Sentence Intonation Research Articles

Sentence intonation and syllable stress in speech produced during simultaneous communication

This study investigated prosodic variables of syllable stress and intonation contours in contextual speech produced during simultaneous communication (SC). Ten normal-hearing, experienced sign language users were recorded under SC and speech only (SO) conditions speaking a set of sentences containing stressed versus unstressed versions of the same syllables and a set of sentences containing interrogative versus declarative versions of the same words. Results indicated longer sentence durations for SC than SO for all speech materials. Vowel duration and fundamental frequency differences between stressed and unstressed syllables as well as intonation contour differences between declarative and interrogative sentences were essentially the same in both SC and SO conditions. The conclusion that prosodic rules were not violated in SC is consistent with previous research indicating that temporal alterations produced by simultaneous communication do not involve violations of other temporal rules of English speech.

Spectral Distribution of Prosodic Information

Prosodic speech cues for rhythm, stress, and intonation are related primarily to variations in intensity, duration, and fundamental frequency. Because these cues make use of temporal properties of the speech waveform they are likely to be represented broadly across the speech spectrum. In order to determine the relative importance of different frequency regions for the recognition of prosodic cues, identification of four prosodic features, syllable number, syllabic stress, sentence intonation, and phrase boundary location, was evaluated under six filter conditions spanning the range from 200-6100 Hz. Each filter condition had equal articulation index (AI) weights, AI = 0.01; p(C)isolated words approximately equal to 0.40. Results obtained with normally hearing subjects showed that there was an interaction between filter condition and the identification of specific prosodic features. For example, information from high-frequency regions of speech was particularly useful in the identification of syllable number and stress, whereas information from low-frequency regions was helpful in identifying intonation patterns. In spite of these spectral differences, overall listeners performed remarkably well in identifying prosodic patterns, although individual differences were apparent. For some subjects, equivalent levels of performance across the six filter conditions were achieved. These results are discussed in relation to auditory and auditory-visual speech recognition.

Computational modeling of accent and intonation in declarative sentences of Spanish

Although Spanish is commonly known to have stress accent, preliminary study shows that the word accent in Spanish has a clear and consistent manifestation in the voice fundamental frequency contour (henceforth the F0 contour). The present study is aimed at obtaining analytic and quantitative description of the prosodic characteristics of Spanish useful both for basic understanding and for speech synthesis. The speech material consists of a set of 21 declarative sentences of various lengths, designed to examine effects of word accent, syntactic, and discourse structure as well as speaker’s intention. These sentences were read by two native speakers of Castilian Spanish. The F0 contours were extracted by computer and were further analyzed using a quantitative model of the process of F0 contour generation, originally proposed for Japanese [Fujisaki etal., Proc. 7th ICA 3, 133–136 (1971)]. The results indicate that the model applies quite well to F0 contours of Spanish as far as the present material is concerned. An F0 contour can be represented by the sum of phrase components and accent components, the former being correlated to the syntactic structure and the latter to the lexical word accent. Focus is indicated by an increase in the magnitude of the accent component, while speaker’s intention is indicated by the presence of another component occurring toward the end of a phrase or a sentence. These results provide a powerful tool for high-quality text-to-speech synthesis.

On the intonation of sinusoidal sentences: contour and pitch height.

A sinusoidal replica of a sentence evokes a clear impression of intonation despite the absence of the primary acoustic correlate of intonation, the fundamental frequency. Our previous studies employed a test of differential similarity to determine that the tone analog of the first formant is a probable acoustic correlate of sinusoidal sentence intonation. Though the typical acoustic and perceptual effects of the fundamental frequency and the first formant differ greatly, our finding was anticipated by reports that harmonics of the fundamental within the dominance region provide the basis for impressions of pitch more generally. The frequency extent of the dominance region roughly matches the range of variability typical of the first formant. Here, we report two additional tests with sinusoidal replicas to identify the relevant physical attributes of the first formant analog that figure in the perception of intonation. These experiments determined (1) that listeners represent sinusoidal intonation as a pattern of relative pitch changes correlated with the frequency of the tonal replica of the first formant, and (2) that sinusoidal sentence intonation is probably a close match to the pitch height of the first formant tone. These findings show that some aspects of auditory pitch perception apply to the perception of intonation; and, that impressions of pitch of a multicomponent nonharmonic signal can be derived from the component within the dominance region.

Single- versus multichannel vibrotactile supplements to intonation and stress by normal-hearing and hearing-impaired adults

The possible benefit of a wearable, single-channel versus eight-channel tactile aid for conveying voice fundamental frequency (F0) was estimated in three experiments. Severely or profoundly hearing-impaired (HI) and normal-hearing (NH) adults identified position of stressed words and rising versus falling intonation in sentences previously recorded for this purpose by Bernstein etal. [J. Acoust. Soc. Am. 85, 397–405 (1989)]. In experiment 1, NH subjects performed the identification task in counter-balanced visual-alone (VA) and visual-tactile (VT) conditions. Both tactile configurations conveyed intonation but neither conveyed stress. In experiment 2, NH subjects performed the task tactile alone. Both stress and intonation were conveyed. In experiment 3, pre- and post-lingually HI subjects demonstrated effects of the aid for identification of intonation but not of stress. As in the previous study (Bernstein etal., 1989), visual stress was highly accurate in all VA conditions and tactile information shown to be present in experiment 2 did not improve identification in VT conditions.

The transmission of prosodic information via selected spectral regions of speech

In a recent article [K. W. Grant and L. D. Braida, J. Acoust. Soc. Am. 89, 2952–2960 (1991)] it was demonstrated that spectrally different bands of speech with equal articulation index (AI) scores provided approximately equal auditory-visual sentence recognition when combined with speechreading. Given that different parts of the frequency spectrum provide different segmental cues for consonant and vowel recognition, current models of auditory-visual integration [e.g., L. D. Braida, J. Exp. Psychol. 43A, 647–677 (1991)] would predict that some spectral regions of speech are more complementary to speechreading than others. This raises the possibility that the findings of Grant and Braida are attributable, at least in part, to suprasegmental cues that were transmitted equally well by the different spectral bands tested. To test this possibility, the identification of syllable number, syllable stress, sentence intonation, and phrase boundary location was assessed under six approximately equal AI filter conditions similar to those evaluated by Grant and Braida. The results indicate that syllable number and syllable stress are perceived best through high-frequency bands, intonation is perceived best through low-frequency bands, and phrase boundary location is perceived equally well throughout the speech spectrum. These results are discussed in terms of the importance of different spectral regions of speech for the recognition of suprasegmental cues, and how this may relate to overall speech intelligibility. [Work supported by NIH Grant No. DC00792.]

Functional lateralisation of pitch accents and intonation in Norwegian: Monrad-Krohn's study of an aphasic patient with altered “melody of speech”

The present article provides a linguistic analysis of Monrad-Krohn's famous description of a patient with deviant prosody (1947). Monrad-Krohn found it particularly striking that the patient's “melody of speech” had been damaged although her musical abilities showed no impairment. The patient had suffered a traumatic motor aphasia as the result of a shrapnel wound in Broca's area. In East Norwegian, accented syllables are associated with one of two pitch patterns, either a fall in pitch or a low-level pitch. The patient had trouble producing the distinction between these two accents. Monrad-Krohn's description also suggests that she may have accented words in contexts where they should have been unaccented, with the result that these syllables (erroneously) received one of the two pitch patterns. It is suggested that her apparently deviant sentence intonation could, in some cases, be the “secondary” result of a tendency to produce full vowels in positions where a reduced vowel would be expected, thus abnormally prolonging an otherwise appropriate rising pitch pattern.

The Prosody of Mandarin Chinese

Through acoustic analysis of Mandarin Chinese intonation, the author finds that the intonation baseline moves up when intonation is shifted from assertive to interrogative; therefore, two baselines and two intonation layers must be reckoned with. Sentence intonation affects the tonal values and the tonal shapes of intrinsic lexical tones, though not beyond recognition. Tonal changes prove to be closely related to sentence intonation, which is superimposed simultaneously onto the utterance as a whole. The author's findings support the position of the movability of the intonation baseline and rectify some widely spread traditional claims concerning Mandarin Chinese prosody.

Improvement of synthetic speech quality through syntactic information

Many words in Japanese have identical written expression but different pronunciation. Natural synthetic speech therefore requires selection of the correct pronunciation for words and optimized prosodic features, including accent position and level, sentence intonation and length of pause, through the use of syntactic features. This paper describes (1) a new method of determining phrase accent level, based on accentual phrase boundary location and compound word structure, and (2) a newly proposed syntactic class of phrase boundaries. The results of the automatic determination of pronunciation, and opinion tests of intelligibility and naturalness are also described. About 10 000 words are assigned to syntactic and semantic features to determine correct pronunciation, representing about 20% of the total vocabulary. Pronunciation of 99% of the words in a Japanese economic daily were correct, and naturalness of the synthetic speech was 1.1 grades higher under the five-grade opinion test.

Prosodic Parameters in a Variety of Regional Danish Standard Languages, with a View towards Swedish and German

Abstract Acoustic analysis of six different regional variants of Standard Danish, two forms of Swedish and two of German divides the ten into two major categories: those that signal sentence intonation function by local cues and those that do it by global means. The purely global types are also those that have no default sentence accents, and which signal focus by reduction of the surroundings rather than by boosting of the focussed element (a focal accent proper). A specific final completion cue is restricted to the local intonation types and is there an entity separate from that which cues prosodic terminal or nonterminal status. The implication for models of intonation is discussed.

Can the start of an epileptic seizure be described as a transition to a low dimensional chaotic attractor?

Fourteen native speakers of German heard normal sentences, sentences which were either lacking dynamic pitch variation (flattened speech), or comprised of intonation contour exclusively (degraded speech). Participants were to listen carefully to the sentences and to perform a rehearsal task. Passive listening to flattened speech compared to normal speech produced strong brain responses in right cortical areas, particularly in the posterior superior temporal gyrus (pSTG). Passive listening to degraded speech compared to either normal or flattened speech particularly involved fronto-opercular and subcortical (Putamen, Caudate Nucleus) regions bilaterally. Additionally the Rolandic operculum (premotor cortex) in the right hemisphere subserved processing of neat sentence intonation. As a function of explicit rehearsing sentence intonation we found several activation foci in the left inferior frontal gyrus (Broca’s area), the left inferior precentral sulcus, and the left Rolandic fissure. The data allow several suggestions: First, both flattened and degraded speech evoked differential brain responses in the pSTG, particularly in the planum temporale (PT) bilaterally indicating that this region mediates integration of slowly and rapidly changing acoustic cues during comprehension of spoken language. Second, the bilateral circuit active whilst participants receive degraded speech reflects general effort allocation. Third, the differential finding for passive perception and explicit rehearsal of intonation contour suggests a right fronto-lateral network for processing and a left fronto-lateral network for producing prosodic information. Finally, it appears that brain areas which subserve speech (frontal operculum) and premotor functions (Rolandic operculum) coincidently support the processing of intonation contour in spoken sentence comprehension.

Characterizing sentence intonation in a right hemisphere-damaged population

Current production studies present a mixed view of right hemisphere-damaged (RHD) patients' ability to produce normal sentence intonation. The present study characterized the sentence intonation of RHD patients, focusing on a greater number of acoustic parameters than past works, and relying on more naturally elicited speech samples through use of a story completion task. Eight RHD speakers and seven nonneurological control subjects produced declarative and imperative sentences as well as yes-no and wh-questions. Slope of F 0 change, linearity of pitch contour, and variance of F 0 points were calculated for each utterance as a whole, as well as for the preterminal and the terminal contour separately. RHD contours were less linear and flatter in F 0 decline than normal controls for the declarative sentences. The patients' yes-no questions also differed from normal productions, displaying smaller F 0 dispersion around a mean F 0. Preterminal range values were more restricted for patients' utterances of yes-no questions, while terminal properties between groups differed for three of the four sentence types examined. The present results suggest some disturbance in the patients' ability to manipulate fundamental frequency across sentential domains. These data are discussed in terms of current theories of a general dysprosody in RHD patients.

Stress group patterns, sentence accents and sentence intonation in southern Jutland (Sønderborg and Tønder) - with a view to German

This paper investigates prosodic stress group patterns, the presence and manifestation of default and focal sentence accents and the nature of sentence intonation signalling in Standard Danish spoken on a substratum of South Jutland dialects, viz. Sønderborg and Tønder, and in two varieties of German, Standard North German and Flensburg. The following facts appear: sentence intonation (understood to encompass both utterance function and utterance juncture) is signalled globally in Tønder, locally in Sønderborg, and with a mixture of global and local signalling in German. Default accents are nonexistent in the two Danish varieties, optional in German. Focus is signalled, optionally (and never in final position), by stress reduction of the surroundings in the Danish regions, but is compulsory and takes the shape of a proper sentence accent, though modest, in German. Sønderborg and German have unambiguous final lengthening, whereas both lengthening and shortening finally occurs in Tønder. Prosodic stress group patterns suffer a clean truncation when their duration is shortened in the Danish regions, but a mixture of compression and truncation in German. Finally, Tønder has stød, Sønderborg and (of course) German do not.

An analysis of the characteristics of speech perception ability by the combined use of cochlear implant and lipreading

In this study, test materials for evaluating speech perception ability by a combined use of cochlear implants and lipreading were first selected, taking into account various acoustical parameters such as fundamental frequency, intensity, and formant frequencies that can be transmitted through a speech signal processor in a cochlear implant, as well as supplemental speech information that can be obtained through mouth shape in lipreading. The test materials included the short versus long vowels, unvoiced versus voiced consonants. Japanese word accent and sentence intonation, and the 5 Japanese vowels, 50 monosyllables, 50 words, and 29 sentences that consisted of 100 words. Then, both the speech signal and the mouth shape for utterances of the test materials by a female speaker were recorded on videotapes and were presented to the subjects who had cochlear implants in Tokyo. The results showed that the test materials were useful for a detailed analysis of the characteristics of speech perception ability through the cochlear implants and lipreading.

Towards a quantified, focus-based model for synthesizing English sentence intonation

An algorithm for assigning information focus within an English text on the basis of an interaction of grammatical functions and contextual coreferential relationships is phonetically quantified with respect to the parameter of pitch (F 0) and situated within a more embracing model of sentence prosody. The model is readily adaptable for implementation in a text-to-speech program. The algorithm for assigning focal prominences serves as a basis for accounting for English sentence intonation. Levels of focal prominence are defined within an empirically determined sloping grid for declarative sentences. The grid consists of two parallel lines whose direction represents the course of the intonation and whose intervening distance represents the nonemphatic register (F 0 range at a given point in time) of a given speaker. An informal experiment based on analysis by synthesis is used to test the focus assigning model. The placement of prefocal phrasal prominences within the grid is also discussed and situated in the rule system of the prosody model. The resultant rules are then applied to a fragment of discourse. Derivations and synthesized F 0 curves are presented and discussed.

The role of intonation cues in aphasic patients' performance of the grammaticality judgment task

This study investigated the possibility that the reported success of agrammatic aphasic patients in performing auditory grammaticality judgments results from their use of intonational cues to sentence well-formedness. Two agrammatic aphasics, two anomic aphasics, and two nonagrammatic patients with comprehension deficits for semantically reversible sentences were asked to judge grammatical well-formedness in three conditions. Results in a “natural” listening condition replicated the finding of M. C. Linebarger, M. F. Schwartz, and E. M. Saffran (1983, Cognition, 13, 361–392) that agrammatic patients are sensitive to grammaticality despite poor ability to use syntactic cues in comprehension tasks. Two additional listening conditions employed signal-processed auditory stimuli in which the fundamental frequency was normal or was set at a constant level. Right-hemisphere-damaged and normal controls showed no performance deficit for judgments of monotone sentences. The aphasic patients' performance was slightly worse for both signal-processed conditions, but there was little apparent effect of removing sentence intonation on their ability to judge sentence grammaticality. These results indicate that the high levels of performance on grammaticality judgment tasks have not resulted from patients' detection of intonational cues in ungrammatical sentences.

Stress group patterns, focus signalling and sentence intonation in two regional Danish standard languages: Aalborg and Næstved

This paper investigates stress group patterns, prosodic focus signalling and sentence intonation contours in Standard Danish as spoken on a substratum of Northern Jutland (Aalborg) and South Zealand (Næstved) dialect. The major difference between these two regions (and the difference to Standard Copenhagen Danish) is to be found in the way segments and Fo patterns are aligned in the prosodic stress group. Both regions share with Copenhagen the global signalling of sentence intonation contours, as well as a pronounced reluctance to indicate, with prosodic means, in this style of delivery, the pragmatic, contextually invited focus of the utterance.

Intonation on Bornholm - between Danish and Swedish

Acoustic investigations of seven speakers on the island of Bornholm and two speakers from Copenhagen, Malmö and Stockholm, respectively, have proved Bornholm to be an interesting compound, prosodically, between Standard Danish and Swedish. A prosodic continuum can be established from Standard Danish, via Skania, over Bornholm, to Standard Swedish. The parameters investigated are (1) manifestation of sentence accent, (2) manifestation of sentence intonation, (3) alignment of fundamental frequency with syllables and segments at the level of the prosodic stress group, and (4) final lengthening. One particularly interesting implication of the results is the division, both according to their function and their form, of sentence accents into (1) prosodically or syntactically determined final default accents and (2) contextually or pragmatically determined focal accents. Default accents are nonexistent in Standard Danish and Skanian, optional in Bornholm and obligatory in Stockholm. Focal accents are non-existent in Standard Danish, optional - but rather rare - in Skanian, optional - but rather frequent - in Bornholm and obligatory in Stockholm.

Use of the tracheostoma valve in the marking of contrastive stress and sentence intonation

Two proficient tracheoesophageal speakers experienced in use of the tracheostoma valve (TSV) produced a total of 64 examples of “Bev loves Bob.” These productions varied with location of contrastive stress, type of sentence intonation, and use of TSV. Thirty-four listeners judged the utterances in terms of stress placement and intonation contour simultaneously. An analysis of variance of the perceptual results indicated that TSV use reduced speaker ability to signal the two suprasegmental features; however, locus of the suprasegmental features, particularly stress placement, appeared to exert a stronger effect.

Transformations of voice fundamental frequency for a vibrotactile device to aid lipreading

As part of the development of a wearable vibrotactile device to convey voice fundamental frequency F0 to hearing-impaired lipreaders, a study was conducted to compare transforms of F0. Rothenberg and Moliter [J. Acoust. Soc. Am. 66, 1029–1038 (1979)] reported that most errors in a tactile-alone task of judging stress and intonation in sentences were mislocations of the stressed word. We found that contrastive stress can be detected visually. The present experiment compared F0 transforms in a tactual-visual task. Transforms included: (1) direct glottal to tactile pulse; (2) linear transform with lowering of center frequency and shift in range; (3) similar to (2) with log of the normalized glottal period; (4) similar to (2) with the difference of logs of the current period and a weighted sum of previous periods; and (5) a constant pulse train during voicing. Subjects received tactile stimulation to the index finger by an AV-6 vibrator, and responses were labeled “question” or “statement” and one of three stress positions. Effects of the transforms and conditions of vision-only versus tactual-visual presentation will be reported. [Work supported by NIH.]