<<

Scat syllables and markedness theory*

Patricia A. Shaw University of British Columbia

here could be no more appropriate dedication to Jack: T Th ou Swell (scat solo)

Th ou swell 1927. Words by Lorenz Hart, Music by Richard Rodgers. Scat solo by Betty Carter, transcribed by William R. Bauer (2002a: 251). A highly creative domain between the prosodies of human language and the riff s of inst rumental is the dynamic vocal jazz idiom of scat. Th e present analysis proceeds from the observation that, despite the dist inct ly individualist ic approaches to scatting by renowned jazz mast ers such as Louis Armst rong, Betty Carter, and Chet Baker, the inventory of the semantically empty syllables used in scat is extremely limited in comparison to the rich range of combinatorial possibilities that defi ne well-formed syllables in English. Th is paper explores the degree to which the form of scat syllables in the performance repertoire of various artist s conforms to post ulated universal markedness const raints on natural language syllables. Signifi cantly, markedness theory plausibly accounts for a considerable range of the data. Nonetheless, certain syst ematic deviations occur. It is proposed that the relative markedness of such properties may be genre-dependent, funct ioning in scat to enhance musical form or modality.

1. Introduction

Like the majority of human languages in the world, which evolved and persist as strictly oral traditions, scat emerged in the realm of musical genres as a vibrant, expressive, and exclusively oral idiom. However, unlike human languages, scat does not build on a consistent, conventionalized relationship between sound and meaning. Its essence is creative, improvisational vocal tract sound. Its syllables and sequences are evocative and emotive, but not denotative. There is no standardized or systematic interpretability to the musically parsed cadences of scat syllables. For example, the title of ’s 1926 hit Heebie Jeebies has a consistent interpretation, verifiable across different speakers, as * I am deeply indebted to Mike Fitzgerald, Kate Hammett-Vaughn, Ted Moore, Tyler Peterson, Suzanne Pittson, Fred Stride, and particularly Bill Bauer and Alan Matheson for their generous guidance. Special thanks to Walter Pedersen for his enthusiastic assistance with transcription and in tracking recordings.

Toronto Working Papers in Linguist ics 27: 145–191 Copyright © 2008 Patricia Shaw Patricia A. Shaw

refering to a kind of nervous energy or a scattered uneasy feeling, the ‘jitters’. However, the sequence of syllables in the scat line in (1) would elicit no coherent consensual meaning.¹ (1) Bars 5–7 of the scat solo by Louis Armstrong in Heebie Jeebies (1926)² WRB: | duw Œ daw diy duw də | ‰ diy də də dow diy | dow di dow duw ‰ duw– | | duw Œ daw diy duw də | ‰ diy də də dow diy | dow dɩ dow duw ‰ duw– | In a formal linguistic sense, then, scat syllables are semantically ‘empty’. Nonetheless, of considerable linguistic interest is their form. The present analysis proceeds from the observation that, despite the distinctly individualistic approaches to scatting by great vocal jazz masters, the repertoire of the semantically empty syllables used in scat is extremely limited in comparison to the rich range of combinatorial possibilities that define well-formed syllables in English. For example, two properties of the excerpt in (1) are immediately noteworthy, and, as it turns out, are robustly characteristic of scat vocables produced by a broad diversity of performers. First, consider the onset and coda³ structure of the scat syllables in (1): of the 15 syllables, all have a single consonant as onset—there are no clusters, no onsetless syllables, and none has even a single consonant as coda. In other words, all are ‘open’ CV or CVG syllables, despite the fact that English words are built on an inventory of combinatorial possibilities that readily sanctions codas, and allows quite extensive complexity within both onset and coda clusters, e.g. as [str...] and [...ŋkθs] in ‘strengths’ [strɛŋkθs]. Secondly, not only do all the scat syllables in (1) have a non-complex onset, but in fact they all have the same consonant [d] as the syllable onset. As a means of comparison, now consider in (2) the structure of the syllables in another scat solo by Louis Armstrong from Hotter Than That, recorded 3 years later (cited from Reeves 2001 by Bauer 2002b: 308). Just as in (1), all the syllables in (2) are canonical open syllables: all have a single segment onset, none has a complex onset, and none has a coda. However, in contrast to (1), there are no [d]s. Rather, here all 16 syllables have [b] as their onset. (2) Bars 49–54 of the scat solo by Louis Armstrong in Hotter Than That (1929) WRB: | | | | | | | boh ‰ bə Œ boh ‰ ba Œ bə ‰ biy Œ bə ‰ biy Œ bow ‰ bə Œ bow ‰ bə Œ ba ‰ biy Œ ba ‰ biy Œ | | | | | | | bɔ ‰ bə Œ bɔ ‰ ba Œ bə ‰ biy Œ bə ‰ biy Œ bow ‰ bə Œ bow ‰ bə Œ ba ‰ biy Œ ba ‰ biy Œ

¹ An independent measure of what has—or has not—conventionalized semantic interpretability is reflected by which sequences of sounds are accorded entry as ‘words’ in standard English dictionaries. Consistent with the particular example chosen here, ‘heebie-jeebies’ is listed as a word in the American Heritage dictionary: “slang. A feeling of uneasiness or nervousness; the jitters.”. However, none of the various potential spellings of the scat syllables (de, dee, deh, di, dih, du, duh, doo, etc. ) are. ² The transcription line labelled WRB is by Bauer 2002b: 308; the transliteration beneath it follows the principles of phonemic interpretation in Appendix 1. ³ The use of the terms onset and coda here does not entail the attribution of category or constituency status within a formal theory of prosodic structure. Rather, these are simply cover terms to reference (i) as onset, the string (possibly null) of segments between the left edge of a syllable and the Nucleus, and (ii) as coda, the string (possibly null) of segments between the Nucleus and the right edge of a syllable. The Nucleus is assumed to be an independent category node, which in English dominates a short vowel (V), long vowel (Vː), or diphthong (VG). C abbreviates ‘consonant’, V ‘vowel’, G ‘glide’, σ ‘syllable’.

146 Scat syllables and markedness theory

In sum, two generalizations are strikingly evident from the data in (1) and (2). First, of the 24 consonants available in the English phonemic inventory, the only two used in these excerpts are [b] and [d]. Secondly, the syllable structure is consistently open, i.e. not closed by a coda consonant. To an extreme then, Louis Armstrong’s repertoire in these citations exemplifies the fundamental premise of this research: scat draws on a very limited subset of the sounds and of the syllabic groupings that are regularly used in English. However, how representative are these generalizations? Is the favouring of [b] and [d] part of Satchmo’s own particular idiosyncratic style, or is this genuinely something that is broadly characteristic of scat? What—you are doubtless wanting to interject—about the [š] in shoo be doo? And to what extent do other scat singers use a more diversified and complex range of syllable shapes? What about codas? After all, who put the bop in the bop shoo bop shoo bop?⁴ A diverse sampling of vocal scat is investigated here, ranging from classic jazz icons like Louis Armstrong, Betty Carter, and Chet Baker to pop music song-writers/recording artists like Johnny Cymbal and , who in the early 60s wittily transported the playful and unmistakably sexy edginess of scat directly into their rock’n’roll lyrics. Across these artists, generations, and genres, the basic introductory observations about scat are consistently affirmed: the inventory of sounds used and their syllabic organization constitute a significantly small subset of the full diversity of available English options. The principal goal is to identify just what generalizations about phonological form hold within this body of scat data, and to explore various hypotheses that might plausibly explain why the particular patterns that are attested emerge in scat. From the perspective of linguistic theory, the observations are evaluated in the context of postulated universal constraints on articulatory phonetics and phonological markedness. Interestingly, a considerable range of the data is plausibly accounted for by markedness theory. Equally interesting is the finding that certain systematically attested scat properties run directly counter to markedness expectations. The highly marked, yet robustly attested status of these characteristics suggests that over-riding the body of linguistic constraints on the scat phonological system are competing constraints on scat as a musical performance genre, constraints that function to enhance the melodic pitch contour, the musical phrasing, the auditory interpretation, or the distinctive trademarking of individual artistic style. What results from this analysis is a unique perspective into the structural and performative interface of two complex systems of human vocal expression—music and language—each subject to distinct sets of constraints and conventions, sometimes convergent and sometimes conflicting, but ultimately combining in the creative exuberance of scat.

1.1. Purview

The analyses in §2 below are sequenced with respect to the recording date chronology. Beginning with the seminal 1926 Heebie Jeebies recording, the full context of the Louis Armstrong scat solo from which the three-bar excerpt in (1) was drawn is explored

⁴ Barry Mann and Gerry Goffin did, in their 1961 hit single, Who Put the Bomp?

147 Patricia A. Shaw

in §2.1. This is then compared to the phonological properties of a 1929 rendition of Hotter Than That. In §2.2, the focus shifts to Chet Baker (1955; 1989), an icon of consonantal minimalism. In contrast, Betty Carter’s repertoire, representatively examined (1955; 1979) in §2.3, introduces a considerably expanded consonantal inventory. Through the subsequent decades, these two artists—Chet Baker and Betty Carter—remained committed to the vocal jazz idiom of scat despite a significant shift in the general public’s musical interests away from . For each, a comparison of performances recorded nearly a quarter century apart provides an interesting measure of individual creative evolution, as well as of particular consistencies despite dramatic shifts in the musical and cultural backdrop of the latter half of the twentieth century. Although the popularity of bebop—the jazz medium that had become virtually synonymous with scat—had significantly declined by the 60s, the vocabulary of scat itself surged into a different realm of wide-spread prominence in that same period: the American Hit Parade. As seen in §3.1 and §3.2, in major hits by recording artists like Barry Mann (1961) and Johnny Cymbal (1963; re-recorded by ShaNaNa in 1980), canonical scat syllables are directly imported into lyrics like Who Put the Bomp in the Bomp bah Bomp bah Bomp? Here scat is explicitly objectified, transported, and incorporated into a different and evolving musical genre. Although bereft of its improvisational core, this “embedded scat” phenomenon carries forward the continuing identification of scat as infectiously fun and irresistibly seductive. Despite rife competition for cornering the sex appeal market from a burgeoning and rapidly diversifying popular music scene in America, it was scat (the bop, the dip, and the rama lama ding dong) that “made my baby fall in love with me, yeah!!” By 1963, Mr. Bass Man’s “baw bə bə baw bə baw bə baw baw” had elevated him to being “the hidden King of Rock’n’Roll” (Johnny Cymbal 1963), and scat had clearly spread from bebop jazz to become established in the R’n’R mainstream as eminently cool.

1.2. Methodology

Transcriptions of the body of scat data that informs the present study are presented in Appendix 2. With a few notable exceptions, particularly Bauer (2002a, b), there is a paucity of formal documentation of scat, and the diverse original sources that have been drawn on here differ considerably in transcription conventions and rigour. Bauer’s work constitutes an immensely detailed and valuable resource: in the extensive Appendix (2002a: 245–343) to his outstanding contribution to the study of Betty Carter’s musical genius, Bauer provides a full transcription in musical notation of Carter’s melodic line, synchronized with the lyrics, for 15 tunes. Of these, six incorporate scat vocables, phonologically transcribed by Bauer in Trager-Smith notation. The two chosen for the analysis in §2.3 allow a comparison across a 24-year time frame stretching from 1955 to 1979. As well as Bauer’s Betty Carter material, the present analysis also incorporates his transcription (2002b) of Louis Armstrong’s Heebie Jeebies and his

148 Scat syllables and markedness theory

citation of Reeve’s (2001)⁵ transcription of bars 49–55 of Louis Armstrong’s Hotter Than That scat solo. Note, however, that the Trager-Smith system adopted by Bauer has been transliterated here, following the transcription conventions detailed in Appendix 1. Two other helpful sources were Kernfeld’s (1995) transcription of Armstrong’s Hotter Than That and Bastian’s (Bastian and Alexander 1995) transcriptions of Chet Baker’s scat solos. As both these writers used different non-standardized representations (duh, day, doe, etc.) that were ambiguously interpretable, these were re-transcribed⁶ from audio files of the original recordings, following the principles in Appendix 1. This re- transcription is directly paired with the source transcriptions in Appendix 2. For the other songs (§3.1, §3.2), the transcriptions presented here are novel. It is worth foregrounding the complexity and relatively narrow focus of this task. Because the goal is to relate the articulatory expression of these singers to the range of phonological parameters that typologically characterize natural language systems, many features of the sophisticated manipulations of vocal tract sound are not represented in the relatively broad transcription system adopted here. Further, individual perceptions of the appropriate categorization of a constantly mutating cadence of vocables into segmental values may differ, as discussed in detail in Appendix 1. Given the paucity of literature on linguistic properties of scat, this preliminary study will hopefully “open the door” to further research into the nature of this interface.

2. The Phonological Properties of Scat

The analytical goal in this section is to examine the phonological inventory of onsets and codas⁷ in the scat syllables of the tunes documented in the database in Appendix 2, as well as to determine general properties of syllable shape in the output. Some challenges related to the fluidity of the medium or of individual expression are raised in the discussion of particular performances below. More general methodological issues pertaining to the classification of syllabic form are presented in Appendix 1.

2.1. Louis Armstrong

Of Louis Armstrong’s vast repertoire, an examination of two of his recordings from the early heydey of jazz in the 1920s serves here to establish a frame of reference both for Armstrong’s own style and for subsequent diachronic developments in scat.

⁵ Reeves, Scott. 2001. Creative Jazz Improvisation. 3d ed. Upper Saddle River, J.J.: Prentice-Hall. This resource was not available to me, and hence is cited only through Bauer’s (2002b) reference. ⁶ These were re-transcribed independently by myself and by a research assistant with both musical and linguistic training. Where there was variance in the transcriptions, either between us and/or with cited sources (e.g. §2.2), I assume sole responsibility for the interpretation adopted in this analysis. ⁷ Thus, the present focus is on consonantal patterns. For analysis of vowel quality in scat syllable nuclei, the interested reader is referred to Bauer (2002b), which presents detailed discussion of vocalic ‘timbre’.

149 Patricia A. Shaw

2.1.1. Louis Armstrong, Heebie Jeebies (1926)

Even a cursory look at the first 4 non-lexical syllables ([eə iyf gæf əmf]) that lead into the scat solo of Heebie Jeebies (see Appendix 2.1.1) suffices to identify them as unusual in comparison with the syllabic patterns which follow. Therefore, the analysis below focuses first on the subsequent 48 syllable tokens. The chart in (3) summarizes the findings about simplex syllable onsets. Consonants which are attested in onset position are in white cells, along with their raw frequency count. Possible, but unattested, onset consonants appear in shaded cells. Additional information about onsetless syllables and cluster behaviour is on the right.

(3) Onsets: ptč k ʔ No Onset: / b =  d =  ǰ g Onset clusters: sk =  f Ɵ sš h vðz ž mn l =  r =  yw Viewed against the full backdrop of the 24 consonants which can function as syllable onsets in English, the fact that 20 (83.3) are not used at all (viz. the shaded grey cells) clearly underscores the initial premise that scat is highly selective in its segmental inventory. Of the four segments [d, b, l, r] that do appear as onsets, [d] is the clear favourite, initiating 37 of the 48 syllables (77.1). As one might expect from the discussion in §1, the runner-up is [b] and although it trails far behind with only five appearances (10.4), its occurrence is nonetheless salient. The liquids [l] and [r] make an early appearance in syllables 5 and 7 respectively of this set of 48, followed very shortly (beginning with σ3 of bar 4) by a running stream of 19 consecutive [d]-initial syllables. Markedly heralding the start of a new phrase in bar 8, an initial [b] breaks the [d]-only alliteration, leading into an alternating b-d-b-d-d sequence. Then, after this cascade of 22 [d] onsets with only two [b] onsets having disrupted the auditory flow, in bar 9 the only consonant cluster hits: [sk]. Its alliterative sequencing (three in a row), its timing, and its composition all contribute to its striking impact. Nothing has primed the listener for an [sk] cluster. Although [sk] is not at all an uncommon English onset, in the context of the segmental composition of Louis Armstrong’s scat sequence here, it is totally deviant: neither [s] nor [k] occur anywhere else, either before or after, and it has unique status as the only onset cluster. Frequency, then, is significant—not only at the high end in terms of ascertaining what segments might most commonly appear in scat vocalization, but also at the low end in terms of observing what segments and/or combinations are drawn on only very rarely, to powerful effect. Although the vast majority of syllables in the Heebie Jeebies solo are open (39/48 = 81), the identity and frequency of the attested coda consonants is shown in the chart in (4). Of the 21 possibilities,⁸ only three appear, with [p] being the most common. Note that there is no overlap at all in the identity of the consonants that occur as onsets

⁸ As post-vocalic [w] and [y] appear only in diphthongs, they are not counted as possible codas.

150 Scat syllables and markedness theory

[d, b, l, r, sk] and those that occur as codas [p, m, t]. This is patently not an inherent characteristic of English (cf. words like pad, tab, mask, etc.), but will be seen to be a common characteristic, particularly of obstruents, in scat.

(4) Codas: p =  t =  č k No Coda: / bdǰ g Coda clusters: Ø f Ɵ sš vðz ž m =  n[ŋ]

l r (y) (w)

A final question is whether any particular syllabic forms, from a wholistic perspective, are preferred. In this 48-syllable sample, there are three favoured shapes: nine tokens of [də], eight each of [diy] and [duw]. Aside from these, there is remarkably little repetition of exactly the same phonological form in the residual 23 syllables. The frequency counts of the particular scat shapes are given in the following table:

(5) Frequency/ Syllable form Frequency/ Syllable form

 (18.8) də  (4.2) dɩp, daw, bə  (16.7) diy, duw biy, bam, bəp, duwt,  (2.1)  (6.2) dey, la, rɩp, ɩp, skiyp dɩ, dow skæm, skɩ

Having established this body of generalizations about onsets (3), codas (4), and over- all syllabic form (5), let us return to formally consider the properties of the introductory four syllables: [eə iyf gæf əmf]. Clearly the initial impression that these four syllables are unusual in the context of the entire scat sequence is indeed validated. Three of the four are onsetless, compared to only one of the 48 syllables that follow. The only onset consonant, [g], is unique: this segment appears nowhere else in the full scat database examined here. With respect to codas, note that there are no coda clusters anywhere else in the work, whereas this introductory sequence ends emphatically with an [mf] cluster. Moreover, the last three of these four syllables reiterate the coda [f]: not only are codas relatively infrequent in the rest of the work (there are only nine codas in 48 syllables: 18.7), but the particular segment [f] is unattested elsewhere as either a coda or an onset. Louis Armstrong’s choice of such unusual scat form in this quadra-syllabic bridge functions dramatically to grab the listener’s attention as Armstrong moves from the preceding English lyrics invoking everyone to “c’mon and do the Heebie Jeebies dance” to settle into the full-blown canonical scat syllables that follow.

2.1.2. Louis Armstrong, Hotter Than That (1929) The second Louis Armstrong tune analyzed here is the much longer 165 syllable scat solo from Hotter Than That (see Appendix 2.1.2), from which the excerpt cited earlier

151 Patricia A. Shaw in (2) was taken. Whereas bars 49–54, as seen in (2), draw exclusively on a sequence of [b]-initial syllables, a full count of onsets throughout the solo shows that [d] (= 78) is in fact used more frequently than [b] (= 57). [d] and [b] are by far the most prevalent onset consonants, with [b] exceeding the next ranked candidate [w] by a difference of 49. (6) Onsets:⁹ d (), b (), w (), l (), r (), n (), m (), y (), h (), t () No Onset: / Onset clusters: zw (), mw (), bw () Even in this work where the inventory of onsets stretches to 10 different segments, consistent patterns recur. For example, the four onsets attested in Heebie Jeebies, viz. [d, b, l, r] are all included within this larger set. Of the residual segments, all are attested— though with low frequency—in the other scat data investigated here, except [t]. The occurrence of [t] as an onset is unique not only in this song (in the second syllable of the otherwise uniform [d]-initial syllables in line BK⁵), but also in the entire sample of scat repertoire studied here. Moving to a consideration of the onset clusters attested in Hotter Than That, we encounter an interesting trio: [zw] (time 2:02), followed in the same line by [mw] and shortly thereafter by [bw]. Not only are none of these found elsewhere in the present database, but none of these /Cw/ sequences is part of the standard repertoire of English syllable onsets. Louis Armstrong here is clearly deviating from the canonical constraints on English well-formedness, and Native English listeners would, of course, attend to such novelty immediately. The hypothesis to be advanced here is that such cases illustrate a domain of tension between linguistic form and musical expression, where enhancement of the latter is achieved by violation of markedness constraints on the former. Consider next the coda inventory:

(7) Codas: p (), t (), m (), ṃ (), n (), l (), g () No Coda: / Coda clusters: Ø As was the case in Heebie Jeebies, most (76.4) of the syllables in this tune too are open. Although there is somewhat greater segmental diversity in the coda repertoire, it is still very limited: only six of the 21 possible consonantal codas are attested. There are no coda clusters. Interestingly, the three coda segments ([p, t, m]) that appear in Heebie Jeebies constitute a proper subset of the larger coda inventory here, with [p] again being significantly more frequent (2.75 times more; 52.4 of coda attestations) than its closest contender [t] (19.0). Three syllables in this work are realized exclusively as a “syllabic” [ṃ]. Apart from these cases, [m] functions once as an onset (see (6)) and three times as a post-vocalic coda. Interestingly all instances of [ṃ] follow a coda [w] or a [u] in the preceding syllable. The shared labial gesture across this sequence is a kind of harmonic pattern which recurs in various forms in other case studies below. Although [d] is attested as an onset segment 20 more times than [b], when one looks at which full syllable shapes recur most frequently, the two are pretty comparable:

⁹ For space reasons, for the rest of the discussion attested consonants will not be contextualized within the full inventory of English as in (3) and (4), but will simply be listed in rank order of frequency.

152 Scat syllables and markedness theory

[ba] edges out [də] by a count of 16 to 15. [bi] in its variant realizations (i.e. with length and/or homorganic glide) is tied with [da] at twelve occurrences each, then the favoured [d] takes over in the next most frequent syllables [di] and [du].

(8) Frequency/ Syllable form Frequency/ Syllable form

 (9.7) ba  (6.7) di ~ diː ~ diy

 (9.1) də  (4.8) du ~ duː ~ duw

 (7.3) bi ~ biː ~ biy, da Note that none of the most common syllables here have front/back lax or front/back mid vowels.

2.2. Chet Baker

Among the major scat artists through the decades, Chet Baker is renowned for the extreme minimalism of the consonant set that forms the basis for his scat improvisations. A comparison of different takes of the same tune, Everything Happens to Me, recorded more than three decades apart (1955 compared with 1989), illustrates remarkable consistency in the consonantal repertoire employed, despite major differences in the melodic and rhythmic structure. Transcriptions of the eight-bar scat bridge in these two versions are given in Appendix 2.2.1 and 2.2.2. Although Jim Bastian’s transcriptions (labelled JB) and my own (labelled PAS) differ in orthographic form,¹⁰ they are generally consistent in those features relevant to the present focus.¹¹ However, two domains of difference merit comment. One pertains to vowel quality: Chet Baker’s vocalization is extraordinarily mobile. The looseness and fluidity of movement in Baker’s vocalic articulation present significant challenges, such that the transcribed values that I propose are at best an approximation of a nuclear target range within the interconsonantal domain. What emerges most reliably is a general pattern of lax quality (primarily [ɩ ɛ ʋ æ ə]) and the predominant openness of syllabic form. The second notable difference between Bastian’s notation and mine pertains to consonants. Whereas Bastian remarks on the fact that Chet’s ‘scat vocabulary made predominant use of syllables beginning with the letter “D” ’ (Bastian and Alexander 1995: 4), not all [d]s are distinctly articulated with a full stop closure. In a number of cases, what is phonetically realized is the corresponding fricative [ð]. For example, the AIF wave file in (9) from the 1955 version (time = [2:37.6–2.38.4]) shows a sequence of two syllables,

¹⁰ Whereas Bastian’s orthographic interpretation is English-like, e.g. “ee” for [iː], the transcription I offer follows the principles in Appendix 1, with explicit representation of the more prominent glides but otherwise just length on the tense vowels. ¹¹ A discrepancy in bar 6 of the 1989 version is that JB documents 2 more syllables than I am able to discriminate. The present analysis is based on my total count of 53 syllables vs. Bastian’s 55. However, the strength of the generalizations is statistically robust, regardless of the difference in syllable count.

153 Patricia A. Shaw

the first with a clear [d] stop closure attack in comparison with the lack of full closure [ð] in the onset of the second syllable: (9)

JB: d eh d eh PAS: d ə ð ɛ This tendency is much more prevalent in the 1955 version, where of the 42 “D” onsets, 13 are realized as [ð]. In the 1989 version, only one of 45 “D”s is. It is entirely plausible that the phonological “target” in cases like the second consonantal onset in (9) is indeed a /d/, as consistently represented in Bastian’s transcriptions, but that its lenition to the smooth, non-punctuated continuant [ð] may reflect Chet Baker’s “airy”, “almost weightless”, “romantic crooner” style, disarmingly characterized as “being sweet talked by the void” (Bastian and Alexander 1995: 4). Invoking Sapir’s (1933) “psychological reality of the phoneme” argument, the hypothesis advanced here is that Bastian’s perceived “D” is interpretable as a more abstract level of representation, i.e. phonemic /d/, and that its sometimes lenited non-plosive phonetic realization as [ð] is a phonologically non-distinctive, surface level articulation. Consistent with this interpretation is the broad-based generalization in §1 that [d] is part of the standard scat repertoire; [ð] is not otherwise attested in any of the scat pieces by other artists studied here. In the analyses that follow, then, Baker’s [ð] articulations are taken to be epiphenomenal and are not independently represented in his scat inventory.

2.2.1. Chet Baker, Everything Happens to Me (1955) The onset repertoire of the early (1955) version of Everything Happens to Me reveals a highly skewed frequency distribution:

(10) Onsets: d (), y (), b (), h () No Onset:¹² / Onset clusters: Ø Ambisyllabic [tṇ] coda/onset: / Similar to what was seen in Louis Armstrong’s rendition of Heebie Jeebies (§2.1.1), where [d] initiates 37 of the 48 syllables (77), here /d/ accounts for 77.8 (42/54) of the onsets. Concomitantly, the relative infrequency of the residual segments raises questions as to their distribution and functional load. The next most frequent onset is [y]; it occurs only three times (3/54). Interestingly, the distribution of these markedly less frequent segments is often melodically significant. For example, both [h] and [b]—which occur only once each—

¹² The evaluation of No Onset status is challenged by Baker’s fluidity of articulation. Specifically, there are six cases in Baker 1955 and three in Baker 1989 where a coda [t] precedes a syllabic [ṇ]: as the [t] is interpretable as an ambisyllabic transition creating an onset for [ṇ], these are not counted as No Onset.

154 Scat syllables and markedness theory

appear in particularly prominent prosodic positions. Each is phrase-initial: the only occurrence of [h] introduces the second major phrase in bar 3, and the sole instance of [b], in the up-take into bar 7, initiates the final phrase of the scat bridge. Summarized in (11), the coda inventory is even more minimal.

(11) Codas: t (), ṇ (), n () No Coda: / Coda clusters: Ø Combining the consonantal repertoires of (10) and (11), we see that Baker’s 1955 improvisation utilizes a mere six segments from the full English set of 24 options: 25 of the available inventory. As observed in the previous works, here too there is a strong preference for open syllables (39/54 = 72.2). However, in contrast to Louis Armstrong, for whom [p] was the most frequent coda, Chet Baker does not use [p] at all, in either of the two scat performances examined here. Rather, his codas are exclusively alveolar [t, n], with [t] being the more prevalent. Somewhat parallel to the trans-syllabic gestural continuity of the feature [labial] leading into [ṃ] in Louis Armstrong’s Hotter Than That, there is a consistent homorganic pattern observed in the distribution of [ṇ] in Baker’s scat. Specifically, all instances of [ṇ] are immediately preceded by a homorganic coda [t]. Further, all cases of coda [n] or [ṇ] are followed by a homorganic /d/ onset of the subsequent syllable. In terms of syllable shape, Chet Baker’s preferred forms in this 1955 take are syllables where his near-ubiquitous /d/ combines with a non-low, non-high lax vowel:

(12) Frequency/ Syllable form  (35.2) dɛ(ː) ~ ðɛ(ː)  (16.7) də(ː) ~ ðə(ː)

2.2.2. Chet Baker, Everything Happens to Me (1989) Although by no means identical in rhythmic, melodic, or expressive form,¹³ the 1989 performance of this same song is remarkably consistent in its consonantal inventory. The most transparent difference in the onset repertoire is the fact that [b], used only once in the 1955 version, is completely absent in the 1989 take.

(13) Onsets: d (), y (), h () No Onset: / Onset clusters: Ø Ambisyllabic [tṇ] coda/onset: /

As seen in (13), the prevalence of /d/ in the 1989 version emerges as even more disproportionate, accounting for 85 (45/53) of the onsets. Clearly, /d/ in and of itself constitutes the core of Chet Baker’s consonantal inventory. Again, where another segment is used by Baker, it functions through its very uniqueness to demarcate a prosodically ¹³ For example, a very rudimentary comparison shows the opening bar in the 1955 version has 6 syllables moving from Ebm to A b+ towards Db ∆, whereas bar 1 in the 1989 recording has 10 syllables moduating from Fm through Bb towards Eb ∆. (Note: ∆ = major 7).

155 Patricia A. Shaw

prominent position. Thus, the sole occurrence of [h] introduces what is arguably the most prosodically salient position: the very first syllable of the first phrase of the scat bridge. In the 1989 version, Baker’s sparse and tightly restrictive treatment of codas is remarkably consistent with his 1955 repertoire, though their particular distribution in the scat melodic lines is entirely divergent.

(14) Codas: t (), ṇ (), n () No Coda: / Coda clusters: Ø As documented in (14), the same 2 segmental values are attested as in Baker’s 1955 coda chart in (11). Again, all three instances of [ṇ] are introduced by a dual function coda/onset [t], and are followed by a homorganic onset [d]. Given the fluid mobility of Chet Baker’s vowel articulations, a characterization of his favoured syllable shapes unequivocably identifies an open syllable with a [d] onset but is much less definitive in terms of vowel quality. Most generally, as in the 1955 version (see (12)), his articulation meanders around a mid lax vowel, either schwa [ə] or a ‘neutral position’ [ɛ], identified for English as the characteristic articulatory setting for the onset of speech (Chomsky and Halle 1967). However, on notes of longer duration, his resonant crooning often ascends to a tenser high back [u]. Based on the transcriptions in Appendix 2, there is considerable consistency between the 1955 and 1969 versions in terms of a frequency of use ranking:

(15) 1955 Frequency/ Syllable form 1969 Frequency/ Syllable form

 (35.2) dɛ  (17) dɛ  (16.7) də  each (13.2) də, du, dʋ However, as is evident from the lower frequency numbers and the three-way tie for second place in the 1969 count, there are no strongly identifiable constraints on his wide- ranging vocalic diversity.

2.3. Betty Carter ¹⁴

Even as the repertoire of scat vocabulary expanded through the creatively explosive bebop rush of the 1940s, [d] and [b] remained particularly prominent. For example, although Betty Carter was a major innovative force in extending the repertoire of jazz vocables, Bauer notes that in Carter’s short scat solo in Babe’s Blues (1958), of the nine consonants which are used as syllable onsets, /b/ and /d/ together “initiate more than half of the vocable classes used in the solo” (2002b: 312). Other Betty Carter songs attest to this same generalization: in my count of the 197 syllables in her 36-bar scat solo rendition of You’re Driving Me Crazy (1958; transcribed by Bauer 2002a: 252–254), the most frequent onset consonant is [d] (in 80 of the 197 syllables) and the next most frequent is [b] (in 36 of the 197 syllables). Thus, although Carter uses 10 different consonants as onsets in

¹⁴ My commentary on Betty Carter is deeply indebted to Bauer’s (2002a,b) insightful and superbly documented interpretation of her life and work.

156 Scat syllables and markedness theory

this solo, the two segments [b] and [d] together comprise the majority (58.9) of onset choices. In the following sections, we look at two of her other tunes to broaden the base of comparison further.

2.3.1. Betty Carter, Thou Swell (1955) Recorded the same year (1955) as the early version of Chet Baker’s Everything Happens to Me that was considered in §2.2.1, Betty Carter’s scat rendition of the original 1927 classic Thou Swell draws on the following inventory of eight consonants as simplex onsets (see Appendix 2.3.1).

(16) Onsets: d (), b (), l (), y (), w (), h (), r (), š () Onset clusters: ly (), dl (), sp () No Onset: /

Constituting a combined total of 75/114 (=65.8), the consonants [d] and [b] are reaffirmed as incontestably at the core of Carter’s—and everyone else’s—stock of scat resources. Although less frequently drawn on, the consonants [l, r, y, w, h] are all familiar as staple scat segments that have been attested in the work of Louis Armstrong and Chet Baker examined in the preceding sections. The innovative element in (16) is Carter’s once-only exploitation of [š] (bar 13, coupled with the unique attestation of [r] in the sequence [šiy ra]). The use of [š] is rare in Betty Carter’s scat, although it figured prominently in the influential repertoire of and became a flagship marker of 1950s doo wop motifs like “shoo bee doo” and “sha na na”.¹⁵ Despite the collective recognition among jazz artists of certain segments being standard communal property in the scat arsenal, other specific sounds acquired the status of individual trademarks. Carter reportedly admonished a young vocalist in 1978: “Why are you using scat syllables like ‘shoo-bee-doo-bee’? Those belong to Sarah, and they belong to the fifties.” (Berliner 1994: 254, 804, cited by Bauer 2002b: 314–315) At the heart of improvisional creativity in music, as in language, is the challenge of innovation under the constraints of structural limitations, critically the inventory of segments and restrictions on their combination. Given the very small set of sounds that came to be established as the “conventional” scat inventory in the works of the early artists, to then have certain consonants among these evolve into sound symbolic associations with a particular singer and/or decade effectively heightens the challenge for new artists to create an individualistic scat voice. Onset clusters are generally quite rare in scat. Of the four that occur in this work, only one [sp] conforms to standard well-formedness constraints of English. The other two, [ly] and [dl], draw on segments that are very common in the scat inventory of onsets, but in bundling them into tauto-syllabic onset sequences Carter pushes beyond the canonical bounds of regular English. Just as [š] became a Sarah Vaughan scat trademark, the [dl] onset is a strong candidate for a Betty Carter signature: jumping ahead 27 years ¹⁵ It was from the vocals in the Silhouettes’ 1957 hit song Get a Job that the 50s revival group, Sha Na Na, took its name. ¹⁶ What a Little Moonlight Can Do (1982) Whatever Happened to Love? Verve/Polygram 835 683–1; see transcription by Bauer (2002a: 310–343).

157 Patricia A. Shaw

to her 1982 recording¹⁶ of What a Little Moonlight Can Do, this same highly marked onset appears eleven times, most strikingly in a sequence of six syllables in the climactic scat line of bars 189–190 (WB line as transcribed by Bauer 2002a: 317; transliteration (2nd line) as in Appendix 1):

(17)  WB: | ‰ ə weh– dlow dlow |– dlə dle dle | dlow dow Œ | ... | ‰ ə wɛə– dlow dlow |– dlə dle dlɛ | dlow dow Œ | ...

Carter’s usage of codas in Thou Swell is infrequent, as seen in (18), and the observed patterns are familiar. She draws strictly on the resonants [m, n, l]. Both of the syllabic segments are alveolar, and follow a homorganic onset [d]. (18) Codas: m (), ṇ (), l (), ḷ () No Coda: / Coda clusters: Ø The syllable shapes which surface most frequently in this piece are not at all surprising either:

(19) Frequency / Syllable form Frequency / Syllable form

 (13.2) bə  (10.5) ba

 (12.3) duw  (8.8) də In sum, despite the creative uniqueness of how her scat artistry uses them, Carter’s arsenal of tools as represented by this acclaimed 1955 performance draws on a markedly standard repertoire.

2.3.2. Betty Carter, Open the Door (1979) Based on his intimate and broad-based musical insights into the full body of Betty Carter’s “relatively small recorded output”, Bauer (2002a: xi) contends “that the defining features of Carter’s style remained consistent even as her approach kept changing.” From the linguistic perspective of the present study, a comparative analysis of the scat interludes in Carter’s 1979 version of Open the Door (see Appendix 2.3.2), recorded 24 years later than the 1955 work discussed above, reveals a tightly focussed phonological repertoire. The five onset segments that appear in the 1979 version, documented in (20), are a subset of the eight that were used in Thou Swell (see (16)).

(20) Onsets: d (), y (), w (), l (), h () Onset clusters: Ø No Onset: /

Notably absent from the attested onsets in (20) is [b]. However, the ubiquitous scat onset [d] is not only present, but strongly dominant, introducing 19 of the 27 syllables (= 70.4). The other onset segments here, viz. [l, y, w, h], are all scat basics, not just in Carter’s earlier work, but in that of other scat vocalists. Of particular interest in (21) is the total absence of post-vocalic coda consonants.

158 Scat syllables and markedness theory

(21) Codas: ṇ (), ṃ () No Coda: / Coda clusters: Ø The only syllables that are not open CV or CVG structures are the 3 cases where there is a syllabic nasal. A comparison of the first three scat lines (cf. bars 9, 14, and 16, respectively, in Appendix 2.3.2) reveals a striking and doubtless strategic parallelism of form and function where these three syllabic nasals occur. Specifically, each is in absolute phrase-initial position of the first three scat lines, with each new cycle entailing some minimal variation from the preceding one: labial [ṃ] in the first phrase shifts place of articulation to alveolar [ṇ] in the second phrase, which itself is repeated in the third line but differentiated by the introduction of an [h] onset. Abstracting away from rhythm, duration, and pitch, the segmental content of these three lines is reproduced below:

(22) ṃ də dow ... ṇ duw duw duw ... hṇ duw duw diy duw ... What this short prosodic progression illustrates is that far from scat being comprised of randomly articulated sequences of a delimited set of nonsense syllables, the skill of a brilliant scat artist like Betty Carter entails masterful structuring of content and sequence: here, each nasal syllable introduces an iteration of exclusively [d]-initial syllables, and each line builds substance and momentum by adding one more syllable. Finally, in determining which syllable shapes are most prevalent, there are two that clearly emerge as most frequent:

(23) Frequency/ Syllable form

 (29.6) du(w)  (25.9) dey While [du(w)] figures prominently in the repertoire of her other work (cf. (19)) and that of the other singers sampled here, [dey] is less favoured, though not unattested (cf. (5)).

2.4. Syllable Structure Generalizations

Having documented specific aspects of syllable content and form in two different works from each of three renowned jazz vocalists, spanning the 63 years between 1926 and 1989, we are now in a position to determine what generalizations, if any, hold across this sample, despite each artist’s highly individualistic musicianship and distinctly unique approach to the idiom. The initial question posed in §1 was to what extent the delimitation of onsets to [d] and [b], as exemplified by the brief excerpts in (1) and (2), is representative of a broader database of scat. The onset tabulations from each previous section (viz. (3), (6), (10), (13), (16), (20)) are summarized in the table in (24) below. Note in (24) that the frequencies of [d] and [b] are given both as a token count and as a percentage value of the number

159 Patricia A. Shaw of scat syllables in each piece. There are three particularly interesting facts revealed by these results. First, none of the six works studied here—including the full texts of each of the classic performances from which (1) and (2) were drawn—uses exclusively [d] and/or [b] onsets. In every case, the vocalist has chosen some scat syllables with other onset consonants, however minimal this extended range may be. For example, in cases like Louis Armstrong’s 1926 recording of Heebie Jeebies (§2.1.1) and Chet Baker’s 1989 version of Everything Happens to Me, there is only one occurrence of each of two other onsets. (24) Simplex Onsets: comparative usage by different scat vocalists

dbyhlwrmnšt 2.1.1. Armstrong 1926 37 = 77.1 5 = 10.4   2.1.2. Armstrong 1929 78 = 47.3 57 = 34.5         2.2.1. Baker 1955 42 = 77.8  = 1.9 3  2.2.2. Baker 1989 45 = 84.9   2.3.1. Carter 1955 40 = 35.1 35 = 30.7       2.3.2. Carter 1979 19 = 70.4     Total syllables: 461 261 = 56.6 98 = 21.3          At the other end of the spectrum, Louis Armstrong’s Hotter Than That employs the greatest diversity: ten different consonants. A further observation is that there are only nine consonants other than [d] and [b] which comprise the full set of onsets that are collectively utilized by these artists. Together, these latter two facts affirm the initial premise of this research: of the full complement of 24 consonants that can potentially function as syllable onsets in English, scat draws on a very limited, and largely recurrent, subset. The third conclusion that emerges from (24) is that there is a consistent asymmetry in the relative frequency of [d] over [b]. In two of the songs (§2.2.2, §2.3.2), there is no [b] at all; in a third (§2.2.1), there is a single attestation; in the remaining three, though the degree of imbalance differs, the direction of difference is constant. In contrast to the robust generalizations about simplex onsets, the usage patterns with respect to complex onsets, as summarized in (25), do not at all cohere. (25) Complex Onsets: comparative usage by different scat vocalists

sk zw, mw, bw ly, sp, dl σ’s with CmplxOns 2.1.1. Armstrong 1926  / 6.25 2.1.2. Armstrong 1929 , ,  / 1.8 2.2.1. Baker 1955 / --- 2.2.2. Baker 1989 / --- 2.3.1. Carter 1955 , ,  / 3.5 2.3.2. Carter 1979 / --- / 2.17

160 Scat syllables and markedness theory

Clusters appear in only three of the six pieces, with an extremely low frequency count (averaging just over 2). Significantly, there is no overlap at all in the specific clusters used in each of the works, even by the same singer. Moreover, with respect to the identity of segments involved in these clusters, it is patently not the case that these sequences are compositionally built from the simplex onset consonant inventory: [k], [p], [s], and [z] in the clusters of (25) are not part of the repertoire of (24). Most striking is that a majority (5/7) of the attested clusters violate canonical English patterns: although all of the individual segments involved are legitimate potential simplex onsets, none of [zw], [mw], [bw], [ly], or [dl] conform to standard “well-formed” sequences in English. Across these diverse observations, there is in fact a consistent generalization, namely: complex onsets are highly marked. In terms of frequency they are rare, and in terms of content they are often exceptional. Consider now the properties of codas, summarized in (26). Whereas the excerpts in (1) and (2) in §1 were comprised exclusively of open syllables, this generalization does not hold of any single work considered in its entirety. Nonetheless, open syllables are unequivocably dominant, ranging from 94 to 72 in individual works and with the overall average being 82.65. (26) Codas: comparative usage by different scat vocalists

ptnṇ m ṃ l ḷ g σ’s with no Coda 2.1.1. Armstrong 1926   / 81.25 2.1.2. Armstrong 1929        / 74.5 2.2.1. Baker 1955    / 72.2 2.2.2. Baker 1955   / 86.8 2.3.1. Carter 1955     / 93.9 2.3.2. Carter 1979   / 88.9 Totals:          / 82.0

Moreover, there were no complex codas. With respect to segmental identity, of the 21 potential English coda consonants, only six different segments appear. Compared with the inventory of scat onsets in (24), it is interesting to note that there is overlap in the resonant repertoire /m, n, l/, but complementarity in the obstruent stops: onset /b, d/ vs. coda /p, t/.¹⁷ Once again, Louis Armstrong is the king of segmental diversity in his Hotter Than That rendition (§2.2.1), which draws on seven different codas, whereas the other artists employ a much more restricted range of between two and four. Across the artists, the most favoured segments are [t] and [n/ṇ], although Armstrong’s clear personal favourite is [p]. Finally, consider in (27) the generalizations that hold regarding the overall form of scat syllables that are used by these diverse singers. Only syllables which occurred at least three times, and with greater than 7 frequency in each song are included in the

¹⁷ Thus, the unique instances of onset /t/ and coda /g/, both in Armstrong’s Hotter Than That (§2.1.2) appear anomalous: the /g/ in terms of both place and voicing, and the /t/ in terms of voicing.

161 Patricia A. Shaw table below. Because the overall syllable count differed considerably across the different selections, the most frequent syllables for each artist are simply ranked, with 1 being the most frequent. Ties are represented by the same number. The scale descends for each artist, but may stop at either 2, 3, or 4 depending on the actual frequency values (as detailed in the corresponding tables in each individual section above). Thus, for example, for each of Chet Baker in §2.2.1 and Betty Carter in §2.3.2, the very high frequency of two particular syllables results in no others exceeding the criterion level. (27) Syllables: comparative usage by different scat vocalists (<7, Ranked =high)

də du(w) ba dɛ da bi(y) di(y) dʋ dey bə 2.1.1. Armstrong 1926   2.1.2. Armstrong 1929  2.2.1. Baker 1955  2.2.2. Baker 1989    2.3.1. Carter 1955   2.3.2. Carter 1979  Comparing the scat choices in the two different recordings by each vocalist shows Baker to be the most consistent, despite the 34-year time interval between these performances: [dɛ] and [də] rank 1 and 2, respectively, in both. In contrast, in each of the two recordings by Armstrong and Carter, different syllables rank 1 and 2, and for each of them, the top- ranked syllable in one of their pieces does not even reach criterion in the other, viz. [ba] in Armstrong, and [bə] in Carter. Clearly, there is no single “favoured” syllable shape: in this sample of six scat performances, there are five different syllables that emerge as the most frequently used in any given piece. Nonetheless, the tabulations in (27) provide striking confirmation of the two generalizations originally observed in the brief Louis Armstrong extracts in (1) and (2). First, all these favoured syllables are open. Secondly, they all start with [d] or [b]. The over-all favourite is [də], with [du(w)]—as in “doo-wop”—coming in second.

3. Beyond Bebop

As bebop morphed into hard bop and doo-wop in the 1950s, and classical jazz of the previous decades diversified under multiple influences, particularly R&B and the explosive impact of rock’n’roll, jazz scat began to wane in popularity. A few exemplary jazz vocalists continued the bebop scat tradition, but other genres had come to dominate the pop music scene. Not until the uniquely versatile creative talents of Al Jarreau and Bobby McFerrin emerged in the late 70s did vocal improvisation once again top the charts. The next two tunes come not from core vocal jazz repertoire, but from the heart of the early 60s Hit Parade era, in the decade following the bebop heyday. What makes these works substantially different from the preceding ones is that scat is formally scripted into the lyrics, not improvised: in the first example (§3.1), scat syllables are explicitly

162 Scat syllables and markedness theory

referenced in the English text, and in the second example (§3.2), more extensive scat lines alternate with English. These case studies are of interest in two respects: first, for interrogating the extent to which the segmental content and shape of these select scat tokens conform to the generalizations established for the classical scat vocables examined in the preceding vocal jazz tunes; and secondly, for the insights that this phenomenon provides from a historical perspective on the evolving diversification of the cultural impact of scat. Despite bebop itself having shifted out of the popular mainstream at that time, the fact that very young¹⁸ creative songwriters chose to incorporate scat syllables into their lyrics in the 1960s reflects its strong formative influence on their own musical identities and its enduring legacy in the broad-based musical culture of the era.

3.1. Barry Mann, Who Put the Bomp? (1961)

The infectiously popular music and words of this 1961 hit were co-written by Barry Mann and Gerry Goffin, with Barry Mann as the original recording artist. Because the lyrics here are not improvised, but rather are composed in conformity with a tightly structured, fixed melodic and rhythmic framework, the methodology of previous sections—namely, a frequency count of attested segmental tokens in a stream of spontaneously improvised scat—is less revealing than simply the inventory of segments and syllable shapes that are drawn on. That is, what is particularly significant is just which scat vocables are chosen for the lyrics, as this very choice implies that these particular forms already (in 1961) had significant currency in the general public domain as cool and hip. Archetypal and high-profile scat syllables here (see Appendix 3.1) include the [šu] ~ [šə] attributed to Sarah Vaughan (§2.3.1), the [dɩp] that surfaces as early as Heebie Jeebies (see (5)), as well as the [bap] that not only persists to this day as the name of the genre, but that had become the basis of Betty Carter’s moniker: Betty Bebop.¹⁹ The rhythmically alternating syllables [bə] (line 2) and [dɩ] (line 8) are clearly canonical scat form, adhering to both the preference for [b]/[d] onsets and No Coda (open syllable). Although it was noted in every improvisational jazz sample investigated earlier that open syllables were much more frequent than closed syllables, a superficially inconsistent observation is that the reverse is the case in Who Put the Bomp?. What this illustrates, I would suggest, is the potential over-riding effect of prosodic constraints when scripting lyrics to a fixed melodic line and rhythmic beat. The lyrics for the lines with the scat syllables [bam], [bap], [dɩp] are basically structured as follows, with the CVC closed syllables out-numbering the open CV syllables four to two. Each of the underlined syllables in (28) is directly synchronized with a rhythmic beat. (28) Who put the [CVC] in the [CVC] [CV] [CVC] [CV] [CVC] ? Because closed CVC syllables are prosodically heavier, aligning a closed syllable with a rhythmically strong position functions to enhance the prominence of the beat. ¹⁸ Barry Mann and Gerry Goffin were 22 when their co-written success Who Put the Bomp? was released, and teen idol Johnny Cymbal was 18 when he wrote and recorded Mr. Bass Man. ¹⁹ This was ’s nickname for her, despite her expressed dislike of it (Bauer 2002a: 4–5).

163 Patricia A. Shaw

Note that the initial ‘who’ [huw], even though open, is also heavy by virtue of the long/ tense diphthong. Further enhancing the strong rhythmic stability of these lines is the fact that the light open scat syllables [bə], [šə], and [dɩ] are never aligned with the beat. While this kind of prosodic alignment of heavy syllables with positions of rhythmic prominence, and the complementary preference for light syllables in weak rhythmic positions, most certainly occurs in improvised scat as well, it would appear to be a significantly less dominant factor, perhaps since rhythm itself is also subject to improvisation. Of further interest in the lyrics of this song is that there is a category of forms that are neither standard English lexical items nor syllables that conform to the characteristics of scat. Concatenations of essentially semantically empty compounds that rhyme or alliterate, such as “rama lama”, or that carry some onomatopoetic value like “ding dong”, or that live on a hip fringe of the English lexicon like “boogity boogity” were also drawn from the pop music scene of the 50s, namely the Edsels’ major doo-wop hit Rama Lama Ding Dong, originally released in 1958 on Dub Records and re-released on Twin Records in 1961,²⁰ and the Quincy Jones composition Boogity Boogity, recorded on Milt Jackson’s 1958 Plenty, Plenty Soul. Unlike scat, these sequences each pattern basically as a lexicalized unit, without independent freedom of realization of the constituent syllables. The form of all 3 of these expressions is essentially reduplicative, with the nature of any deviance from full identity falling directly within recognized cross-linguistic patterns of reduplication (e.g. Moravcsik 1978, McCarthy and Prince 1986, Hurch 2005). Finally, based on the generalizations established in §2, some of the segmental content in these examples falls markedly outside of that found in core scat, viz. the [ŋ] codas in “ding dong”, and the [g] onset in “boogity”. In sum, Who Put the Bomp? is highly syncretic in its explicit references to many of the rapidly evolving musical influences of the era. The lyrics integrate unmistakably identifiable scat syllables from the classical vocal jazz tradition, with references from the progression into doo-wop, along with the blues-based “modern jazz” sophistication of Quincy Jones and Milt Jackson. What this tells us is that although the pure jazz scat genre itself isn’t charting in the mainstream at this point in time, it remains a major foundational force in the broader musical scene. Moreover, of all the diverse genres referenced in these lyrics, it is a scat line that is attributed with ultimate success in the conquest of love: “When my baby heard bam bə bə bam bə bam bə bam bam, every word went right into her heart...”

3.2. Johnny Cymbal, as recorded by Sha Na Na, Mr. Bass Man (1963)

The second example illustrating the continuing legacy of scat in the pop scene of the early 60s is Johnny Cymbal’s signature song, Mr. Bass Man. Sha Na Na’s re-recording of it in ²⁰ Although the Edsels, like the ill-fated car model they named themselves after, were defunct as a group by the time their version of Rama Lama Ding Dong rose to prominence on the national charts, the song itself attained significantly greater longevity as the title song of Sha Na Na’s 1980 album. Note too that in the historical context of the 50s “Ding Dong” itself carried an established frame of reference from the title and lyrics of Louis Armstrong’s early 1930s hit, I’m a Ding Dong Daddy From Dumas (on The Best of Louis Armstrong and His Orchestra: 1930-31. Classics B000001NJB).

164 Scat syllables and markedness theory

1980 stands both as a tribute to its enduring popularity and as a major contribution to ensuring its continued exposure to subsequent generations. The transcription in Appendix 3.2 is based on the Sha Na Na version, and differentiates the scat lines that are sung by Mr. Bass Man himself (abbreviated BM in Appendix 2) from the fledgling attempts of the “Wanna-be” guy (abbreviated W in Appendix 2) who sings, following line 9, “I wanna be a bass man too”. Interestingly, this separation reveals some fascinating differences. As seen in (29), Mr. Bass Man himself uses exclusively [b] onsets. In contrast, the majority of Wanna-be’s onsets are [b], but his inventory also includes a substantial number of [d]s and [y]s, both of which accord with the standard scat onsets documented in (24). Although [ʔ] is not included in (24), its absence is directly attributable to the transcriptional principles outlined in Appendix 1, so the two attestations of [ʔ] here are not anomalous. The unique occurrence of [s] at the beginning of line 5 is odd, given the generalizations of (24), but may be explicable as perseverance of the final sibilant of the immediately preceding word “songs”, across the juncture from English lyrics to scat. (29) a. Mr. Bass Man’s scat lines (including back-up line and joint BM/W lines): Onsets: b () Onset clusters: Ø No Onset: Ø/ b. Wanna-be’s scat lines: Onsets: b (), d (), y (), ʔ (), s () Onset clusters: Ø No Onset: Ø/ Not only is the greater diversity of segments in the novice’s attempts of interest, so too is the distribution of these segments. For example, in three lines (lines 5, 6, 17), Wanna- be switches in mid-sequence from [d]-onsets to [b]-onsets (significantly, a switch to the correct target), but never does he switch in the opposite direction. All other lines are either exclusively [d] (lines 13, 14, 25, 26) or exclusively [b] (3, 7, 10, 12, 22, 24, 29). There is also a marked discrepancy in coda patterns between Mr. Bass Man and Wanna-be. Mr. Bass Man uses exclusively [m]/[ṃ] codas, whereas Wanna-be models [m] most frequently, to be sure, but he also draws on the 3 most favoured scat codas that were documented in (26): [p, t, ṇ]. Nonetheless, note that Wanna-be’s very last solo line achieves perfect canonical form as defined by Mr. Bass Man: exclusively [b] onsets and exclusively [m] codas. (30) a. Mr. Bass Man’s scat lines (including back-up line and joint BM/W lines): Codas: m (), ṃ () Coda clusters: Ø No Coda: / b. Wanna-be’s scat lines: Codas: m (), t (), p (), ṇ () Coda clusters: Ø No Coda: / Although Wanna-be uses a broader inventory of both onsets and codas, these segments are significantly constrained in their distribution, in that a consistent pattern of syllable-internal consonant harmony obtains with respect to place of articulation in closed syllables. That is, a labial [b] onset is followed by a labial [m] or [p] coda, regardless of the vowel quality in the nucleus, e.g.: bam, bum, bəm, bom, bʋm, bɩp. Similarly, an

165 Patricia A. Shaw

alveolar [d] onset is closed by [ṇ] or [t].²¹ Given that none of the other onsets /y, ʔ, s/ occur in closed syllables, this generalization regarding intra-syllabic consonant harmony holds throughout the entire work.

4. Explanatory Hypotheses

The analyses of these several examples of scat show that, across the diversity of musical styles and individual expressions, the repertoire of sounds and syllable shapes is remarkably consistent and extremely limited in comparison to the extensive range of segments and combinatorial possibilities that are used in English, let alone available within the articulatory range of the human vocal apparatus. To address the question of what might account for these patterns, three hypotheses are explored: that vocal scat is essentially imitative of instrumental jazz (§4.1); that the repertoire of sounds in scat are constrained by phonological markedness theory (§4.2); and that scat production is subject to independent constraints on musical form and vocal performance (§4.3). Although each of these, among other cognitive and performative factors, doubtless contributes to shaping the output of scat, the argumentation to follow suggests that specific tenets of phonological markedness theory interacting with the melodic imperative for a voice line to carry pitch contribute substantially to broadening our understanding of the attested patterns.

4.1. The Imitative Hypothesis

A number of theorists within the musical literature have hypothesized that scat vocalization is essentially imitative of jazz instrumental expression. For example, Robinson (2002: 515) attributes the origin of scat to “singers imitat[ing] the sounds of jazz instrumentalists”. Bauer (2002b: 303) cites Milton Stewart (1987: 65, 68, 74) as showing that “the vocables used by such notable exponents of scat as Ella Fitzgerald and Sarah Vaughan often mimic the tonguing, phrasing, and articulation of instrumentalists.” Stoloff (2003: 4) notes that Louis Armstrong, “like many other ‘instru-vocalists’ who followed, unconsciously used scat syllables that emanated from his trumpet style”. The core question in considering the “Imitative Hypothesis” is to what extent such comparisons are based on essentially arbitrary associations, as opposed to qualities of instrumental sound production that are directly reproduced in the choice of consonants and vowels in a scat syllable. That is, are there consistent, independently verifiable articulatory correlations between an instrumental rendition and a particular scat vocalization? Or, like the arbitrariness of the sound-meaning correspondences in natural language, is the seemingly “imitative” association based on fundamentally arbitrary, conventionalized interpretations? One type of case is illustrated by the fact that sometimes hand gestures lent an explicit instrumental identity to the vocables. Stoloff (2003: 5) points out that “Ella, for example, often used trombone-like hand motions while scatting “du-wah” type syllables”.

²¹ All 3 instances have the same vowel: [dɩt].

166 Scat syllables and markedness theory

The question here then is whether there is anything inherent in the phonetic properties of the syllables “du-wah” [du wa] that is uniquely representative of the production or perception of trombone sound, or whether the explicitly iconic identification established by Ella’s hand gestures substantially contributes to creating a conventionalized significance. Weighing against a one-to-one interpretation of the Imitative Hypothesis is the fact, noted earlier in (27), that [du(w)] is the second most frequent syllable used by Louis Armstrong in Heebie Jeebies, Chet Baker in Everything Happens to Me (1989), and Betty Carter in Thou Swell. In other words, the documentation in §2 establishes that throughout the scat repertoire, [du(w)] is simply an extremely common syllable. What seems most plausible, then, is that a “du-wah”/trombone sound-meaning connection evolved into a conventionalized relationship, with the explicit interpretive overlay of hand gestures contributing significantly to establishing this as a semi-lexicalized associative correspondence. A second type of case exemplifying the frequent interpretation of scat as directly representative of instrumental effects is illustrated by Robinson’s (2002: 515) identification of the following line from Louis Armstrong’s Hotter Than That (1927, OK 8535) as one “which illustrates his clear imitation of a trumpet rip”: (31) From L. Armstrong Hotter Than That (1927); transcription J.B. Robinson:

A basic question here is: How much of the interpretation of this phrase being a “trumpet rip” follows from the initial monosyllabic identity tag “rip”? First, the research documentation in §2 establishes that “rip” is not in the common inventory of scat syllables. In fact, it is a unique attestation in the database of 461 scat syllables. Secondly, “rip” is a recognizable English word, with a particularized semantic interpretation specifically within the jazz lexicon. Thirdly, this word is positioned strategically at the very beginning of the scat sequence that is interpreted by Robinson as “a trumpet rip.” In terms of perceptual salience, initial position is the locus of greatest prosodic prominence in the phrasal domain. Moreover, note in (31) that “rip” bears the highest pitch level and its rhythmic value (a quarter note) is twice the value of each individual note in the sequence of eighth notes that follows. Collectively these prosodic cues of position, pitch level, and duration converge to focus the listener’s attention on this entry, which is realized not by a familiar scat syllable, but rather by the lexically informative label that this is a “rip”. Finally, a complementary question stemming from Robinson’s characterization of this sequence as a “trumpet” rip, is whether there is anything in the choice of the particular scat syllables—independently of the lead signifier “rip”—that is uniquely associated with a trumpet, as opposed to a sax, bass, or any other instrument. Again the collective evidence in §2 establishes that the specific syllables that follow “rip” in (31) are all unequivocably canonical scat, used by a diversity of singers across a diversity of melodies, chord progressions, tempos, and rhythms. Nonetheless, the fact that it is Louis Armstrong himself, one of the most virtuoso jazz trumpeters of all time, who is scatting in (31) unquestionably establishes an association

167 Patricia A. Shaw between his vocal and instrumental expression. Of course, a particular musician’s primary instrumental identity would not preclude scat excursions into imitative or evocative effects of other instrumentation. However, one might ask: given that Chet Baker and Louis Armstrong are both jazz trumpeters and scat vocalists, is there any significant parallelism between them in the choice of scat repertoire? Comparison of their use of onsets in the chart in (24) and of codas in (26) not only provides a distinct profile for each, but also establishes no greater similarity between them than between either one of them and Betty Carter, who was not a trumpet player. In short, the research evidence here argues that the specific choice of scat syllables for each of these performers follows a canon of phonological constraints on scat repertoire that are independent of trumpet—or any other—instrumental realization. Most fundamentally, I would submit, it is the musical individuality of each of these artists and their unique creative mastery of the cognitive systems involved that transcends defined conventions on the essential form of notes and syllables, and systemic constraints on their patterning. However, to explore the empirical bases of the Imitative Hypothesis yet further, consider commentary such as that advanced by Stewart (1987: 65–66), who interprets Ella Fitzgerald’s 1949 performance of Flying Home as follows: Fitzgerald alternates the bilabial ‘b’ and ‘p’ plosives with the lingua-alveolar ‘d’ plosives. The ‘b’ and ‘p’ sounds are formed similarly to the sounds of jazz wind instruments, which sound by the release of built-up mouth air pressure onto the reed, while the ‘d’ sound is similar to the tonguing on jazz brass instruments.

On the basis of a phonological model of natural language sound production, my hypotheses about the articulatory correlations entailed in initiating and modifying air flow on reed and brass wind instruments differ from Stewart’s. Specifically, pitch-based sound on a trumpet or any other brass instrument is produced by bilabial constriction: labial is the primary articulator. As well, tonguing effects—most commonly coronal, but also dorsal—function significantly to modify the stream of sound in terms of attacks, closures, trills, duration, phrasing, tonal quality, etc. Less frequent, but certainly available within the repertory of articulatory modifications, are uvular and laryngealization effects. Consequently, under an articulatorily-based Imitative Hypothesis, trumpet-denotative scat would liberally draw on a inventory of both labial and coronal consonants, but could also include other articulatory effects. In contrast, in producing the primary sound on a reed instrument, like a sax or clarinet, the player’s lips and upper teeth hold the mouthpiece: although lip compression can modify pitch, tone, or timbre, labial is not a primary articulator in the way that it is with brass instruments. However, the range of tonguing effects and other articulatory modifications would be similar. The Imitative Hypothesis implication that follows from this comparison would be that sax- or clarinet- imitative scat should have no [p]s or [b]s (contra Stewart’s interpretation above), whereas brass-imitative scat could. Essential to testing such articulatory-modeling claims would be a body of data where the intentionality of the scat singer is unambiguous. As none of the references drawn on here provide adequate documentation to explore these hypotheses more definitively, they are left for future research.

168 Scat syllables and markedness theory

In summary, despite various approaches to the hypothesis that scat vocalization is essentially imitative of jazz instrumental expression, what has been shown is that there is in fact little empirical evidence to sustain a non-arbitrary relationship in the form of realization across the two modalities. Moreover, compared with the huge range of distinct combinatorial possibilities in jazz instrumentation, whether articulated by mouth, hand, valve, slide, bow, or mallet, the exceedingly small set of segments in the core repertoire of scat presents a striking contrast. What the Imitative Hypothesis fundamentally fails to explain is why the rich diversity of instrumental sound is not more extensively mirrored in scat. The possible articulatory range of the human vocal apparatus far exceeds what is found in human language systems, let alone in scat. Moreover, even the much more limited range of segmental and combinatory possibilities in the English phonological system significantly exceeds what is found in scat. The fundamental question then is what hypotheses might offer a more insightful and constrained explanation for the small and remarkably consistent inventory of segments and syllable shapes that characterize scat. In the next section it is argued that phonological markedness theory constitutes a productive basis of inquiry.

4.2. Markedness Theory

From a linguistic perspective, the framework of phonological markedness theory embodies a number of hypotheses against which these empirical generalizations about scat can be evaluated. It is markedness theory that negotiates the interface of fundamental questions regarding linguistic diversity vs. universality, seeking to understand across the manifest differences of human languages just what properties of language may be universally attested, what properties may be correlated with or implicated by another property, and what properties are rare or may in fact never be attested. The basic premise to be evaluated in the context of specific constraints identified in the discussion to follow is that the phonological form and content of scat are relatively “unmarked” along various diverse, independent measures of markedness.

4.2.1. Markedness Hypotheses about Syllabic Shape

Consider first syllabic form. Evidence from several diverse domains of natural language— cross-linguistic studies of canonical syllable structures, phonological epenthesis, cluster simplification processes, language acquisition, prosodic morphology, etc.—independently identify CV syllables as the most basic and the single universally attested syllable shape, justifying the characterization of CV as the ‘core’ syllable. In accord with this empirical generalization, all of the diverse approaches to markedness theory (cf. Jakobsen 1941/1968; Trubetzkoy 1939; Chomsky and Halle 1968; Greenberg 1966; Kaye and Lowenstamm 1984; Prince and Smolensky 1993; McCarthy and Prince 1994; de Lacy 2002 among others) converge on a recognition of open CV syllables as the least marked syllable type. Within the framework of Optimality Theory (Prince and Smolensky 1993, McCarthy and Prince 1995, Kager 1999, etc.), the relative markedness of an output sequence is

169 Patricia A. Shaw

determined with respect to its violation of each of a ranked set of universal constraints on phonological structure. Constraints relevant to syllable shape properties are stated in (32), adapted from Kager (1999: 93, 94, 97):

(32) a. Onset *σ[V A syllable must have an onset. b. NoCoda *C]σ A syllable must not have a coda. c. *ComplexOnset *σ[CC Onsets are simple. d. *ComplexCoda *CC]σ Codas are simple. The optimization of CV results from the fact that this syllable shape violates none of the constraints in (32). The emergence of core CV syllables as ubiquitously preferred in scat is therefore entirely in conformity with markedness predictions about syllable shape. Different measures confirm their special status, from the lead observation that the scat excerpts in (1) and (2) contain exclusively ‘core’ syllables to the accumulated evidence in (27) that the 10 most frequently used syllable shapes are all open CV syllables. Although the survey of scat in §2 sustains the generalization that the vast majority of scat syllables adhere to the simplex onset plus no coda pattern, it also reveals that not one of the six pieces analyzed here consists only of such syllables. Deviation from this optimally unmarked canon falls into two categories: §4.2.2. violations of (32c) *ComplexOnset, and §4.2.3, violations of (32b) NoCoda. Notably, there are no syllables documented in the present database that violate the *ComplexCoda constraint in (32d): all codas in the tunes sampled here consist of a single consonant.

4.2.2. Complex Onsets A very small set of syllables (an overall total of 2.17 of the sample, as shown in (25)) have two consonants as opposed to one in the onset. Such cases violate the constraint *ComplexOnset in (32c), and fall into two subtypes, dependent on specific segmental content. First are the clusters [sk] and [sp]. What differentiates these from the second subtype of *ComplexOnset violations is that [sk] and [sp] are familiar, frequent, well-formed clusters of English. Interestingly, however, they are not common in scat. Only Armstrong (1926: bar 9-10 in §2.1.1) uses [sk], and it occurs only in the alliterative sequence [skiyp skæm skɩ]. Similarly, only Carter uses [sp], and it occurs only once (1955: bar 17 in §2.3.1). Thus, not only are these clusters marked cross-linguistically by virtue of being structurally complex onsets, but they are also foregrounded in terms of perceptual salience within the scat repetoire by virtue of being so infrequent. A final observation is that outside of their occurrence in these clusters, nowhere else in this scat database do any of the individual segments [s], [k], or [p] occur as simplex onsets. As a consequence, these sequences do not conform to the basic generalization that complex margins in natural language phonological systems are characteristically compositional.²² That is, the well-formedness of an [sk] or an [sp] onset cluster in English builds on the independent

²² As stated by Greenberg (1963: 263): “If syllables containing sequences of n consonants in a language are to be found..., then sequences of n-1 consonants are also to be found in the corresponding position (prevocalic or postvocalic).”

170 Scat syllables and markedness theory

availability of each of [s], [k], and [p] as a simplex onset. Thus, on yet another dimension of general properties of phonological systems, these clusters are marked. In short, despite their being entirely within the well-formedness constraints of English, the rare injection of an [sp] or [sk] cluster into a stream of the more limited consonantal playing field of scat syllables will effectively cause them to stand out as highly unusual. In contrast, the second subtype of violations of the *ComplexOnset constraint in (32c) consists of clusters that deviate from standard English: [bw], [mw], and [zw] in Armstrong (1929: 2:02, 2:15 in §2.1.2), and [dl] and [ly] in Carter (1955: bar 6, 7, 16 in §2.3.1). Interestingly, although these segmental concatenations are not well-formed English onsets, they differ from the first subtype in that they are basically compositional within the scat repertoire of onset consonants. That is, with the exception of [z], each of the components of these clusters—viz. [b], [d], [m], [l], [w], and [y]—occurs as a simplex onset in the scat database, as charted in (24). There are two other ways that this second set of clusters differs from the [sk] and [sp] clusters. First, they comprise exclusively voiced segments. The fact that the segments in these clusters agree in voice conforms with Greenberg’s (1978: 252) markedness generalization that “combinations which are homogeneous in respect to voicing” are favoured “over those which are heterogeneous.” Secondly, drawing on the Sonority Hierarchy in (33a), note that each of these onset sequences conforms to the Sonority Sequencing Principle in (33b), in that there is an increasing sonority cline between the first consonant and the second. (33) a. Sonority Hierarchy (< indicates ‘less sonorant than’) Obstruent (O) < Nasal (N) < Liquid (L) < Glide (G) < Vowel (V) b. Sonority Sequencing Principle: (Clements 1990: 285) Between any member of a syllable and the syllable peak, only sounds of higher sonority rank are permitted. To summarize, although these clusters are not part of the familiar English repertoire, there are three general cross-linguistic markedness measures to which they conform: they are compositional; they are homogeneously voiced; and they obey the Sonority Sequencing Principle. What sets this subset of onsets apart from standard English clusters as well as from general cross-linguistic expectations is their relatively marked status with respect to two other constraints on segmental sequencing, both of which fall within the broad purview of the Obligatory Contour Principle (OCP). First, the systematic absence of Liquid-Glide sequences in English reflects a general constraint on minimal sonority distance (34a). In standard English, all Liquid-Glide onset clusters are prohibited: *[ly-, *[lw-, *[ry-, *[rw-. In Betty Carter’s scat, however, [ly- slips past the *[Liquid-Glide constraint. Secondly, militating against various assimilatory forces within the grammar are certain context-sensitive pressures to avoid homorganic place. In standard English, there are no Labial (*Lab–Lab) onset sequences: *[bw-, *[mw-, *[pw-, *[fw-, *[vw-, but in Louis Armstrong’s scat [bw- and [mw- occur, these being the two that transition from a voiced [-continuant] attack into the [w]. Similarly, with Betty Carter, it is the voiced [-continuant] [d] that releases into a liquid [l] that violates the prohibition in standard English against the *Cor–Cor sequences, *[dl- and *[tl-.

171 Patricia A. Shaw

(34) Obligatory Contour Principle (OCP): a. Minimal Sonority Distance (cf. Vennemann 1988, Clements 1990, Zec 2007): *[Liquid-Glide: *[ly- b. Avoidance of homorganicity in consonant-resonant onset clusters: *Lab–Lab: *[bw-, *[mw- *Cor–Cor: *[dl-

None of these constraints characterizes the other non-English cluster, [zw], that Armstrong uses. On a cline of relative markedness, *[zw- is not strongly deviant: it is not subject to repair strategies in the pronunciation of proper names like “Zwicky”; and its voiceless onset counterpart [sw], as in sweet, sway, swan, swoon..., has well-established familiarity in the non-scat lexicon of the romantic lyricists of this same era. Nonetheless, [zw] is outside the boundaries of standard English phonotactics, and will be recognized as such by the listener. The hypothesis developed in §4.3 below is that such violations of the phonological system are not arbitrary: rather, they are strategic manipulations of the dynamic constraints that conventionally delimit linguistic structure, functioning to enhance a range of performative musical effects. To summarize thus far, the argumentation in this section illustrates how phonological markedness theory provides an insightful framework for characterizing why certain overwhelmingly common patterns emerge in the scat syllables of different artists. At the same time, the discussion reveals that this theoretical approach also functions to identify what properties of the empirical residue are not amenable to general linguistic explanation. Based just on an examination of syllable onsets, the fact that this residue is extremely narrow in scope and in realization is itself an interesting finding. In the next section, the relative markedness of coda realization is explored.

4.2.3. Coda Constraints Although the vast majority of scat syllables in the repertoire here do not have a coda, 17 do, as tabulated in (26). However, like onsets, their realization is very restricted. Of the 21 possible coda segments in English (see (4)), only six different segments appear: there are multiple occurrences of [p, t, n, m, l] and a single occurrence of [g]. As the transcribed value of this latter segment (§2.1.2, [1:56]) varies between [g] and [v]—either one of which would be a unique attestation—it will not be incorporated into the following discussion. In markedness terms, there are several cross-linguistic generalizations that characterize the identity and distribution of the five other segments. Note among the obstruents that there are no fricatives or affricates. There are only the two plain anterior stops [p] and [t] which, in terms of frequency (see (26)), account for 59 (49/83) of all attested codas. Given that these are the voiceless counterparts of [b] and [d], which clearly emerge as the overwhelming segmental favourites in onsets, a major question relates to why the value of [voice] is in complementary distribution between onset and coda? Markedness theory offers a straightforward account of the coda behaviour, in that the preference for obstruents to be voiceless in syllable-final position (alternatively, at the end of a word or before another obstruent) is a widely

172 Scat syllables and markedness theory attested cross-linguistic phenomenon. This contextual neutralization underlies the OT formalization of the positional markedness constraint in (35): (35) *Voiced Coda (Kager 1999; cf. Steriade 1999, Gordon 2007, Zec 2007) Obstruents must not be marked for [voice] in coda position. This constraint is unviolated in the entire scat corpus documented here, and effectively captures the relevant generalization: if a coda is an obstruent, then it must be voiceless. Not all the attested codas are obstruents, however. The residual codas [m], [n], and [l] are all sonorants. On the basis of the cross-linguistic observation that some languages, like Chinese, allow only sonorants in coda position, Pepperkamp (2003) proposes the markedness constraint in (36): (36) *Obstruent Coda Codas cannot be obstruents. The postulated constraint in (36) makes two predictions. First, a phonological system could have only sonorant codas, as Pepperkamp argues for Chinese. Secondly, a phonological system could not have exclusively obstruent codas: that is, if it has obstruent codas, then it also must have sonorant codas. This second type of system is exactly what is documented for both tunes analyzed for each of Louis Armstrong and Chet Baker (see (26)). Of particular interest, however, is the fact that this is not what has emerged for either of the Betty Carter recordings. As summarized in (26), her inventory of codas is precisely the system characterized by the first prediction: there are only sonorant codas. This is really quite striking confirmation of the role of universal markedness constraints in governing the strictly delimited inventory of scat. Moving to a consideration of place of articulation properties of codas, we note that the limitation of the set of attested scat codas {p, t, n, m, l} to Labials and Coronals is also systematically derivable from general tenets of markedness theory. Drawing on various observed asymmetries in inventories, epenthesis, neutralization, etc., the markedness hierarchy in (37) identifies Dorsal place as the most highly marked: (37) Place Markedness Hierarchy (de Lacy 2007: 23) *Dorsal »» *Labial »» *Coronal Hence, the non-attestation of Dorsals and, concomitantly, the preferred status of Coronals and Labials follow from this markedness generalization. Finally, it is important to consider not just the distinctive properties of segments in a particular prosodic position, but also aspects of their sequential relation to their neighbours. As a dramatic example of harmonic assimilation, all nine instances of [ṇ] in Chet Baker’s minimally contrastive articulatory flow are preceded by homorganic [t] and followed by [d]. Thus, a single coronal non-continuant gesture is sustained across the tri- segmental sequence, modulated only by velic movement for the oral-nasal contrast and laryngeal voicing. Even in the context of the much more diverse articulatory repertoire in Louis Armstrong’s Hotter Than That, an examination of trans-syllabic properties in it reveals that the place of articulation in the vast majority of the 42 codas is homorganic with the place of articulation of the following onset. Specifically, all eight cases of coda [t]

173 Patricia A. Shaw are followed by a [d] onset. Similarly both [n] codas precede [d]. All three post-vocalic coda [m]s are also homorganic, in one case to [b] and in the other two cases to [w]. All three tokens of syllabic [ṃ] follow a comparable pattern, preceding onsets [w], [m], and cluster [mw]. Aside from the unique instance of a [g], the only coda segment that is ever independent of this assimilatory effect is Louis Armstrong’s favoured coda in these works, [p]. Still, the majority of [p] codas (13/22 = 59.1) precede homorganic [b]. The residual nine—all of which occur before [d]—are the only non-homorganic codas in this entire scat set. Again, these coda-onset assimilatory patterns constitute further evidence of a remarkably consistent and delimited range of vocal behaviours in scat that are systematically correlated with a broadly motivated positional markedness constraint, the Coda-Condition: (38) Coda-Condition (Itô 1989; Kager 1999) A coda cannot have a place feature different from the following onset. Note that (38), which fosters adjacent labial-labial or coronal-coronal articulations, is differentiated from (34b), which militates against labial-labial or coronal-coronal sequences, by virtue of prosodic context. The former applies across a coda-onset sequence whereas the latter obtains between segments within a complex onset. What has been argued in this section is that all the defining properties of scat codas in the current sample fall directly within the explanatory framework of the independently movitated theory of phonological markedness. They may be exclusively sonorant (36); if obstruent, they are voiceless (35); they are solely coronal and labial (37); and they are overwhelmingly homorganic with a following onset (38).

4.2.4. Why are [d] and [b] the favoured onsets in scat? Having examined the markedness properties of syllable shape, of complex onsets, and of codas, let us now return to two fundamental questions raised in the introduction, where it was noted that, in the Louis Armstrong excerpts in (1) and (2), [d] and [b] are the only onsets. The first question was whether this observation is genuinely representative of scat or whether it is essentially accidental, attributable perhaps to this particular artist, to selective sampling, to the stylistic phraseology of these brief excerpts, or to some other factor. The present analysis clearly affirms that these two segments are indeed the most prevalent onset consonants. As summarized in (24), [d] is incontestably the favoured scat onset in every tune by every artist examined in §2. The next most frequently attested onset is [b]. However, as noted earlier, there is an evident asymmetry in their usage. In two of the six songs, [b] is not used at all; in a third, it appears only once. In the other three songs, it ranks below [d] with a broad range of variance: from 66.7 difference (in Armstrong §2.1.1) to 12.8 (Armstrong §2.1.2) to 4.4 (Carter §2.3.1). Nonetheless, despite this imbalance between [d] and [b], they both clearly stand out as more prominent than any of the other nine consonants that appear in onsets. The second question is what the explanatory basis of this generalization might be. Notably, it does not mirror standard English frequency patterns. According to the

174 Scat syllables and markedness theory

Francis and Kucera (1982) count of the token frequency of all word-initial onsets in a corpus of over a million English words, neither [d] nor [b] stands at the head of the relative frequency ranking of single consonant onsets, summarized in (39a).²³ In fact, in terms of the absolute measures cited in (39b), [b] is almost twice as frequent as [d] in this extensive corpus of standard English usage, a result that is opposite to the consistently greater frequency and breadth of distribution of [d] over [b] in scat.

(39) a. ð > w > h > b > t > s > k, m > f > d > ... b. [b] = .05335, [d] = .02762 In short, standard English frequency measures do not account for the two most robust generalizations that have emerged in the scat data: (1) the preference of [d] over [b]; and (2) the prevalence of [d] and [b] over all other consonants in the English inventory. Phonological markedness theory contributes significantly to a principled interpretation of these results. First, consider place of articulation. The Place Markedness Hierarchy already introduced in (37) effects an internal ranking of the coronal place as the most optimal (least marked), of labial as an intermediate class, and of dorsal as the least optimal (most marked). Two major empirical findings about scat onsets follow directly from this hypothesis of a fixed place hierarchy: one is the preferred status of coronal [d] over labial [b]; and the second is the absense of dorsal [k] or [g]. Dorsals, the most marked of the English stops, are simply unattested as scat onsets. Thus, the place features of [d] and [b] are clearly consistent with fundamental markedness tenets. But, what about their manner and laryngeal properties? With respect to manner, the fact that [d] and [b] are both obstruents accords with the cross-linguistic preference for low sonority onsets, captured by the fixed positional markedness constraint hierarchy in (40a). Further, within the class of obstruents, the articulated subcategorizations of (40b) identify stops as less sonorant than fricatives. (40) a. Optimal Onset Sonority (de Lacy 2001; Prince and Smolensky 2004) *Onset/L » *Onset/N » *Onset/O b. Obstruent Sonority (Dell and Elmedlaoui 1985; Prince and Smolensky 2004) *voicedFric » *voicelessFric » *voicedStop » *voicelessStop The combined effect of the markedness relations in (40a) and (40b) accords directly with the notable paucity (a single attestation: see (24)) of fricatives as a simplex onset (Sarah Vaughn’s and Sha Na Na’s trademark [š] notwithstanding). We conclude then that, for manner features, phonological markedness theory provides considerable explanatory coverage of the favoured status of [d] and [b] in the full context of the highly constrained scat inventory. However, a major anomaly persists with respect to laryngeal markedness: the privileged status of the voiced obstruents [d] and [b] and, correlatively, the extreme rarity of their voiceless counterparts [p] and [t] as scat onsets are directly counter to the predictions of markedness theory. That is, based on cross-linguistic generalizations

²³ This rank order is constructed from the Francis-Kucera token frequencies for single consonant onsets cited in the appendix to Stemberger 1990: 157. The cited values in (39b) are from this same source.

175 Patricia A. Shaw

from a variety of perspectives (including the typology of sound systems,²⁴ natural classes, direction of neutralization, direction of language change, segmental complexity, perceptual and articulatory contrast, and other factors), voiceless obstruents are the unmarked series, this generalization motivating the markedness constraint in (41): (41) Voicing markedness (de Lacy 2002) *[+voice, -sonorant] Obstruents must be voiceless.

As shown in (24), [d] and [b] together comprise 77.9 of scat syllable onsets. In contrast, there are no instances of [p] in onset position, and only a single occurrence of [t] (in Louis Armstrong’s Hotter Than That). Beyond the database of tunes analyzed in §2, a full examination of Bauer’s (2002a: 245–343) prodigious body of transcriptions of Betty Carter’s scat corroborates this generalization: in the entire collection, [t] is unattested as an onset and [p] is exceedingly uncommon.²⁵ Given robust cross-linguistic support for (41), it can only be concluded that the overwhelming preponderance of the voiced obstruents [d] and [b] as onsets in scat, combined with the virtual absence of voiceless [t] and [p], is distinctly odd from a markedness perspective. The very fact that this pervasive asymmetry is clearly defined in markedness terms as deviant is a productive consequence of the theory, and is pursued further in §4.3.

4.2.5. The Contributions of Markedness Theory To summarize, the goal of §4.2 has been to explore the degree to which phonological markedness theory provides an explanatory framework for the observed patterns in syllable shape and segmental inventory in scat. The results of this approach are of considerable interest, I believe, to deepening our understanding of the interface of natural language systems and musical vocal performance. Although couched in an optimality theoretic framework, the markedness generalizations invoked here essentially derive from the confluence of a diversity of insightful conceptual approaches to markedness issues that have spanned many decades of linguistic research. The breadth of empirical coverage offered by an essentially small body of tightly constrained and independently motivated hypotheses is considerable. First, in §4.2.1 it is seen that the robust preference for open CV syllables in scat directly accords with the four universal markedness constraints in (32) that govern syllabic form. There are two kinds of deviations from this basic canon: a very small set (2) of syllables have complex onsets and a larger set (17) have simplex codas. There are no complex codas.

²⁴ Obstruents are exclusively voiceless in the phonological inventories of many languages, e.g. hənʼqʼəminʼəmʼ (Salish), Nuu-chah-nulth (Wakashan), Hawaiian (Austronesian), Korean, etc. Further, the presence of voiced obstruents in a language characteristically entails the presence of their voiceless counterparts, as in English. ²⁵ Two tokens of [p] are in You’re Driving Me Crazy (Bauer 2002a: 252; bars 7, 9); a third token is in the initial syllable of bar 57 in the 1979 take of I Could Write a Book (Bauer 2002a: 307). These last two are plausibly interpretable as an ambisyllabic coda-onset from the preceding [bap] syllable.

176 Scat syllables and markedness theory

Then, to extend the analysis beyond syllable shape, the particular properties that “mark” the attested complex onsets are examined in §4.2.2, the properties of codas are analyzed in §4.2.3, and the observed featural asymmetries of simplex onsets are investigated in §4.2.4. As shown in §4.2.3, the identity and distribution of the limited repertoire of codas conform to positional markedness constraints on voice (35), manner (36), and coda-onset place agreement (38), as well as to the fixed place hierarchy in (37). In §4.2.4, it is seen that all but one dimension of the featural identity of the restricted inventory of scat onsets follows markedness patterns. Specifically, they comply with the fixed place hierarchy of (37), with the positional markedness constraints governing the intersection of sonority and manner in onsets in (40a), as well as with the fixed hierarchy in (40b) that optimizes non-continuant manner and sonority. Onsets deviate on only one—albeit a perceptually highly salient—measure: voice, as formalized by the constraint in (41). An alternate hypothesis was therefore tested, namely, that this “anti-markedness” result might correlate with identified frequency patterns for standard English onsets. However, comparative evaluation of the evidence shows no systematic relationship to support a frequency hypothesis. In sum, the explanatory power of a markedness explanation for these diverse and strikingly consistent factors is substantial. However, a further empirical strength of applying markedness theory to the analysis of scat is that the theory characterizes not only what corresponds with English and/or universal language patterns, but it also functions to define the specific nature and locus of deviance. In essence, markedness theory effectively subcategorizes the residue of scat properties that fall outside of the predictions of markedness-governed sound patterns into two domains. One is absolutely pervasive across all singers, namely: the consistent realization of voiced [d] and [b] onsets, to the virtual exclusion of their voiceless unmarked counterparts [t] and [p]. The other is a much less coherent set of often unique attestations of highly marked segments, such as Louis Armstrong’s once-only insertion of a triple sequence of [f] codas and later of [sk] onsets in Heebie Jeebies. The fact that the empirical residue that is not amenable to a linguistic markedness explanation is extremely narrow in scope is theoretically interesting, and suggests that quite specific competing functional forces external to the linguistic system may be at play. Scat is, after all, a component of each jazz artist’s musical repertoire and individual creativity. In the following section, the musical interface of vocal performance with the “marked” linguistic residue is examined.

4.3. The Performative Interface between Vocal Music and Phonological Markedness

Given that scat is at the interface of constraints on linguistic sound structure and the exigencies of vocal music production, the central issue to be addressed here is whether there are factors of musical performance or creative expression that conflict with and override phonological markedness, thus providing a systematic explanation for the deviant residue identified in §4.2.

177 Patricia A. Shaw

4.3.1. Voice The most salient property of scat syllables that breaches markedness patterns is the pervasive preference for [d] and [b] as onsets, to the virtual exclusion of [t] and [p]. In natural language systems, there is a robustly attested cross-linguistic asymmetry: a phonological inventory may have only voiceless obstruents, or both voiced and voiceless, but not only voiced obstruents. This asymmetry is formalized in (41) by hypothesizing a markedness constraint that identifies *[+voice, -sonorant] segments as “marked”, in the absence of a corresponding constraint prohibiting [-voice, -sonorant] segments. What has been documented in §2 is that in scat this asymmetry is reversed. The fundamental question is why: what properties obtain in the performative context of scat that would create pressure to systematically violate this constraint? We have seen the role played by the canonical syllable structure constraints in (32) to optimize a repetitive CV alternation in natural language. Further, it is hypothesized, through constraints like (40), that the optimal CV contour alternates between minimal sonority onsets (voiceless stops) and maximal sonority nuclei ([a]), thus enhancing perceptual contrast of the peaks and troughs. What is happening in scat is that the optimal voiceless stop onset is being compromised just one notch up the Obstruent Sonority hierarchy in (40b) to the category of voiced stops. The articulatory factor that differentiates these two categories is vocal cord vibration. The functional relevance of this difference is that vocal cord vibration is essential to carry pitch. In natural language, pitch plays a criterial role in marking tone, intonation, and often stress. However, linguistically significant pitch is characteristically carried not by the onset, but by the syllabic nucleus, and sometimes by subsequent coda elements dependent on sonority. In music, pitch is fundamental to the expression of melody. Unlike language, the melodic line of music is not structured by constraints that optimize a voiceless-voiced contour. To the contrary, although periods of relative silence (rests) contribute to phrasing a melody, the foundation of a melodic line is a continuous soundwave that allows pitch realization and modification in order to create a succession of tones. Consequently, in vocal music that incorporates natural language, there is an inevitable tension between the articulation of voiceless sounds and the fluidity of the melodic line, since voiceless segments cannot bear pitch. Research on singing in Tashlhiyt Berber (Dell and Elmedlaoui 2008), a language with extensive sequences of voiceless obstruents, reveals two strategies for the realization of the melodic line across such stretches. One strategy is to prolong a preceding vowel so that it carries not only its own tone, but also the tone that metrical scansion would normally assign to the following voiceless syllable. A second strategy is to epenthesize an unmarked schwa vowel to carry the pitch of the associated melody. What I hypothesize is happening in scat is a third strategy, namely: the musical melodic imperative for voiced realization overrides the natural language markedness constraint in (40b) that optimizes voiceless stops in onsets. Because voiced stops can, however briefly, carry some pitch realization, they satisfy the high-ranked musical constraint on melodic voicing and therefore emerge as the most prevalent onsets in scat. This interface between the vocal music “Melodic Voicing” constraint, formalized in

178 Scat syllables and markedness theory

(42a), and the markedness constraints on onset sonority (deconstructing *Onset/O from (40a) into the more refined hierarchy of (40b)) is illustrated in the tableau in (42b):

(42) a. Vocal Music Melodic Voicing Constraint: *[-voice]/ ♫ melody Sung segments must be voiced.

b. Music System: *[-voice]/ ♫ melody Language System: *Onset/voicedStop *Onset/voiceelessStop

a. [ta] * ! *

☞ b. [da] *

Similarly, the absence of [p] onsets, despite the occurrence of [b] onsets, follows from this same interface where the musical constraint (42a) outranks the language constraints.

4.3.2. Performative Markedness

The second kind of deviance identified by markedness theory does not cohere in a single identifiable characteristic. Rather, there is an assortment of infrequent—indeed, often unique—attestations of sounds that are relatively marked in terms of general cross-linguistic patterns identified in §4.2. The question to be explored in this section is whether there might be an independent functional motivation for the inclusion of marked structure. Consider, for example, the diverse array of complex onset clusters listed in (25) and discussed in §4.2.2. A variety of different frequency measures related to these onset clusters attest to their rarity. Clusters occur in only three of the six scat tunes, and in the output of only two of the three artists: there are no clusters in Chet Baker’s work. In total, 10 cluster tokens are attested in the database of 461 syllables: thus, they mark a mere 2.17 of onsets. Louis Armstrong’s output contains six of the ten clusters; even within his scat corpus of 213 syllables, the percentage of complex onsets is only marginally higher: 2.8 (6/213). In terms of individual frequency, five of the seven different types of documented clusters are unique attestations. As shown in the summary onset chart in (24), even certain simplex onsets are attested only once in the present sample. The basic generalization that the full body of frequency measures reported on here sustains is that the inventory of segments and combinatorial possibilities in scat is extremely limited in comparison to the full inventory of sounds in English, not to mention the even more extensive repertoire of phones in other natural language systems or, in fact, in the extraordinary array of oral articulations that the human vocal apparatus is capable of. A further claim has been advanced that the very limited inventory of scat as defined in terms of frequency correlates very strongly with segments and structures that are characterized as relatively unmarked according to basic tenets of phonological markedness theory. Consider then the cognitive impact of deviation from what has been identified as the high frequency, phonologically unmarked norm for scat. The introduction of novelty into an established, familiar sequence will command immediate auditory attention and interest.

179 Patricia A. Shaw

For example, as noted earlier (§1.2.1) in the discussion of Heebie Jeebies, it is after a extended auditory train of 22 [d]-initial syllables, minimally broken by two [b]-initial syllables, that Louis Armstrong interjects the alliterative sequence of three syllables with an [sk] onset cluster. The auditory impact—phonologically and musically—is undeniably powerful. A variation on Armstrong’s creative use of deviant phonology to musical effect occurs in Hotter Than That, where the trio of non-English clusters [zw], [mw], and [bw] occur within a few bars of one another, establishing a brief articulatory leitmotif that through its disruption of the familiar scat canon simultaneously produces auditory tension and artistic interest. Betty Carter’s repertoire, as observed in §2.3.1, is similarly enriched and the listener’s attention engaged by unanticipated occurrences of the highly marked [dl] or [ly] clusters. Chet Baker’s implementation of phonological deviance is illustrated by the distribution of [h]. As pointed out in §2.2, in each of the 1955 and the 1989 versions of Everything Happens to Me, [h] occurs only once, in different places, but in both cases its uniqueness functions to demarcate the initial syllable, a particularly prominent prosodic position, of a musically important phrase. The hypothesis, then, is that phonological deviance and low frequency may functionally conspire in scat to enhance perceptual salience. By challenging the limits of phonological markedness constraints manifest in scat, a jazz artist can effectively arrest the listener’s attention, strategically manipulate the cognitive tension of phonological deviance, and creatively expand the expressive repertoire at the interface of language and vocal music.

5. Conclusions

Scat extends the vibrant improvisational genre of instrumental jazz to the human voice. As an oral idiom, scat draws on the same articulatory apparatus as natural human languages do. Because it is uniquely situated at the interface of the cognitive and performative systems that underlie both music and language, scat can potentially deepen our insight into the complex organizational structure of each of these particularly human creative systems, as well as of their interaction. Through investigation of the form of scat syllables used by three renowned jazz vocalists—Louis Armstrong, Chet Baker, and Betty Carter—in performances that range across time from 1926 to 1989, the analysis in §2 establishes that, despite their markedly different musical styles and individual creativity, the set of consonantal sounds that these diverse performers draw on in the creation of scat syllables is strikingly consistent and is extremely limited in comparison to the extensive range of segments and combinatorial possibilities that define the inventory of syllables in English. Moreover, as shown in §3, the observed generalizations apply not only within the classical scat canon of these scat masters, but also in scat-derived pop music lyrics of the early 60s rock’n’roll era. Given that the articulatory range of the human vocal apparatus far exceeds what is found in scat, the fundamental goal pursued in §4 is to explore what might account for the attested sound patterns in scat. Three hypotheses are investigated. The first (§4.1), familiar from the jazz literature, holds that jazz vocalization is essentially imitative of

180 Scat syllables and markedness theory

jazz instrumental expression. Although this hypothesis holds considerable popular appeal, it is difficult to substantiate from an empirical perspective. The second hypothesis (§4.2) is that phonological markedness theory provides an insightful framework for characterizing why certain overwhelmingly common patterns emerge in the scat syllables of different artists. It is argued that there is substantial support for this approach. Specifically, all the defining properties of scat codas in the current sample fall directly within the explanatory framework of independently movitated markedness constraints: codas may be exclusively sonorant, in accordance with (36) *Obstruent Coda; if obstruent, they are voiceless, following (35) *Voiced Coda; they are solely coronal and labial, to the exclusion of dorsals, in conformity with (37) the Place Markedness Hierarchy; and they are overwhelmingly homorganic with a following onset, in adherence to (38) the Coda-Condition positional markedness constraint. With respect to onsets, the clear preference for stops accords with the cross-linguistic preference for low sonority onsets, captured by the positional markedness constraint in (40a) that defines Optimal Onset Sonority, combined with the obstruent sonority hierarchy in (40b). The fact that dorsals, the most marked of the English stops, are unattested as scat onsets, as well as the preferred status of coronal /d/ over labial /b/ follow, as in the case with codas, from the Place Markedness Hierarchy in (37). In sum, phonological markedness theory provides considerable explanatory coverage of the highly constrained scat inventory. Not all of the observed scat data is interpretable in terms of phonological markedness, however. A valuable strength of the theory is precisely this consequence of its identifying two sets of empirical residue that are not amenable to general linguistic explanation. In §5 it is argued that such violations of phonological markedness are systematic, and function to enhance performative components at the interface of vocal music and the phonological system. The most striking and consistent property of scat syllables that challenges markedness theory is the overwhelming preference for the voiced obstruents [d] and [b] as onsets, in conjunction with the extreme rarity of their voiceless counterparts [p] and [t]. Whereas in a natural language system, the perceptual contrast between onset and nucleus is optimized by a [-voice] onsets, in a vocal music system, the pitch level of the melodic line can only be carried by [+voice] segments. What is hypothesized here is that the musical imperative for melodic/voiced realization over-rides (in optimality terms, outranks) the natural language markedness constraint. The second type of scat anomaly involves a diversity of infrequent attestations of relatively marked sounds. Because the introduction of deviation into a stream of high frequency, phonologically unmarked scat has considerable cognitive impact, it is hypothesized that phonological deviance and low frequency may functionally conspire in scat to enhance perceptual salience. By defying the familiar bounds of the scat inventory, a jazz singer can effectively capture the listener’s attention, extend the articulatory repertoire that he can creatively work with, and transcend the familiar. At the heart of improvisional creativity in music, as in language, is the challenge of innovation under the constraints of structural limitations. Although an intricate variety of cognitive and performative factors contribute to scat improvisation, what has been argued in this paper is that phonological markedness theory provides an explanatory

181 Patricia A. Shaw framework of the structural constraints that largely define the articulatory domain of scat. Interfacing with this phonological framework, and sometimes over-riding specific constraints within it, are performative exigencies of melodic realization and the ineffable creative workings of improvisational genius.

References

Bauer, William R. 2002a. Open The Door: The Life and Music of Betty Carter. Ann Arbor: University of Michigan Press. Bauer, William R. 2002b. : A Timbral and Phonemic Analysis. Current Musicology 71–73: 303–322. Bastian, Jim, and John Alexander. 1995. Chet Baker’s Greatest Scat Solos. Smithfield, RI: Coastal Publishing and Educational Resources. Berliner, Paul. 1994. Thinking in Jazz: The Infinite Art of Improvisation. Chicago: University of Chicago Press. Chomsky, Noam, and Morris Halle 1968. The Sound Pattern of English. NY: Harper and Row. Clements, G.N. 1990. The Role of the sonority cycle in core syllabification. In Papers in Laboratory Phonology I: Between the grammar and physics of speech, ed. J. Kingston and Mary E. Beckman. Cambridge: Cambridge University Press. 283–333. de Lacy, Paul. 2001. Markedness in prominent positions. In MITWPL 40: HUMIT 2000., eds. O. Matushansky et al. Cambridge, MA. 53–66. [also ROA 542] de Lacy, Paul. 2002. The formal expression of markedness. University of Massachusetts, Amherst: Doctoral dissertation. de Lacy, Paul. 2007. Themes in phonology. In The Cambridge Handbook of Phonology, ed. Paul de Lacy. Cambridge: Cambridge University Press. Dell, F. and M. Elmedlaoui. 1985. Syllabic consonants and syllabification in Imdlawn Tashlhiyt Berber. Journal of African Languages and Linguistics 7: 105–130. Dell, F. and M. Elmedlaoui. 2008. Poetic meter and musical form in Tashlhiyt Berber songs. Cologne: Rüdiger Köppe. Francis, W.N. and H. Kucera. 1982. Frequency analysis of English usage: Lexicon and grammar. Boston: Houghton Mifflin. Gordon, Matthew. 2007. Functionalism in phonology. In The Cambridge Handbook of Phonology, ed. Paul de Lacy. Cambridge: Cambridge University Press. Greenberg, Joseph H. 1963. Memorandum concerning language universals. In Universals of Language, ed. J.H. Greenberg. Cambridge, MA. Greenberg, Joseph H. 1966. Universals of Language. Cambridge: MIT Press. Greenberg, J.H. 1978. Some generalizations concerning initial and final consonant clusters. In Universals of human language 2: Phonology, ed. J.H. Greenberg. 243–280. Stanford: Stanford University Press. Hurch, Bernard. 2005. Studies on reduplication. Empirical approaches to language typology, No. 28. Berlin: Mouton de Gruyter.

182 Scat syllables and markedness theory

Ito, Junko. 1989. A prosodic theory of epenthesis. Natural Language and Linguistic Theory 7: 217–259. Jakobsen, Roman. 1941/1968. Child language, aphasia and phonological universals. The Hague and Paris: Mouton. Kager, R. 1999. Optimality Theory. Cambridge: Cambridge University Press. Kaye, Jonathan, and Jean Lowenstamm. 1984. De la syllabicité. In Forme sonore du language: Structure des représentations en phonologie, ed. F. Dell, D. Hirst and J.-R. Vergnaud. Paris: Hermann. 123–159. Kernfeld, Barry. 1995. What to Listen for in Jazz. New Haven, Conn.: Yale University McCarthy, John, and Alan Prince. 1986. Prosodic morphology. Technical report 32. Rutger University Center for Cognitive Science. (online revised version: http://ruccs. rutgers.edu/pub/papers/pm86all.pdf ) McCarthy, John, and Alan Prince. 1994. The emergence of the unmarked: Optimality in prosodic morphology. In Proceedings of the North East Linguistic Society 24, ed. Merce Bonzalez. Amherst, MA: GLSA Publications. 333–379. McCarthy, John, and Alan Prince. 1995. Faithfulness and Reduplicative Identity. In University of Massachusetts Occasional Papers in Linguistics 18, eds. Jill Beckman, Laura Walsh Dickey, and Suzanne Urbanczyk. 249–384. Amherst, MA; GLSA Moravcsik, Edith. 1978. Reduplicative constructions. In Universals of human language 3: Word structure, ed. J. H. Greenberg. 297–334. Stanford, CA: Stanford University Press. Peperkamp, S. 2003. Phonological Acquisition: Recent Attainments and New Challenges. Language and Speech 46, 2–3: 78–113. Prince, Alan, and Paul Smolensky. 2004. Optimality Theory: Constraint Interaction in Generative Grammar. Oxford: Basil Blackwell. [1993. ROA 537] Robinson, J. Bradford. 2002. Scat Singing. In New Grove Dictionary of Jazz, Vol. 3, ed. Barry Kernfeld. 515–516. Sapir, Edward. 1933. The Psychological Reality of the Phoneme. In Selected Writings of Edward Sapir in Language, Culture and Personality, ed. David Mandelbaum. 1986. Berkeley: University of California Press. Stemberger, Joseph Paul. 1990. Wordshape errors in language. Cognition 35: 123–157. Steriade, Donca. 1999. Phonetics in phonology: The case of laryngeal neutralization. In Papers in Phonology 3, ed. Matthew Gordon. UCLS Working Papers in Linguistics 2. UCLA. 25–146. Stewart, Milton L. 1987. Stylistic Environment and the Scat Singing Styles of Ella Fitzgerald and Sarah Vaughan. Jazzforschung/Jazz Research 19: 61–76. Stoloff, Robert. 2003. Blues Scatitudes. : Gerard and Sarzin Publishing Co. Trubetzkoy, Nikolaj. 1939. Grundzuge der Phonologie. Gottingen: Vandenhoeck & Ruprecht. Vennemann, Theo. 1988. Preference laws for syllable structure. Berlin: Mouton de Gruyter. Zec, Draga. 2007. The Syllable. In The Cambridge Handbook of Phonology, ed. Paul de Lacy. Cambridge: Cambridge University Press.

183 Patricia A. Shaw Appendix 1. Transcription conventions

Reducing the dynamic auditory flux of articulatory movement in scat vocalization to discrete transcriptional conventions that are defined in terms of phonologically independent unitary segments and syllables entails both informed choice and compromise. Some of the major factors impacting on the transcriptions presented in Appendix 2 are discussed here. First, there are some different notational symbols that are aligned with different transcription traditions. Given the American roots of jazz, certain “Americanist” symbols, like [š] and [č], are here adopted, rather than their corresponding IPA counterparts [∫] and [t∫]. Nasalization is indicated by a tilda over the vowel, e.g. [ã]. A “syllabic” resonant is marked with a subscript dot, e.g. [ṇ]. More nuanced are issues related to levels of abstraction away from phonetic detail. The transcriptions in Appendix 2 are basically very broad, ignoring most aspects of phonetic realization that are entirely regular, such as pre-tonic aspiration of stops. However, certain other features that are normally non-contrastive in English, but that surface prominently in unpredictable environments are explicitly marked, for example, the nasalization in Louis Armstrong’s string of syllables initiated in bar 49 of §2.1.2. A particularly complex domain is the representation of vowels, as their articulation tends to be highly mobile. Bauer opts for a relatively abstract transcription, based on the Trager-Smith system for phonemicization of English. Bauer defines the interpretation of vowels with reference to the words in the table below (see Chart 1 in Bauer (2002a: 238); and Table 3 in Bauer (2002b: 306–307)). In order to standardize the transcription system used for all the scat data considered in the present study, the Trager-Smith (abbreviated T-S) representations are here interpreted as in the “PAS” column below. The major differences are in the representation of lax vowels and of the T-S post-vocalic /h/. Bauer’s transcriptions reproduced in Appendix 2 include both his T-S notation and a transliteration in terms of the general correspondences set out in (1).

(1) English Word T-S PAS English Word T-S PAS

beat iy iy boot uw uw pit i ɩ put u ʋ bait ey ey boat ow ow pail eh ɛə caught oh ɔ pet e ɛ pot aa pat ææ cut əə Both systems neutralize the considerable variation in vowel realization that may relate to an individual singer’s articulation, phonological context, or melodic interpretation. An inherent limitation of the Trager-Smith transcriptions that impacts on the present analysis derives from the fact that the T-S system does not include glottal stop, since [ʔ] is non-phonemic in standard English. As a consequence, there is systematic

184 Scat syllables and markedness theory ambiguity in the onsetless status of syllables that are transcribed by Bauer as vowel-initial. There are three different types of contexts where this observation is relevant. First, consider the post-rest “ə” in what is transcribed as / ‰ ə duw .../ in bar 6 of Betty Carter’s Thou Swell (§2.3.1). In the context of the present evaluation of the relative markedness of scat syllable structure, the question is whether this syllable is truly onsetless, i.e. simply [ə], which would be a marked syllable structure, or whether there is a sub-phonemic epenthetic glottal stop functioning as an onset and creating an unmarked structure, i.e. [ʔə]. My auditory assessment of the phonetic realization of these contexts basically accords with Bauer’s phonemic transcriptions: generally Carter’s mellifluous voice transitions very smoothly into a vocalic realization both phrase-initially and in phrase-internal vowel-vowel sequences, as in [... lyə a əm ...] in bar 7. Despite the musical appropriateness of these seamless transitions to different vowel targets, within the linguistic analysis such syllables are tallied as onsetless, and hence “marked”. The second context is exemplified in the last two syllables of bar 6 of §2.3.1 Thou Swell, where Bauer’s transcription /duw uw/ implies the second syllable /uw/ has no onset. Here there are two other potential interpretations: (i) a glottal onset, or (ii) a trans- syllabic perseveration from the preceding glide into an onset role. The retranscription [du wuw], adopted here, interprets the intervocalic glide as an onset. Alternatively the [w] closure may be considered ambisyllabic. Either way, such cases are not onsetless, and hence not categorized as “marked” in the present analysis. A third and similar type of case where there plausibly is dual functionality of an intervening consonant is the situation where a ‘syllabic’ nasal is preceded by a syllable headed by a lax vowel and closed by a stop that is tautosyllabic to the following nasal, e.g. the sequences [dɛt ṇ] and [dɩt ṇ] in line 7 of Chet Baker’s Everything Happens to Me (§2.2.2). At a very surface level, such [t]s are arguably ambisyllabic, functioning as both coda to the preceding syllable and onset to the following one. Consequently, the syllabic resonant in such cases is not classified in the present analysis as onsetless. Space limitations here unfortunately preclude inclusion of the corresponding musical transcription for the full repetoire of scat renditions analyzed here (but see citations to musical notation by Bauer (2002a, b) and Bastian and Alexander (1995)). However, because there are, as one might expect, certain phonological correlations in positions of prosodic prominence, both bar divisions and rests are encoded in some (but not all) of the transcriptions here. Bar divisions are represented by | . Rests are in standard musical notation: sixteenth note ≈, eighth note ‰, quarter note Œ, and half note Ó. A hold, where a syllable is held across a bar line, is indicated by a dash on both sides of the bar, e.g. [... dip–|– bə ...]. The reality, of course, is that a wealth of auditory information that springs from this improvisational conjunction of creative and physical forces—the finessed range of articulatory movement, the rich and highly individualistic molding of acoustic shapes and tonalities—is not captured in conventional phonetic transcription. For the present purposes, however, the notation adopted in Appendix 2 provides considerable insight into the linguistic issues under investigation.

185 Patricia A. Shaw

Appendix 2: Scat Transcriptions

2.1.1. Armstrong, Louis. Heebie Jeebies scat solo (February 26, 1926) Okeh Records 8300. Transcription by W.R. Bauer (2002b: 308). Transliteration (2nd line) follows the principles outlined in Appendix 1. [Note: see Bauer for a full musical transcription of the melodic line.]

 WRB: eh | iyf Œ gæf Œ | əmf ‰ diy bə ‰ | diy də la bam Œ | rip ip ‰ di duw diy duwt | eə | iyf Œ gæf Œ | əmf ‰ diy bə ‰ | diy də la bam Œ | rɩp ɩp ‰ dɩ duw diy duwt |  WRB: duw Œ duw diy duw də | ‰ diy də də dow diy | dow di dow duw ‰ duw– | duw Œ duw diy duw də | ‰ diy də də dow diy | dow dɩ dow duw ‰ duw– |  WRB: –bə duw biy dey də | skiyp Œ skæm Œ | ski bəp ‰ diy də di də | –bə duw biy dey də | skiyp Œ skæm Œ | skɩ bəp ‰ diy də dɩ də |  WRB: dip dæw diy ‰ dip– | –duw də dæw də ... dɩp dæw diy ‰ dɩp– | –duw də dæw də ... 2.1.2. Armstrong, Louis. “Hotter Than That” scat solo (December 13, 1929). Hotter than That, Track 1, 1:18. Okeh Records 8535. Transcription BK by Barry Kernfeld (1995: 168). Transcription PAS and time markers by Patricia A. Shaw. Transcription WB/R by Bauer (2002b: 308), citing Reeves (2001): bars 49–54.

 [1:19] BK1: Dip deh doop da, doe doe doe doe. PAS: dɩp diː duːp də da do do do  BK2a: Dah dew dah doot doot dew, da dee dee doot, PAS: da daw da duːt duːt duː də diː diː duːt  BK2b: daw bee do bee dup baw lahp baw. PAS: da biː duː biː duːp ba la(p-) baw [1:26] BK3a: Bah bee boop, buh dee bee doop bee, PAS: ba biː buːp bə diː biː duːp biː BK3b: hew law bah de bohm, bah bah bah bough PAS: hə lo ba di bom ba ba ba baw [1:30] BK4a: Wah bee bah bee bee, low bah dah-oh-ah, PAS: wa biː ba biː biː loː ba də o wa BK4b: lah dah bee bop bah deep bah feh. PAS: la da biː bap bə diːp ba bɛ̃ [1:35] BK5: Dah to dit dit dew dup, dee duh doe. PAS: da tu dɩt dɩt duː dəp diː də doːl

186 Scat syllables and markedness theory

BK6a: Rip dee duh duh dew dah daw-ee-ya doe doe dip dip, PAS: rɩp diː də də duː da do wiː yo də dɩt dɩp  [1:40] BK6b: baw buh bah bah baw beep bah beep baw bah baw bah bah beep bah beep WRB: | | | | | | | boh ‰ bə Œ boh ‰ ba Œ bə ‰ biy Œ bə ‰ biy Œ bow ‰ bə Œ bow ‰ bə Œ ba ‰ biy Œ ba ‰ biy Œ | | | | | | | bɔ ‰ bə Œ bɔ ‰ ba Œ bə ‰ biy Œ bə ‰ biy Œ bow ‰ bə Œ bow ‰ bə Œ ba ‰ biy Œ ba ‰ biy Œ PAS: bõ ba bã bõ bãw biːp bãw biːp bõ ba bõ bõ bã biːp bã biːp  BK6c: thiz dit duh duh. PAS: di dɩt də də [1:49] BK7: Reap dew dit done, dah nah naw naw deep dah dee, dah done dah dew. PAS: rip diw duːt dən da də na nə diːp də diː da dan də dɛl [1:52] BK8: Bah bah dah beep bew. PAS: ba bə da biːp biw BK9: Bah bee dut zuh bow. PAS: ba bi dəp da bõ [1:56] BK10: Wah-oh dove dew, duh boop bee dew the boop, wah-oo-lough. PAS: wæuw̃ dag duː də buːp biː duː diː buːm wæuː læw [guitar] [2:02] BK11: Zwee boo bee dew um-wow dah-dah-wow. PAS: zwiː buː bi duː ṃ mwaw də də waw [guitar] [2:06] BK12: Oooooo dah-dum-wah um-mough hmaf hwow. PAS: ũː ː da dəm wæw̃ ṃ mæw̃ ṃ waw [guitar ... ] [2:11] [2:16] BK13: Reap deh diddle dee tih duh, boo wuh buh bow. PAS: rɩp di du dəl di bə dæp bwa bæ bo 2.2.1. Baker, Chet. Everything Happens to Me. (1955) Verve Jazz Masters 32, Verve CD 314 516 939–2. Track = 3.31 minutes, Scat bridge = [2:27–2:58]. Transcription JB by Jim Bastian (Bastian and Alexander 1995: 14). Transcription PAS and time markers by Patricia A. Shaw.  [2:27] JB: | Œ ‰ Det ’n deh dah ’n duh– | – dit dah dah dah– Œ PAS: | Œ ‰ dɛt ṇ dɛ dʋt ṇ dɛː– | – dɩt dɛ dɛ ðɛː– Œ  [2:34] JB: | ‰ ah det ’n deh deh deh– | – det dee yah dah– ‰ PAS: | ‰ hɛː– dɛt ṇ dɛ də ðɛː– | – dət diː yə dɛː– ‰  [2:42] JB: ‰ yah dah deh | deh ‰ PAS: ‰ yʋ dən dæ | dɛː ‰

187 Patricia A. Shaw

 [2:44] JB: ‰ dah ee dah dut ’n dah dee– dah | dah– yah dah deh deh– ‰ ‰ PAS: ‰ də iː dɛ dʋt ṇ də deː– də | dəː– yə də ðɛ ðɛ– ‰ ‰  [2:50] JB: bah | deh– deh deh deh dah dah dah det ’n deh dah dah dah– |– det ’n deh Œ PAS: bə | ðɛː– ðɩ dɛ də ðɛ ðɛ ðɛ ðʋt ṇ də də ðɛ dɛː– |– ðɛt ṇ ðɛə Œ 2.2.2. Baker, Chet. Everything Happens to Me. (1989) Chet Baker Sings and Plays from the Film ‘Let’s Get Lost’, Novus CD3054-2-N. Track = 5:15 minutes, Scat bridge = [3:47–4:21]. Transcription JB by Jim Bastian (Bastian and Alexander 1995: 15). Transcription PAS and time markers by Patricia A. Shaw.  [3:47] JB: ‰ ≈ Hoo day dut ’n dah dah deh– deh deh deh | deh– ‰ Ó PAS: ‰ ≈ huː deː dʋt ṇ de yə deyː– də də də | ðə– ‰ Ó  [3:57] JB: Œ ‰ ≈ yah deh deh dah deh– | – Œ PAS: Œ ‰ ≈ dʋː də deːy – dʋ ̃ diː– | – Œ  [4:06] JB: Œ dah deh dah deh deh deh deh deh deh deh deh dee– | – PAS: Œ dəw dɛ dʋ dɛ dʋ də duː dɛw duː dʋ də diː – | –  [4:10] JB: – dee dee dee dee dee dee dee dee dee dee deh det ≈ deh– | – PAS: – diː duː dɩː de du dɛ dʋ duːː dʋ dɛt ≈ dɛː – | –  [4:15] JB: ‰ ‰ ≈ yeh deh det ’n deh dit ’n deh dee ee– | – day doo ee doo– Œ PAS: ‰ ‰ ≈ dɛ dɛ dɛt ṇ dɛ dɩt ṇ dɛ du–iy– | – dɛ duː ʋː– ʋː Œ 2.3.1. Carter, Betty (née Lillie May Jones) 1929–1998. Thou Swell. (1955) Meet Betty Carter and Bryant. Columbia/Legacy CK 64936, 6. (from A Connecticut Yankee. 1927. Rodgers and Hart, WB Music Corp.) Transcription by W. R. Bauer (2002a: 251; includes musical transcription). Transliteration (2nd line) follows the principles outlined in Appendix 1.  (nasal) WB: ‰ hə lə dow di dl ə yə də | hiy– də də ba yow bow || ba ba duw bə ba ‰ biy | ‰ hə lə dow dɩ dḷ ə yə də | hiy– də də ba yow bow || ba ba duw bə ba ‰ biy |  WB: – ə di dn di də yow bow | ba ba duw bə ba ba duw bə | – ə dɩ dṇ dɩ də yow bow | ba ba duw bə ba ba duw bə |  WB: uw də ‰ ə duw lyə duw uw – | – uw ə duw lyə a əm | bey dow– ‰ ə | uw də ‰ ə duw lyə du wuw – | – u wə duw lyə a əm | bey dow– ‰ ə |  WB: la əm biy bə duw bə duw wiy – | – da ə yuw duw bə | du du di – duw bə | la əm biy bə duw bə duw wiy – | – da ə yuw duw bə | du du dɩ – duw bə |  WB: di dəl də i– duw əm | ba bə də šiy – ra | ‰ ə la əm biy bə duw bə | dɩ dəl də i– du wəm | ba bə də šiy – ra | ‰ ə la əm biy bə duw bə |

188 Scat syllables and markedness theory

 WB: iy bə duw bə ‰ ə ba bə | di dliy ə duw bæw– Œ | ‰ hey ba bə spi də lə di də lə | iy bə duw bə ‰ ə ba bə | dɩ dli yə duw bæw– Œ | ‰ hey ba bə spɩ də lə dɩ də lə |  WB: ba bə di də lə bey (ay || fiyl ) Œ Œ || ba bə dɩ də lə bey ‘I feel’

2.3.2. Carter, Betty. Open the Door (1979) Words and music by Betty Carter, 1964. MyKag Publ. Co. The Audience with Betty Carter. Bet-Car MK 1003; reissue Verve 835 684–1. Transcription by W.R. Bauer (2002a: 294–303; includes musical transcription). Transliteration (2nd line) follows the principles outlined in Appendix 1.

 WB: Œ ṃ– | – də | dow | ... | ... |  WB: Œ Œ ṇ duw– duw– | duw |  WB: Ó Œ ‰ hṇ | duw– duw– | – diy | duw | ... | ... |  WB: Œ ‰ duw iy uw iy uw | duw – | – Ó | ... | ... | Œ ‰ du wi yu wi yuw | duw – | – Ó | ... | ... |  WB: Œ le də dey dey– | – dey dey– dey dey | dey– | – Œ Ó | ... Œ lɛ də dey dey– | – dey dey– dey dey | dey– | – Œ Ó | ...

3.1. Barry Mann. “Who Put the Bomp?” (1961) Words and Music by Barry Mann and Gerry Goffin. ABC-Paramount 45 NK-10237. Lyrics from The Official Barry Mann and Cynthia Weil Website: http://www.spectropop.com/hmannandweil.html. Transcription by Patricia A. Shaw. Due to space limitations, only lines that have scat are included below, and only scat syllables are transcribed. For reference, lines are numbered at the left. 1. Who put the bomp 7. Who put the dip bamp dɩp 2. In the bomp bah bomp bah bomp? 8. In the dip da dip da dip? baːm bə baːm bə bam dɩp dɩ dɩp dɩ dɩp 3. Who put the ram . . . ræm 4. In the rama lama ding dong? 9. Boogity boogity boogity ræ mæ læ mæ dɩŋ dãŋ bu gɩ di bu gɩ di bu gɩ di 5. Who put the bop 10. Boogity boogity boogity shoo bap bu gɩ di bu gɩ di bu gɩ di šu 6. In the bop shoo bop shoo bop? . . . bap šə bap šə bap

189 Patricia A. Shaw

3.2. Sha Na Na “Mr. Bass Man” Words and Music by Johnny Cymbal (1963). Original release: Kapp 503. Re-recorded by Sha Na Na, Rama Lama Ding Dong (1980). Transcription by Patricia A. Shaw. [Note: The lyrical lines are marked for BM “Mr. Bass Man”, W “I wanna be...”, and Back for the back-up singers’ line. English is in italics and in standard orthography. Only scat syllables are transcribed.] For reference, scat lines are numbered at the left. BM: 1. baw bə bə baw bə baw bə baw baw 2. bə bə baw bə bə baw bə baw bə baw bəm bəm W: Mr. Bass Man, you’ve got that certain something... Mr. Bass Man, you set that music thumpin’ To you it’s easy when you go 1-2-3 3. bə bʋ bə bam ba? BM: You mean: 4. bʋ bə baw bʋ bə bʋ baw bə baw baw ba ? W: Yeah! Mr. Bass Man, you’re on all the songs 5. sə dɩ də bɩ bə bum bum 6. And the dɩ dɩt ba ba bæw Hey Mr. Bass Man, you’re the hidden King of Rock ‘n’ Roll 7. bə bə bə ba bæw?? [0:33] BM: No, no! 8. bə bə ba bə bə bu ba bə ba ba bæw W: Oh, it don’t mean a thing when the leader’s singin’ Or when he goes [0:41] 9. ʔay yay yay yay yæ yæ Hey Mr. Bass Man, I’m askin’ just one thing Will you teach me, mmm, yeah, the way you sing ‘Cause Mr. Bass Man, I wanna be a bass man too 10. bɩ bʋ bə ba bɑ?

BM: Try this: 11. bʋ bə bæ bu bu bə ba bə bæw [1:00] W: Oh Mr Bass Man, I really think I’m winnin’ 12. With the bɩp bum bõ 13. And a dɩ dɩt dɩ dæ dæ Oh Mr. Bass Man, now I’m a bass man too 14. dɩ də dɩ dæ BM: That’s it! 15. bu bə bæ bu bə bu ba bə ba Back: 16. bəm bəm bəm bə bə bʋm bʋm bə bəm bə bə bə bʋm bʋm ba bə bʋm BM: Now you! [1:18] W: 17. də dɩ dṇ dɩ bə bum bum bo bəm bə bə bə bə bom bʋm BM: With me

190 Scat syllables and markedness theory

BM/W: 18. bʋm bə bə bɔm bə bə bɔm ba bum ba bəm 19. bʋm bə bə bəm bə bə bəm _ bʋm bəm 20. bəm bə bə bum bə bə bəm bə bə bəm bə bə bʋm bəm 21. bʋm bə bəm bə bə bu bə bu bʋ bə bəm [1:30] W: Oh, it don’t mean a thing oh when the leader’s singin’ Or when he goes [1:35] 21. ʔay yay yay yay yæ ya Hey Mr. Bass Man, I’m askin’ just one thing Will you teach me, mmm, yeah, the way you sing ‘Cause Mr. Bass Man, I wanna be a bass man too 22. ba bɑ BM: Try this: 23. bʋ bə bæ bu bu bə ba bʋ bæw W: Oh Mr. Bass Man, I really think I’m winnin’ With the 24. bɩp bum bõ And a 25. dɩ dɩt dɩ dæ da(w) Oh Mr. Bass Man, now I’m a bass man too [2:05] 26. dɩ də dɩ dɩ də dæw BM: That’s it. 27. bu bə bæ bu bə bu ba bə bæw [2:08] 28. bəm bəm bəm bə bə bə bʋm bʋm bʋm bə bə bə bə bə bʋm bʋm bə bə bʋm Now you! [2:12] W: 29. bə bə bom bə bə bom bə bum BM: Soundin’ good ... With me [2:16] BM/W: 30. bʋm bə bə bʋm bə bə bʋm ba bʋm ba bʋm 31. bʋm bə bə bəm bə bə bəm _ bə bəm 32. bəm bəm bə bə bəm bʋm bə bə bəm bə bə bəm bə bə ba bṃ 33. bə bəm bə bə bə bə bʋm bə bəm 34. _ bum bum bə bʋm bə bə bə bə bʋm 35. bə bə bə bə bʋm bʋm bə bʋm bə bə ...

191