Intonation and Speech Act Type

March 26, 2018 | Author: joe1974 | Category: Tone (Linguistics), Syllable, Oral Communication, Human Voice, Communication


Comments



Description

Journal of Pragmatics North-Holland11(1987) 483494 483 INTONATION An Experimental AND SPEECH ACT TYPE Approach to Rising Intonation in Queclaratives Ronald GELUYKENS * It is hypothesized that the role of rising intonation for the recognition of so-called ‘ Queclaratives’ is overrated. Via a perception experiment, it is shown that pragmatic factors, more specifically Searle’ s (1969) ‘ felicity conditions’ , play a decisive role in the recognition of an utterance with declarative form as a question. Provided pragmatic cues are sufficiently strong to determine speech act status, rising intonation is shown to be virtually without impact; if, on the other hand, pragmatic cues do not favor any particular speech act type, intonation may, but need not, act as a cue for determining question-status. 1. Introduction The present study investigates more closely the relevance of rising intonation contours for the perception of so-called ‘ queclaratives’ (Sadock (1974)) in English. Queclaratives are defined here as utterances having the form of a declarative sentence but functioning as requests for information. Previous research (see Geluykens (1985)) has led us to believe that the link between rising intonation and the question-status of an utterance is less systematic than might be expected. In a corpus’ of 60 queclaratives, only 33% were found to have an intonation contour ending with a rise in pitch; the vast majority (63%) ended in a falling pitch. Nevertheless, in quite a number of descriptions of British English intonation - notably Kingdon (1958) and O’ Connor and Arnold (1961) - it is claimed that the use of certain rising contours can tranform a declarative sentence into a question. Though most authors admit the possibility of a falling contour on a queclarative (e.g. Halliday (1970: 27)) the implication that rising intonation is more ‘ neutral’ is often present. * Thanks are due to Rent: Collier, David Crystal and an anonymous Journal of Pragmatics referee for comments on earlier versions of this paper, and to Hans ‘ t Hart and Jan Roelof De Pijper (IPO, Eindhoven) for their assistance in preparing the data. The test was carried out at Reading University with financial assistance from the British Council. Author’ s address: R. Geluykens, Groeningerstraat 4, 2210 Borsbeek, Belgium, and Trinity College, Cambridge CB2 ITQ, UK. 1 The corpus used was the Survey of English Usage (University College London). We are indebted to its Director, Sidney Greenbaum, for making these data available. 037%2166/87/$3.50 0 1987, Elsevier Science Publishers B.V. (North-Holland) Uldall (1962) notes a tendency to interpret utterances with a contour ending high in the range as being questions.1.5) see also Simon and Fourcin (1976). Take for instance Searle’ s (1969) Preparatory Rule ‘ speaker does not know the answer’ . The utterance is therefore likely to be interpreted as a genuine request for information. ‘ high rise’ and ‘ fall plus rise’ . whenever it is uttered. at least in short (one-word) utterances. observes a correlation between rises and questions. but she acknowledges the importance of factors such as voice-quality. irrespective of its intonation contour). Geluykens / Rising intonation and gueclaratives Experimental studies on this issue are few and far between. they are devised in such a way that minimal changes in the lexical material have maximal pragmatic effects. The explanation for the question-proneness can be found in the fact that the ‘ felicity conditions’ for ‘ directives’ (Searle (1969)) are being met. The present experiment tries to fill that gap. It is stated by HaddingKoch and Studdert-Kennedy (1964) that “listeners may make use of the entire F” contour” (1964: 181) in deciding on the question-status of an utterance.484 R. We will argue that the same holds. consider the sentence ‘ You feel ill’ . despite its declarative form (and. 2. Fourcin (197. since the verb ‘ feel’ refers to the addressee’ s mental state. We will label this type of utterance ‘ question-prone’ . The pragmatic variable In order to change the pragmatic implications of our test utterances in the simplest possible way. the preparatory rule will not be violated. As an example. Bolinger (195 1) reports on the distinction between ‘ low rise’ . provided other factors do not block such an interpretation. The same kind of reasoning holds for the other felicity conditions. We will start from the following basic hypothesis: Cues of a pragmatic-contextual nature are the determining factor in recognizing an utterance with declarative form as a queclarative. for question-recognition. by and large. It is obvious that. finally. Schaffer. no attempt has ever been made to investigate systematically the possible relationship between pragmatic and prosodic factors in the recognition of questions. gesture and facial expression. the use of a rising intonation contour does not necessarily turn a declarative utterance into a question. tempo. concludes that “there are few consistent relationships holding between the intonation characteristics present in an item and what kind of judgment it receives” (1983: 253). writing on the role of intonation in the turn taking system. . Procedure 2. to which the speaker has no direct access. it is more likely to be interpreted as a question. To our knowledge. as we will try to show. as it is very unlikely that someone would request information concerning his/her own mental state. Y2. the contours generated by this model have been successfully tested as regards their ‘ naturalness’ . is decidedly question-prone. Of each of the resulting three types. 3 The data were prepared at the Institute for Perception Research (IPO). in (3) what is involved declaration of intention concerning reasoning applies here. Since the verb ‘ feel’ in this instance refers to the mental state of the speaker rather than of the addressee. Geluykens / Rising intonation and queclaratives 485 The utterance ‘ I feel ill’ . H3) (3) I’ I/you/he like(s) the apples a & (14. in the following manner. For this purpose. but a some future action.4 * Underscoring: & lines indicate Tonic position. Eindhoven. interrupted lines indicate secondary (the latter is relevant for contour 5 only). Use was made of an analysis-resynthesis procedure. is not so much a mental state. H4) (4) (15. H2) m/you’ re/he’ s going to the &atre tomoxrow (13. cf. we can effect large changes in pragmatic implications. H5) (5) I/you/he do(es) not like Italians Note that. This utterance is thus ‘ statement-prone’ rather than ‘ question-prone’ . respectively) (2) I’d/you’d/he’d love to see the girJ (12. Y4. The same line of statement-prone owing to the fact that about one’ s own intentions. Y5. Y3. stylized intonation contours were used to replace those. five token utterances were recorded. First of all. Tonic . it appears from our examples. 2. We have therefore constructed our test utterances along the same lines. The intonational variable3 to as 11. Much more likely. The recordings were then analyzed and re-synthesized with elimination of the original F” values. is pragmatically totally different from the first one. the utterances were recorded. De Pijper (1983). Through a simple change in pronoun. use was made of the model for British English intonation devised by De Pijper (1983). the felicity conditions for directives are not met. 4 For more details. on the other hand.R. on the Five types of intonation contour were superimposed on the test utterances. Contours used are represented schematically in figure 1. though: 13 is one does not normally ask questions other hand.2. adding utterances of the type ‘ He feels ill’ as supposedly ‘ neutral’ cases without any special bias towards questionstatus or statement-status. s/he is merely conveying this piece of information to the addressee. spoken in a monotonous voice and with careful avoidance of any ‘ abnormal’ loudness and duration features. yielding 15 test utterances in all:* (1) And I’ m/you’ re/he’ s not feeling very wvll (henceforth referred Yl and Hl. Y3. In the test utterances. 3. Contours 2 and 3 are both rising tones. changing into a steep half rise to high about 80 msec before the end of voicing. to ensure good comprehension. a steep rise to high sets in. Contour 5. secondly. yielding 75 test utterances in all. with one Australian exception. and falls to low on the Tonic (i. the pretonic remains high rather than falling gradually.486 R.e. subjects were . Contour 3 differs from contour 2 in two respects. The experiment Fifteen subjects were asked to participate in the experiment. followed by a fall starting about 80 msec after the Tonic vowel onset. In contour 2. Tonics occurring at 710 msec for the shortest. The pitch then stays low until the end of the contour. The contour then rises gradually. Eight subjects were female.197O) ‘ tone 1’ . at the Onset. more specifically a fall-rise (Halliday’ s ‘ tone 4’ ). immediately after the Onset there is a gradual fall to mid position in the range. It is simply a combination of contour 1 and the final part of contour 3. The rise is completed at the vowel onset of the Onset syllable. pretonics started in the region of 270400 msec after the beginning of the utterance. viz. Contour 4 is a complex tone. the most prominent syllable). and each utterance was recorded two times. Utterances were recorded in random order. is a compound tone. representing Halliday’ s (1968. as the ‘ Onset’ of the tone unit). 15 with a falling one. a steep rise to high takes place. then rising to high at the start of the pretonic (we will refer to this point. most of them had no previous knowledge about intonation. seven were male. respectively. corresponding to Halliday’ s ‘ tone 2’ and ‘ tone 3’ . There were 2 second intervals between each utterance. all of them native speakers’ of British (more specifically. 1040 msec for the longest utterance. 487). ‘ English’ ) English. The pitch then stays low until the steep final rise occurring on the final syllable. 60 with a rising contour. which ends 40 msec after the Tonic vowel onset. On the Tonic. a fall followed by a rise (‘ tone 13’ ). These 5 countours were combined with the 15 utterances (l)-(5). First of all. finally. The contour starts in the middle of the pitch range. apart from a slight quarter tone dip just before the tonic. Contours used are represented graphically in figure 1 (see p. The pitch then stays high in the range. Geluykens / Rising intonation and queclaratives Contour 1 is a falling tone. It starts at mid level. followed by a gradual fall to mid extending to the vowel onset of the Tonic. 40 msec before the Tonic vowel onset this changes into a steep fall. Total length of the utterances varied from 1660 up to 2020 msec. following Crystal (1969). this fall starts 50 msec after the vowel onset of the Tonic syllable. the first part of the final rise is even less steep. )“ . ‘ more question than statement’ (Q”). in other words: whether he has uttered a statement or a question (. asked not to rewind the tape. Stylized intonation contours. tonic 1. on a four-point scale. you have to decide whether he [the original voice was a male one] has asked you something. Subjects were asked to make judgments concerning the question-status of the test utterances. or simply told you something. Geluykens / Rising intonation and gueclaratives 487 Cl c2 c3 c4 CS onset Fig.R. . The four categories were: ‘ definite question” (represented as Q’ in table 1). . The actual utterances were preceded by the following instructions: “Suppose the person on the tape is talking to you. . and figure 2 (table 2 conflates the four original categories into two. Some of them complained about the lack of time afforded to respond to each individual utterance. did not excuse them from having to decide on its question-status. ‘ definite statement’ (S’ ). The proportion of question-responses for each category of utterances (‘ I’ . the discussion. is based on 60 utterances rather than on the original 75. None of the subjects had any difficulties with the test. ‘ you’ and ‘ he’ . Geluykens / Rising intonation and queclaratives ‘ more statement than question’ (~3”). No ‘ neutral’ category was included. however. this. therefore. they were excluded from our results. respectively) is shown in figure 2 (see p. either opting for question-status (Q’ and Q”) or for statement-status (S’ and S”).488 R. Results are represented in tables 1 and 2. Table 1 Division of responses on four-point scale. In addition. subjects were asked to mark a sentence as ‘ odd’ if it sounded unnatural to them in some way. viz. but this was consciously built into the experiment. Q 11-U with: Cl c2 c3 c4 C5 Total Cl-C5 Yl-Y5 Cl c2 c3 c4 c5 with: 4 13 13 4 6 (7%) (22%) (22%) (7%) (10%) 0 (0%) 0 (0%) 2 (3%) 1 (2%) 2 (3%) 5 (2%) Q” 0 (0%) 10 (17%) 9 (15%) 6 (10%) 1(12%) 32 (12%) s” 6 19 24 17 19 (10%) (32%) (40%) (28%) (32%) s 54 31 25 36 32 (90%) (51%) (42%) (60%) (53%) 85 (28%) 178 (58%) 26 26 19 26 22 (43%) (43%) (32%) (43%) (36%) 11 10 22 20 18 (18%) (17%) (36%) (33%) (30%) 19 11 6 10 14 (32%) (18%) (10%) (17%) (24%) Total Cl-C5 HlLH5 Cl c2 c3 c4 c5 with: 40 (13%) 119 (40%) 81 (27%) 60 (20%) 0 2 2 0 0 (0%) (3%) (3%) (0%) (0%) 1 18 12 12 10 (2%) (30%) (20%) (20%) (17%) 14 25 22 20 26 (24%) (42%) (36%) (33%) (43%) 45 15 24 28 24 (75%) (25%) (40%) (47%) (40%) Total Cl-C5 4 (1%) 53 (18%) 107 (36%) 136 (45%) . ‘ question’ and ‘ statement’ ). which forced the listeners to decide one way or the other. 490). Since utterances I-Y5H5 received a markedly high number of ‘ odd’ judgments. so as to obtain spontaneous reactions. In the Y-utterances. C2 vs. whereas the same holds for a mere 12% of the Iutterances.tailed) tests for the 4-point scale (for more details on tests. Cl vs. C2. The significance level is set at 0. .). The H-category differs significantly from both the I. cf. C3. Question Statement II-15 with: Cl c2 c3 c4 c5 Total Cl-C5 Y l-Y5 Cl c2 c3 c4 c5 with: 30 39 32 30 28 (50%) (65%) (53%) (50%) (47%) 30 21 28 30 32 (50%) (35%) (47%) (50%) (53%) 0 10 11 7 9 (0%) (17%) (18%) (12%) (15%) 60 50 49 53 51 (100%) (83%) (82%) (88%) (85%) 37 (12%) 263 (88%) Total Cl-C5 HlLH5 Cl c2 c3 c4 c5 with: 159 (53%) 141 (47%) 1 20 14 12 10 (2%) (33%) (23%) (20%) (17%) 59 40 46 48 50 (98%) (67%) (77%) (80%) (83%) Total Cl-C5 57 (19%) 243 (8 1%) 4. Results 5 Our results show that the pragmatic variable contributes significantly to the perception of two different speech act types: 53% of the Y-utterances are perceived as being questions. Siegel (1956)). there are no significant differences whatsoever between falling contour Cl and the total of the other. etc. The same holds for Cl compared to every individual rising contour (Cl vs. C3. Geluykens / Rising intonation and queclaratives 489 Table 2 Division of responses on two-point scale.001 throughout. Kolmogorov-Smirnov (2.and the Y-category.R. neither on a two-point nor on a four-point scale. ’ Statistical tests carried out on the material are the following: X2-tests to calculate significance levels on the 2-point scale. rising contours C2 to C5. Likewise. there are no significant differences within the group of rising contours (C2 vs. Y. differences between Cl and each individual rising contour are all significant. Put differently: Rising intonation contributes somewhat to question-recognition. In absolute terms. Geluykens / Rising intonation and queclaratives 90 70 50 30 10 c 1 2 3 4 5 contour (see also table 2).and H-utterances alike). etc. As for H-utterances. Fig. % of question-responses per utterance-type and per contour C4. however. irrespective of the precise form of the rising contour. The situation is very different for the ‘ definite statement’ and ‘ more statement than question’ categories (S’ and S” in table 1). Comparing Cl with the rising contours. The number of question-responses is always slightly higher than is the case for I-utterances. there are no significant differences. Among the rising contours.). the number of question-responses is very low. Put differently: Intonation does not ‘ strengthen’ or ‘ weaken’ the question-status of an utterance (this holds for I-. we . Comparing Q’ and Q” (table 1) yields no significant differences between the ‘ definite question’ and the ‘ more question than statement’ categories.490 R. similar observation as for the I-utterances apply: There are significant differences between falling contour Cl and each rising contour. In the I-utterances. even for rising contours (18% for C3 being the highest ratio). 2. but not significantly so. All this supports strongly our hypothesis that the contour used is irrelevant for question-recognition. on the other hand. In other words. C2 and C5 for the Y-category. Our general conclusion concerning the role of rising intonation in queclaratives may be summed up as follows: . rising intonation does influence the statement-status of an utterance. as there is a marked variation in the number of questionresponses across the I-. We will discuss this in more detail in the next section. Discussion As pointed out in the previous section. other prosodic factors. there is a marked tendency to interpret the utterances as statements rather than questions (the maximum proportion of question-responses being 33% for H-utterances with C2). 5. First. that a number of other factors contribute to an utterance having question-status in actual discourse. loudness and tempo might be relevant.R. whereas there are no significant differences among the rising contours. C4 for the H-category). intonation might be even less important. pragmatic factors are more important than intonational factors. the pragmatic variable proves to be very effective.and H-categories (12%. The scepticism shown by Crystal (1969) and Ladd (1980). as shown by the increase in question-responses. In other words. even with rising contours. Geluykens 1 Rising intonation and queclaratives 491 get significant levels of difference almost throughout (exceptions being Cl vs. since here the pragmatic variable contributes to their question-status. pauses. However. Facial expression and gestures are obvious candidates. which turns out to be not quite as questionprone as expected. there is a high frequency of statement-responses throughout.and H-utterances. one might have expected a higher frequency of question-responses for H-utterances. Y-utterances are the most important test-case. in actual conversation. two observations have to be made. among others. especially for the H-utterances (the supposedly neutral category). even for the Y-category (47%). even here. considering the number of rising contours used in the experiment. respectively). rising intonation does seem to have some effect. The number of question-responses is surprisingly low. As for our hypothesis. As it is. and Cl vs. In I. viz. 53% and 19%. the test exploits only one pragmatic factor. the preceding linguistic context is bound to be important. Results are very striking: The intonation contour appears to be of no importance for recognizing these utterances as questions. it makes statements less ‘ definite’ . such as pitch range. the claim made in the literature that rising intonation is a deciding factor for the recognition of queclaratives lacks any empirical support. Y. Secondly. where there are several pragmatic cues. therefore. this suggests that. This offers indirect support to our hypothesis: If rising intonation were really that important for questioning. concerning the role of intonation is therefore fully justified. It seems highly likely. on the contrary. sir? Could you pass me the salt? Is everything all right. we have not taken into account the possible attitudinal implications of the intonation contours (see Cutler (1977) for further discussion). but need not. Liberman and Sag (1974) for a holistic view). the kind of rising contour is irrelevant. Furthermore. but only the final rise. independent ‘ meaningful elements’ . Once more. C3 (rise). be used to turn such an utterance into a queclarative.and H-utterances). By and large. however. I. rising intonation may. in all three categories (I-. Since the utterances are clearly not interpreted as queclaratives. there are no significant differences between C2 (rise). For all the utterances. C4 (fall-rise) and C5 (fall plus rise). we have noticed a difference in S’ and S” responses as a function of the contour used. resulting in ‘ tentative statements’ . Consider the following conversational exchanges: (6) A: B: (7) A: B: (8) A: B: Is everything all right. provided pragmatic factors contribute to the utterance’ s question-status. we must conclude that the statement-force is simply weakened to some extent by the final rise. You’ re not feeling very well? . This area is clearly under-investigated (but see the work carried out by Sag and Liberman (1975)). for instance. For one thing. Our experiment is limited in other respects as well. and is obviously related to the ‘ openness’ or ‘ non-finality’ meaning of Rises referred to in the literature (see. statements with a rising contour are interpreted as being ‘ less definite’ statements than those with a falling contour. Y. even for the cases where rises do score some effect (viz. Cruttenden (198 1)).492 R. Within the statement-responses. with separate ‘ meanings’ attached to each contour as a whole (cf. This seems to suggest that it is not the entire contour which is relevant. please. it lends support to Bolinger’ s (1982) case for a decomposition of contours into smaller. The kind of rising contour used appears not to make much difference as regards the question-status of an utterance. are too limited to draw any firm conclusions about the ‘ meaning’ of rising intonation. this argues against a ‘ holistic’ approach to intonation. we have not touched on the possible role of intonation in distinguishing genuine questions from ‘ Indirect Speech Acts’ (Searle (1975)) such as requests.and H-). If pragmatic cues fail to make the utterance question-prone. Geluykens / Rising intonation and queclaratives Rising intonation is irrelevant for the recognition of a declarative utterance as a queclarative. I think I’ ll take an aspirin. sir? I could do with some salt. Our data. This seems to be a consistent feature of rising intonation throughout our test. are probably different speech acts. it is not able to overrule the pragmatic cues present in an utterance. requests).H. It can. Dwight L. the most rising inonation can do is to weaken the statement-force to some extent.R. 1951. Bolinger. Intonation and its parts. Our experiment leads to questions concerning the role of intonation in other languages. it would not be realistic to expect simple correlations between prosodic form and pragmatic function. Greenberg. despite their formal differences. Ferguson and E. Language 58: 505-533. more research along the same lines is in order to evaluate the role of prosody in questioning. Alan. However. Cambridge: Cambridge University Press. Intonation: Levels vs. the replies in (7) and (8). CA: Stanford University Press. depite their shared declarative form. In: J. on the basis of the present experiment.A. Comparative studies are. Dwight L. Moravcsik. eds.. Stanford. Dwight L.A. Cruttenden. 1982. Geluykens / Rising intonation and queclaratives 493 It is clear that the replies in (6) and (7) are similar speech acts (viz. With this in mind. For instance. ‘ Intonation across languages’ . Bolinger. Word 7: 199-210. C. it might be worthwhile to investigate contours which start higher than the baseline. pitch range has not been varied in this experiment. attitudinal factors suggest themselves as worthy of more research. Prosodic systems and intonation in English.. Rising intonation does not contribute significantly to the recognition of queclaratives. 1981. Crystal. a notable exception being Bolinger (1978) who pleads for “fuller descriptions of how and when particular intonations of yes-no questions are used” (1978: 503) [Bolinger includes queclaratives in his category ‘ yes-no question’ ]. Universals of human language. Clearly. 6.. Journal of Linguistics 17: 77799. contribute towards question-recognition if pragmatic cues are insufficient for a straightforward interpretation. When strong pragmatic factors are present to favor either the question-status of an utterance or its statement-status. David. however. 1969. Conclusion It has been shown that. however. at least for the rising contours tested here. on the other hand. very scarce. Falls and rises: Meanings and universals. Similar claims about queclaratives as those for English are sometimes made (French is a case in point). configurations. We might add that there is also a need for fuller descriptions of the pragmatic environment in which intonation patterns are used.. On the pragmatic side. all contours rise from the baseline to the upper limit of the pitch range. References Bolinger. 1978. . the role of rising intonation in queclaratives is overestimated. J. 1974. Ambiguity: Question or statement? or “Are you asking me or telling me?“. 59982. and A. Morgan.. John R. and G. eds. (Ms. 1970. Intonation and grammar in British English. Language. 1985. 1968.K.J. O’ Connor. Ronald. Arnold. Fourcin. Elisabeth. Robert Jr.F.. C. Speech acts: An essay in the philosophy of language. Ladd. Chicago Linguistic Society: 416427. Vol. London: Longmans. Simon. Studdert-Kennedy. 1962. and retardation. Bloomington. ed. New York: Academic Press.A. 1980. 1975. Anne. cognitive deficits.J. Phonetica 11: 1755185. Jeremy M.) Hadding-Koch. An experimental study of some intonation contours. 1975. Geluykens / Rising intonation and queclaratives Cutler. The groundwork of English intonation. Sag. In: N. D. 3346.. London: Longmans.. Cambridge: Cambridge University Press.. Dordrecht: Foris Publications. Speech and Hearing. M. John R. Proceedings of the Fourth International Congress on Phonetic Science: 779-783. 1983. New York: Academic Press. ‘ Indirect speech acts’ . Modelling British English intonation.K. 1958.494 R. O’ Connor. 1969. Liberman. Syntax and semantics. Reading University. Kingdon. In: P. The role of intonation as a cue to turn taking in conversation.A. The context-dependence of ‘ intonational meanings’ . Searle. Papers from the Eleventh Regional Meeting. Oxford: Oxford University Press. Halliday. pp. Geluykens. 1977. 1975. Mark and Ivan Sag. London: Butterworths. Searle. ‘ Differences between individual listeners in their comprehension of speech and perception of sound patterns’ . 11. Papers from the Thirteenth Regional Meeting. Uldall. Ivan and Mark Liberman.. A course in spoken English: Intonation. 1961. Schaffer. Fourcin. The intonational disambiguation of indirect speech acts. Jan Roelof. Roger. Cole and J. The intonation of colloquial English.. Papers from the Tenth Regional Meeting.. Journal of Phonetics 11: 243-257. pp. Chicago Linguistic Society: 1044137. New York: Academic Press. Deborah. 1983. S. 1976. De Pijper. Nonparametric statistics for the behavioral sciences. The structure of intonational meaning. Oxford: Oxford University Press. An empirical study into the prosodic feature ‘ rising intonation’ and its relevance for the production and recognition of questions. Halliday. Toward a linguistic theory of speech acts. University College London. and M. . 1974.. Siegel. Speech acts. K.D. Sadock. M. Chicago Linguistic Society: 487-497. ‘ Speech perception in the absence of speech productive ability’ . Prosodic form and discourse function. 1964. 1956. IN: Indiana University Press. A..
Copyright © 2024 DOKUMEN.SITE Inc.