vowels, but also about conservative and advanced sound systems. The method of demonstration is simple: a comparison between two recordings of the same subject saying the same carrier sentence with the same open syllables embedded in it with a time lag (in this case six years) between the recordings. Nonsense words were not used. Methods
The subjects Two female members of the permanent staff of the Department of English, University of Aarhus, were used, SL and AR. SL is British, AR Australian, and both hold a university degree. In 1979 they were in their middle forties . From the point of view of pronunciation there is no great difference between their speech, but SL's high vowels are conservative, whereas those of AR are advanced, the auditory impression of her /u:/s being in the region of [ii]. The linguistic material This consisted of five groups of sentences, viz. all English words consisting of unvoiced consonants (except /h/) and /sp, st, sk/ and the five long vowels. The resulting open syllables were embedded in the carrier sentence I say · · · today. First the group of sentences with /i:/ was spoken once, then the sentences with the other long vowels. Mter this the subject started afresh and read the five groups ten times in all. From this body of words the open syllables treated below were selected. On each occasion, the subjects were instructed to speak at their normal rate of delivery and with their mouths at the same distance from the microphone, preferably 30- 40 em. It is clear that even with the above instruction in mind, the subjects could not possibly speak at the same rate of delivery in 1973 and 1979. The effects of the differences in speed are studied below. Instrumentation The sentences were recorded on a semi-professional tape recorder in a soundproof room at the Institute of Phonetics, University of Aarhus, and analysed by means of Fr~kjrer-J ensen 's Trans Ptichmeter and Intensity Meter, whose output - a duplex oscillogram, two intensity curves and (in 1973 only) fundamental frequency - was registered by a mingograph. Of the two intensity curves one was logarithmic and provided with a high pass filtering away the energy below 500Hz. Its integration time was 2.5 ms. The other was linear and unfiltered with an integration time of 10 ms. The speed of the mingograph was 10 em/sec. The material selected for treatment comprised 542 words. Parameters The consonants were measured from the offset of vowel-like vibrations in say to the onset of such vibrations in the long vowels. In their turn, these vowels were measured from the said point to the offset of vowel-like vibrations before /t/ in today. The open interval was thus included in the stop. Duration was measured in cs.
A dichotomy in the English sound system
441
List of symbols w, w 1 =word spoken in 1973 and 1979, respectively; c, c~> v, v 1 = consonant and vowel , corresponding to w, w 1 ; sd = standard deviation ; N, N 1 = number , corresponding to w, w 1 ; c/v = (Nc 7 Nv) x 100 ; percentages greater than 100 are given in integers; significance : · = 5 > p > 2.5, · · = 2.5 > p > 1, · · · = p < 1, 0 =not significant; spc, SPv =relative speed of production of c and v, respectively; the difference SPc- SPv is a relative expression of the decelerating/accelerating effect on the changes of rate of delivery; negative sign signifies a decelerating effect of the consonant, positive sign an accelerating effect; SPc - SPv > 1201: high effect, 1201> SPc- SPv > 1101: medium effect, spc - SPv < 1101: slight effect, all three limits are arbitrary; d = difference; the term physiological dimension comprises movements of articulators, but especially (and in most cases) differences in height of the articulating portion of the tongue. The effects of changes in this dimension are seen in minimal pairs in the same experiment and are termed synchronic ; the effects of changes in the time/speed dimension are seen in the same word spoken in 1973 and 1979, respectively, and are called diachronic. SL spoke faster in 1973 than in 1979, and the word par had to be rejected in 1973. AR spoke faster in 1979 than in 1973. Results Durational aspects On the basis of Tables I-III the salient durational features constituting the dichotomy in the English sound system can be outlined as follows: SL (conservative system) v < v 1 ; c > c 1 (exceptions: purr, car, cur); cjv > cdv 1 AR (advanced system), combinations with /i:, u:J v>v 1 ;c>c 1 ;cfvv 1 ;c>c 1 ;c/v>c 1 /v 1 in tar, caw;cfv v in such words. AR . The numerical values of cfv never exceed 100 in combinations with /i:/, i.e., c < v in such words . SL. The numerical values of cfv never exceed 100 in combinations with /u:/, but are always higher than the corresponding values of cfv in the minimal pairs with /a :,;);,:) :/ , i.e., in all such words c < v.
H. Petersen
442
/p/ +long vowels: duration, sd and cjv
Table I
SL pea par purr paw pooh
sd
w
WI
41.9 54.1 59.5 47.3
46.7 58.3 57.8 60.8 52.9
2.8/4.4 -/5.1 2.1/2.7 6.7/3 .7 2.3/2.5
42.9 44.9 43.2 41.5 38.4
32.5 37.4 37 .2 36.4 32.0
2.5/1.2 1.7/1.0 1.7/2.2 2.7/2.9 1.9/2.5
-
c
c1
24.0 21.2 24.9 22.3
23.8 19 .5 21.3 24.4 21.8
18.3 18.6 17.2 18 . 1 18.0
15.3 15.5 16.5 15.8 15.4
v
VI
1.6/2.3 -/1.5 1.3/1.4 2.0/2.0 0.9 /1.4
17.8 33 .0 34.6 25.0
22.8 38.8 36 .5 36.4 31.1
1.5/2.5 -/4.5 1.3/2.0 6.7/3.4 2.9/2.3
1.8/0.8 1.3/1.3 1.4/1.0 0.8/1.0 0.8/2 .0
24.6 26.3 26.0 23.5 20.4
17.3 22.0 20.7 20.6 16.6
2.7/1.0 1.5/1.2 1.4/2.3 2.1/2.8 1.6/0.9
sd
sd
cjv
cifv 1
135 64 .1 72.1 89.2
105 50.2 58.5 67.1 69.9
74.2 70.7 66.2 76.9 88.2
88.3 70.4 79 .8 76.5 92.8
-
AR pea par purr paw pooh
SL:N= 40,N1 =50, AR : N = 47,N 1 =50.
/t/ + long vowels: duration, sd and cjv
Table II
SL tea tar taw two
w
WI
sd
c
cl
sd
v
VI
sd
cjv
cdv1
42.6 55.9 59.3 47.6
46.2 62.6 60.5 52.0
2.0/3.4 2.4/6.1 6.8/3.1 3.0/3 .5
25.5 23 .0 25.2 22.4
23.9 21.6 23.7 20.6
2.1/1.6 3.2/2.0 2.8/3.0 1.7/1.8
17 .1 32.9 34.1 25.2
22.4 41.0 36.8 31.4
1.6/2.9 1.7/5.0 5.0/5.2 3.7/2 .0
149 69.8 73.7 88.7
107 52.6 64.6 65.6
45.7 48 .7 41.9 37.3
33.3 38.7 36.4 31.8
2.7/1.5 2.3/2.0 2.1/3.0 1.7/1.7
20.8 20.7 17.6 18 .9
16.0 15.9 15.8 15.8
1.4/1.0 1.2/1.2 1.4/1.0 1.1/1.6
24.9 28.0 24.3 18.4
17 .3 22.8 20.6 15.9
1.8/0.8 2.2/1.6 1.0/2 .8 0.9/1.1
83 .3 73.8 72.4 103
92.4 69.4 76.6 99.1
sd
cjv
cifvi
AR tea tar taw two
SL:N=N, =40,AR:N=37,N 1 =40.
Table III
sd
w
WI
42.6 55.1 52.8 62 .8 50.1
46.8 62.6 60.0 62 .5 54.9
2.0/2.5 2.3/4 .5 3.3/3 .5 7.7/3.7 3.2/2.2
key 45.8 car 47 .1 cur 41.3 caw 41.8 coo 37.1
34.7 38.8 36.0 37.6 30.8
2.3/2.7 2.0/1.8 2.5/1.6 1.8/2.3 1.6/ 1. 2
SL key car cur caw coo
/k/ + long vowels: duration, sd and cjv cl
sd
v
24.1 22.4 21.4 26.9 23 .9
23.6 22.5 23.3 24.8 22 .5
1.5/1.0 3.0/1.6 1. 7/2.3 3.8/1. 7 2.2/1.4
18.5 32.8 31.3 35.9 26.2
23.2 40 .1 36.6 37.7 32.4
1.7/2.1 2.0/3 .2 2.3/2.5 4.6/3.1 2.7/1.9
130 68.4 68.3 74.9 91.2
101 56 .0 63 .7 65.8 69 .3
21.6 19 .9 18.4 19.2 17 .6
17 .1 16 .6 15 .9 16 .7 15.3
1.1/1.7 1.6/1.4 1.2/0.8 1.5/0.7 1.3/1.1
24.2 27.3 22.9 22.7 19.5
17.5 22.2 20.1 20 .9 15.5
2.2/1.6 2.1/0.9 1.9/1.8 1.2/2.0 0.7/0 .6
89.6 72 .9 80.2 84.3 90.2
97.6 74.7 78.8 80.0 98.8
c
VI
AR
SL: N=N, =50, AR:N=48,N, = 50.
A dichotomy in the English sound system
443
AR. The numerical values of cjv sometimes exceed 100 in combinations with ju:j and are always higher than the corresponding values of cjv in minimal pairs with /i:, a:,;}:,:> :j, i.e., in combinations with ju: / c >·v (rarely) or c < v; in all other words c < v. With one exception (c/ v pea < cjv paw), cjv fi:l > cjv fa:,':" ·''/· SL. The values of cjv show that there is no possibility of direct proportionality in the system as a whole. AR. The values of cjv show that there is hardly any possibility of direct proportionality in combinations with the two high vowels, whereas such a contingency would seem to be possible in the three low vowels and in fact actualizes in the pairs par, paw, car, cur, in which the differences between the cjvs are less than 2%. I shall call such pairs ideal pairs, and my arbitrary definition of this phenomenon is cjv- cdv 1 ~ 2. The prerequisitie of direct proportionality is well-known: cjv = c 1 /v 1 , or, adapted to the conditions of physiological measurements : cjv "'=' ctfv 1 • We shall insert the values in par, the "best" ideal pair: 18.6/26.3"" 15.5/22.0; 0.707 "'=' 0.705, and in car, the "worst" ideal pair: 19.9/27.3 "'=' 16.6/22.2; 0.729 ""0.748. Interaction of c and v The three stops interact differently with the five long vowels as expressed by cjv. With SL we find, for instance, that /k/ attains its maximum duration in the combinations with /:>:/ in both tempos, whereas with AR it reaches its maximum duration in the combinations with /i:/. In combinations with /k/ the maximum duration of a vowel in the slow tempo is attained by /a :/ with SL and AR. On the other hand, in the fast tempo /3:/ is the longest vowel with SL, but /a:/ with AR. This leads to the question: which factors enter into the making of cjv? The height of the articulating portion of the tongue is obviously one such factor, but it seems irrelevant which portion (front, mid, back) is the active articulator: SL has her highest cjvs in the high front region, AR has hers in the high back region. With SL it is an almost invariable rule that the numerical value of cjv drops with the lowering of the articulating portion of the tongue, very much in accordance with Gimson's vowel chart (Gimson, 1980). Although with AR there is a certain approximation to this pattern, there are several exceptions. Another factor to be considered is rate of delivery (speed). With SL this is quite uncomplicated: the higher the speed, the higher is cjv in the same word. This matter is more intricate with AR: the same rule obtains in the two high vowels as in SL's system. (Excep· tion two), but the differences between the cjvs in the same word are not nearly as great as in the conservative system. The pattern, however, breaks down in the low-vowel region. An isolation of each of the factors would clearly be of great interest and will be attempted below (effects of changes in the time/speed and physiological dimensions). Relative speed In order to illustrate what is to be treated below, two examples from Table II are cited in Table IV. The figures obtain under conditions of increasing rate of delivery. If all the 12 signs are changed, the effects of decreasing speed are seen. SL's figures show clearly that the consonant counteracts a change of speed, whereas, on the face of it, it might seem that with AR consonant and vowel cooperate in bringing about a change of speed, at least to some degree . I shall assume that if on two occasions a subject suceeded in speaking the sentences of this experiment at exactly the same rate of delivery, then the equations (c/c 1) x
H. Petersen
444
Table IV
w- WI c - ci V - VI
Two examples of relative speed (cs) SL tea
SL tar
AR tea
AR tar
-3.6 + 1.6 -5.3
-6.7 +1.4 - 8.1
- 12.4 -4.8 -7 .6
- 10 .0 -4.8 -5 .2
100 = 100, (vfvi) x 100 = 100 would obtain. On this assumption rest the reflections below. With two sets of the same words in the same phonetic surrounding, the relative time of production of consonant and vowel can easily be calculated in per cent. The percentage expression of relative speed of production (sp c• spv) is the reciprocal of this quantity . As we are in no need of time relations, only speed relations are stated in Table V, which illustrates these relations under conditions of increasing rate of delivery. Reciprocation of the figures would show the relations under conditions of decreasing rate of delivery. Table V
Speed relations with increasing rate of delivery* AR
SL SPw
SPc
SPv
SPc- SPv
SPw
SPc
SPv
SPc - SPv
99 .2
128
-28.8
10 1 98.0 97.6
110 105 125
-9.0 -7.0 -27.4
132 120 116 114 120
120 120 104 115 117
143 120 126 114 123
-23.0 0.0 -22.0 + 1.0 -6.0
93.5 94.0 94.4 92.1
131 125 108 125
-37.5 -31.0 - 13.6 -32.9
137 126 115 117
130 130 I ll 119
144 123 118 115
- 14.0 + 7.0 - 7.0 + 4.0
97.7 100 109 92.3 94.2
125 122 117 105 124
-27.3 -22.0 -8.0 - 12.0 -29.8
132 121 115 Ill 120
126 120 116 115 115
138 123 114 109 126
- 12.0 -3.0 + 2.0 + 6.0 - 11.0
/PI + long vowels pea par purr paw pooh
Ill
107 102 112 ft / + long vowels tea 108 tar 112 taw 102 two 109 /k/ + long vowels key 110 car 113 cur 114 caw 99.5 coo 110
*For example, when the rate of delivery of the word "pea" (SL) increases from 100 to 111, the speed at which the lips perform their articulation decreases from 100 to 99.2, and the speed of adaptation of the vocal chords increases from 100 to 128 ; spc- SPv = - 28.8 means that the consonant has a high decelerating effect on the process, which is accelerated by the vowel only. Another example, "tar" (AR): SPw increases from 100 to 126, spc from 100 to l30,spv from 100 to 123;spc -SPv = + 7.0, which means that the consonant has a slight accelerating effect, whereas now, the vowel acts as decelerator. The general picture of the conservative system is a harmonious one : all three stops have a high decelerating effect in combinations with the vowels /i:, a:, u :/. Their decelerating effect is slight to mediu m in combinations with the remaining two vowels. In all cases the vowels act as accelerators.
A dichotomy in the English sound system
445
The general picture of the advanced system is a chaotic one: not even in combinations with the high vowels / i:, u: / can an exceptionless rule be laid down, although in these combinations the stops in most cases have a decelerating effect (exception: /t/ in two). Now the same consonant (/p/) has a high decelerating effect in pea, now a slight accelerating effect in paw, and now its effect is neutralized in par. The vowels act as accelerators in eight cases, as decelerators in five, and in one case the effect of the vowel is neutralized. As might be expected, in the four ideal pairs par, paw, car, cur, the effects of both consonants and vowels are neutralized or nearly so. Synchronic and diachronic effects The observable differences in the duration of consonants in the same word in the two experiments and in minimal pairs beginning with the same consonant in the same experiment (Tables I-III, etc.) are attributable to effects of changes in either the time/speed or the physiological dimensions- or perhaps in both together (Tables VI-VII). The words tea and the minimal pairs teajtar with AR will serve as examples of the method employed in separating the effects on the duration on /t/ of changes in either dimension. Synchronic differences. In the two minimal pairs tea/tar, the difference between the two sets of /t/s is 0.1 cs in either experiment. This leads to three mutually exclusive contingencies. (1) The differences are significant in both sets. If so, this is taken as a proof that the changes in the physiological dimension resulted in non-accidental synchronic effects of general validity on the duration of /t/. (2) The differences are not significant in both sets. If so, this is taken as a indication that the changes in the physiological dimension have no appreciable effect on the duration of /t/, that their effect is accidental only. However, on account of the consistency of the results, they must be supposed to be of general validity. (3) The result in one set is inconsistent with that in the other, e.g., significant in one, not significant in the other. If so, the effects of the changes in the physiological dimension on the duration of the consonants are unpredictable. Diachronic differences. The difference in duration between the two /t/s in 1973 and 1979 is 4.8 cs. If this difference is significant, this is taken as a proof that the change in the time/speed dimension resulted in a non-accidental diachronic effect of general validity on the duration of /t/. If the difference is not significant, this is taken as an indication that the diachronic effect brought about by the change in the time/speed dimension is accidental, predictable and of no appreciable effect on the duration of jtj. In the production of /p/, the active articulators (lips and tongue) move as independently of each other as articulators can. Consequently, in the /p/- words we can expect the clearest separation of the effects of the positions and movements of either articulator on the duration of the consonant. Effects of changes in the time/speed dimension Diachronically, the effects on the duration of consonants of changes in the time/speed dimension are very different in the two systems: with SL, out of 13(14) words there are significant effects only in three (tea, two, cur). In the remaining 10(11) words the effects are not significant. With AR, out of 14 words there are highly significant effects in 13. In purr, the effects are not significant. This means that in the conservative system the duration of the great majority of stops is not sensitive to changes in rate of delivery, whereas in the advanced system the duration of an even greater majority of stops is highly sensitive to such changes.
""" """
0\
Table VI
The effects on the duration of /p/ of changes in the time/speed and physiological dimensions*
SL1973 pea par purr paw pooh
pea
par
-
-
SL1979
purr
paw pooh
...
...
. ..
-
0 . .. -
...
0
...
. ..
-
-
...
pea 0
. .. 0
..
par
purr
paw pooh
...
. ..
0
. .. 0 . .. . .. . ..
. ..
0
0
. ..
0
AR 1979
AR 1973
0
pea pea par purr paw pooh
par purr
-
0
-
0 0 0 0
0
..
.. 0 0
paw pooh 0 0
-
0
0
0 0 0 0
-
pea
par
purr paw pooh
...
0
. ..
0 0
0 0 0
0 0 0
...
0
.. . 0 0
...
0 0 0 0
~
§
;:::
0
*The former are given diagonally in the 1979 tabulations. The latter are stated horizontally and vertically. See list of symbols for explanation of entries.
Table VII
The effects on the duration of /t/ of changes in the time/speed and physiological dimensions*
SL 1973 tea tea tar taw two
-
tar
taw 0
-
-
0
. ..
0
. ..
SL 1979 two
... 0
...
-
tea
. ..
tar
taw
...
0
0
0
. ..
two
. .. 0
0
0
.. .
AR 1979
AR 1973
. ..
..
tea tea tar taw two
-
0
... . ..
tar 0
-
. .. . ..
taw
two
... ...
. ..
..
-
-
. .. ..
tea
... 0 0 0
tar
taw
two
0
0 0
0 0 0
. .. 0 0
. .. 0
*The former are given diagonally in the 1979 tabulations. The latter are stated horizontally and vertically. See list of symbols for explanation of entries.
;:t:
~ Q..
;::;·
;::,-
()
Table VIII
SL 1973
key key car cur caw coo
-
. ..
car
0
cur
. .. 0
-
. . . .. . .. 0 0 .. .
SL 1979 caw
..
C)
The effects on the duration of /k/ of changes in the time/speed and physiological dimensions*
coo
key
0
0
. .. 0 . . . . .. .. .. -
0
car
0 0
. .. 0
AR 1973
cur
caw
coo
0 0
...
0 0
0 0
0 0
...
key
... 0
car cur caw coo
key
car
..
-
..
... .. . .. 0 . .. . . .
cur
AR 1979
caw
. . . . .. 0 ..
-
0 0
~ s·
0 -
..
coo
key
. . . . .. . .. 0 0
..
-
0
...
car
caw
0
0 0
0
. .. 0 0 ..
;;t.
cur
. ..
..
0
*The former are given diagonally in the 1979 tabulations. The latter are stated horizontally and vertically. See list of symbols for explanation of entries.
coo
0
"'~ ~ g: "":;::
()
;:s
Q..
~
""
~
""'""'""
.....:]
448
H. Petersen
Effects of changes in the physiological dimension Synchronically, the effects on the duration of consonants of changes in the physiological dimension are also very different : With SL, out of 26 minimal pairs there are significant effects in 16 (eight pairs in /p/- words, four in /t/· words and only four in /k/- words. In five pairs the effects are consistently 0 (one pair in /p/- words, two pairs in /t/- and /k/- words, respectively). The effects are "mixed" in five pairs (one pair in /p/- words, four pairs in /k/words). With AR, out of 26 minimal pairs there are significant effects in five pairs (one pair in /p/- words, four pairs in /k/- words). In 10 pairs the effects are consistently 0 (7 pairs in /p/words, 1 pair in /t/- words and 2 pairs in /k/- words). The effects are "mixed" in 11 pairs (2 in /p/- words, 5 in /t/- words and 4 in /k/- words). This means that in the conservative system the duration of the majority of stops is highly sensitive to differences in tongue height, whereas in the advanced system only a small minority is sensitive to such changes. For reasons stated above, the results in the /p/- words are particularly interesting. Out of 10 possible minimal pairs in the conservative system, tongue height has decisive influence on the duration of /p/ in eight minimal pairs, no, but predictable influence in one pair, unpredictable influence in one. Out of 10 possible minimal pairs in the advanced system, differences in height of articulating portion of the tongue have decisive influence on the duration of /p/ in one pair, but no predictable influence in seven pairs, unpredictable influence in two. The theme of the influence of vocalic oppositions on the duration of consonants could, of course, be elaborated in much greater detail than above. I shall, however, content myself with maintaining that in this area as well as the other areas treated in this paper, the dichotomy in the English sound system stands out clearly. The articulation of /i: , u:/
/i:/ SL. The course of F 2 is fairly horizontal, and the displacement, if any, of its former and latter halves is small (Fig. 1). Corresponding to this, the most characteristic shape of the
... 1.
r
+..
...
~i' ,1. . - ~fl'th
[p Figure 1
i :]
A dichotomy in the English sound system
449
"top" of her intensity curve is a rounded, almost semi-circular configuration (Fig. l(a)), i.e., the intensity is strongest a little before mid-vowel. The crux of the matter is that the tip of the tongue does not move (much), so that the escape channel of the airstream retains its shape, and a conservative (i:] -sound results. AR. There is a distinct displacement in F 2 , its former half being lower and of higher intensity than the latter. The transition from one to the other is of a lighter black than the two halves (Fig. 2). Corresponding to this, the most characteristic shape of the intensity curve is a two-humped configuration, the former hump being higher than the latter and separated from it by a deep "depression" (Fig. 2(a)).
[P
ij]
Figure 2
There are thus three discernible stages in the production of this /i:/: (a) high position of the tip of the tongue , narrow escape channel and maximum intensity, (b) lowering of the tip of the tongue (is it "blown away" by the powerful airflow?) with decreasing intensity, and (c) the tip of the tongue "swishes back" (with its own elasticity?) to an even higher position than in stage (a), the intensity increases, but not to its former height, because the airflow has diminished during stage (b). The result is the advanced diphthong (ij]. ju:j SL. The main course of F 2 is fairly horizontal with a minor displacement in mid-vowel (Fig. 3). This means that the articulating portion of the tongue does not move (much). Corresponding to this, the intensity curve (Fig. 3(a)) shows a fairly steady state before and about mid-vowel, after which the intensity decreases gradually. The resulting sound is a conservative [u:]. AR. F 2 slopes downwards in its entire extent, signifying, probably, a gradual widening of the escape channel of the airstream in that the articulating portion of the tongue glides gradually down towards the central area (Fig. 4). The intensity curve shows a fairly steady state in mid-vowel (Fig. 4(a)). The resulting sound is an advanced (i.i:].
H. Petersen
450
[ p
u :]
Figure 3
---.----,._----._- ;:·;.- -·_ + 1
l liill_-~lli- )~lr~---
r---- ----~ 1 - - ._ :·;.~ -- - - ~ - -----•• 1 ~ · ······ ---·--·- ··
...
[ a,;
tr
D 1.
i.i :
··-··· · ···------- ---· . ·-··- - -·· &
a d
••
11,, ,, ..
s1 ]
[P
u:]
Figure 4
The diachronic factor
In my article in English Studies (Petersen, 1980), I quoted the terms forward scan and anticipatory mechanism, thereby labelling a phenomenon whose nature was unknown to me, but which was clear enough in outline: a built-in instability of the English vowel system manifesting itself in the fact that consonants tend to be longer in an /i:/-environment than in an /a:f-environment. After having now described the external manifestations of the phenomenon in some detail (the internal ones are inaccessible to me), I would like to rename "the thing" and to call it the diachronic factor, because I think that it may be the factor underlying the chains of vowel shifts discussed briefly in the opening of this article.
A dichotomy in the English sound sy stem
451
My reasons are these : (1) The PE conservative system seems to me to be respresentative of the English sound system as it was when the great vowel shift had run its course. It has attained to some unstable equilibrium. It s unstable element is (and perhaps was) the great decelerating effects of the consonants, especially in combinations with the peripheral high and low vowels, perhaps not unlike the state of the system in Chaucer's days . (2) On the other hand , the PE advanced system is in a state of chaos and imbalance. Seen from the point of view of duration, the only feature it has in common with the conservative system is the medium to high decelerating effect on the consonants in combinations with the high front vowel /i: /, which, however, is itself in a state of change as is the high back vowel /u:/. Perhaps the state of this system is not unlike that of the transitional system between Chaucer and Shakespeare. I think that in its present state of disequilibrium it may well have passed the starting point of a fresh evolutionary cycle. It may also happen that the evolution will be arrested by the great conservative powers radio , TV, cinema, etc. References Gimson, A. C. (1970). A n Introduction to th e Pronunciation of English, pp. 114-115. London : Arnold. Gimson, A. C. (1980). Everyman 's English Pronouncing Dictionary (14th edition) p. XV. London: Dent. Kisbye, T. (1980). A Sh ort History of the English Language, p. 111. Aarhus : Aarhus University. Petersen, H. (1980) . Interaction between segments of open syllables. English Studies, 61, 156- 170.