Perceptual Effects on English Consonant Interpretation
5.3 Asymmetric mappings of stop voicing and aspiration
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
172
5.2.4 Interim summary
In this section, we have identified the hierarchy of phonetic salience in terms of syllable position, based on plentiful acoustic facts and cross-linguistic evidence. It seems to make obvious sense that, everything else being equal, perceptual cueing power is the strongest for onsets, intermediate for second/third onsets, and the weakest for codas. The results of the survey confirm that positional differences are statistically significant (χ2 = 717.2, df = 2, p < .0001). Following Steriade’s (2008) formula, if a speaker’s judgment of phonological similarity is deduced from observations of confusion, the salience hierarchy of the same consonant in different positional contexts can be sketched in (116), denoting that the perceptual salience between a consonant and silence in the simplex onset position is greater than that in the second or third onset position, which in turn is greater than that in the coda position.
(116) Salience hierarchy by position ( = less confusable, more distinctive than) C vs. Ø/_V C vs. Ø/(C)(C)_V C vs. Ø/V_
5.3 Asymmetric mappings of stop voicing and aspiration
As introduced in Chapter 3, English stops contrast in both aspiration and voicing (e.g. [p, p , b]), yet TM stops contrast in aspiration only (e.g. [p, p ]). There is a question that naturally arises in light of such a systematic distinction between the two languages: how do the English contrasts in aspiration and voicing map to TM, where no voiced obstruents are allowed in any position? This entails either of two potential mappings below.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
173
(117) Potential mappings of English voicing and aspiration contrasts to Mandarin a. L2 [+voiced] [–voiced] b. L2 [–aspirated] [+aspirated]
b p p b p p p p p p
L1 [–aspirated] [+aspirated] L1 [–aspirated] [+aspirated]
In (117a), English voiced [b] is mapped to TM unaspirated [p], and English voiceless [p] and [p ] are both mapped to TM aspirated [p ]. In (117b), however, English unaspirated [b] and [p] are both mapped to unaspirated [p], while L2 aspirated [p ] is mapped to TM aspirated [p ]. Intuitively, (117b) seems to be the better choice, since two of the stops in English, [p ] and [p], are mapped to their identical counterparts in TM, respectively, and by doing this the English “[+aspirated]
vs. [–aspirated]” status remains. In (117a), however, only English [p ] is faithfully mapped to an identical TM aspirated stop. Surprisingly, what is predominantly attested in the corpus is the mapping in (117a). The numerical facts are given in what follows.
In 4.3.1, English aspirated stops as the simplex onsets are 64.24% (282/439) likely to be realized as the TM aspirated counterparts, and 93.15% (408/438) of the voiced stop onsets in the same position are mapped to their unaspirated counterparts.
In 4.4.4, where the stop resides in the “stop-[l]” sequence, aspirated stops are 100%
(23/23) mapped to the identical aspirated stops, while voiced stops are 100% (6/6) mapped to the unaspirated counterparts. For another structure of onset cluster in 4.4.5, where the stop precedes [ ], 87.93% (51/58) of the aspirated stops are mapped to the same aspirated stops or affricates, while 97.5% (39/40) of the voiced stops are interpreted as their unaspirated counterparts or unaspirated affricates. In the coda position, moreover, the mappings of stops reveal no significant difference from those in the onset position. In 4.5.1, being simplex codas, English unaspirated voiceless
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
174
stops are majorly interpreted as aspirated stops in TM with the percentage of 93.06%
(67/72), and voiced stops are mapped to unaspirated stops 87.5% (21/24) of the time.
In 4.6.1, where the unaspirated voiceless stop is preceded by [s], it is realized as the aspirated counterpart 100% of the time. In the reverse order, namely the “stop-[s]”
sequence, the unaspirated voiceless stops in English are mostly mapped to the aspirated counterparts or merged with the following [s] to form an aspirated affricate in TM, as found in 68.42% (13/19) of the cases. If the stop is voiced, it is 100% (5/5) likely to be mapped to their unaspirated counterparts or merged with the following [s]
to form an unaspirated affricate. In 4.6.2, where the stop follows [s], the English unaspirated voiceless stop is 100% mapped to aspirated stops in TM. In the “stop-stop”
sequence in 4.6.3, the only observable case shows that the retained unaspirated voiceless stop is interpreted as the aspirated counterpart. In 4.6.4, where the stop follows a nasal, unaspirated voiceless stops are predominantly mapped to the aspirated counterparts, found in 95.24% of the cases, and voiced stops are 100%
realized as the correspondent unaspirated stops. Finally, in 4.6.7, where the stop follows a liquid, 96.43% (27/28) of the unaspirated voiceless are mapped to the corresponding aspirated stops, and 95% (19/20) of the voiced stops are realized as their unaspirated counterparts.
Given the fairly consistent adaptation tendencies, the only seemingly exceptional pattern is found in the “[s]-stop” sequence in 4.4.1, where the English unaspirated voiceless stops are majorly interpreted as the identical unaspirated stops in TM, as found in 61.36% (27/44) of the items, rather than the expected aspirated counterparts, as those observed above. The answer to the “exception” will be given later.
The adaptation patterns show that in general, English aspirated and voiceless unaspirated stops both tend to map to aspirated stops in TM, while voiced stops are interpreted as voiceless unaspirated stops in most cases, lending support to Paradis
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
175
and Tremblay’s (2009) finding, in which the Mandarin under discussion refers to the variety used in Mainland China. However, as observed in 4.4.1, a majority of English voiceless unaspirated stops following [s] in the onset position do not correspond to Paradis and Tremblay’s (2009) finding and map to voiceless unaspirated stops in TM , unlike the voiceless unaspirated stops in coda. In this respect, Ladefoged (2001) observes that a voiceless unaspirated stop preceded by [s] sounds nearly identical to the voiced counterpart; namely, they have similar VOTs around zero millisecond, judged by the waveforms of “a buy” and “a spy” in his book (pp 133).
VOT (Voice Onset Time) refers to the amount of time, measured in milliseconds (ms), between the release of the consonant closure and the onset of the vocal-fold vibration, usually visualized by reference to the waveform of a sound. Oftentimes, a voiced stop has a negative VOT since the voicing begins during the oral closure (i.e.
before the release), a voiceless unaspirated stop has a zero VOT because voicing starts just about when the oral closure is released, and an aspirated stop has a positive VOT because voicing starts after the oral closure is released. Through the observation of VOT, voicing and aspiration as features are no longer a plus-minus matter, but a continuum the value of which may vary in different positions within a syllable or word and in different languages for a single stop. For example, the voiced stops of Sindhi are fully voiced and have a large negative VOT, but in English they have little or zero VOT. For another example, aspirated stops in Sindhi have a VOT of only 50 ms, but those in Navajo have a VOT of up to 150 ms (Ladefoged 2001).
The following table summarizes the mapping patterns of stops from English to TM.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
176
(118) Mapping pattern of stops TM English
aspirated unaspirated
aspirated 68.46% (356/520) 31.54% (164/520)
voiced 6.6% (34/515) 93.4% (481/515)
voiceless unaspirated [_.] 91.45% (139/152) 8.55% (13/152) [.s_] 38.64% (17/44) 61.36% (27/44)
A couple of obvious conclusions concerning the asymmetric mapping of aspiration and voicing from English to TM are thus drawn below. First, English aspirated stops are majorly mapped to TM aspirated stops across the board (68.46%, 356/520), and English voiced stops are chiefly interpreted as voiceless unaspirated stops (93.4%, 481/515). The mapping from an English voiced stop to a voiceless unaspirated stop in TM, which is devoid of voicing contrast in consonants, is due to their perceptual similarity in VOT. The effect of VOT on the perceptual mapping is statistically significant (χ2 = 421.6, df = 1, p < .0001). Second, despite being both symbolized as [p, t, k] with IPA, the English voiceless unaspirated stops appearing after [s] in onset and those appearing in coda behave distinctly with respect to their mapping patterns of voicing and aspiration in TM. Specifically, when the same L2 voiceless unaspirated stop appears in coda position, up to 91.45% (139/152) are interpreted as its aspirated counterpart (with an epenthetic vowel) in TM (e.g.
[.(æ.l p.] Gallup → [.kai.lwo.p u.] 蓋洛普), with the minor rest 8.55% (13/152) to the TM voiceless unaspirated counterpart (e.g. [.t% p.l n.] Chaplin → [.t wo.pje.lin.]
卓別林). On the contrary, when an English voiceless unaspirated stop appears after [s]
in onset, 61.36% (27/44) of them are mapped to the TM voiceless unaspiated counterpart (e.g. [.stæn.l .] Stanley → [. !.tan.li.] 史丹利), and the rest 38.64%
(17/44) are to the TM aspirated counterpart (e.g. [.st k.t n.] Stockton → [. !.t a.k .tw n.] 史塔克頓). The effect of such a positional difference on the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
177
mapping of aspiration and voicing is significant in statistics (χ2 = 255.9, df = 4, p
< .0001). It is then assumed that a voiceless unaspirated stop in coda behaves more similarly to its aspirated counterpart in acoustics (VOT), as it may occasionally be released with aspiration when the word is uttered in an emphatic way. When an English aspirated stop follows [s] in the onset and undergoes the process of deaspiration, however, its VOT is closer to that of a voiced stop (around zero ms) In our terms, for the voiceless unaspirated stop that ends a syllable, the absence of auditory voicing weighs heavier than the absence of auditory aspiration noise. For the ones that follow [s], conversely, the absence of auditory aspiration noise weighs more than the absence of the auditory voicing. This generalization may explain the asymmetric mapping of voicing and aspiration between the two languages.