Phonetic and Phonological Analysis of Focal Accents of Disyllabic ...
Phonetic and Phonological Analysis of Focal Accents of Disyllabic Words in Standard Chinese 1
Yuan Jia1, Ziyu Xiong2, Aijun Li2
English Department, Nankai Univerisity Institute of Linguitics, Chinese Aademy of Social Sciences [email protected], [email protected], [email protected] 2
Abstract. The article investigates the phonetic and phonological property of focal accents conveyed by disyllabic focused words with various tonal combinations in Standard Chinese. Phonetically, the effect of focal accents upon f0 resides in two aspects: the manner and the condition of focal accents. Phonologically, the distribution of focal accents is mainly concerned. Acoustic and perceptual experiments and the underlying tonal target of focused constituents are employed in both phonetic realization and phonological analysis. Major findings are that: f0 ranges of focused words are expanded as the H tones of both focused syllables are raised; the f0 of the post-focus syllables are compressed obviously in the way the H tones of Tone1 and Tone2 are lowered; the realization of accents is closely related to the tonal target of the focused words; specifically, accents influence the acoustic performances of tones; furthermore, the combination of H/L determines the distribution of accents. Key Words: focal accents; phonetic realization; phonological analysis; tonal target
1
Introduction
In previous literature on English focus, in the phonetic aspect, Xu [1] proposed that narrow focus is realized by expanding the pitch range of the on-focus stress syllables and by suppressing the pitch range of the post-focus syllables while leaving the pitch range of the pre-focus syllables intact. In the phonological aspect, Ladd has highlighted the relation between focus and intonational nucleus placement [2] [3]. To this relation, Ladd adds the role played by abstract prominence patterns: the nucleus signals the focus of the utterance, and the nucleus is assigned to the element that bears the sentence stress in the sentence-level prominence pattern. Studies of focus in Standard Chinese concentrate on the phonetic realization of focused constituents. Xu [4] investigates the formation of f0 under the influence of focus by examining short Mandarin sentences with systematically varied tonal components and focuses. Results of his experiment demonstrate that the f0 range is expanded by focus: the high points of H tones are raised while the low points of L tones are lowered. Further, the f0 of all the words following the focus are substantially lowered no matter whether the f0 of the focused words are raised or lowered. In contrast with the on-focus raising and post-focus lowering, the f0 of the pre-focus
syllables are barely changed. Although the existing studies have reported changes in f0 triggered by focus, phonetic and phonological explorations on the Chinese disyllabic words as a whole entities bearing focus, especially on the underlying factors constraining the distribution of these focal accents, have still been far less than those on English focus, and even absent at all. In this regard, we intend to examine the phonetic and phonological property of focal accents conveyed by Chinese disyllabic focused words with exhaustively various tonal combinations, and to offer a description of the effect of focal accents upon f0 by the very means to reduce the tonal combination to tones in terms of H and/or L. We further endeavor to disclose the underlying forces constraining the distribution of the accent patterns.
2 2.1
Methodology Experiment designing
Focused words adopted in the acoustic and perceptual experiments are all disyllabic and with the morphosyntactic structure represented as Modifier-Head 1 , or MH for short. Sixteen kinds of tonal combinations for the focused word of MH structure emerge from the formula of “[Tone1+Tone1] … [Tone4+Tone4]”. And the composing syllables fall into a couple of sets: {Gan1, Tang2, Biao3, Da4} as the set of the first syllables, and {Ge1, Yi2, Jie3, Mei4} the set of the second. A certain focused word is thus constituted by the combination of each member from the first set and each from the second. All these words are set in four patterns of target sentences: 1) Jin1 Tian1 + [focused words] + Fei1 Dong1 Jing1; 2) Jin1 Chen2 + [focused words] + Hui2 Dong1 Jing1; 3) Jin1 Wan3 + [focused words] + Fan3 Dong1 Jing1; 4) Jin1 Ye4 + [focused words] + Qu4 Dong1 Jing1. It has been made apparent in previous studies that focal accents exert certain effect upon f0 of the adjacent syllables. In the present study, however, the adjacent syllables are allocated with the four tones for the purpose of deliberating the effects imposed by the focal accents upon f0 of these adjacent syllables under every tonal circumstance. The tones of the syllables apart from the focused words and the adjacent syllables are set invariably in Tone1. To approach to the different types of focus, we employ guide sentences which are mainly composed of wh-question equivalents. The guide sentence for all broad focus expressions in this article is invariably the interrogative of “Fa1 Sheng1 Le0 Shen2 Me0 Shi4? (What happened?)” And for narrow focus sentences we employ the guide sentences generally formulated as “Jin1 Tian1/Chen2/Wan3/Ye4 Shei2 Fei1/Hui2/Fan3/Qu4 Dong1 Jing1? (Who fly to/go back to/go back to/go to Tokyo today/this morning/this evening/tonight?)”.
1
Phonetic realization of focused words is different due to the different morphosyntactic structure of focused words (Jia, Xiong and Li) [5].
2.2
Experiment procedures
Four Standard Chinese speakers, one female and three males, ranging from 20 to 45 of age, were invited as the subjects. The recording was conducted in the sound-treated booth at the Institute of Linguistics, Chinese Academy of Social Sciences. The subjects were seated comfortably in front of the computer screen, and the microphone was placed a hand away from mouth. During the recording, each sentence appeared on the screen three times in random order; meanwhile, the guide sentences were broadcast and the subject was asked to read aloud the displayed sentences as the response to the question in normal speed without any irregular pause. Sounds were recorded and saved directly into computer through sound recording software as “wav” file. The target sentences in which the clause-middle focused words have 16 kinds of tonal combinations pronounced by a male and a female speaker were chosen for perceptual experiment samples. As has been mentioned above, each target sentence was recorded three times; only one was collected for each target sentence so as to reduce the unnecessarily huge amount of data. Since in this experiment the subjects are only required to give judgments over the patterns of the focused words in terms of Weak or Strong, and hardly any correlation exists between the circumferential tones and the focused words, thus, we render the tones both preceding and following the focused words as being Tone1. Totally three females and five males were invited to participate in the perceptual experiment. They are all standard Chinese speakers and show rather sensitive and stable perception. In order to minimize the negative analogical effect and perceptual fickleness, all the sentences involved were broadcast in sheer random order through perceptual software, and each sentence was perceived by as many repetitions as the subjects required in order to confirm their judgment. Thus, for each focused word we had two target sentences to provide to the subjects for judgment, and finally we got 8×2=16 samples for the word with the same tonal combination. Each subject was asked to finish the experiment individually, without any interpersonal consultation. During the experiment, four options appeared with the icons worded as: “the first syllable strong 2 “, “the second syllable strong”, “both strong” and “both not strong”. The subject was expected to choose one of these options according to their decisive judgment of the weight of the focused words in target sentences. 2.3
Data labeling and extraction
Speech was first labeled by automatic segmentation software, and then the syllable boundaries were modified by hand. Before extracting the data, the manual refinement of the pitch tier was conducted in order to ensure the accuracy of the data. The data were retrieved by praat script with each syllable for 10 points, and the duration of the utterances was normalized. Finally, SPSS10.0 was used to get the means of f0 for each target sentence.
2
Since each focused words in our experiment consists of a couple of syllables, it is sensible to require the subjects to tell which of the two syllables or whether both are strong, or not strong.
3
Phonetic realization of focal accents
In this part, two aspects of the effect of focal accents upon f0 will be discussed: 1) the manner of effect, specifically taken to mean rising or lowering the f0, and 2) the condition of effect, which means whether the lowering or rising imposes the effect on H or L tones. The fundamental unit we apply to describe these effects is the underlying tonal target, H or L; specifically, we render each focused syllable separately to divide it into its original tones and in this manner to observe the performance, rather than deal with a single syllable as the descriptive unit. The following two figures are the mean f0 curves of sentences: “Jin1 Tian1 [Gan1] × [Ge1, Yi2, Jie3, Mei4] Fei1 Dong1 Jing1”, focused words [Gan1] × [Ge1, Yi2, Jie3, Mei4] are under both broad and narrow focus conditions. These sentences are adopted for the comparison among the pitch range changes of the second syllables of the focused words.
Fig.1& 2. f0 means of “Jin1 Tian1 [Gan1]×[Ge1, Yi2, Jie3, Mei4] Fei1 Dong1 Jing1”. [Broad and Narrow Focus Conditions] MH indicates that the morphosyntactic structure of focused word is Modifier-Head, “0” means the focus type is broad and “2” is narrow focus, the following “2” illustrates the position of the focused word is middle, the following “11” demonstrates the tonal combination is Tone1+Tone1 and “12” is Tone1+Tone2, etc, the last “1, 2, 3 and 4” show the tones adjacent to the focused word are Tone1, Tone2, Tone3 and Tone4. Comparison of the two graphs shows that the pitch range of the second syllable is expanded by the focal accent, from the specific values, the pitch range of the broad focus syllable is 104Hz-215Hz and the difference of 215 minus 104 equals to 111Hz, while this difference of the second syllable range under narrow focus condition presents 137 Hz, which demonstrates that the pitch range of the second syllable of the focused words is expanded by the focal accent for 25Hz. The following two graphs read “Jin1 Tian1 [Gan1, Tang2, Biao3, Da4]×[Ge1] Fei1 Dong1 Jing1” under both broad and narrow focus conditions from which the changes of pitch range of the first focused syllables can be seen.
Fig. 3 & 4. f0 means of “Jin1 Tian1 [Gan1, Tang2, Biao3, Da4]×[Ge1] Fei1 Dong1 Jing1”. [Broad and Narrow Focus Conditions] The above two figures illustrate that the pitch ranges of the first syllables are expanded by the focal accent, specifically, the pitch ranges of the first syllables of the broad focus words are within 114Hz-218Hz, and the pitch ranges of the narrow focused syllables are 122Hz-240Hz, meaning that the pith range of the first syllable is enlarged by 14Hz. The above paragraph demonstrates that the pitch range is expanded by focal accents and the internal cause for the pitch range changes is analyzed through the direct comparison of the focused words under different focus conditions in the following figures. The following figure is the mean f0 curves of sentence: “Jin1 Tian1 Gan1 Ge1/Gan Mei4 Fei1 Dong1 Jing1”, focused word “Gan1 Ge1” and “Gan1 Mei4” are under both narrow and broad focus conditions.
Fig. 5 & 6. f0 means of “Jin1 Tian1 Gan1 Ge1/Gan1 Mei4 Fei1 Dong1 Jing1”. Examination of the above two figures reveals that focal accents lift the f0 of the H tones of the two syllables of the focused words while the L tone of the focused syllable resembles that of the broad focused syllable. The realization of focal accents manifests on the two focused syllables differs from English that the raising of the pitch is on the stressed syllable at the sentential level. In general, the manners and conditions of the effect of focal accents upon f0 in under-focus domain can be generalized in three aspects: firstly, the pitch range is expanded by the focal accent; secondly, the essential cause for the expansion of the f0 range is the increase of the f0 of the H tones, which indicates that the effect imposed by focal accents upon pitch mainly manifests itself on the H tones; thirdly, the effect exerted by focal accent on L tones is by no means obvious and systematic; fourthly, the manifestation of the effect of the focal accent on both of the focused syllables characterizes the feature as being language-specific of Chinese. Figure 7 and 8 illustrate the effect of focal accents on f0 range of the syllables in pre-focus and post-focus domain. Target sentences in Figure 7 are under broad focus
atmosphere while Figure 8 under narrow one. The model of target sentences presents: “Jin1 Tian1 (Chen2/Wan3/Ye4)+[Gan1 Ge1]+ Fei1 (Hui2/Fan3/Qu4) Dong1 Jing1”. As the pitch changes imposed by focal accents are under direct discussion, the tones of the focused words are set as Tone1.
Fig.7 & 8. f0 means of “Jin1 Tian1 (Chen2/Wan3/Ye4)+[Gan1 Ge1]+Fei1(Hui2/Fan3/Qu4) Dong1 Jing1”. [Broad and Narrow Focus Conditions] Figure 7 and 8 demonstrate that the pitch range of the syllables following the focused word is compressed remarkably by focal accents while the pitch range of the syllables preceding the focused word remain much the same. The intrinsic cause for the compressing of the pitch range of post-focus syllables can be obtained from Figure 5 and the following two graphs. The follow figures are the f0 curves of “Jin1 Chen2/Ye4 Gan1 Ge1 Hui2/Qu4 Dong1 Jing1”.
Fig.9 & 10. f0 means of “Jin1 Chen2\Ye4 Gan1 Ge1 Hui2\Qu4 Dong1 Jing1” Generally in the above mentioned three graphs have been discussed the phonetic realization of the focused word “Gan1 Ge1” with the differently assigned tones in surrounding. What is unanimously represented among the four figures is that f0 in the focus position behave in the same manner, i.e., the f0 of the H tones are remarkably raised under the effect of the focal accents. Pre-focus f0 of the four tones, irrespective of whether being H or L tones, show either no perceptible difference or only slight disparity that is too erratic to be of any statistical significance. Post-focus f0, however of the H tones of Tone1 and Tone2 are significantly compressed while those of the H tone of Tone4 show a minor height difference over that under broad focus condition, the reason why the H tone of Tone4 are not affected obviously is that the H tone locates itself in the transition part, and contains the L tone. All the L tones are hardly subject to the effect of the focal accents, with only quite negligible changes that also lack statistical importance.
4
Phonological analysis of focal accents
Within the frame of Structure-based FTA 3 theory (Ladd & Gussenhoven)[2][6], the linguistic description of accent patterns involves two complementary but essentially separate aspects: a statement about which parts of an utterance are focused, and a statement about how a given pattern of focus is determined by the location of the accent. They argue that the speaker’s decision about what to focus is subject to all kinds of contextual influences that are difficult to identify. They also notice that once the focused part of the utterance is specified, the accent pattern follows more or less automatically by language-specific rules or structural principles. Therefore, in this part, the accent patterns of focused words at the sentential level with various tonal combinations will be described through the theoretical model of Metrical Phonology [7] based on the data obtained from the perceptual and experiments, further, the language-specific rules for restricting the distribution of the focal accents will be analyzed. Table 1 demonstrates the perceptual results of accent patterns for each focused words. Table 1.
Tonal Combinations
Perceptual results of focused words
Both not strong
Both strong
First Strong
Second Strong
Total
Tone1+Tone1 3 13 16 Tone1+Tone2 2 11 3 16 Tone1+Tone3 1 4 7 4 16 Tone1+Tone4 6 8 2 16 Tone2+Tone1 5 11 16 Tone2+Tone2 6 2 8 16 Tone2+Tone3 2 4 6 4 16 Tone2+Tone4 6 1 9 16 Tone3+Tone1 3 13 16 Tone3+Tone2 1 3 2 10 16 Tone3+Tone3 4 7 5 16 Tone3+Tone4 5 1 10 16 Tone4+Tone1 5 3 8 16 Tone4+Tone2 1 6 9 16 Tone4+Tone3 1 10 5 16 Tone4+Tone4 5 1 10 16 It is can be seen from Table 1 that the first syllables of 8 focused words with the tonal combination of “Tone1+Tone4” were judged by the subjects to be “strong”, which occupies 50% of the total number 16, whereas the “both strong” 37.5% and “second strong” 12.5%. In light of the phonetic realization shown by Figure 6, the accent pattern of “Tone1+Tone4” displays “s-w” relationship. Similar method can be adopted to approach the accent 3
Gussenhoven proposes the “Focus-to-Accent” (FTA) approach. In very general terms, the FTA theory is that words and constituents in utterances can be focused for various reasons, and that focused words and constituents are marked by pitch accents.
patterns of the other focused words with different tonal combinations, specifically, the accent patterns of the focused words of “Tone1+Tone2”, “Tone1+Tone3”, “Tone1+Tone4”, “Tone2+Tone3”, and “Tone3+Tone3” reflect “s-w” relationship; while the tonal combinations of “Tone2+Tone1”, “Tone3+Tone1”, “Tone3+Tone2”, “Tone3+Tone4”, “Tone4+Tone1”, and “Tone4+Tone3” show the opposite accent relationship of “w-s”. The remaining five tonal combinations, “Tone1+Tone1”, “Tone2+Tone2”, “Tone2+Tone4”, “Tone4+Tone2”, and “Tone4+Tone4”, invariably show the “w-s” pattern through the perceptual experiment. The explanation for this strong tendency is to be made in Table 6. The following tables show the accents distribution pattern of 16 different tonal combinations of MH structure from the word level to the sentential level through metrical grids (in the grids below, Line 0 stands for syllabic level, Line 1 for lexical level, and Line 2, sentential, “()” indicates syllable boundary): Table 2. Gan1 Line2 Line1 Line0
× (s)
Phonological representation of [Gan1]×[Ge1, Yi2, Jie3, Mei4] Gan1
Ge1
× × (s)
Line2 Line1 Line0
× × (s)
Yi2
× (s)
Gan1 Line2 Line1 Line0
× × (s)
Jie3
× (s)
Line2 Line1 Line0
Gan1
Mei4
× × (s)
× (s)
Lin [8] investigates the stress pattern of Mandarin disyllabic words in citation forms through perceptual experiment, and proposes that when disyllabic words are in normal stress, there is no absolute fixed stress pattern. This indicates that each syllable of MH structure words can bear word stress in citation forms. As can be seen from the above table, the sentential accent dwells on the second syllable “Ge1” in “Gan1 Ge1”, and the accents dwell on the first syllable in “Gan1 Yi2”, the first syllable in “Gan1 Jie3”, and the first syllable in “Gan1 Mei4”. Table 3.
Line2 Line1 Line0
Tang2
Ge1
× (s)
× × (s)
Phonological representation of [Tang2]×[Ge1, Yi2, Jie3, Mei4]
Line2 Line1 Line0
Tang2
Yi2
× (s)
× × (s)
Tang2 Line2 Line1 Line0
× × (s)
Jie3
× (s)
Line2 Line1 Line0
Tang2
Mei4
× (s)
× × (s)
And the table above shows that the syllables on which the sentential accents occur are “Ge1” in “Tang2 Ge1”, “Yi2” in “Tang2 Yi2”, “Tang2” in “Tang2 Jie3”, and “Mei4” in “Tang2 Mei4”. Table 4. Biao3 Line2 Line1 Line0
× (s)
Phonological representation of [Biao3]×[Ge1, Yi2, Jie3, Mei4]
Ge1
× × (s)
Biao3 Line2 Line1 Line0
× (s)
Yi2
× × (s)
Biao3 Line2 Line1 Line0
× × (s)
Jie3
× (s)
Line2 Line1 Line0
Biao3
Mei4
× (s)
× × (s)
From the above table can be seen that “Ge1” in “Biao3 Ge1”, “Yi2” in “Biao3 Yi2”, “Biao3” in “Biao3 Jie3”, and “Mei4” in “Biao3 Mei4” are the four syllables on which the sentential accents are located. Table 5.
Line2 Line1 Line0
Da4
Ge1
× (s)
× × (s)
Phonological representation of [Da4]×[Ge1, Yi2, Jie3, Mei4] Da4 Line2 Line1 Line0
× (s)
Da4
Yi2
× × (s)
Line2 Line1 Line0
× × (s)
Jie3
× (s)
Line2 Line1 Line0
Da4
Mei4
× (s)
× × (s)
And this table shows that the focal accents occupy the four positions of “Ge1” in “Da4 Ge1”, “Yi2” in “Da4 Yi2”, “Da4” in “Da4 Jie3”, and “Mei4” in “Da4 Mei4”. We would like to explore the more unified explanation for the underlying causes for the shifting of the stress and the specific distribution patterns shown by the tables above. Sharing the very morphosyntactic structure of MH, the accent distribution patterns of these words, however, vary with the various tonal combinations. The pair of “Tang2 Ge1” and “Tang2 Jie3”, for instance, gives the accent patterns of “strong-final” and “strong-initial” respectively. The cause for the shifting of the stress from the first syllables to the second at the sentential level can seemingly be reduced to the simple fact that with the first syllable being the same in both words the only difference between them lies in the tones of the second syllables, Tone1 and Tone3 respectively. The accent distribution of above minimal pair of “Tang2 Ge1” and “Tang2 Jie3” implies that the explanations can be sought if the traditionally termed four tones have been reduced to the original tones marked duly by the very permutations of H and L, that is, Tone1 is transformed into HH, Tone2 LH, Tone3 LL, and Tone4, HL. Concretely, therefore, in “Tang2 Ge1” the syllable “Ge1” claims the accent just because “Ge1” can be re-rendered as “HH”, compared with “Tang2” as “LH”. “HH” overtakes “LH” in number of “H” and therefore appears strong. Similar situation holds true for “Tang2 Jie3” in which “H” in the “LH” pattern of “Tang2” outnumbers that of “LL” of “Jie3” and thus claims the accent. This correlation can be attested for the patterns of the other words listed below: “Strong-initial” patterns are found for “Gan1 Yi2”, “Gan1 Jie3”, “Gan1 Mei4” and “Da4 Jie3” because they can be re-written respectively as “HH LH”, “HH LL”, “HH HL” and “HL LL”; another “Strong-initial” pattern notably exists for “Biao3 Jie3”, which can be attributed to the tone sandhi of “Biao3 Jie3 → Biao2 Jie3”, that is, “LH LL”. Correspondingly, “Strong-final” patterns occur for “Biao3 Ge1”, “Biao3 Yi2”, “Biao3 Mei4” and “Da4 Ge1”, with the respective renderings of “LL HH”, “LL LH”, “LL HL” and “HL HH”. However, a seemingly disturbing case emerges for five words in each of which the combination pattern in terms of “H/L” for the first syllables is totally identical in number of H to the corresponding second ones, as in “Gan1 Ge1” re-rendered as “HH HH”, “Tang2 Yi2” as “LH LH”, “Da4 Mei4” as “HL HL”, “Tang2 Mei4” as “LH HL”, and “Da4 Yi2” as “HL LH”. Solely from the perspective of “H/L” pattern the above five words should all have the pattern in which the first and the second syllables are equally strong. The fact both revealed by the perceptual experiment as well as
phonetic graphs 4 , however, turns out to be that it is the second syllables of the five words that bear the sentential accents and they fall into the “strong-final” pattern. Therefore, by permutation, then, three patterns of contrast in terms of the amounts of H tones come into being. And they are, of the two syllables: 1) the first with more H tones, 2) the second with more H tones, and 3) each containing equally numbered H tones. The result of “more Hs claiming strong” can be explained through the phonetic analysis of focal accents that the realization of the focal accents mainly manifested on the H tones of the focused syllables. What, then, accounts for the pattern with “H/L” pattern being equal? Pierrehumbert [9] proposes that when two stressed syllables sound equal in pitch, the second is actually lower and when the perception of pitch is equal in height the second syllable bears more stress. In view of this, the height of the pitch of the H tones is the crucial factor accounting for the accent patterns of the word with equal number of H tones. The following table provides the mean maximum pitch values of the H tones of the focused words under narrow focus condition with the tonal combinations of Tone1+Tone1, Tone2+Tone2, Tone2+Tone4, Tone4+Tone2 and Tone4+Tone4 (H1 indicates the mean pitch values of the H tones of the first focused syllable and H2 is the second, the unit for these values is Hz). Table 6. Tonal Combination T1+T1 T2+T2 T2+T4 T4+T2 T4+T4
Maximum pitch values of H tones
Tonal Feature HH + HH LH + LH LH + HL HL + LH HL + HL
H1 (Hz)
H2 (Hz)
237 222 179 244 242
242 231 265 242 240
The extraction of the above values of f0 under the circumstances of focus is achieved by smoothing off the influences exerted by the tones preceding and following the focused words, and thus by obtaining the means of the 16 focused words, each of which contains 48 samples. The puzzling fact can be explained from the above values that the mean maximum pitch values of the second focused syllable are higher than the first ones in the tonal combinations of Tone1+Tone1, Tone2+Tone2 and Tone2+Tone4. And in the tonal combinations of Tone4+Tone2 and Tone4+Tone4 the heights of the H tones of the two syllables nearly equals each other. All these data demonstrate that the second syllables of the focused words of these tonal combinations bear stronger accent. Therefore, in this part, we provide the phonological description of focused word with various tonal combinations through the theoretical model of Metrical Phonology. From the representation of the accent pattern of these focused words we offer the unified explanation for the underlying causes for the distribution of the accents by the means of re-rendering the tones of each syllable of focused word into the original H/L tones or the combinations of these two tones. When the underlying tonal combinations are different (this difference lies in the tonal combinations of each syllable between the focused words), the determining role of restricting the 4
As has been discussed above, the raising of f0 is mainly manifested on the H tones.
distribution of the accents is attributed to the amount of the H tones; however, when the underlying tonal combinations are identical, a predominant tendency exists of being always “strong-final”. These results strongly indicate that the phonological composition of the focused syllable is the primary cause for restricting the distribution of the accents. 5
Conclusion
In this study, an integrated work of phonetic and phonological analysis for focal accents conveyed by disyllabic words in Standard Chinese has largely been done. Major findings are achieved from all the above analyses are expressed in both the phonetic and phonological aspects: for the phonetic perspective we have: 1) in the focus position under narrow focus condition, the pitch range of f0 is enlarged as the H tones under narrow focus condition is significantly higher than under broad focus condition, and L tones show on obvious differences in f0 under narrow focus condition, if there were, from those under broad focus condition, but the differences are by no means systematic and worthy of note. This finding, however, differs from that made by Yi Xu [4], who argues that under focus conditions f0 of H tones becomes higher and of L tones, lower. 2) In the positions preceding the focus, f0 range remains much the same, whether H tones or L tones, f0 is but slightly subject to the effect exerted by the focus, and this finding bears no considerable divergence from that of Xu [4]. 3) Extremely significant effects imposed by the focal accents are found on the post-focus pitch range of syllables, which slightly but notably differs from the finding made by Xu [4] that the pitch range of f0 of the post-focus syllables is considerably compressed by the focus. Our study, however, offers a more refined discovery that the focal accents do lower the f0 of the H tones of Tone1 and Tone2, but exert only insignificant effects on the f0 of the L tones of Tone2, Tone3 and Tone4 and of the H tones of Tone4. For phonological perspective, the underlying causes for the distribution of the focal accent are explored through the phonological description of the accent patterns of the focused words. The manifestation and distribution of focal accents are closely related with the underlying tonal target of the focused words that the realization of the tones is influenced by accents and the phonological composition of the focused constituents determines the distribution of the focal accents. We would therefore agree with Ladd [3] that once focus is identified in accordance with the speaker’s intent and with the context, it is the language-specific structural factors that determine the accent distribution. In Standard Chinese, the phonological combination of the focused constituents is the primary causes that determine the distribution of the focal accents, this analysis also identical with Chomsky [10] that phonological characteristics determine the nature of the accents. 6
Acknowledgements
This research is supported by CASS key projects “Fundamental Speech Corpus and Pitch Pattern of Standard Chinese” YZDN50-05050 and “Research on Phonetics and Spontaneous Speech Processing”.
References 1．Xu, Yi. 2005. Phonetic realization of focus in English declarative intonation. Journal of Phonetics. 2．Ladd, D. Robert. 1980. The structure of Intonational Meaning: evidence from English. Bloomington, Indiana: Indiana University Press. 3．Ladd, D. Robert. 1996. Intonational phonology. Cambridge University Press. 4．Xu, Yi. 1999. Effects of tone and focus on the formation and alignment of F0 contours. Journal of Phonetics. 5．Jia, Yuan, Xiong Ziyu, Li Aijun. 2006. Narrow focus patterns of disyllabic words with different morphosyntactic structures in Standard Chinese. TAL 2006 LA ROCHELLE (France). 6．Gussenhoven, Carlos. 1983. Focus, mode and nucleus. Journal of Linguistics 19:377-417. 7．Liberman, Mark and Alan Prince. 1977. On the Stress and Linguistic Rhythm. LI 8:249-336. 8．Lin, Maocan. 1984. Qingzhong Yin. In Wu Zongji and Lin Maocan (eds.), Shiyan Yuyinxue Gaiyao (An introduction to experimental phonetics). Beijing: Commercial Press. 9. Pierrehumbert, Janet. 1979. The perception of fundamental frequency declination. Journal of the Acoustical Society of America, 66(2): 363-369. 10．Chomsky, Noam. 1971. Deep structure, surface structure, and semantic interpretation. In D. Steinberd and L. Jakobovits (eds.) Semantics-an Interdisciplinary Reader in Philosophy, Linguistic and Psychology, 183-216. Cambridge: Cambridge University Press.
Yuan Jia1, Ziyu Xiong2, Aijun Li2
English Department, Nankai Univerisity Institute of Linguitics, Chinese Aademy of Social Sciences [email protected], [email protected], [email protected] 2
Abstract. The article investigates the phonetic and phonological property of focal accents conveyed by disyllabic focused words with various tonal combinations in Standard Chinese. Phonetically, the effect of focal accents upon f0 resides in two aspects: the manner and the condition of focal accents. Phonologically, the distribution of focal accents is mainly concerned. Acoustic and perceptual experiments and the underlying tonal target of focused constituents are employed in both phonetic realization and phonological analysis. Major findings are that: f0 ranges of focused words are expanded as the H tones of both focused syllables are raised; the f0 of the post-focus syllables are compressed obviously in the way the H tones of Tone1 and Tone2 are lowered; the realization of accents is closely related to the tonal target of the focused words; specifically, accents influence the acoustic performances of tones; furthermore, the combination of H/L determines the distribution of accents. Key Words: focal accents; phonetic realization; phonological analysis; tonal target
1
Introduction
In previous literature on English focus, in the phonetic aspect, Xu [1] proposed that narrow focus is realized by expanding the pitch range of the on-focus stress syllables and by suppressing the pitch range of the post-focus syllables while leaving the pitch range of the pre-focus syllables intact. In the phonological aspect, Ladd has highlighted the relation between focus and intonational nucleus placement [2] [3]. To this relation, Ladd adds the role played by abstract prominence patterns: the nucleus signals the focus of the utterance, and the nucleus is assigned to the element that bears the sentence stress in the sentence-level prominence pattern. Studies of focus in Standard Chinese concentrate on the phonetic realization of focused constituents. Xu [4] investigates the formation of f0 under the influence of focus by examining short Mandarin sentences with systematically varied tonal components and focuses. Results of his experiment demonstrate that the f0 range is expanded by focus: the high points of H tones are raised while the low points of L tones are lowered. Further, the f0 of all the words following the focus are substantially lowered no matter whether the f0 of the focused words are raised or lowered. In contrast with the on-focus raising and post-focus lowering, the f0 of the pre-focus
syllables are barely changed. Although the existing studies have reported changes in f0 triggered by focus, phonetic and phonological explorations on the Chinese disyllabic words as a whole entities bearing focus, especially on the underlying factors constraining the distribution of these focal accents, have still been far less than those on English focus, and even absent at all. In this regard, we intend to examine the phonetic and phonological property of focal accents conveyed by Chinese disyllabic focused words with exhaustively various tonal combinations, and to offer a description of the effect of focal accents upon f0 by the very means to reduce the tonal combination to tones in terms of H and/or L. We further endeavor to disclose the underlying forces constraining the distribution of the accent patterns.
2 2.1
Methodology Experiment designing
Focused words adopted in the acoustic and perceptual experiments are all disyllabic and with the morphosyntactic structure represented as Modifier-Head 1 , or MH for short. Sixteen kinds of tonal combinations for the focused word of MH structure emerge from the formula of “[Tone1+Tone1] … [Tone4+Tone4]”. And the composing syllables fall into a couple of sets: {Gan1, Tang2, Biao3, Da4} as the set of the first syllables, and {Ge1, Yi2, Jie3, Mei4} the set of the second. A certain focused word is thus constituted by the combination of each member from the first set and each from the second. All these words are set in four patterns of target sentences: 1) Jin1 Tian1 + [focused words] + Fei1 Dong1 Jing1; 2) Jin1 Chen2 + [focused words] + Hui2 Dong1 Jing1; 3) Jin1 Wan3 + [focused words] + Fan3 Dong1 Jing1; 4) Jin1 Ye4 + [focused words] + Qu4 Dong1 Jing1. It has been made apparent in previous studies that focal accents exert certain effect upon f0 of the adjacent syllables. In the present study, however, the adjacent syllables are allocated with the four tones for the purpose of deliberating the effects imposed by the focal accents upon f0 of these adjacent syllables under every tonal circumstance. The tones of the syllables apart from the focused words and the adjacent syllables are set invariably in Tone1. To approach to the different types of focus, we employ guide sentences which are mainly composed of wh-question equivalents. The guide sentence for all broad focus expressions in this article is invariably the interrogative of “Fa1 Sheng1 Le0 Shen2 Me0 Shi4? (What happened?)” And for narrow focus sentences we employ the guide sentences generally formulated as “Jin1 Tian1/Chen2/Wan3/Ye4 Shei2 Fei1/Hui2/Fan3/Qu4 Dong1 Jing1? (Who fly to/go back to/go back to/go to Tokyo today/this morning/this evening/tonight?)”.
1
Phonetic realization of focused words is different due to the different morphosyntactic structure of focused words (Jia, Xiong and Li) [5].
2.2
Experiment procedures
Four Standard Chinese speakers, one female and three males, ranging from 20 to 45 of age, were invited as the subjects. The recording was conducted in the sound-treated booth at the Institute of Linguistics, Chinese Academy of Social Sciences. The subjects were seated comfortably in front of the computer screen, and the microphone was placed a hand away from mouth. During the recording, each sentence appeared on the screen three times in random order; meanwhile, the guide sentences were broadcast and the subject was asked to read aloud the displayed sentences as the response to the question in normal speed without any irregular pause. Sounds were recorded and saved directly into computer through sound recording software as “wav” file. The target sentences in which the clause-middle focused words have 16 kinds of tonal combinations pronounced by a male and a female speaker were chosen for perceptual experiment samples. As has been mentioned above, each target sentence was recorded three times; only one was collected for each target sentence so as to reduce the unnecessarily huge amount of data. Since in this experiment the subjects are only required to give judgments over the patterns of the focused words in terms of Weak or Strong, and hardly any correlation exists between the circumferential tones and the focused words, thus, we render the tones both preceding and following the focused words as being Tone1. Totally three females and five males were invited to participate in the perceptual experiment. They are all standard Chinese speakers and show rather sensitive and stable perception. In order to minimize the negative analogical effect and perceptual fickleness, all the sentences involved were broadcast in sheer random order through perceptual software, and each sentence was perceived by as many repetitions as the subjects required in order to confirm their judgment. Thus, for each focused word we had two target sentences to provide to the subjects for judgment, and finally we got 8×2=16 samples for the word with the same tonal combination. Each subject was asked to finish the experiment individually, without any interpersonal consultation. During the experiment, four options appeared with the icons worded as: “the first syllable strong 2 “, “the second syllable strong”, “both strong” and “both not strong”. The subject was expected to choose one of these options according to their decisive judgment of the weight of the focused words in target sentences. 2.3
Data labeling and extraction
Speech was first labeled by automatic segmentation software, and then the syllable boundaries were modified by hand. Before extracting the data, the manual refinement of the pitch tier was conducted in order to ensure the accuracy of the data. The data were retrieved by praat script with each syllable for 10 points, and the duration of the utterances was normalized. Finally, SPSS10.0 was used to get the means of f0 for each target sentence.
2
Since each focused words in our experiment consists of a couple of syllables, it is sensible to require the subjects to tell which of the two syllables or whether both are strong, or not strong.
3
Phonetic realization of focal accents
In this part, two aspects of the effect of focal accents upon f0 will be discussed: 1) the manner of effect, specifically taken to mean rising or lowering the f0, and 2) the condition of effect, which means whether the lowering or rising imposes the effect on H or L tones. The fundamental unit we apply to describe these effects is the underlying tonal target, H or L; specifically, we render each focused syllable separately to divide it into its original tones and in this manner to observe the performance, rather than deal with a single syllable as the descriptive unit. The following two figures are the mean f0 curves of sentences: “Jin1 Tian1 [Gan1] × [Ge1, Yi2, Jie3, Mei4] Fei1 Dong1 Jing1”, focused words [Gan1] × [Ge1, Yi2, Jie3, Mei4] are under both broad and narrow focus conditions. These sentences are adopted for the comparison among the pitch range changes of the second syllables of the focused words.
Fig.1& 2. f0 means of “Jin1 Tian1 [Gan1]×[Ge1, Yi2, Jie3, Mei4] Fei1 Dong1 Jing1”. [Broad and Narrow Focus Conditions] MH indicates that the morphosyntactic structure of focused word is Modifier-Head, “0” means the focus type is broad and “2” is narrow focus, the following “2” illustrates the position of the focused word is middle, the following “11” demonstrates the tonal combination is Tone1+Tone1 and “12” is Tone1+Tone2, etc, the last “1, 2, 3 and 4” show the tones adjacent to the focused word are Tone1, Tone2, Tone3 and Tone4. Comparison of the two graphs shows that the pitch range of the second syllable is expanded by the focal accent, from the specific values, the pitch range of the broad focus syllable is 104Hz-215Hz and the difference of 215 minus 104 equals to 111Hz, while this difference of the second syllable range under narrow focus condition presents 137 Hz, which demonstrates that the pitch range of the second syllable of the focused words is expanded by the focal accent for 25Hz. The following two graphs read “Jin1 Tian1 [Gan1, Tang2, Biao3, Da4]×[Ge1] Fei1 Dong1 Jing1” under both broad and narrow focus conditions from which the changes of pitch range of the first focused syllables can be seen.
Fig. 3 & 4. f0 means of “Jin1 Tian1 [Gan1, Tang2, Biao3, Da4]×[Ge1] Fei1 Dong1 Jing1”. [Broad and Narrow Focus Conditions] The above two figures illustrate that the pitch ranges of the first syllables are expanded by the focal accent, specifically, the pitch ranges of the first syllables of the broad focus words are within 114Hz-218Hz, and the pitch ranges of the narrow focused syllables are 122Hz-240Hz, meaning that the pith range of the first syllable is enlarged by 14Hz. The above paragraph demonstrates that the pitch range is expanded by focal accents and the internal cause for the pitch range changes is analyzed through the direct comparison of the focused words under different focus conditions in the following figures. The following figure is the mean f0 curves of sentence: “Jin1 Tian1 Gan1 Ge1/Gan Mei4 Fei1 Dong1 Jing1”, focused word “Gan1 Ge1” and “Gan1 Mei4” are under both narrow and broad focus conditions.
Fig. 5 & 6. f0 means of “Jin1 Tian1 Gan1 Ge1/Gan1 Mei4 Fei1 Dong1 Jing1”. Examination of the above two figures reveals that focal accents lift the f0 of the H tones of the two syllables of the focused words while the L tone of the focused syllable resembles that of the broad focused syllable. The realization of focal accents manifests on the two focused syllables differs from English that the raising of the pitch is on the stressed syllable at the sentential level. In general, the manners and conditions of the effect of focal accents upon f0 in under-focus domain can be generalized in three aspects: firstly, the pitch range is expanded by the focal accent; secondly, the essential cause for the expansion of the f0 range is the increase of the f0 of the H tones, which indicates that the effect imposed by focal accents upon pitch mainly manifests itself on the H tones; thirdly, the effect exerted by focal accent on L tones is by no means obvious and systematic; fourthly, the manifestation of the effect of the focal accent on both of the focused syllables characterizes the feature as being language-specific of Chinese. Figure 7 and 8 illustrate the effect of focal accents on f0 range of the syllables in pre-focus and post-focus domain. Target sentences in Figure 7 are under broad focus
atmosphere while Figure 8 under narrow one. The model of target sentences presents: “Jin1 Tian1 (Chen2/Wan3/Ye4)+[Gan1 Ge1]+ Fei1 (Hui2/Fan3/Qu4) Dong1 Jing1”. As the pitch changes imposed by focal accents are under direct discussion, the tones of the focused words are set as Tone1.
Fig.7 & 8. f0 means of “Jin1 Tian1 (Chen2/Wan3/Ye4)+[Gan1 Ge1]+Fei1(Hui2/Fan3/Qu4) Dong1 Jing1”. [Broad and Narrow Focus Conditions] Figure 7 and 8 demonstrate that the pitch range of the syllables following the focused word is compressed remarkably by focal accents while the pitch range of the syllables preceding the focused word remain much the same. The intrinsic cause for the compressing of the pitch range of post-focus syllables can be obtained from Figure 5 and the following two graphs. The follow figures are the f0 curves of “Jin1 Chen2/Ye4 Gan1 Ge1 Hui2/Qu4 Dong1 Jing1”.
Fig.9 & 10. f0 means of “Jin1 Chen2\Ye4 Gan1 Ge1 Hui2\Qu4 Dong1 Jing1” Generally in the above mentioned three graphs have been discussed the phonetic realization of the focused word “Gan1 Ge1” with the differently assigned tones in surrounding. What is unanimously represented among the four figures is that f0 in the focus position behave in the same manner, i.e., the f0 of the H tones are remarkably raised under the effect of the focal accents. Pre-focus f0 of the four tones, irrespective of whether being H or L tones, show either no perceptible difference or only slight disparity that is too erratic to be of any statistical significance. Post-focus f0, however of the H tones of Tone1 and Tone2 are significantly compressed while those of the H tone of Tone4 show a minor height difference over that under broad focus condition, the reason why the H tone of Tone4 are not affected obviously is that the H tone locates itself in the transition part, and contains the L tone. All the L tones are hardly subject to the effect of the focal accents, with only quite negligible changes that also lack statistical importance.
4
Phonological analysis of focal accents
Within the frame of Structure-based FTA 3 theory (Ladd & Gussenhoven)[2][6], the linguistic description of accent patterns involves two complementary but essentially separate aspects: a statement about which parts of an utterance are focused, and a statement about how a given pattern of focus is determined by the location of the accent. They argue that the speaker’s decision about what to focus is subject to all kinds of contextual influences that are difficult to identify. They also notice that once the focused part of the utterance is specified, the accent pattern follows more or less automatically by language-specific rules or structural principles. Therefore, in this part, the accent patterns of focused words at the sentential level with various tonal combinations will be described through the theoretical model of Metrical Phonology [7] based on the data obtained from the perceptual and experiments, further, the language-specific rules for restricting the distribution of the focal accents will be analyzed. Table 1 demonstrates the perceptual results of accent patterns for each focused words. Table 1.
Tonal Combinations
Perceptual results of focused words
Both not strong
Both strong
First Strong
Second Strong
Total
Tone1+Tone1 3 13 16 Tone1+Tone2 2 11 3 16 Tone1+Tone3 1 4 7 4 16 Tone1+Tone4 6 8 2 16 Tone2+Tone1 5 11 16 Tone2+Tone2 6 2 8 16 Tone2+Tone3 2 4 6 4 16 Tone2+Tone4 6 1 9 16 Tone3+Tone1 3 13 16 Tone3+Tone2 1 3 2 10 16 Tone3+Tone3 4 7 5 16 Tone3+Tone4 5 1 10 16 Tone4+Tone1 5 3 8 16 Tone4+Tone2 1 6 9 16 Tone4+Tone3 1 10 5 16 Tone4+Tone4 5 1 10 16 It is can be seen from Table 1 that the first syllables of 8 focused words with the tonal combination of “Tone1+Tone4” were judged by the subjects to be “strong”, which occupies 50% of the total number 16, whereas the “both strong” 37.5% and “second strong” 12.5%. In light of the phonetic realization shown by Figure 6, the accent pattern of “Tone1+Tone4” displays “s-w” relationship. Similar method can be adopted to approach the accent 3
Gussenhoven proposes the “Focus-to-Accent” (FTA) approach. In very general terms, the FTA theory is that words and constituents in utterances can be focused for various reasons, and that focused words and constituents are marked by pitch accents.
patterns of the other focused words with different tonal combinations, specifically, the accent patterns of the focused words of “Tone1+Tone2”, “Tone1+Tone3”, “Tone1+Tone4”, “Tone2+Tone3”, and “Tone3+Tone3” reflect “s-w” relationship; while the tonal combinations of “Tone2+Tone1”, “Tone3+Tone1”, “Tone3+Tone2”, “Tone3+Tone4”, “Tone4+Tone1”, and “Tone4+Tone3” show the opposite accent relationship of “w-s”. The remaining five tonal combinations, “Tone1+Tone1”, “Tone2+Tone2”, “Tone2+Tone4”, “Tone4+Tone2”, and “Tone4+Tone4”, invariably show the “w-s” pattern through the perceptual experiment. The explanation for this strong tendency is to be made in Table 6. The following tables show the accents distribution pattern of 16 different tonal combinations of MH structure from the word level to the sentential level through metrical grids (in the grids below, Line 0 stands for syllabic level, Line 1 for lexical level, and Line 2, sentential, “()” indicates syllable boundary): Table 2. Gan1 Line2 Line1 Line0
× (s)
Phonological representation of [Gan1]×[Ge1, Yi2, Jie3, Mei4] Gan1
Ge1
× × (s)
Line2 Line1 Line0
× × (s)
Yi2
× (s)
Gan1 Line2 Line1 Line0
× × (s)
Jie3
× (s)
Line2 Line1 Line0
Gan1
Mei4
× × (s)
× (s)
Lin [8] investigates the stress pattern of Mandarin disyllabic words in citation forms through perceptual experiment, and proposes that when disyllabic words are in normal stress, there is no absolute fixed stress pattern. This indicates that each syllable of MH structure words can bear word stress in citation forms. As can be seen from the above table, the sentential accent dwells on the second syllable “Ge1” in “Gan1 Ge1”, and the accents dwell on the first syllable in “Gan1 Yi2”, the first syllable in “Gan1 Jie3”, and the first syllable in “Gan1 Mei4”. Table 3.
Line2 Line1 Line0
Tang2
Ge1
× (s)
× × (s)
Phonological representation of [Tang2]×[Ge1, Yi2, Jie3, Mei4]
Line2 Line1 Line0
Tang2
Yi2
× (s)
× × (s)
Tang2 Line2 Line1 Line0
× × (s)
Jie3
× (s)
Line2 Line1 Line0
Tang2
Mei4
× (s)
× × (s)
And the table above shows that the syllables on which the sentential accents occur are “Ge1” in “Tang2 Ge1”, “Yi2” in “Tang2 Yi2”, “Tang2” in “Tang2 Jie3”, and “Mei4” in “Tang2 Mei4”. Table 4. Biao3 Line2 Line1 Line0
× (s)
Phonological representation of [Biao3]×[Ge1, Yi2, Jie3, Mei4]
Ge1
× × (s)
Biao3 Line2 Line1 Line0
× (s)
Yi2
× × (s)
Biao3 Line2 Line1 Line0
× × (s)
Jie3
× (s)
Line2 Line1 Line0
Biao3
Mei4
× (s)
× × (s)
From the above table can be seen that “Ge1” in “Biao3 Ge1”, “Yi2” in “Biao3 Yi2”, “Biao3” in “Biao3 Jie3”, and “Mei4” in “Biao3 Mei4” are the four syllables on which the sentential accents are located. Table 5.
Line2 Line1 Line0
Da4
Ge1
× (s)
× × (s)
Phonological representation of [Da4]×[Ge1, Yi2, Jie3, Mei4] Da4 Line2 Line1 Line0
× (s)
Da4
Yi2
× × (s)
Line2 Line1 Line0
× × (s)
Jie3
× (s)
Line2 Line1 Line0
Da4
Mei4
× (s)
× × (s)
And this table shows that the focal accents occupy the four positions of “Ge1” in “Da4 Ge1”, “Yi2” in “Da4 Yi2”, “Da4” in “Da4 Jie3”, and “Mei4” in “Da4 Mei4”. We would like to explore the more unified explanation for the underlying causes for the shifting of the stress and the specific distribution patterns shown by the tables above. Sharing the very morphosyntactic structure of MH, the accent distribution patterns of these words, however, vary with the various tonal combinations. The pair of “Tang2 Ge1” and “Tang2 Jie3”, for instance, gives the accent patterns of “strong-final” and “strong-initial” respectively. The cause for the shifting of the stress from the first syllables to the second at the sentential level can seemingly be reduced to the simple fact that with the first syllable being the same in both words the only difference between them lies in the tones of the second syllables, Tone1 and Tone3 respectively. The accent distribution of above minimal pair of “Tang2 Ge1” and “Tang2 Jie3” implies that the explanations can be sought if the traditionally termed four tones have been reduced to the original tones marked duly by the very permutations of H and L, that is, Tone1 is transformed into HH, Tone2 LH, Tone3 LL, and Tone4, HL. Concretely, therefore, in “Tang2 Ge1” the syllable “Ge1” claims the accent just because “Ge1” can be re-rendered as “HH”, compared with “Tang2” as “LH”. “HH” overtakes “LH” in number of “H” and therefore appears strong. Similar situation holds true for “Tang2 Jie3” in which “H” in the “LH” pattern of “Tang2” outnumbers that of “LL” of “Jie3” and thus claims the accent. This correlation can be attested for the patterns of the other words listed below: “Strong-initial” patterns are found for “Gan1 Yi2”, “Gan1 Jie3”, “Gan1 Mei4” and “Da4 Jie3” because they can be re-written respectively as “HH LH”, “HH LL”, “HH HL” and “HL LL”; another “Strong-initial” pattern notably exists for “Biao3 Jie3”, which can be attributed to the tone sandhi of “Biao3 Jie3 → Biao2 Jie3”, that is, “LH LL”. Correspondingly, “Strong-final” patterns occur for “Biao3 Ge1”, “Biao3 Yi2”, “Biao3 Mei4” and “Da4 Ge1”, with the respective renderings of “LL HH”, “LL LH”, “LL HL” and “HL HH”. However, a seemingly disturbing case emerges for five words in each of which the combination pattern in terms of “H/L” for the first syllables is totally identical in number of H to the corresponding second ones, as in “Gan1 Ge1” re-rendered as “HH HH”, “Tang2 Yi2” as “LH LH”, “Da4 Mei4” as “HL HL”, “Tang2 Mei4” as “LH HL”, and “Da4 Yi2” as “HL LH”. Solely from the perspective of “H/L” pattern the above five words should all have the pattern in which the first and the second syllables are equally strong. The fact both revealed by the perceptual experiment as well as
phonetic graphs 4 , however, turns out to be that it is the second syllables of the five words that bear the sentential accents and they fall into the “strong-final” pattern. Therefore, by permutation, then, three patterns of contrast in terms of the amounts of H tones come into being. And they are, of the two syllables: 1) the first with more H tones, 2) the second with more H tones, and 3) each containing equally numbered H tones. The result of “more Hs claiming strong” can be explained through the phonetic analysis of focal accents that the realization of the focal accents mainly manifested on the H tones of the focused syllables. What, then, accounts for the pattern with “H/L” pattern being equal? Pierrehumbert [9] proposes that when two stressed syllables sound equal in pitch, the second is actually lower and when the perception of pitch is equal in height the second syllable bears more stress. In view of this, the height of the pitch of the H tones is the crucial factor accounting for the accent patterns of the word with equal number of H tones. The following table provides the mean maximum pitch values of the H tones of the focused words under narrow focus condition with the tonal combinations of Tone1+Tone1, Tone2+Tone2, Tone2+Tone4, Tone4+Tone2 and Tone4+Tone4 (H1 indicates the mean pitch values of the H tones of the first focused syllable and H2 is the second, the unit for these values is Hz). Table 6. Tonal Combination T1+T1 T2+T2 T2+T4 T4+T2 T4+T4
Maximum pitch values of H tones
Tonal Feature HH + HH LH + LH LH + HL HL + LH HL + HL
H1 (Hz)
H2 (Hz)
237 222 179 244 242
242 231 265 242 240
The extraction of the above values of f0 under the circumstances of focus is achieved by smoothing off the influences exerted by the tones preceding and following the focused words, and thus by obtaining the means of the 16 focused words, each of which contains 48 samples. The puzzling fact can be explained from the above values that the mean maximum pitch values of the second focused syllable are higher than the first ones in the tonal combinations of Tone1+Tone1, Tone2+Tone2 and Tone2+Tone4. And in the tonal combinations of Tone4+Tone2 and Tone4+Tone4 the heights of the H tones of the two syllables nearly equals each other. All these data demonstrate that the second syllables of the focused words of these tonal combinations bear stronger accent. Therefore, in this part, we provide the phonological description of focused word with various tonal combinations through the theoretical model of Metrical Phonology. From the representation of the accent pattern of these focused words we offer the unified explanation for the underlying causes for the distribution of the accents by the means of re-rendering the tones of each syllable of focused word into the original H/L tones or the combinations of these two tones. When the underlying tonal combinations are different (this difference lies in the tonal combinations of each syllable between the focused words), the determining role of restricting the 4
As has been discussed above, the raising of f0 is mainly manifested on the H tones.
distribution of the accents is attributed to the amount of the H tones; however, when the underlying tonal combinations are identical, a predominant tendency exists of being always “strong-final”. These results strongly indicate that the phonological composition of the focused syllable is the primary cause for restricting the distribution of the accents. 5
Conclusion
In this study, an integrated work of phonetic and phonological analysis for focal accents conveyed by disyllabic words in Standard Chinese has largely been done. Major findings are achieved from all the above analyses are expressed in both the phonetic and phonological aspects: for the phonetic perspective we have: 1) in the focus position under narrow focus condition, the pitch range of f0 is enlarged as the H tones under narrow focus condition is significantly higher than under broad focus condition, and L tones show on obvious differences in f0 under narrow focus condition, if there were, from those under broad focus condition, but the differences are by no means systematic and worthy of note. This finding, however, differs from that made by Yi Xu [4], who argues that under focus conditions f0 of H tones becomes higher and of L tones, lower. 2) In the positions preceding the focus, f0 range remains much the same, whether H tones or L tones, f0 is but slightly subject to the effect exerted by the focus, and this finding bears no considerable divergence from that of Xu [4]. 3) Extremely significant effects imposed by the focal accents are found on the post-focus pitch range of syllables, which slightly but notably differs from the finding made by Xu [4] that the pitch range of f0 of the post-focus syllables is considerably compressed by the focus. Our study, however, offers a more refined discovery that the focal accents do lower the f0 of the H tones of Tone1 and Tone2, but exert only insignificant effects on the f0 of the L tones of Tone2, Tone3 and Tone4 and of the H tones of Tone4. For phonological perspective, the underlying causes for the distribution of the focal accent are explored through the phonological description of the accent patterns of the focused words. The manifestation and distribution of focal accents are closely related with the underlying tonal target of the focused words that the realization of the tones is influenced by accents and the phonological composition of the focused constituents determines the distribution of the focal accents. We would therefore agree with Ladd [3] that once focus is identified in accordance with the speaker’s intent and with the context, it is the language-specific structural factors that determine the accent distribution. In Standard Chinese, the phonological combination of the focused constituents is the primary causes that determine the distribution of the focal accents, this analysis also identical with Chomsky [10] that phonological characteristics determine the nature of the accents. 6
Acknowledgements
This research is supported by CASS key projects “Fundamental Speech Corpus and Pitch Pattern of Standard Chinese” YZDN50-05050 and “Research on Phonetics and Spontaneous Speech Processing”.
References 1．Xu, Yi. 2005. Phonetic realization of focus in English declarative intonation. Journal of Phonetics. 2．Ladd, D. Robert. 1980. The structure of Intonational Meaning: evidence from English. Bloomington, Indiana: Indiana University Press. 3．Ladd, D. Robert. 1996. Intonational phonology. Cambridge University Press. 4．Xu, Yi. 1999. Effects of tone and focus on the formation and alignment of F0 contours. Journal of Phonetics. 5．Jia, Yuan, Xiong Ziyu, Li Aijun. 2006. Narrow focus patterns of disyllabic words with different morphosyntactic structures in Standard Chinese. TAL 2006 LA ROCHELLE (France). 6．Gussenhoven, Carlos. 1983. Focus, mode and nucleus. Journal of Linguistics 19:377-417. 7．Liberman, Mark and Alan Prince. 1977. On the Stress and Linguistic Rhythm. LI 8:249-336. 8．Lin, Maocan. 1984. Qingzhong Yin. In Wu Zongji and Lin Maocan (eds.), Shiyan Yuyinxue Gaiyao (An introduction to experimental phonetics). Beijing: Commercial Press. 9. Pierrehumbert, Janet. 1979. The perception of fundamental frequency declination. Journal of the Acoustical Society of America, 66(2): 363-369. 10．Chomsky, Noam. 1971. Deep structure, surface structure, and semantic interpretation. In D. Steinberd and L. Jakobovits (eds.) Semantics-an Interdisciplinary Reader in Philosophy, Linguistic and Psychology, 183-216. Cambridge: Cambridge University Press.
