Читать книгу Stress Variation in English - Alexander Tokar - Страница 13

Finally, we briefly discuss differences between stressed and unstressed syllables without specifically considering whether the former bear not only word but also phrasal stress. According to van der Hulstvan der Hulst (2010b: 9), the prototypical characteristics of a stressed syllable are 1) greater durationduration, 2) a balanced spectral tilt, 3) a higher fundamental frequencyfundamental frequency/pitchpitch, 4) greater precision or extra phonetic traits (such as full vowel quality, aspirationaspiration), 5) extra phonotactic possibilities/greater complexity, 6) marking sites for morphological processes, and 7) serving as an anchor for intonational tones.

Indeed, as already mentioned in the previous sections, syllables bearing stress are longer and louder (than unstressed ones) and, if they bear not only word but also phrasal stress, also higher-pitched.

What is meant by “a balanced spectral tilt” is “a more even intensityintensity distribution across the frequency spectrum for accented vowels” (van der Hulstvan der Hulst 2010b: 4). This is because “the vowel of an accented syllable is more precisely articulated (less-centralized) and/or longer” (van der Hulst 2010b: 4). E.g., Kent & NetsellKent & Netsell (1971) studied the production of the words convoy, commune, compact, escort, impact, increase, insight (incite), insult, protest, and suspect, which, depending upon the location of stress in them, function as instances of different morphosyntactic categories. The central finding is that the production of stressed syllables is characterized by “a relative lowering of the jaw and a greater opening of the lips” as well as the “[displacement of] the tongue points in the direction of the presumed articulatory target” (Kent & Netsell 1971: 33, 36). With regard to the latter, Kent & Netsell (1971: 36) note that, for example, “the tongue-marker positions during stressed /i/ are displaced upward and forward relative to the tongue-marker positions during unstressed /i/.” In other words, the stressed /i/ is higher and fronter than /i/ that occurs in non-stressed positions.

In a more recent article GayGay (1978: 348) also mentioned a greater electromyographic activityelectromyographic activity (i.e., the activity of the muscles) that underlies the production of stressed vowels, and even more recently, Mücke & GriceMücke & Grice (2014: 49) have observed that “[a]rticulation is more distinct in prominent positions, resulting in temporal and/or spatial expansiontemporal and/or spatial expansion of articulatory movements.” The two strategies employed are sonority expansionsonority expansion and hyperarticulationhyperarticulation:

The Sonority Expansion Hypothesis claims that the intrinsic sonority of a vowel is enhanced to strengthen syntagmatic contrasts, i.e. the contrast between vowels and consonants. Under accent, the speaker intends to produce a louder vowel by opening the mouth wider over a longer time. A more open oral cavity allows for greater radiation from the lips, leading to an increase of the overall acoustic energy. […]

The Strategy of Hyperarticulation involves the enhancement of contrastive features as well as of sonority. A low vowel is produced with a lower tongue position, a front vowel with a more fronted tongue position […] and a back vowel with a more retracted tongue position (Mücke & GriceMücke & Grice 2014: 49)

Of note is also an investigation by Scarborough et al.Scarborough et al. (2009), which deals with the so-called “optical phonetics” of both word and phrasal stress in English. Similar to earlier contributions, Scarborough et al. (2009) have studied the production of forms that, depending upon their stress contours, function as instances of different morphosyntactic categories: discharge, discount, pervert, and subject. Scarborough et al. (2009) have addressed the question of whether the production of syllables receiving either word or phrasal stress differs from the production of unstressed/unaccented syllables with regard to the following parameters: eyebrow displacement, head displacement, lip distance, lip displacement in opening and closing gestures, chin displacement in opening and closing gestures, lip opening and lip closing velocities, and chin opening and chin closing velocities (Scarborough et al. 2009: 145).

As far as word stress is concerned, syllables bearing it have turned out to be different from unstressed ones with regard to the parameters of head displacement, lip distance, lip opening displacement, lip opening velocity, chin opening displacement, and chin opening velocity: These measures are all larger and faster in stressed syllables (Scarborough et al.Scarborough et al. 2009: 146). The parameters of lip closing displacement, lip closing velocity, chin closing displacement, and chin closing velocity exhibit the same pattern, but as the authors report, they fail to achieve statistical significance (Scarborough et al. 2009: 146–147). The only parameter that apparently has nothing to do with the difference between stressed and unstressed syllables is eyebrow displacement: As Scarborough et al. (2009: 147) observe, “the eyebrows showed no movement at all.”

As for phrasal stress, each of the parameters named above has achieved statistical significance, i.e., the measures for them are larger and faster in accented syllables. Of particular interest are the results for eyebrow and head displacement. According to Scarborough et al.Scarborough et al. (2009: 151–152), these parameters

differ from the others in that they are not directly associated with segmental articulation. Therefore, their movements could be purely prosodic, free to cue stress without being constrained by segmental context. As noted above, the typical movement of the head associated with stress was in a downward direction, while brow movements were typically upward. From our qualitative observation of the tokens, these movements were coordinated such that the eyebrows (or at least one of them) rose just before the stressed word and fell slowly during the word, and the head fell, or nodded, beginning during the closure of the initial stop in stressed words […] (Scarborough et al.Scarborough et al. 2009: 151–152)

To conclude, Scarborough et al.Scarborough et al.’s (2009) results indicate that not only is the optical phonetics different for stressed and unstressed syllables as well as for accented and non-accented words. In addition, stressed and accented syllables differ from each other with regard to eyebrow movements. These play an important role in the production of accented syllables, but they play no role at all when a stressed syllable is produced.

The claim that stressed units are generally produced more intensively than unstressed ones seems to be true not only of spoken but also of sign languages. For example, a study by Wilbur & SchickWilbur & Schick (1987) addressed the articulatory nature of stress in American Sign Language. Since it is primarily a monosyllabic language, Wilbur & Schick (1987: 303) were concerned with the production of stress at the sentence level. According to the authors, signers who participated in their study

[…] frequently used strong facial features, such as brow furrowing, eyebrow raising, puffed cheeks, labial plosions, tongue thrusts, or clenched teeth, to indicate that a particular sign or phrase was stressed. […]

In addition to facial behavior, in many stressed signs, subjects would use body shifting, head movement and tilting, and body and head jerks (body shifts) to add emphasis to a sign or its phrasal group. (Wilbur & SchickWilbur & Schick 1987: 310)

Note also that reduction of vowels in unstressed positions, which is so typical of the English language, can likewise be attributed to production differences between stressed and unstressed syllables. Because stressed syllables are longer than unstressed ones,

[…] for destressed (or fast) speech, commands for successive articulatory movements are issued at a rate that is too fast for the targets to be attained. The result is articulatory (and acoustic) undershoot, the extent of which is directly proportional to the durationduration of the vowel. In other words, spectral reduction is a natural consequence of a decrease in vowel duration. (GayGay 1978: 348)

With regard to marking sites for morphological processes, consider the following three transcriptions: /ˈti:ni/, /ˈmɑrɾi/, and /mɑr/. According to Carter & ClopperCarter & Clopper (2002: 333), these are three possible American English shortenings of the noun martini. The obvious structural differences between them are as follows. In the variant /ˈti:ni/ the stressed syllable /ˈti:/ is the preserved stressed syllable /ˈtiː/ of the base form martini. In the variant /ˈmɑrɾi/ the original stressed syllable appears in a reduced, unstressed form /ɾi/, with the original /t/ being replaced by a flap. Finally, the shorteningshortening /mɑr/ does not contain the original stressed syllable /ˈtiː/ in any form. As the production experiment by Carter & Clopper (2002: 334) has demonstrated, of these three clipping possibilities, the first possibility, exemplified by /ˈti:ni/, has more often been made use of than the other two possibilities, exemplified by the forms /ˈmɑrɾi/ and /mɑr/. In other words, when English speakers shorten words, they tend to preserve the original stressed syllables of longer input forms as stressed syllables of shorter output forms. Further examples from Carter & Clopper (2002: 321–322) include Bécca for Rebécca, cáuse for becáuse, ráffe for giráffe, nána for banána, etc. Examples like these provide a good illustration of the fact that stress can be an important factor determining the outcome of a morphological process such as shortening, or as van der Hulstvan der Hulst (2010b: 9) notes, stress can “mark sites for morphological processes.”

As for the last characteristic—serving as an anchor for intonational tones—recall once again that

the big role of (changes in) fundamental frequencyfundamental frequency results from the fact that the pitchpitch properties of ‘stressed’ syllables are due to the fact that the examined words are uttered in isolation. This causes their ‘stressed’ syllable to be bearers of intonational pitch movement. (van der Hulstvan der Hulst 2012: 1508)

Thus, because in the English language, a falling tonetone is usually “associated with finality, completeness, and definiteness” (GutGut 2009: 122), an individual English word cited in isolation (e.g., in online dictionaries, in which a standard pronunciation of a word can often be listened to) is as a rule pronounced with a high-falling tonehigh-falling tone (CruttendenCruttenden 2008: 282); the stressed syllable of such a word is therefore higher-pitched than the other syllables occurring in the same word.

In addition to the aforementioned prototypical features of stressedness, provided by van der Hulstvan der Hulst (2010b: 9), for purely practical purposes of determining which syllable in a word (that occurs in a corpus of spoken data) bears stress, any researcher can also rely upon his or her own auditory impressions of some syllable in the word under analysis being more prominent than other syllables constituting the same word. At first glance this may seem fairly subjective, but as HayesHayes (1995: 22) observes:

The relevant facts of segmental phonetics and pitchpitch contours are clear to anyone with a reasonably good ear and a little practice. Since the auditory data involve discrete patterns rather than physical quantities, they are more reliable and easier to interpret. (HayesHayes 1995: 22)

The assertion that “the auditory data […] are more reliable and easier to interpret” is, among other things, justified by the parasitic nature of stressparasitic nature of stress, which means that “it invokes phonetic resources that serve other phonological ends” (HayesHayes 1995: 7). Consider, for instance, the role of vowel durationduration as a cue of word stress. As TorsuevTorsuev (1960) demonstrates, durational differences between English vowels can, apart from the presence/absence of stress, occur due to several other reasons. For example, the fact that in pairs of words such as bead–bid, pool–pull, naught–not, etc., the phonetically very similar vowels /iː/ and /ɪ/, /uː/ and /ʊ/, /ɔː/ and /ɒ/ exhibit durational differences has obviously nothing to do with the location of stress in these monosyllabic words. It is simply a peculiarity of the English language that it has vowel contrasts such as /iː/ vs. /ɪ/, /uː/ vs. /ʊ/, /ɔː/ vs. /ɒ/, accounting thereby for a semantic difference between the words bead and bid, pool and pull, and naught and not; according to Torsuev (1960: 45), in cases like these we are dealing with durational differences due to the historical length of English vowels. Another well-known fact is that vowels in English that are immediately followed by voiced consonants are longer than those that are followed by voiceless ones (which is also completely unrelated to stress): E.g., the length of the short vowel /ɪ/ of bid, where the following sound is the voiced /b/, can, according to Torsuev (1960: 45), be identical to or even exceed the length of the long vowel /iː/ of beat, where it is followed by the voiceless /t/; similarly, the long vowel /iː/ of see is longer than the same vowel in seat and seen because /siː/ is an open syllable, whereas both /siːt/ and /siːn/ end in consonants (Torsuev 1960: 45).

Likewise, reliance of stress on loudnessloudness as one of its cues is parasitic in HayesHayes’ (1995: 7) sense because in phonology the primary function of loudness is the organization of segments into syllables. Thus there is the famous Sonority Sequencing PrincipleSonority Sequencing Principle (see, e.g., GandourGandour 1998: 210; HallHall 2006: 330; YavaşYavaş 2011: 142; ColinaColina 2012: 137), which explains why English has words such as plate, blame, clue, glue, fly, sly but not *lpate, *lbame, *lkue, *lgue, *lfy, *lsy. The latter are illicit because the Sonority Sequencing Principle requires that onsets that are made up of multiple segments be characterized by an increasing sonority. That is, the following sound must be more sonorous (i.e., louder) than the preceding one. The words plate, blame, clue, glue, fly, sly are in compliance with this requirement because the following sound /l/ is more sonorous than the preceding sounds /p/, /b/, /k/, /g/, /f/, /s/. In the asterisked words, by contrast, the more sonorous sound /l/ is followed by the less sonorous /p/, /b/, /k/, /g/, /f/, /s/, which violates the Sonority Sequencing Principle. What matters, however, in connection with stress is that just like vowels may exhibit a particular durationduration irrespective of whether they occur in stressed syllables or not, so is every sound characterized by a particular degree of loudness irrespective of whether the syllable in which that sound occurs bears stress. According to Yavaş (2011: 135), the two main factors that determine the intrinsic sonority of a sound are the vibration of the vocal folds (voicing) and the degree of opening of the vocal tract during the articulation of that sound. The loudest sounds are obviously low vowels, such as, e.g., /æ/, /ɑ:/, /ɒ/ (Yavaş 2011: 135). Their articulation involves the maximum opening of the vocal tract and like all vowels, they are voiced. The least sonorous sounds are voiceless plosives, such as /p/, /t/, /k/ (Yavaş 2011: 136). Their articulation involves a complete obstruction to the airflow, and, additionally, the vocal folds do not vibrate.

Finally, the parasitic nature of reliance on pitchpitch as a cue of stress is perhaps even more obvious because apart from its association with stress, pitch is to a much larger extent associated with intonation, one of whose aspects is tonetone. Traditionally, tones are classified into falling tones, rising tones, and combinations of falling and rising tones. (A more refined classification will also distinguish between low and high falling and rising tones). As pointed out above, a falling tone in English is “associated with finality, completeness, and definiteness” (GutGut 2009: 122), and this is precisely the reason why such tones occur in words cited in isolation (CruttendenCruttenden 2008: 282).

To summarize, every syllable is characterized by a particular pitchpitch height, a particular durationduration, and a particular degree of loudnessloudness. These can be measured relatively easily with a speech analysis software, such as Praat (Boersma & Weenink 2014). But as the values that syllables exhibit with respect to these parameters might be caused by things other than stress, stress contours, as HayesHayes (1995: 8) argues, cannot be “read off” from the phonetic record. That is, when we measure syllables’ fundamental frequencies, durations, and intensities, the values we obtain reflect not only the syllables’ stress contours, but also things such as the historical/combinatorial/syllabic lengthhistorical/combinatorial/syllabic length of English vowels, the intrinsic loudness of segments constituting a syllable, or the intonational patterns employed by a speaker when uttering a particular word containing a particular syllable, etc. The latter factors do of course also contribute to the acoustic effect of the stressedness/stresslessness of a syllable, but they are, strictly speaking, not directly related to stress.