The Journal of Seventeenth-Century Music

Volume 9 (2003) No. 1

Beyond Isomorphism toward a Better Theory of Recitative

John Walter Hill*

Abstract

Jacopo Peri aimed to portray dramatic characters through musical control of declamation. Modern commentary acknowledges this but merely asserts isomorphism between his music and speech. This only displaces the question from the musical to the linguistic sphere and provides neither methods nor vocabulary to account for the results. The autosegmental-metrical theory of intonational and metric phonology provides for a more precise account of the means by which inflection, accent, stress, and prominence correlate with syntax, grammar, logical relations, affect, and character. Progress toward a better theory of recitative will coordinate with advancement within in the field of intonational phonology.

1.  The Intentions of Peri and Other Florentines in Composing Recitative

2.  The Autosegmental-Metrical Theory of Intonational and Metric Phonology

3.  Intonational Phonology Applied to the Analysis of Peri’s Recitative

4.  Conclusions: Toward a Better Theory of Recitative

References

Audio Examples

Illustrations

1. The Intentions of Peri and Other Florentines in Composing Recitative

1.1 I believe that it is generally agreed among specialists that, in composing the opera Euridice, “Peri was trying to imitate the ideal of the poet/actor reciting in heightened declamation,” as Tim Carter wrote in his 1989 book on the composer.¹ Yet little has been done to demonstrate whether, to what extent, and how Peri succeeded in this. If Peri imitates an actor, what does that actor communicate and how?

1.2 In its attempts to account in detail for the artistic results of declamation in early recitative, even the best critical commentary suffers from an ad hoc, narrative, discursive, and rambling quality, which are faults I find with my own attempts in this area. The problem, I think, lies in our lack of a technical vocabulary and an underlying theory. My view is that a specific and detailed discussion of music always has the potential to enhance aesthetic perception, and is worthwhile for that reason alone.

1.3 Peri, himself, began to articulate a theory of recitative when he wrote, “I recognized that in our speech certain words are intoned in such a way that one can base harmony upon them, and, in the course of speaking, one passes through many others that are not so intoned, until one returns to another capable of motivating a new consonance.”² And “I held [the bass] fixed through both false and good proportions [i.e., intervals] until, having run through various notes, the voice of he who speaks arrived at one [syllable] that, being intoned in ordinary speech, opens the way to a new harmony.”³ Here, in modern terms Peri recognizes what modern phonologists call “pitch accent,” and he says that syllables that are pitch accents can be accompanied by a change of harmony and that transitional syllables can be sung as dissonances against a sustained bass note. But in practice, Peri changes bass notes on many unaccented syllables, notably the last syllables of many versi piani, where he frequently places the resolution of a cadence. And there are also many accented syllables that do not receive a new harmony. Almost any page of the score provides examples of this. Although Peri’s recognition of the special importance of pitch accents is of great interest, as we shall see, the idea that these actually control the movement of the bass is mistaken.

1.4 More important, in my view, is Peri’s reference to “those manners and accents that serve us in our grief and joy and similar states.”⁴ In writing this, Peri was echoing the words that Vincenzo Galilei put in the mouth of Giovanni Bardi: “To express the conceptions of the mind by means of words,” let musicians go to the theater and “observe the manner in which one quiet gentleman speaks to another, with regard to the highness or lowness of the voice, the volume of sound, the kinds of accents and gestures, and how the words are uttered with relation to fastness and slowness. Let them observe a little the difference which occurs among all those things when one of these [gentlemen] speaks with one of his servants, or one of these servants speaks with another servant. Let them consider the prince, when he has occasion to converse with one of his subjects and vassals or when he speaks with a petitioner who is presenting his petition. Let them observe how an infuriated or excited man speaks, and also the married woman, the young girl, and the mere child. Let them also observe how the clever strumpet speaks, how the lover speaks to his beloved while seeking to bend her to his will, how those who are mourning speak, also those who cry out, how a timid man speaks, and, finally, how those speak who are rejoicing. From these diverse situations, provided they observe them attentively and examine them carefully, they can select the norm of what is suitable for the expression of any concept at all with which they may be required to deal.”⁵ And along the same lines, also reformulating what Peri wrote, here is Giovanni Battista Doni’s prescription for composing in stile recitativo from his Trattato della musica scenica (1635–39):

He who would like, therefore, precisely to express those very accents and inflections of pitch that are naturally uttered in speaking … will need to pay great attention, to practice at length, and to use suitable means; but above all [he will need] an extremely sensitive ear. For this reason he should diligently observe which syllables are intoned with a uniform and steady accent, and on which ones the voice is raised or lowered, and to what note or interval, considering those rapid transitions that occur between the accented syllables, and all the varieties that are made principally by artful and expert speakers, according to the manner, affection, and sentiment of that which is spoken, for example, interrogations, threats, and all kinds of interjections and rhetorical speech.⁶

1.5 Doni recognizes that in ordinary speech pitches are not sustained but usually slide and that it requires a violin without frets to imitate this effect. Nevertheless, certain musical intervals are approximated by speech, he believes, as for example the rising perfect fifth by the word “verrò” under specific circumstances.

1.6 As for rhythm, Galilei’s idea, learned from Mei, that ancient Greek recitational singing followed the rhythms of speech rather than those of measured music seems to have been stated as an implied prescription for modern singing. Caccini, who quoted Plato to the effect that “music is naught but speech, with rhythm and pitch [following it] in the last place,”⁷ claimed that the essence of the Florentine recitational style was “singing with nonchalance like a new speaking in music without observing measured time.”⁸

1.7 It is clear from their own testimony that Peri, Caccini, and their followers and successors, including Monteverdi, Schütz, Henry Lawes, and Purcell, intended to interpret their texts and to project affect, in large part, through their musical control of declamation—inflection, accentuation, pacing, volume, tessitura. Most modern specialists seem to agree that, by and large, they succeeded in this. But their success has been difficult to demonstrate. Even more difficult is an analysis of exactly what interpretations and affects are promoted by their musical declamation, and exactly how their musical declamation achieves those ends. Either their success is merely asserted, or the interpretations and affects are merely described or named.⁹ Or, at most, some demonstration is offered showing that they achieve some degree of isomorphism with speech.¹⁰ But what, exactly, that speech—that declamation—communicates, and how, has been very difficult to demonstrate and to discuss in a compact, readable fashion.

2. The Autosegmental-Metrical Theory of Intonational and Metric Phonology

2.1 In my search for better means to assess and account for Peri’s success in attaining the goal he and his fellow Florentines set for recitative style, I have turned to linguistics, particularly to intonational and metrical phonology and some related and underlying research in psychology, phonetics, and acoustics.

2.2 I find that these resources have not been explored very much in musicology. Raymond Monelle’s 1992 book, Linguistics and Semiotics in Music, devotes three very general paragraphs to this matter before discussing features of abstract musical design that seem to reflect speech patterns. Joseph Swain’s Musical Languages seems to discuss every parallel between music and language except delivery.¹¹

2.3 I have used as an entry point into current linguistic research D. Robert Ladd’s book Intonational Phonology,¹² published by Cambridge in 1996. The summary of this branch of research given below is derived from this book unless otherwise noted.

2.4 Phonology studies postlexical (sentence-level) intonation as a linguistic phenomenon deriving from the factors of pitch (F0), intensity, and duration. Phonology was revolutionized toward beginning in the 1980s by the introduction of autosegmental-metrical theory, whose four basic tenets are:

1. The tonal structure of speech consists of a string of pitch accents and edge tones.
2. Pitch accent can be distinguished from stress.
3. Pitch accents and edge tones consist of pitch targets, linguistically classified as either High or Low.
4. The phonetic realization of High or Low pitch accents depends upon local context; overall contour mostly reflects and is determined by local features.

2.5 These basic concepts are illustrated in an example, reproduced by Ladd. I have adapted it from Janet Pierrehumbert’s ground-breaking 1980 dissertation, which introduced the autosegmental-metrical theory of intonational phonology.¹³ In this example (figure 1), the pitch traces from a spectrograph have been stretched vertically for the sake of clarity, and the text has been added. Thus High and Low accents are marked by an asterisk next to the letters H or L. Notice how they are defined by local activity, not by their elevation, even in relation to one another, but by how they are approached. The overall descent of pitch in this sentence, called “declination,” is considered normal in most intonational languages. Autosegmental-metrical theory finds that this declination is the result of a series of downstep accents. As a consequence, it usually happens that High accents later in a sentence are lower in absolute pitch than Low accents that occurred earlier in the utterance.

The meaning and implication of the sentence can be changed by shifting accentuation, or “focus”:

I REALLY believe Ebenezer was a dealer in Magnesium. I really BELIEVE Ebenezer was a dealer in Magnesium. I really believe EBENEZER was a dealer in Magnesium. I really believe Ebenezer was a DEALER in Magnesium. I really believe Ebenezer was a dealer in MAGNESIUM.

The intonation represented on Pierrehumbert’s chart is taken to be neutral, and is called the “citation form” of the sentence (audio 1).

2.6 In autosegmental- metrical phonology, a monotone series of syllables, when one occurs, normally forms a transition between brief “events.” Turning points in the contour of utterances are more likely to produce pitch accent than are transitions. (Jacopo Peri’s understanding of this aspect of speech, mentioned in paragraph 1.3, above, corresponds to this concept of “events” and “transitions” remarkably, and so does the usual practice in much recitative, of interspersing distinctive melodic events with strings of syllables set in a relatively neutral manner, as, for example, on a single pitch .) In any case, a tone needs metrical strength (usually stress) to become a pitch accent. Other cues to prominence can be duration and vowel quality.

2.7 The three distinguishable facets of stress are:

1. Abstract prominence relations (or “metrical strength”)
2. Concrete acoustic prominence or salience
3. Location of prominence-related intonational events (i.e., “pitch accent”).

2.8 Phonologists have found that Europeans tend to think of high pitch as exclusively producing prominence, but they are often wrong in this. For example, native Italian speakers are often place the location of the main accent in who-, what-, where-, when-questions on the first word because it is higher, as in “Dove vai?” whereas phonologists place it on the second word because of intensity, duration, and metrical strength.

2.9 Although the normal contour of speech is downward, “metrical analysis also makes sense of ‘declination reset’, the upward modification of the pitch range at the beginning of a new stretch of declination, [before] ‘final lowering’, the corresponding downward modification of the pitch range at the end” (p. 279). This is shown in another example from Pierrehumbert’s dissertation reproduced in Ladd’s book (figure 2; audio 2).

2.10 In their search for universals, phonologists point to three phenomena that are found in most languages:

1. Sentence declination, with low or falling pitch to show completion.
2. High or rising pitch for many questions and for non-finality.
3. Presence of local pitch movements on new or otherwise informative words.

3. Intonational Phonology Applied to the Analysis of Peri’s Recitative

3.1 In all Romance languages, statement intonation normally involves downstepped final accents. This, of course, is why all but a few declarative sentences in Peri’s Euridice conclude with stepwise descent to the locally prevailing modal final. At important articulations within sentences, Peri carefully avoids that kind of stepwise descent, preferring instead either two long notes on the same pitch or else a stepwise descent to a non-final pitch.

3.2 As in ordinary speech, Peri’s recitative reserves upward pitch inflections for certain types of questions, while most questions of what linguists call the “WH” category (who, what, where, and why) are naturalistically inflected downward.

3.3 Ladd points out that “In Italian, an additional emphatic falling declarative accent (H* or H*+L) may be placed on a lexically unstressed utterance-final syllable, so that the final word as two accents. This adds emphasis or conveys some sort of special emotional involvement on the part of the speaker” (p. 131). He finds this to be part of a “general strategy for increasing the emotional content of the contour by suspending the normal constraints on tune-text association” (p. 131). Ladd’s examples of this are his transcriptions of two utterances: “Mi fai male!” and “… una terrazza da dove si vedeva tutta Roma!” This emphatic falling declarative accent seems to be represented in Peri’s Euridice when a complete sentence ends with an “8-7-8” melodic motion to and from a leading tone. I count ten instances of this at the conclusion of sentences in the opera. In almost all cases, the printed text alignment shows that the final, unaccented syllable is to be initiated on the leading tone and slurred upward into the final note of the cadence, imitating very closely the pitch contour of this special kind of final emphatic declarative accent. In every case of this, I would judge that the utterance in its context could appropriately be delivered with this additional emotional content, although there are many other emotionally charged sentences that receive the normal declarative downstep treatment.

3.4 Several more general points are illustrated in a spoken rendition of Orfeo’s expression of joy (“O mio fedel”) in his love of Euridice. My wife, Laura Callegari Hill, a native speaker of Italian, made this recording (audio 3a) on her first try, without consulting Peri’s musical setting (audio 3b).

3.5 The speech sound tracing of Laura’s spoken rendition of “O mio fedel” (figure 3) and the following charts were made with the program called “Praat” created and kindly supplied to me personally by Paul Boersma, Professor in the Institute of Phonetic Sciences at the University of Amsterdam, to whom I am very grateful. This chart shows pitch traces in the heavy lines, intensity in the upper thin line, and harmonicity (that is, degree of intoning or signal-to-noise ratio) in the lower thin line. As you can see and hear, Laura divided the sentence into three phrases. The incompletion of the first two is signaled by upwardly inflected phrase tones, notated as H%. The completion of the sentence is marked, as usual, by a downwardly inflected edge tone at the end, notated as L%. Each of the three phrases contains its own declination, and the two declination resets that make this possible result from High-accent focus on the “informative” words “pur” and “infinito.” Likewise the other High accents fall on informative words, like “tuoi,” which creates the distinction between what is disclosed to your eyes versus what is distilled in my heart. In this example, Peri’s recitative (audio2nd 3b) is notably close to the recorded speech (audio 3a), and where the two diverge—as at the word “stilla” and at Peri’s ascending scale segments on “dell’infinito mare” and “che di dolcezza”—the spoken version could have conformed to Peri’s setting with little difficulty. Notice that Peri approaches the two Low accents, on “fedel” and “traspare” from the step above, whereas all the High accents are approached from below or from the same pitch. The syllables in the spoken version that are the most nearly intoned, as shown by the highest spikes of the lower thin line, are “mio,” “pur” “stil-la,” “tuoi,” “A-mor,” and “cor.” Peri meets five of these six syllables with a consonant note in the bass. But he also places bass notes under “O,” “fe-del,” “tra-spar-re,” “del-l’in-fi-ni-to” (oddly), ma-re,” dol-cez-za,” and the last syllable of “stilla,” which is not accented. In general, these syllables are of the longest durations in the spoken rendition, and they might have been intoned in a more elevated style of declamation, such as can still be heard in some performances of Shakespeare’s plays. Above the upper voice staff, I have added the approximate note values that result from rounding off the actual durations of the spoken syllables. They are surprisingly close to Peri’s rhythms. However, my method of deriving rhythms from the speech is not exact, only relative; and the range of durations in Peri’s setting is stretched with respect to the actual durations of the spoken sample.

3.6 Beneath the transcription of Peri’s musical setting I have placed the parallel one by Giulio Caccini (figure 3), which I find notably inferior. Caccini, for example, treats the accents on “fedel” and “traspare” as High accents and, thus, loses the contented and kindly tone that the speech has in both Laura’s spoken version and Peri’s setting. Caccini’s lack of accent on “pur” forfeits one of the declination resets and further reduces the warmth of the utterance.

3.7 As shown in each of the analyses given so far, pitch accents can be made of one or two tones; if there are two, one of them is considered to be central. A phonological “tune” consists of a series of pitch accents, possibly articulated by phrase accents, framed by an initial and concluding boundary tone. Some researchers have attempted to name and identify specific tunes, such as “neutral declarative intonation,” “interrogative intonation,” “contradiction contour,” “surprise-redundancy contour,” “major continuation,” “minor continuation,” and “implication.” A well-known tune is the “calling contour,” featuring a falling minor third. Ladd concludes that “One central goal of theories of intonational phonology is to be able to provide an explicit phonetic characterization of all the tunes of a given language” (p. 206), but this has not yet been accomplished for any of them.

3.8 Phonological tunes represent linguistic, syntactic meaning, and are thought to retain their identity in many different renditions, as long as the pattern of pitch accents is somehow preserved. To illustrate this point, I offer two spoken renditions of the first clause of the sentence delivered by the Pastore del Coro at the beginning of Scene 1 in Euridice.

3.9 Here, I should insert that, while Laura’s first attempt to read Orfeo’s speech “O mio fedel” matched Peri’s setting surprisingly well, her five recordings of the opening lines of the Pastore del Coro, including several that made use of Peri’s score, did not come very close to Peri’s treatment of the text. As possible reasons for this, I would observe that the shepherd’s opening sentence is far more complex and extended, its syntax stretched. But also, Orfeo’s speech “O mio fedel” is delivered in a relatively calm, sincere, and intimate tone, perhaps resembling modern familiar speech more than the theatrical and exaggerated welcoming speech of the Shepherd, which perhaps has its echoes in today’s announcements of circus ringmasters or the harangues of American country preachers.

Ninfe ch’i bei crin d’oro sciogliete liete a lo scherzar de’ venti, e voi ch’almo tesoro dentro chiudete a’ bei rubini ardenti, e voi ch’a l’alba in ciel togliete i vanti, tutte venite, o pastorelle amanti; e per queste fiorite alme contrade risuonin liete voci e liete canti		Nymphs whose beautiful golden hair is loosened happily to the play of the breezes, and you whose sweet treasure is enclosed within beautiful glowing rubies, and you who steal the pride of the dawn in the sky, come all, oh loving shepherdesses; and throught these sweet, flowing fields may happy voices and happy songs resound.

3.10 figure 4 gives two speech sound tracings of this text as read by my wife, Laura, and the musical setting by Peri. The first spoken rendition can be heard in audio 4a, and the second (take 4) in audio 4b. Peri’s musical setting is heard in audio 4c.

3.11 Although these two utterances sound quite different, their phonological analysis diverges in only one place, the accented syllable of “liete.” The High accent in figure 4, take 4, and audio 4b replaces the Low accent of figure 4, take 1, and audio 4a, creating a new focus on the concept of happiness and a declination reset that alters the contour of the utterance. Audio 4a was recorded before Laura had seen Peri’s music, while audio 4b represents her attempt to reflect Peri’s vocal line by emphasizing “liete” and eliminating the low closure on the second syllable of “Ninfe.” However, even audio 4b retains the defect of a Low phrase tone at the end of the word “venti”—a defect because this is just the first of three dependent subject-clauses. Peri’s setting, as you see, avoids that defect with a pitch repetition that represents a High, non-final, phrase tone. And Peri also avoids a phrase closure after “Ninfe,” because what follows is a restrictive clause, since the women referenced in the following two subject clauses are also nymphs.

3.12 There are two further places where Peri’s pitch contour is slightly at odds with Laura’s spoken renditions: he sets the words “crin” a step lower than the word “oro,” but it is almost impossible to reflect that in speaking, because “crin” is a transitional syllable following a high accent and because of the intrinsically high quality of the vowel in the word “crin.” And the same discrepancy is found at the word “de’” at the end of the penultimate measure. Peri’s rhythm, too, does not match the syllable durations of the spoken examples as well as it did in Orfeo’s speech previously discussed. I think that these differences are partly the result of the more difficult syntax of the present text and partly a consequence of the somewhat melodious style of the shepherds’ recitative in this opera.

3.13 Peri’s strong and unusual accent on “liete,” which was deliberately mimicked in audio 4b, raises an issue of isomorphism because Italian is thought to resist displacing the accent from the end of the sentence, even when the last word is redundant. Thus, Italian President SCALfaro said, “[Le inchieste] servono a mettere a POSTO cose andate fuori POSTO,” which would be like saying, “The investigations serve to put in PLACE things gone out of PLACE” Where emphasis needs to be shifted, modern Italian favors changing the word order to put the word requiring emphasis at the end of the sentence. However, because Peri’s accent on “liete” does not eliminate the phrase-ending accent on “venti” and because the accent does not affect the syntactical meaning of the sentence, it would probably be classified by phonologists as a paralinguistic accent. Remember that the high accent on “liete” caused a rise in Peri’s melodic line, a declination reset in phonological terms. Thus, the autosegmental-metrical theory of phonology helps to reconcile the phenomenon of sentence declination with the preponderance of arching phrase contours observed in Peri’s Euridice by Tim Carter. Namely, the mechanism of declination reset can easily replace a normal descent of pitch with an arching contour without necessarily disturbing the phonological “tune” pattern of High and Low accents. When this happens, the exaggerated High accent that causes the declination reset is a paralinguistic cue to attitude or affect. The frequency of these arch contours, with their paralinguistic accents, is simply another indication of the prevailing elevation of emotional tone, which was the deliberate and self-conscious goal of the Florentine participants in the creation of opera.

3.14 The use of pitch excursion, register, and pace as paralinguistic cues to attitude and affect is particularly evident in the contrast between Orfeo’s and Plutone’s modes of speech in Scene 4, where Orfeo pleads with Plutone for the return of Euridice. Orfeo’s words are, of course, also colored by dissonance, chromaticism, and mutations of hexachord and mode. But spectrographs of even my own non-native and amateurish declamation of the intonations of the two interlocutors makes clear that Orfeo (figure 5) is the petitioner and Plutone (figure 6) the aggressive and initially unmoving man of power and authority. The spoken rendition of Orfeo’s speech can be heard in audio 5a, and Peri’s musical setting in audio 5b. Corresponding renditions of Plutone’s response are found in audio 6a and audio 6b.

3.15 Additional features that make audio 5 more emotional than audio 6 are its more exaggerated pitch excursions, higher register, and higher degree of harmonicity. Like unusual accent, these features are classified as “paralinguistic” because they are thought to convey attitude and affect rather than syntactic meaning. A long series of modern linguistic researchers have associated wide pitch excursion with greater emotional involvement and stronger expressive intent.¹⁴ Phonologists have also experimentally demonstrated that higher register is associated with more active emotional states. The same can be said for loudness and the associated quality of harmonicity. The actual emotion conveyed depends partly upon the syntactic meaning of the utterance, and phonologists generally agree that when paralinguistic cues are coupled to individual words, expression transcends the power of words or paralanguage fuctioning separately.

3.16 Girolamo Mei, Giovanni Bardi, and Vincenzo Galilei, of course, all commented upon the effect of register and pace upon expression and characterization. Modern observations run in a parallel direction. For instance, Robert Frick in his 1985 article “Communicating Emotion: The Role of Prosodic Features” (cited in reference 14) reports numerous studies showing that “increased loudness, raised pitch, increased variance in pitch, and increased rate of speech generally correlate with a higher level of emotional activity and arousal [as in Orfeo’s pleas], whereas the prosodic contour for anger [is generally] level, [with] average pitch except for jumps of about a musical fourth or fifth on stressed syllables.… [the] prosodic contour lasting about the length of a breath group” (p. 422). This turns out to be a rather good description of Peri’s setting of Plutone’s words in Scene 4.

3.17 It may reasonably be asked whether a phonological comparison between early recitative and modern spoken Italian problematically begs the question whether pronunciation and intonation have not changed too much during the past four centuries and whether we can know enough about this. My answer would be that the extent to which Peri’s declamation actually does correspond surprisingly well to some reasonable forms of modern speech suggests that whatever changes have occurred have not dissolved all connection between his speech and ours. The marked similarities between my wife, Laura’s, inflections and Peri’s music cannot be the result of chance operations. Further, Peri’s way of portraying grief, pleading, anger, and authority correspond very closely with norms described in modern literature in linguistics and psychology. And though it was in 1528 that Gian Giorgio Trissino measured the pitch peaks in Italian speech,¹⁵ and though it was in 1549 that Pietro Bembo precisely observed the relative length and stress of syllables in poetic verses and sentences,¹⁶ their results still correspond to the findings of modern phonologists.

4. Conclusions: Toward a Better Theory of Recitative

4.1 My study of intonational phonology thus far has caused me to listen to Peri’s Euridice with new ears. In it, I now hear a surprising variety of distinct voices, perhaps a deliberate display of them.

4.2 However, this study has been merely preliminary and explorative. I believe that it has shown that it can be fruitful to analyze recitative prmarily as if it were speech, in addition to analyzing the purely musical characteristics that enhance expression of text—dissonance, mutation and restoration of system and hexachord, text illustration, interruption and completion of melodic processes, and so on—which, in most cases, are features that early recitative inherited from the polyphonic madrigal. The findings of the autosegmental-metrical theory of intonational phonology represents a major advance in the fields of linguistics, perceptual psychology, and communications. It helps us to identify, more clearly and exactly than before, the significant linguistic and paralinguistic features of utterances, and it provides us with a technical vocabulary that can contribute to precision and concision in our analytical reports.

4.3 The linguistic study of the communication of emotion and human character—which are the two aspects of speech that interest us most in the analysis of recitative—has lagged behind the study of the communication of grammar and syntax in “citation forms” of speech. The two reasons for this are that the citation form of an utterance is the usual aim of synthesized speech, which is a goal of the computer industry, a direct or indirect source of much financial support for this sort of research; and linguists prefer to isolate patterns of intonation that correlate with categories of emotion, rather than to take into account the actual content of the utterance. But in spite of these factors, there will be progress in research on the communication of emotion and character in speech patterns. Progress toward a better theory of recitative will be tied to advancement particularly in this aspect of intonation phonology.

---