LIBRARY 
 
 OF THE 
 
 UNIVERSITY OF CALIFORNIA 
 
 OR 
 
 Class 
 
 W \\ 
 
STUDIES 
 
 FROM THE 
 
 Yale Psychological Laboratory 
 
 EDITED BY 
 
 EDWARD W. SCRIPTURE, PH.D. 
 
 Director of the Psychological Laboratory. 
 
 Volume IX. 
 1901 
 
 CONTENTS. 
 
 Researches on the rhythm of speech, by J. E. WALLACE WALLIN, i 
 Note, 143 
 
 YALE UNIVERSITY, NEW HAVEN, CONN. 
 
 Trice, ^i.oo. Obtainable at NEW HAVKN, Psychological Laboratory, 84 High 
 Street; NEW YORK, Gustav K. Stechert, 9 East Sixteenth Street ; LONDON. Williams 
 and Norgate, 14, Henrietta St., Covent Garden; LEIPZIG, Bernhard Liebisch, 
 Kurprinz-Strasse 6 ; PARIS, Schleicher Freres, 15 Rue des Saints-Peres; BOLOGNA, 
 Pietro Virano, Piazza Galvahi. 
 
STUDIES 
 
 FROM THE 
 
 Yale Psychological Laboratory 
 
 EDITED BY 
 
 EDWARD W. SCRIPTURE, PH.D. 
 
 Director of the Psychological Laboratory. 
 
 19O1 
 VOL. IX. 
 
 YALE UNIVERSITY 
 NEW HAVEN, CONN. 
 

 
 
 COPYRIGHT, 1902, 
 
 BY 
 EDWARD W. SCRIPTURE. 
 
 PRESS OF 
 
 THE NEW ERA PRINTING COMPANY, 
 LANCASTER, PA. 
 
CONTENTS 
 
 Researches on the rhythm of speech, by J. E. WALLACE WALLIN, i 
 Note, 143 
 
 134505 
 
OF THE 
 [ UNIVERSITY } 
 
 OF 
 
 RESEARCHES ON THE RHYTHM OF SPEECH 
 
 BY 
 
 J. E. WALLAC^: WALLIN. 
 I. HISTORICAL. 
 
 BRUCKE l employed two methods in investigating the time relations of 
 syllables and feet in verse. 
 
 In the first method the subject of the experiment beat time with the 
 finger upon a key, in unison with the rhythm of the scanning of the 
 particular verse that was read by him. The key was connected with a 
 marker, which was so arranged that the deflections were traced upon a 
 recording drum. The distances between the checks in the record line 
 were regarded as representative of the duration of time between the suc- 
 cessive points of emphasis. 
 
 In the second method a recording lever was attached to the lower lip 
 or to the teeth of the lower jaw, or, though rarely, to the corresponding 
 parts of the upper jaw. A series of monosyllables, consisting of such 
 words as ba, bam, pap, etc., was given to the subject to scan with the 
 utmost regularity in accordance with predetermined types of meter. 
 
 The general conclusions indicated that the intervals between the ac- 
 cented syllables of poetry are exactly equal irrespective of the quality of 
 the material which constitutes the intervals. Whenever the number of 
 unemphatic syllables which occur between two accented syllables exceeds 
 the average, they are uttered with a rapidity sufficient to make the inter- 
 vals equal as to time ; and whenever, on the other hand, the number is 
 deficient, the loss is made up either by prolonging the syllables or by in- 
 troducing pauses. 
 
 BRUCKE' s investigation was concerned with a purely mechanical or ar- 
 tificial method of scanning, different from the natural, free, or artistic read- 
 ing. It is hardly justifiable to apply his conclusions to ordinary spoken 
 verse in support of the principle of equality of feet. 2 The free rhyth- 
 mical flow is the expression of impulses, unconsciously operative, that are 
 in the mind of the poet in writing verse. 3 Mechanical scansion is the 
 
 1 BRUCKE, Physiologische Grundlagen der neuhochdeutschen Verskunst, Wien 1871. 
 
 2 PAUL, Grundziige d. germanischen Philologie, II (i) 909, Strasburg 1893. 
 3 POE, The Rationale of Verse, Works VI 84, 92, Chicago 1895. 
 
 I 
 
2 /. E. Wallace Wallin, 
 
 expression of impulses consciously striving to produce perfect uniformity, 
 for the sake of a certain pleasure which is derived from the satisfaction 
 of expectant attention. In so far as it is perfectly mechanical, it disre- 
 gards the logic and sense of the word, for the fuller gratification of these 
 impulses which are relatively incapable of being intellectualized and which 
 subserve but a limited function. 
 
 BRUCKE'S conclusions are accurate only within wide limits owing to 
 the inaccuracy of the apparatus of that date. The measurements were 
 recorded on a crude form of the kymograph, the rate of rotation of which 
 could not be increased sufficiently to register 1 the minute differences 
 which inevitably existed between the individual measurements. 
 
 In reference to BRUCKE'S second method, we have to consider that 
 the concatenation of syllables in the particular series that was scanned, 
 was arbitrary and artificial. The syllables were uniformly monosyllabic, 
 and of such a character as to require relatively equal strain on the organs 
 of speech in uttering them. A combination of difficult and easy syllables, 
 physiologically considered, would probably have yielded different results. 
 
 The experiments of HURST and McKAY, 2 with more accurate appar- 
 atus, were measurements of the intervals between the beats of the finger 
 made in unison with scanned verse. The method was, in the main, like 
 that of BRUCKE. 
 
 The following conclusions were reached : ( i ) the feet of a given verse 
 are equal in length ; (2) dactyls and trochees are shorter, respectively, 
 than anapaests and iambics ; (3) a radical difference exists between the 
 anapaest and the dactyl, the length of the syllables in the former being of 
 an "ascending," in the latter of a "descending," order; (4) there is 
 a lack of a fixed proportion between syllables the emphatic, however, 
 being longer than the unemphatic. 
 
 The postulates which underlie all similar attempts at measuring the 
 duration and equality of poetical feet, are essentially two : ( i) the rhythm 
 of the scanned verse agrees with the rhythm of the taps of the finger, 
 and (2) neither exercises an influence upon the other and the rhythm of 
 the finger does not tend to regulate the rhythm of the scansion. Neither 
 of these assumptions is legitimate. In the first place, the two types of 
 motor innervation do not exactly correspond. This has been proved 
 by the experiments of MIYAKE, at the Yale laboratory, in which the 
 subject beat upon an electrical key with the finger as the sounds ma, pa, 
 
 1 MEUMANN, Untersuchungen zur Psychologic und Aesthetik des Rhythmus, Philos. 
 Stud., 1894 X 418. 
 
 2 HURST AND McKAY, Experiments on time relations of poetical meters, Univ. Toronto 
 Stud. (Psychol. Series), 1900 157. 
 
Researches on the rhythm of speech. 3 
 
 etc., were spoken into a voice key. The beats of the finger preceded 
 the sounds. A similar result has beeen obtained in the present series of 
 experiments. The subject was required to beat time on the table upon 
 which the phonograph rested in time with the rhythm of the accents of 
 several verses of faultless meter, which were mechanically scanned. The 
 finger beats were transmitted to the phonograph cylinder through the 
 frame of the phonograph. The results showed discrepancies between the 
 two beats, the beat of the finger generally being anterior to the metrical 
 accents. In the second place, the experiments, to which reference was 
 last made, indicate that the movements of the finger serve as regulators 
 of the movements of the larynx. The subject was directed to scan as 
 regularly as possible both with and without the beats of the finger. The 
 result, as may be seen by turning to the E. H. T. records (Tables LXIL 
 and LXXV. below), was that the regularity in the former instance was 
 more than trebled. This explains the high degree of regularity obtained 
 in the measurements of the two preceding series of researches. 
 
 KRAL and MARES* made use of a more direct method of measuring 
 vocal sounds. The apparatus employed was a telephone receiver, in 
 electrical contact with the nerve of a muscle of a frog's leg, the con- 
 tractions of which were registered upon the smoked paper of a kymograph 
 by means of a recording arm. The sounds spoken into the receiver pro- 
 duced electrical vibrations that stimulated the nerve and caused the muscle 
 to contract. 
 
 The experiments included two kinds of measurements: (i) the 
 length of syllables, long and short, uttered separately ; and (2) the 
 measures of scanned verse. The language spoken was Bohemian. 
 
 Four general results were obtained in this investigation: (i) long 
 vowels and diphthongs, though generally longer than short vowels, were 
 often of the same length; '(2) neither of these had absolute time- 
 values, even for the same person ; (3) consonants required a very short 
 time for their utterance, the addition of consonants to syllables not be- 
 ing attended by a proportionate increase in the duration of the sylla- 
 bles, and some additions of consonants tending to shorten rather than 
 lengthen the syllables ; (4) even in attempting to scan verse so as to 
 exactly coordinate the time of the measures, an equality of time-val- 
 ues was not obtainable, whether the scansion was according to the time 
 theory or the emphasis theory of meter. 
 
 While syllables should be measured, not as independent entities but as 
 interrelated elements of a complex group of syllables the conditions 
 
 1 KRAL A MARES, Trvdni hldsek a slabik die objectivnt miry, Listy Filologicke, 1893 
 XX 257. 
 
4 /. E. Wallace Wallin, 
 
 under which they occur in speech these conclusions are important as 
 affecting the comparison of like syllables under like conditions. The 
 conclusion with reference to the time-value of poetical measures, it is 
 interesting to notice, is at variance with the conclusions of the two former 
 experimenters. 
 
 MEYER/ in a series of experiments the publication of which appears 
 never to have been completed, has measured the syllables and feet of 
 spoken verse. An electric tuning fork vibrating 100 times in a second 
 recorded the time line on a smoked drum. A MAREY tambour registered 
 the breath pressure from the mouth. Two trumpet-shaped speaking 
 tubes of paper were attached to the tambour by means of a short glass or 
 metal tube. One of the speaking tubes was made to conveniently fit the 
 mouth cavity ; the other, the nose. The tubes could be used singly or 
 in combination. The result of speaking into the tube was to transmit to 
 the rubber diaphragm of the tambour a series of waves corresponding to 
 the sounds and silences of the utterances. The fluctuation of the dia- 
 phragm were registered by an amplifying straw lever. Thus the duration 
 of the successive explosions of the stream of outflowing air could be 
 measured. The successive puffs of air were hypothetically assumed to 
 extend over the same times as the successive sounds. 
 
 A further development of the apparatus consisted in connecting a tap- 
 ping instrument to a recording lever whose point was adjusted immedi- 
 ately under the point of the tambour. 
 
 The method of taking the records consisted in repeating monosyl- 
 lables singly or as a series of words, at the same time beating with the 
 finger upon the tapping instrument in time with the articulation of the 
 syllable at the moment of greatest emphasis ( ' ' Arsengipfel " ) . The sub- 
 ject was thoroughly trained in coordinating the time of tapping with the 
 time of uttering the words. The attempt was made to make that point 
 of the syllable which represented the climax of the energy exerted by 
 the vocal organs exactly correspond with the tapping time. This was 
 assumed to be possible, for since both processes start from the same 
 center the maximum points of force of the two innervations should occur 
 synchronously. Thus the tapping mark upon the drum, regardless of 
 where it occurred in the course of the sound, represented the point at 
 which the energy of the syllable was the highest. 
 
 The object of the investigation was twofold: (i) to determine the 
 point of highest energy of single syllables beginning with different con- 
 sonants, such as pha, fha, ha, za,fma, etc. ; (2) to investigate the meter 
 of German verse. 
 
 1 MEYER, Beitrage zur deutschen Melrik, Neuere Sprachen, 1898 VI 1-37; 122-140. 
 
Researches on the rhythm of speech. 5 
 
 The author makes the following general deductions : (i) the point of 
 energy occurs in the initial consonant or consonants somewhere shortly 
 before the beginning of the vowel, varying somewhat according to the 
 difficulty of articulating the consonants 1 ; (2) the moment of the least, 
 and not the moment of the highest, energy of speech determines the 
 boundaries of the foot 2 ; (3) whenever the beginning of the arsis (part 
 of syllable preceding point of highest energy) coincides with the begin- 
 ning of a syllable, and the end of the thesis (part following the point of 
 highest energy) coincides with the end of the syllable, the boundaries of 
 the foot coincide with the boundaries of the syllables 2 ; and (4) when- 
 ever unemphatic syllables uniformly coincide with the arsis or the thesis 
 of an accented syllable, the boundaries of the foot of verse and prose 
 (" Sprechtakt ") become identical. 2 
 
 MEYER states 3 that the investigation was subject to the following 
 sources of error, which were incapable of elimination : (i) latent time 
 of the apparatus (" Registrirverzogerung " ), some loss in the time of 
 transmission, and in the quantity of the energy to be recorded in the 
 passage from the receiving to the recording points being inevitable ; 
 (2) errors of measurement (" Wahrnehmungsverzogerung" ) owing to 
 the limited capacity of the visual organ to discriminate small differ- 
 ences. 
 
 The following considerations, as affecting MEYER'S experiments, may 
 be noted: (i) The measurements are essentially measurements of 
 breathing. The air waves may or may not precisely coincide with the 
 sounds. (2) An exact simultaneity in the occurrences within the central 
 organs of the highest points of energy for the innervations of the vocal 
 muscles and the muscles of the hand may perhaps be assumed. The ad- 
 ditional assumption is made that the registering of the impressions are 
 also synchronous. Since, however, the media for transmitting the im- 
 pressions are not the same (the air and the hand), this may be unjustified. 
 Moreover, it has been proved that no muscular movements are capable 
 of exact coordination. Contrary to the supposition (although the move- 
 ments of the hand are not reactions to the movements of the organs of 
 speech) that the movements of the hand and vocal organs are not sub- 
 ordinate but coordinate, the movements of the hand may serve as a 
 regulative concomitant of the movements of the vocal organs. For this 
 reason the results may not apply to the rhythm of free declamation. 
 
 1 MEYER, Beitrdge zur detitschen Metrik, Neuere Sprachen, 1898 VI 134. 
 
 2 MEYER, as before, 138. 
 
 3 MKYER, as before, 26-30. 
 
6 /. E. Wallace Wallin, 
 
 TRIPLETT and SANFORD/ using apparatus similar to that of BRUCKE, 
 HURST and McKAY and MEYER, have measured eleven nursery rhymes of 
 long, common and short meter, scanned by themselves eight or ten times 
 with great regularity, and have studied without apparatus, the patterns 
 adopted. A finger record of taps, a vocal record representing breath 
 puffs, and a time line, made by an interrupter checking off tenths of a 
 second, were traced on the kymograph. The voice record was used for 
 comparing the exactness of the correspondence between the finger taps 
 and the expiratory stresses. The intervals between the finger taps were 
 measured. 
 
 The following conclusions were reached : ( i ) If the pause intervals be- 
 tween the larger rhythmic units are excluded, the measures of scanned 
 nursery rhymes are approximately uniform. (2) There is a tendency to 
 accelerate the speed of scanning the successive measures in the verse 
 as well as the successive verses themselves. (3) No characteristic differ- 
 ences of speed are found between dactylic and trochaic, anapestic and 
 iambic measures. This is contrary to the 'results of previous experi- 
 menters. (4) The most frequent patterns of verse have a characteristic 
 movement, due partly to the distribution of the pauses and the tendency 
 to increase the speed. 
 
 This cursory review of previously existing methods of experimentally 
 investigating the time relations in speech illustrates and enforces the 
 necessity of improved methods of speech investigation. Two types of 
 methods may be used. The former is largely the method which has pre- 
 vailed up to the present time. The sounds of spoken language have 
 been measured by means of finger beats, currents of air and non-repro- 
 ducible sound vibrations. A more direct method consists in measuring 
 directly the sounds recorded in, and reproduced by a talking machine. 
 This method has the following advantages : 
 
 1. The instantaneous action of the recording stylus of the modern 
 talking machines practically eliminates the errors of the " latent time" 
 of the apparatus and the recording of superfluous and irrelevant move- 
 ments. 
 
 2. It is the only method by which the accuracy of the recorded im- 
 pressions can be completely verified, since it alone affords the means of 
 reproducing the sounds that have been recorded. The accuracy of the 
 impressions recorded upon instruments of the non-reproducing kind, 
 must always remain more or less conjectural. 
 
 1 TRIPLETT AND SANFORD, Studies of rhythm and meter, Am. Jour. Psych., 1901 
 XII 361. 
 
OF THE 
 
 UWlVERSIT/ } 
 
 Researches on the rhythm of speech. 
 
 II. APPARATUS. 
 
 The experiments now to be described were begun in the early summer 
 of 1900 and carried out during the academic year 1900-1901. 
 
 The machine used in making records was a LIORET phonograph, re- 
 sembling the graphophone in its construction. The rate of speed was 
 accurately recorded. 
 
 For reproducing purposes an EDISON phonograph was used. A gear 
 wheel with twenty-four teeth having the spaces filled with vulcanized rubber 
 was placed on the axle of the phonograph. A metal spring resting upon 
 the wheel made contact with each tooth. Every revolution of the con- 
 tact wheel made twenty-four closures of the circuit. An EWALD magnetic 
 counter placed in the circuit indicated the number of teeth passing under 
 the end of the spring. 
 
 A telegraph key was placed on each side of the phonograph. By 
 pressing the knob of the key on the left side, the current was sent 
 through the counter ; by pressing the key on the right side the cir- 
 cuit was interrupted. In this way the number of contacts between 
 the movements of the two keys was obtained. A knowledge of the 
 rate of revolution of the phonograph cylinder when the record 
 was taken made it possible to translate the number of contacts into 
 seconds. 
 
 The limit of the greatest possible speed capable of being utilized was 
 determined by two factors : (i) the degree of dexterity developed by the 
 experimenter in reacting upon the telegraph key to the given impressions ; 
 and (2) the limit of accuracy of the magnetic counter. Beyond a cer- 
 tain rate of speed the counter failed to register all the contacts. In none 
 of my measurements was this limit even approached. The highest num- 
 ber of contacts per minute required in the measurements was 2,880. The 
 instrument was capable of registering a much higher number of contacts. 
 Within the limits of speed, however, the accuracy of the measurements 
 depended upon the maintenance of a fairly constant relation between the 
 strength of the electrical current and the tension of the spring of the 
 armature of the counter magnet. The accuracy of the chronoscope was 
 verified by comparing the gross measurement of a large interval (e. g., 
 a verse or a stanza) with the aggregate of all the smaller intervals com- 
 posing it (the measures and pauses); and by comparing the time required 
 to measure off a stanza with the time required to record it, which had 
 been ascertained by the watch. 
 
 The limit of the slowest speed was dependent upon discriminative 
 audibility. At a certain speed the sounds lose their articulate character. 
 
8 /. . Wallace Wallin, 
 
 Except for very special reasons, the speed used in measuring corresponded 
 to the speed used in taking the records. This varied from 80 to 120 
 revolutions per minute. By maintaining the recording speed the records 
 were heard, in respect to speed and pitch, as they were made. The 
 measurements which were made before the experimenter developed a high 
 degree of proficiency were rejected 
 
 The method of measuring consisted in first listening, through the ear 
 tubes, to the record upon the cylinder and choosing the intervals to be 
 measured. The intervals were then listened to repeatedly, and each one 
 was measured from four to five times by pressing one key at the beginning 
 and the other at the end. The average of the different measurements 
 was recorded as the length of the interval. The successive measure- 
 ments of the same interval differed only very slightly, and were frequently 
 identical ; the measurements were thus quite accurate. 
 
 In making the phonograph records upon which this research is based, 
 the procedure was guided by the following considerations : 
 
 1. All mechanical distractions and impediments, such as noises, strain 
 of the muscles of the body or head occasioned by a forced standing or 
 sitting posture, interference with the easy movements of the lips, fatigue, 
 etc., of the subjects, were eliminated as far as possible. 
 
 2. The material contributed for study should come from subjects with- 
 out bias or preconception. 
 
 3. The subjects were uninstructed in respect to the purposes of the ex- 
 periment, except where the nature of the experiment required otherwise, 
 in order to minimize the elements of unnaturalness and intentional change 
 of the manner of speaking. 
 
 4. The subjects were of varied characters, with different languages, from 
 different countries, of different stages and walks of life (elementary and 
 high school pupils, academic and graduate students, professors, poets, 
 orators, musicians, etc.). 
 
 5. The character of each record was made a true copy of the natural 
 characteristics of the mode of delivery of the subject, so as to be typic- 
 ally his own, no matter what the idiosyncrasies. 
 
 By requiring the utterance .to be natural the research could assume the 
 role of being an investigation into the rhythmic peculiarities of actual 
 speech. In a few cases, however, it seemed advisable to omit offering 
 any instructions and the matter was left entirely to the taste of the reciter. 
 Moreover, in proffering suggestions due care was taken lest the advice 
 should tend to make the subject too self-conscious. Cases of mechan- 
 ical scansion were also included for the sake of comparison and complete- 
 ness of scope. 
 
Researches on the rhythm of speech. 
 
 III. THE CENTROID. 
 
 i. Concept of the centroid. A- portion of speech such as ''the cities 
 are full of pride," is physically a vibratory movement of a complicated 
 form of which various portions can be assigned in succession to the 
 sounds t>, 2, s, i, /, /, z, etc., with more or less abrupt changes (glides) 
 between them. The series of sounds represents : ( i ) a continuous succes- 
 sion of vocal movements representing work, (2) a continuous vibration 
 of an air particle, (3) a continuous succession of sensations. Both the 
 speaker and the hearer feel the recurrence of points in the succession at 
 which the impression reaches a maximum. These maxima are felt to 
 have their positions determined not only by the actual maxima of vocal 
 effort and of acoustic impression but also by the preceding sounds and 
 "by the anticipation of following ones. Owing to these circumstances the 
 maxima as felt may differ more or less from the actual vocal or acoustic 
 maxima. Such a maximum of effect may be called a ' centroid.' ' 
 
 The English poetical line may be considered as a certain quan- 
 tity of speech-sound distributed so as to produce an effect equivalent to 
 that of a certain number of points of emphasis at definite intervals. 
 The location of a point of emphasis is determined by the strength of the 
 sounds at and around it. It is like the centroid of a system of forces or 
 the center of gravity of a body in being the point at which we can con- 
 sider all the forces to be concentrated and yet have the same effect. 
 The centroid is not a syllable or a single sound but a point in the 
 course of a sound. 1 
 
 In the stream of vocal sounds the centroid lies in the syllable that 
 relative to adjacent syllables, contains the maximum amount of force or 
 loudness. The syllable containing the centroid may be called the " cen- 
 troid syllable." It has been called the accented, emphatic, stressed, 
 long or loud syllable. 
 
 Every syllable in a verse or sentence may contain a centroid. Whether 
 it is actually to become a centroid syllable when the words are read, will 
 depend upon its logical function and the circumstances of the utterance. 
 
 When the component sounds of speech are considered separately, the 
 centroid represents the maximum of energy of a syllable. This point 
 may, theoretically, either fall midway in the syllable, or precede or follow 
 the middle point. According to BRiiCKE, 2 basing his deduction upon the 
 
 SCRIPTURE, Researches in experimental phonetics (first series}, Stud. Yale Psych. 
 Lab., 1899 VII 101. 
 
 SCRIPTURE, Elements of Experimental Phonetics, New York 1901. 
 
 2 BRCUKE, Physioiogische Grundlagen der neuhochdeutschen Verskunst, Wien 1871. 
 
io /. E. Wallace Wallin, 
 
 measurement of scanned verse, when the vowel of the emphatic syllable 
 is short, the maximum of energy occurs at the end of the vowel or just 
 before the following consonant ; when the vowel is long and not im- 
 mediately followed by a consonant, it lies somewhere in the course of the 
 vowel ; and when the vowel is long, but does not terminate the syllable, 
 the point lies very near the boundary between the vowel and the conso- 
 nant. According to MEYER ' the point of highest energy in syllables con- 
 sisting of a vowel with an initial consonant lies in the early part of the 
 syllable, shortly before the beginning of the vowel sound. 
 
 The physical factors that locate the centroid are increase of amplitude, 
 increase of length and change in pitch. 
 
 Typical cases would be c = . 7 + / . 2 + A.i> or c = a ^ + 4a + A.o or 
 c = # 03 -}- / 04 -f-/o.> etc -> where c indicates the strength of a given cen- 
 troid, #, / and p the factors of amplitude, length and period, and the 
 subscripts the share contributed by each. 
 
 2. Kinds of centroids. Centroids admit of distinctions of kind only 
 from the psychological point of view. We may use the term "phrase 
 centroid" to denote the major or chief centroid of a given phrase. 
 Each phrase should contain one chief centroid, one dominant idea, one 
 specially important word. 
 
 According to the purposes of thought and feeling subserved by the 
 centroid, we obtain a threefold division : rhythmic, pronunciatory and 
 logical. 2 The function of the first is to mark off the different measures 
 of verse ; of the second, to emphasize the root syllables of words ; of the 
 third, to single out prominent words for special emphasis. 
 
 SuNDEN 3 distinguishes two kinds of accent in Swedish: (i) Word 
 accent, or the stress belonging to a separate word, constituting the word 
 when of two or more syllables a sound unity. This is subdivided into 
 (a) compound, or such accent as belongs to polysyllabic words having 
 a grave accent ("gravt's" = accent with low tone) on one syllable and 
 a strong subsidiary accent (" levis"} on the other, e. g., in Anna, 
 konung and mdnsken ; and () simple, or the stress belonging to a mono- 
 syllabic word with acute accent (" akut" = accent with high tone), 
 and a polysyllabic word with a weak subsidiary accent, e.g., in bok t 
 backer and bockerna {akut on first syllable). (2) Phrase or rhetorical 
 ( ' *' oratorisk ' ' ) accent, or the stress of a word in its relation to other words. 
 
 Physically there are no distinctions of kind among centroids. There 
 are no physical signs in the vibrations upon the phonograph cylinder 
 
 1 MEYER, Beitrdge zur deutschen Metrik, Neuere Sprachen, 1898 VI 123. 
 2 LANIER, The Science of English Verse, Chap. IV., New York 1880. 
 3 SUNDEN, Svensk Spraklara, 15, Stockholm 1895. 
 
Researches on the rhythm of speech. 1 1 
 
 whereby the logical varieties can be distinguished. The physical process 
 is a continuity, admitting of distinctions of degree, the general charac- 
 teristics of which are shared by all centroids. 
 
 That we are justified in considering all kinds of emphatic syllables as 
 centroid syllables, also appears from a consideration of the centroid as a 
 process in consciousness. The centroid is an impression which arouses the 
 sense of hearing to a certain pitch of intensity for a certain length of time. 
 It is in respect of intensity that consciousness distinguishes it from other 
 auditory sensations of the same nature. The strong sensations, irrespec- 
 tive of logical distinctions, are designated centroid syllables, the faint 
 sensations, non-centroid ones. 
 
 Hence, as an objective physical occurrence, measureable in amplitude, 
 period and length, and as an event in consciousness, capable of being 
 roughly estimated in respect to intensity, any emphatic syllable may be 
 included in the generic term of centroid syllable. 
 
 The psychological differences in the centroid may be accompanied by 
 correlative differences in tone-color or timbre of voice. 
 
 3. Method of locating the centroid. Two methods are available : (i) 
 We may measure the tracings of sound curves of the talking ma- 
 chines. The centroid will be located at that point along the time-axis of 
 a syllable where the influences of amplitude, length and pitch are the 
 greatest. (2) By listening to a phonograph record, and comparing thef 
 consecutive syllables in respect to the intensity of the auditory sensation 
 which they arouse, we may classify the syllables as weak and strong, j 
 This purely psychological method lies closest to nature, because speech \ 
 centroids, fundamentally considered, are mental quantities. 
 
 The psychological method, according to which the centroids have been 
 located in the present experiment, has been attended with some difficulty. 
 It has sometimes, though rarely, been difficult to determine which of 
 two, or whether any one of two, consecutive syllables is a centroid. 
 Appeal has frequently been made to other persons. Where doubtful cases 
 exist, the fact is indicated in the records by placing a ? directly after the 
 centroid. 
 
 4. The elements of the centroid. The centroid in speech is the result 
 of a number of complex mental forces which constitute an original and 
 inseparable synthesis, and which are the indispensable conditions of the 
 life of the centroid. These elements, constituting the fundamental frame- 
 work of the centroid, are capable of isolation by processes of analysis only ; 
 in any concrete case they constitute an irreducible complexity. 
 
 This complexity of elements is threefold including the factor of time, 
 factor of force, and factor of pitch. Every sound that is uttered is con- 
 
12 /. E. Wallace Wallin, 
 
 ditioned on these three fundamental processes. Every sound is con- 
 crete of a particular pitch, intensity and duration. 
 
 Furthermore, in order to excite the organ of hearing and give rise to a 
 percept, every sound must attain a certain intensity, persist for a certain 
 time, and fall within the limits of the psychological scale of pitch. 
 Hence, both physically and psychologically considered, these three 
 elements form the ultimate elments of the centroid. 
 
 5. The essence of the centroid. The question now arises, which, or 
 how many, of these factors constitute the essence of a centroid ? 
 
 (#) Historical summary. Several answers, somewhat obscure and incon- 
 sistent, have been given to the above questions. To enter into a detailed 
 discussion of these would be out of place. The distinctions that have 
 been introduced have turned either upon differences as existing between 
 different languages or as existing within a given language. 
 
 According to SWEET* and PoE 2 accent is practically non-existent in 
 French poetry. 
 
 Accent in the case of ancient Latin 8 and Greek, 4 as well as in ancient 
 and modern Sanscrit 5 and Persian, has been described in terms of pitch. 
 The nearest modern approaches to this species of emphasis are the Swedish 
 and Norwegian languages {akut accent}, the verse of which, however, is 
 governed by different laws. The Scotch emphatic syllable, on the other 
 hand, has been characterized as having a "grave tone," 6 that is to 
 say, as lacking almost entirely in pitch modification. Theoretically 
 Scotch would afford an instance of a monotone speech. SUNDEN 7 states 
 that grave accent (gravts*) is peculiar to Swedish and Norwegian ; and 
 elsewhere, 8 that in Swedish the elements of pitch (melodic accent, 
 " tonhojd* 1 '} and loudness (expiratory or dynamic accent, tonvikt") 
 coincide, and that an accented 9 syllable (akut or gravis} is always long. 
 
 The property of duration is said to contain the essence of the centroid 
 in Latin, Greek, Arabic, Persian and Sanskrit, by MAYOR 10 and ELLIS. " 
 
 I SWEET, Primer of Phonetics, 45-97, Oxford 1890. 
 
 *PoE, The Rationale of Verse, Works, VI 100, Chicago 1895. 
 3 ELLis, The Quantitative Pronunciation of Latin, 21, London 1874. 
 4 SYMONDS, Sketches and Studies in Southern Europe, II 328, New York 1880. 
 CLARK, Manual of Linguistics, 155 
 
 5 CLARK, Manual of Linguistics, 155. 
 
 6 GUEST, History of English Rhythms, II, London 1882. 
 
 7 SUNDEN, Svensk Spraklara, 15, Stockholm 1895. 
 
 8 As above, 14, 17. 
 
 9 SUNDEN, Kort Ofversigt af Svenska Vitterhetens Historia, 20, Stockholm 1885. 
 10 MAYOR, Chapters on English Metre, London 1886. 
 
 II ELLIS; The Quantitative Pronunciation of Latin, 21, London 1874. 
 
Researches on the rhythm of speech. 13 
 
 In the following two groups, emphasis is resolved into terms of loud- 
 ness or expiratory stress : Modern Greek, Italian, Spanish, German and 
 English; l and old Italic, Keltic, Teutonic and Lithuanian. 2 
 
 Just how far these distinctions are valid, it is not our purpose to dis- 
 cuss. This remains as a problem for future investigation. 
 
 We have reserved the theories of English accent for separate consid- 
 eration. It is in this province that our chief concern lies. 
 
 It is difficult to classify, or bring into any sort of unity, the different 
 utterances on the nature of the centroid in English verse, so indefinite 
 and conflicting are the statements statements that have been charac- 
 terized as more pitiable nonsense than has ever been written on any 
 other subject under the sun. 3 If we divorce the problem now under con- 
 sideration from the connected problem of the quantitative character of 
 English verse, which is reserved for discussion in the following section, 
 the difficulty is increased. We are here concerned with the properties 
 only of the centroid as such, and not of the centroid as a member of a 
 succession of centroids. 
 
 The most prominent feature in the discussions of English prosody is 
 the widespread assumption that the centroid in English is essentially a 
 phenomenon of loudness, force, or stress. This position is maintained 
 by ABBOTT and SEELEY,* MAYOR, 5 WALLis, 6 GUEST, G GURNEY/ ELLIS,* 
 LANIER S and others. A somewhat modified view is to the effect that there 
 is a natural, although not an essential, connection between increase of force 
 and increase of duration. 10 According to this view quantity is secondary ; 
 it is the variable element, while loudness is the stable element. ARNOLD " 
 goes perhaps a step further in affirming that in English poetry, " accented 
 syllables are long, and unaccented short. ' ' POE 12 holds that our start- 
 ing point in the conception of verse is quantity or length ; we begin with 
 the long syllable, and the long syllable is the, encumbered or emphatic one. 
 JOHNSON, 13 the litterateur, transferred the conditions of classical accent to 
 
 1 MAYOR, Chapters on English Metre, London 1886. 
 2 CLARK, Manual of Linguistics, 155. 
 
 3 POE, The Rationale of Verse, Works, VI 59, Chicago 1895. 
 
 4 ABBOTT AND SEELEY, English Lessons for English People, 153-154, Boston 1880. 
 
 5 MAYOR, Chapters on English Metre, III, London 1886. 
 
 * GUEST, A History of English Rhythms. IV, London 1882. 
 
 7 GURNEY, The Power of Sound, 429, 433, 437, London 1880. 
 
 * ELLIS, The Quantitative Pronunciation of Latin, II, London 1874. 
 LANIER, The Science of English Verse, I and II, New York 1880. 
 
 10 SEELEY AND ABBOT, GUEST, MAYOR, ELLIS, SYMONDS. 
 
 1 1 ARNOLD, Manual of English Literature. 
 
 12 POE, The Rationale of Verse, Works, VI 59, 79. 88, Chicago 1895. 
 
 13 GUEST, as above ; CORSON, A Primer of English Verse, IV, Boston 1893. 
 
14 /. E. Wallace Wallin, 
 
 English verse : the emphatic syllable is invariably long. LOTZE 1 ventures 
 the suggestion, without limiting its applicability to any one language, that 
 the emphatic syllable always requires a longer " Sprechzeit." 
 
 MiTFORD, 2 a musician, held that "sharpness" or pitch constituted 
 the essence of the accented syllable; and SCHMIDT/ that there is an 
 almost inseparable connection between pitch and intensity in English. 
 Finally, SWEET* asserts a natural, though not a necessary connection be- 
 tween all the three elements. 
 
 (^) Experimental solution, (i) Method. The method employed in 
 
 determining the essence of the centroid was as follows : Selecting the 
 
 centroids which had been previously located, of any given record, the 
 
 stylus of the recorder was made to pass over each syllable a score or so 
 
 f of times, with varying rates of rotation of the cylinder. By thus listen- 
 
 I ing attentively to the same syllable under varying conditions, it was 
 
 jjfr possible to determine what element, or elements, were predominant in 
 
 / the sense-impression. 
 
 The results were indicated by placing appropriate symbols above the 
 syllables. The next step consisted in verifying, by all available means, 
 the results of mental discrimination. In respect to duration, the method 
 employed was to measure the length of the syllables marked duration 
 centroids ; in respect of pitch, the syllables marked pitch centroids were 
 compared with the pitch of adjacent syllables by aid of the piano. 
 The element of duration and pitch could thus be estimated with 
 great precision as regards their physical aspects. Viewed phycholog- 
 ically, however, that is to say, from the point of view of their importance 
 as elements of intensive states of consciousness, a value could be assigned 
 to each only when compared with other points of emphasis, by way of 
 contrast, similarity, etc. Finally, the estimate of loudness reposes upon 
 the immediate testimony of consciousness. These considerations, how- 
 ever, do not effect the certitude of the determinations with reference to 
 the mere differentiation of the different factors. It is only as respects 
 the degree of loudness that the judgments may be questioned. 
 
 Altogether 513 centroids, selected from eight different records were 
 studied, namely, those of W. L. P., J. W. R., G. A. A., C. O. S., A. 
 D. B., C. O., W. C. and W. W. The results are given in the two fol- 
 lowing tables. 
 
 1 LOTZE, Geschichte des Aesthetik, 301. 
 
 2 GUEST IV, and CORSON IV, as before. 
 
 3 SCHMIDT, Introduction to the Rhythmic and Metric of the Classical Languages, 
 Boston 1878. 
 
 4 SWEET, Primer of Phonetics, 67, Oxford 1890. 
 
Researches on the rhythm of speech. 15 
 
 TABLE I. 
 
 Centroid elements. 
 Principle of synthesis. 
 
 999 
 
 Ol O2 O3 | 1 2 3 ; 1 '2 3 
 
 250 35 3! -5 16 20; 9 7 
 288 1 41 : 16 
 
 
 i 
 
 3 
 
 A ] A 2 AS 
 
 3 6 2 
 
 15 ii 
 
 A A A 
 l 
 I 4 2 
 
 A ] A 2 AS 
 
 4 I 
 
 , number of centroids. 
 
 t, aggregate number of centroids per group. 
 
 j, 2 , 3 , degrees of centroids. 
 
 Number of centroids in entire series of groups, 383. 
 
 TABLE II. 
 
 Centroid elements. 
 Principle of substitution. 
 
 1 2 3 ! 2 3 
 
 15 30 14 28 20 7 
 
 A i A 2 
 
 " 5 
 
 t 59 i 55 I 16 
 
 n = number of centroids. 
 / = aggregate number of centroids per group, 
 u 2> 3' degrees of centroids. 
 Number of centroids in entire series of groups, 130. 
 
 (2) Explanation of symbols. The following symbols are designed to 
 represent the properties, or property, which enter into the make up of 
 the centroid. 
 
 The open circle (o) denotes a relative equality of pitch, loudness and 
 duration emphasis. Ideally this sign should signify that the share con- 
 tributed by anyone of these elements is equal to that of any one of the 
 others. The equality, however, is only relative. In English and per- 
 chance in other modern languages, the element which has the greatest 
 claim to supremacy is that of loudness. Hence the open circle is rep- 
 resentative of a synthesis of three elements, in which the element of 
 loudness is slightly predominant. 
 
 In the dot circle (), or the circle with the dot in the center, the fac- 
 tor of loudness distinctly predominates over the factors of duration and 
 pitch ; in the closed circle () the element of duration preponderates ; 
 and in the inverted triangle the pitch factor overbalances the other two. 
 Of the latter sort two alternatives are possible. The pitch emphasis may 
 be due, in both cases by way of contrast, to a raising (A) or lowering 
 ( v) of the pitch. The latter is the rarer of the two, as will be seen by 
 a study of the records. Theoretically, either is an extreme departure f \^ 
 from the medium pitch of any given record, and tends to an increase of 
 emphasis. 
 
 7' 
 
 OF THE 
 
 UNIVERSITY 
 
1 6 /. E. Wallace Wallin, 
 
 The superposition of one symbol upon another signifies that the symbol 
 which is placed underneath contributes the main part of the total effect. 
 The upper sign indicates the element next in importance. The third 
 element, of course, is subordinate to the element of the uppermost sign. 
 
 In each of these cases one element is predominant ; while the other 
 two are subordinate to this element, they may be equal to each other. 
 The weaker elements are always present as the substratum of the centroid. 
 They are the conditions of its very possibility as a vocal phenomenon. 
 The abolition of any one signifies the abolition of the process of vocal- 
 ization. 
 
 The effect of these several amalgamations, as well as the relations 
 which they sustain to one another in respect to frequency and stress, will 
 be seen in what follows. 
 
 6. Degrees of the centroid. The subscripts to the symbols denote 
 the relative degrees of the force or strength of the centroid as an effect 
 in consciousness. They indicate the degree of the entire centroid and 
 not that of the particular elements to which they are attached. A great 
 number of variations in stress are possible, but the degrees have been 
 limited to the first, second and third. Consciousness can immediately 
 distinguish three degrees of auditory sensations: (i) high, medium, 
 low; (2) loud, medium, weak; (3) long, medium, short. 
 
 ELLIS/ has elaborated a scheme, whose main defect is its over-minute- 
 ness of analysis, in which five different elements (force, length, pitch, 
 weight, and silence) of emphasis are distinguished. To each one of 
 these elements he ascribes nine possible degrees of stress. The theory 
 has been justly criticised, as well as misinterpreted. 2 The misinterpre- 
 tation is due to a misunderstanding of the nature of the centroid. The 
 ] centroid, as already affirmed, is not the resultant of the agglomeration of 
 / isolated fragments superimposed upon one another, but a complexity orig- 
 j inally and inseparably multiplex. None of its elements exists in isolation. 
 * Nothing is added to a third degree centroid, except an intensification of 
 one or all of its elements. Hence the scheme of ELLIS provides for nine 
 degress of stress, and not for forty- five, 3 as has been supposed. 
 
 A like over-minuteness of analysis is characteristic of the system 
 of SWEET. 4 He distinguishes five degrees of quantity (very long, 
 long, half long or medium, short, very short), three degrees of force 
 (level, crescendo, diminuendo), four degrees of stress (very strong, 
 
 v * ELLIS Accent and emphasis, Trans. Eng. Philol. Soc., 1873-74 and June, 1876. 
 
 2 MAYOR, Chapters on English Metre, 57~74, London 1886. 
 
 3 MAYOR, Chapters on English Metre, 69, London 1886. 
 
 4 SWEET, Primer of Phonetics, 43-66, Oxford 1890. 
 
Researches on the rhythm of speech. 1 7 
 
 strong, half strong, weak), and three primary forms of intonation (level, 
 rising, falling). 
 
 In Swedish l the following degrees are recognized : 4th, the principal 
 accent of the acute species, e. g., the first syllable of anden (from and) ; 
 3d, the principal accent of the grave species, e. g., the first syllable of 
 anden (from ande) ; 2d, the strong subsidiary accent, e. g., the second 
 syllable of anden (from ande~}\ ist, the weak subsidiary accent, e. g., 
 the first and third syllable of universitet ; and o, the weakest accent 
 (" tonloshet "), e. g. y the second and fourth syllables of universitet. 
 No word has more than one principal accent (4th or 3d), nor more than 
 one strong subsidiary (2d). 
 
 The groups in Tables I. and II. are arranged according to the order 
 of frequency of occurrence. We shall consider them separately under the 
 principles which they exemplify. 
 
 Principle of synthesis. The several groups of Table I., which result 
 from the different modes of combination of the elements, may be 
 studied from two points of view : frequency of occurrence and degree of 
 intensity. 
 
 The most frequent combination is o, where loudness is slightly more 
 prominent than duration and pitch. This constitutes 75.1% of all 
 the groups. The next, the loudness-pitch fusion (), where loud- 
 ness is distinctly predominant, represents 10.7% ; the loudness-duration 
 (), 4.1% ; the duration-loudness (g), 3.9% ; the pitch-loudness (), 
 2.8%; the duration -pitch (), 1.8%; and the pitch-duration (), 1.3%. 
 The two extremes consist of the o and the (pitch-duration) . Perhaps it 
 may be justifiable to regard the former as the normal, or most natural, and 
 the latter as the abnormal, or most unnatural, mode of fusion. A fact 
 scarcely less striking is that the loudness-pitch (Q ) combination is almost 
 equal in frequency to all the others, exclusive of the o centroid. In 
 respect to frequency, therefore, the element of loudness stands pre- 
 eminent. It is present as a distinctly apprehended element in all except 
 two groups of combinations. 
 
 The following percentages, as to the degree of intensity arising from 
 these fusions, obtain between the different degrees of each mode of syn- 
 thesis : 
 
 2 = 93-3 
 (3= i (3 -48.7 t 3 =43-7 (3= 6.6 
 
 3 = l8.2 (30 = 28.5 (30 = 20 
 
 'SUNDEN, Svensk Spraklara, 14, Stockholm 1895. 
 
1 8 /. E. Wallace Wallin, 
 
 1. Of all the fusions 'the o (normal) centroid alone has the highest 
 percentage in the first degree and the lowest in the third degree. All 
 the others have the smallest percentage in the first degree, with the ex- 
 ception of (pitch-intensity). 
 
 Psychologically, this signifies that the normal or most prevalent mode 
 of synthesis, though excelling in frequency, is vastly inferior in intensity. 
 When an increase in intensity is effected, some one factor gains 
 prominence. 
 
 2. The loudness-pitch () synthesis alone has the highest percentage 
 in the third degree. Hence this is the most effective mode of fusion in 
 respect of intensity. The loudness-duration synthesis (g) is a close 
 second. 
 
 3. All of the last five modes of synthesis of Table I. (loudness-dura- 
 tion, duration-loudness, pitch-loudness, duration-pitch, pitch-duration) 
 have the highest percentage of centroids in the second degree. As 
 second degree centroids, the duration-loudness and pitch-duration stand 
 preeminent. 
 
 Thus far the elements have been considered in conjunction. No one 
 element has been supreme. We now proceed to study the cases in which 
 some one element predominates. 
 
 Principle of substitution. The percentages of frequency are as follows 
 (Table II.): loudness, 45.3 percent.; duration, 42.3 per cent.; and 
 pitch, 12.3 per cent. The proportion, though not as large as for the 
 previous class, is seen to follow the same general order ; the extremes are 
 composed of the loudness and pitch centroids. In so. far as these may 
 be regarded as independent sources of intensity, loudness and duration 
 are each employed nearly four times as frequently as pitch. Loudness is 
 slightly more prevalent than duration. 
 
 The average length of the duration type was 0.28% which exceeds the 
 average length of the normal centroid by 0.09*. That is, it is 1.47 
 times longer than the average length of the ordinary emphatic syllable 
 (falling under the principle of synthesis). 
 
 As to degrees of intensity, the following relations are sustained : 
 
 2=5- 8 i 2 =36.3 
 
 3 =12.7 
 
 Observations and conclusions. A study of the two principles suggests 
 the following conclusions : 
 
 i. The element of loudness ranks highest in the degree of intensity 
 attainable through the employment of any one element as a substitutive 
 
Researches on the rhythm of speech. 1 9 
 
 instrumentality ; duration is equal to about one-half the effectiveness of 
 loudness, while pitch independently is inferior to duration. In some 
 records these elements very rarely play an approximately independent role. 
 This tends to heighten the effect, in harmony with the general principle 
 of contrast, when they are fitly employed. 
 
 2. The chief function of duration and pitch as substitutive elements 
 consists in producing a first degree centroid, or in constituting a doubt- 
 ful syllable, as compared with an adjoining syllable, a centroid syllable. 
 
 Speculation may be indulged in as to the special functions, psycholog- 
 ically, of pitch and duration emphasis, considered as vehicles for express- 
 ing different states of thought and feeling. It may be suggested that the 
 third degree duration emphasis seems specially adapted to express the 
 emotions of awe and grandeur, as well as solemn and grave thoughts. 
 Notice, for example, the frequent use that is made of the duration type 
 of emphasis in the pulpit and on the rostrum (cf. record of C.O.S., 
 below). 
 
 3. The chief function of loudness as a substitutive property, appears- 
 to consist in producing a second degree centroid; a like function seems 
 to be subserved by the duration loudness (), the pitch-duration (? ), the 
 duration-pitch () and pitch-loudness () centroids. Loudness inde- 
 pendently is equal to any one of these integrations in effectiveness of 
 intensity. 
 
 4. The thought is suggested that the function of the coalescence of 
 loudness and pitch (Q ), and probably of loudness and duration (), lies 
 in the production of a third degree, or maximum, centroid effect. It is 
 the former mode of synthesis that displays a highest percentage in the 
 third degree. 
 
 5. The function of the loudness-pitch-duration (o) mode of fusion, 
 in which loudness is slightly the most important element, consists in 
 serving as the substrate of the speech centroid ; and in constituting a 
 syllable a centroid syllable when a constituent of a sequence of words. 
 
 Assuming o to be the threshold of intensity of the centroid, all syl- 
 lables that fall below the threshold may be designated non-centroidal or 
 weak syllables. They are lesser or fractional centroid syllables. There is 
 a natural series of gradations below, as well as above, the threshold of the 
 centroid. The threshold is always relative to any given speech record ; 
 it is not a fixed quantity but is always more or less arbitrary. 
 
 6. A comparison of the two principles shows that the principle of syn- 
 thesis, in respect to frequency of employment, sustains a relation to the 
 principle of substitution, approximately as 3 : i. The integrated centroids 
 are employed about three times as frequently as the substitutive centroids. 
 
2o /. E. Wallace Wallin, 
 
 7. In respect to the third degree of intensity, the proportion between 
 the effectiveness of the mode of synthesis centroids and the mode of sub- 
 stitution centroids is as 1.54 : i. For the second degree, the proportion 
 is as 1.41 : i ; and for the first degree, as i : 2.47. In other words, the 
 second and third degree integrated centroids are each about 50 per cent, 
 more frequent than the second and third degree substitutive centroids. 
 The first degree substitutive centroids, on the other hand, are about two 
 and a half times more frequent than the integrated centroids. The cen- 
 troids consisting of integrated elements are thus seen to be far superior 
 to the centroids consisting of substitutive elements, as regards intensity. 
 
 8. To adequately account 1 for all the variety of force relations obtain- 
 ing among centroids, two principles have to be invoked : the principle 
 of substitution and the principle of synthesis. Both are important, but 
 the last is the more effectively employed, both with reference to fre- 
 quency and strength. The employment of both principles conduces to 
 pleasing variety, richness and melody, and physical and mental economy 
 in speech. A due observance of the two principles contributes no incon- 
 siderable part to the enjoyment derived from listening to a good reading. 
 
 9. Although the essence of the centroid is always a threefold com- 
 plexity, any one of the elements may become predominant, though not 
 entirely supreme. 
 
 7. Relation of the centroid to the parts of speech. A question which is 
 immediately suggested by the foregoing considerations is the relation, 
 both in respect of prevalence and degree of intensity, which the parts of 
 speech sustain to the centroid. To determine this, 336 centroids were 
 classified according to the kinds of words of which they were composed. 
 The results are tabulated in Tables III. and IV. 
 
 TABLE 
 
 III. 
 
 
 Distribution of the centroid among parts 
 
 of speech. 
 
 Kind. 
 
 n 
 
 
 Nouns, 
 
 107 
 
 0.31 
 
 Verbs, 
 
 77 
 
 .22 
 
 Adjectives, 
 
 59 
 
 17 
 
 Adverbs, 
 
 40 
 
 .1 I 
 
 Conjunctions, 
 
 21 
 
 .06 
 
 Prepositions, 
 
 16 
 
 .04 
 
 Personal pronouns, 
 
 ii 
 
 03 
 
 Proper nouns, 
 
 4 
 
 .OI 
 
 Interjections, 
 
 i 
 
 .003 
 
 n, number of occurrences. 
 
 %, percentage of 
 
 all occurrences. 
 
 1 The element of quality, or timbre, has been left out of the account. 
 
Researches on the rhythm of speech. 2 i 
 
 TABLE IV. 
 
 Relation of degrees of cent raids to parts of speech. 
 
 Kind. 1 11 III 
 
 Nouns, 0.54 0.37 009 
 
 Verbs, .54 .35 .10 
 
 Adjectives, .35 .44 .20 
 
 Adverbs, .77 .15 .07 
 
 Conjunctions, .95 .00 .04 
 
 Prepositions, I.oo .00 .00 
 
 Personal pronouns, .54 .27 .18 
 
 Proper nouns, .50 .00 .50 
 
 Interjections, .00 .00 I.oo 
 
 8. Observations and conclusions. While the determinations are largely 
 tentative, they point to a few facts which may be accepted as trustworthy. 
 
 1. The nouns and verbs rank highest, respectively, in grade of fre- 
 quency. The adjectives come third, with a little over one-half the fre- 
 quency of the nouns, and a trifle over three- fourths the frequency of the 
 verbs. The interjections and the proper nouns rank the lowest. 
 
 2. Excluding the interjections, only one case of which was studied, 
 all the different kinds of words have their highest percentage in the first 
 degree, except the adjectives and the proper nouns. The latter are evenly 
 distributed among the first and third degree centroids. The former have 
 the largest percentage in the second degree, and also rank next to the 
 highest of any part of speech in the third degree column. As between 
 nouns and verbs, a parallel proportion obtains. Over half the number 
 are first degree, a little over one-third are second degree, and only one- 
 tenth are third degree centroids. The personal pronouns follow the same 
 general course except that the percentage in the third degree is relatively 
 about twice as high. In the adverbs, on the other hand, the gradation 
 is very rapid from the 1 to the 11 ; the percentage of III is about one- 
 half of 11 adverbs. The conjunctions and prepositions are almost wholly 
 limetd to the first degree. 
 
 In the main, then, it seems that, as regards the intensive or stress as- 
 pect of words, the parts of speech rank in the following order : proper 
 nouns, adjectives, personal pronouns, verbs, nouns, adverbs, conjunctions 
 and prepositions. The proper nouns, however, may be considered doubt- 
 ful, as only four cases were studied. 
 
 The subsumption according to degree of stress differs, it will be seen, 
 from the subsumption according to the frequency of occurrence. 
 
 It is interesting to compare these results with the rules of accentuation 
 propounded by GUEST/ according to which articles are rarely accented ; 
 
 1 GUEST, A History of English Rhythms, 85, London 1882. 
 
22 /. E. Wallace Wallin, 
 
 qualifying words for example, adjectives, adverbs, and possessive pro- 
 nouns receive a fainter accent than the words they qualify ; personal and 
 relative pronouns take a fainter accent than the verbs ; and in words com- 
 pounded of a substantive and adjective, the accent falls on the adjective. 
 This theory is somewhat at variance with the above determinations. 
 
 SUNDEN * holds that articles in Swedish are never accented. Pronouns, 
 auxiliary verbs, prepositions and conjunctions are generally not accented. 
 
 It is quite apparent upon the surface, that, with reference to any part 
 of speech, no rules can be laid down that admit of no exceptions. The 
 same noun, adjective, verb, adverb, pronoun, etc., may at one time be 
 a first, at another a second, and at yet another a third degree centroid, 
 depending upon the relative importance which the word plays as a vehicle 
 for the conveyance of the thought that is uppermost in the mind of the 
 speaker on any given occasion. 
 
 9. Recapitulation. The discussion of the preceding pages has been con- 
 fined to a treatment of the centroid as an isolated element in speech 
 that is, a definition of the concept, the kinds, the elements, degrees and 
 essence (as illustrated in the principles of integration and substitution), 
 and the word-analysis of the centroid, have been given. The next step 
 consists in studying the centroid as a constituent member in a sequence 
 of centroids. This phase of our enquiry diverges in two directions. 
 On the one hand, we have to consider the relations which centroidal and 
 non-centroidal syllables sustain to one another as respects duration, 
 whether that of equivalence, or of simple or complex proportion. This 
 problem concerns the question of the ''quantitative character of verse." 
 On the other hand, we have to consider centroids as recurrences, and 
 determine whether the time intervals between them are equal. This 
 problem deals with the question of the succession of the centroids. 
 
 Both of the above problems are phenomena of time. Much confusion 
 has obtained with reference to their relations, in writings on the "time- 
 theory of verse." The signification of the term "quantity" has cus- 
 tomarily been extended so as to become coextensive with all the time ele- 
 ments of rhythmical phenomena succession, recurrence, duration, etc. 
 " Quantity," however, is a phenomenon of duration ; succession, a phe- 
 nomenon of intermittent recurrence. The two are separate and distinct, 
 and should not be made interchangeable terms in speech. Duration is 
 independent of repetition ; repetition involves duration. Duration, as 
 synonymous with quantity in speech, is an uninterrupted span ; succes- 
 sion, a series of long or short spans. 
 
 1 SUNDEN, Svensk Spraklara, 15, Stockholm 1895. 
 
Researches on the rhythm of speech. 23 
 
 The specific task of the first enquiry is to measure the time occu- 1 
 pancy of elementary sounds, centroidal and non-centroidal, and the \ 
 gaps occurring between them (not pauses). The function of the latter, 
 starting with this given manifold, is to determine the exactness of the 
 coordinations of the intervals which occur between the momenta of 
 rhythmical wholes ; and to disclose the principles governing the rhythm 
 of speech. 
 
 IV. THE MEASUREMENT OF SEPARATE SOUNDS AND SILENCES. 
 
 i. Explanation. Speech, structually considered, is a complex synthe- 
 sis of a number of formal unities. The following may be distinguished : 
 ( i ) the unity of the poem or discourse, as a whole, governed by the 
 principle of the unifying actus of the theme ; (2) the unities of the sec- 
 tions and subdivisions, governed by the principle of the logical subordi- 
 nation of the parts to the whole ; (3) the unities the unities of speech 
 par-excellence of the particular groups of words or syllables which are 
 separated from other groups by pauses, governed by the principle of the 
 pause punctuation-mark or non-punctuation-mark pause and the expi- 
 ration interval; (4) the unities of poetical verses, governed by the prin- 
 ciple of the terminal pause; and (5) the more ultimate unities, consist- 
 ing of separate sounds or syllables, governed by the principle of the 
 silences or glides which separate them, as dependent upon the action of 
 the vocal organs. 
 
 The first two classes of unitary groups are easily recognizable upon the 
 printed page ; in speech they are logically distinguishable. According to 
 SWEET/ the only division actually made in language is that into " breath 
 groups," corresponding to (3) above. 
 
 No account of the first and second kinds of unity has been taken in 
 these experiments. The third (and the second in so far as it coincides 
 with it) and the fourth receive separate treatment elsewhere. The fifth 
 unity is the subject of this section. 
 
 The denial 2 of the existence of the unities of separate sounds and 
 silences is founded upon a natural illusion. The unity of a grain of 
 sand is for unaided discriminative visibility the grain as a whole ; for 
 science it is the particles composing the grain. For the eye it is one 
 thing ; for the microscope another. The only unitary sound groups 
 in speech for unaided discriminative audibility may be the " breath 
 groups." When the reproducing speed of the phonograph cylinder, 
 
 1 SWEET, Primer of Phonetics, 42, Oxford 1890. 
 
 2 SWEET, as before. 
 
24 /. E. Wallace Wallin, 
 
 however, is sufficiently reduced these groups are split up into separate 
 syllable groups. 
 
 The method by which separate sound unities are perceived is thus 
 analogous to those used in perceiving ultimate sight unities, namely, to 
 augment the quantity of the sensations by instrumentation. 
 
 The speed of the phonograph cylinder was reduced until a group of 
 rapidly recurrent sounds, which had previously appeared as a homoge- 
 neity, was split up into discreet elements of sound separated by gaps. 
 That is, psychologically these sounds and silences corresponded to a 
 series of auditory sensations alternating with gaps or silences. It is the 
 time of these several sensations, as well as the time of the intervals of 
 silence between them, that was measured. Hence the terms separate 
 sounds and silences are to be psychologically interpreted. 
 
 Physically, the sounds represent vibrations of given lengths, amplitudes 
 and periods. The gaps or silences probably correspond in some cases 
 to very weak vibrations (glides). Not infrequently, however, they are 
 physically interpretable as absence of vibrations. 
 
 Physiologically, the sounds and silences are interpretable as changes in 
 the condition of the vocal organs. 
 
 Owing to the degree of skill demanded by the difficulty of the work, 
 the measuring was deferred to the later stages of the research. The 
 words sometimes became so indistinct or transitory as to preclude meas- 
 urement ; but the instances were few, for the sounds, in reducing the rate 
 of rotation of the cylinder, tended to become inarticulate before they be- 
 came inaudible. A tone in order to be audible must consist of not less 
 than approximately 16 vibrations per second. Theoretically, therefore, 
 by slowing the speed sufficiently the lower pitches of the tones of the 
 record would fall, as it were, below the threshold of audibility and be- 
 come silent, before the higher pitches. The answer to this objection is 
 that the degree of reduction was determined, not by the threshold of 
 audibility, but by the threshold of articu lability, which is higher than 
 the former. Hence the sound which first became inarticulate determined 
 the particular limit of the reduction. 
 
 The silences are indicated by dashes ( ). The numeral below the 
 dash indicates the duration of the silence in units of contacts ; numerals 
 above the syllables indicate the duration of the syllable to which they 
 pertain. The unit of measurement in all the records is one contact, 
 the value of which is ^, ^-, -fa or ^ of a second, according as the 
 recording speed was 80, 92, 100 or 120 revolutions per minute. 
 
 2. Records. 
 
 Subject, A. R. P. (Palmer, 1901.) 
 
Researches on the rhythm of speech. 25 
 
 o 15 ' o 15 o ii o 25 
 3525234 9 
 My bed is like a little boat ; 
 4243 5 15 
 
 0150 18 o 14 o 19 o 48 
 5 5 86396 ii 
 
 Nurse puts me in and I em bark ; 
 854 3 33 
 
 22 O 23 O 19 O 29 
 2 7666848 
 
 She girds me in my sail ors coat 
 ? 634625 17 
 
 21 Q 14 
 
 And starts me in the dark. 
 ? ? ? ? ? 
 
 (Robert L. Stevenson.) 
 Unit of measurement, -fa of a second. 
 
 Explanation and characterization of record. The subject, an admirer 
 of poetry, was told to declaim so as to make the record express his 
 highest conception of good poetical declamation. No insight was given 
 into the nature of the problem to be investigated. The stanza was re- 
 cited from memory. 
 
 To the ear the record possesses the qualifications of good verse ; the 
 rhythm is markedly smooth, the modulation pleasing, and the centroids 
 are emphasized with due discrimination. 
 
 Subject, W. C. (Churchill, graduate student.) 
 
 20 o 28 -o 9 o 17 o 23 
 5449 4 6 ii 5 
 
 Oh yet we trust that some how good 
 24776 359 
 
 019 o 1 8 o 21 040 
 
 6 4 3 6 4 13 3 J 4 , 
 
 Will be the fi nal goal of ill,4- 
 33627 2 3 h 
 
 23 o 27 o 24 o 45 
 
 3 7,3 68 7 f 4i2 r 
 To pangs-j of nature, sins of will,-} 
 
 2 15 J 3 ' 8 4 24 
 
 23 46 22 
 
 4 7 : 4 12 , 3 8 r 3 *o , 
 De fects of doubt, \ and taints of blood.; 
 
 2 75 126 6 \g 3 
 
 (Tennyson.) 
 
26 J. E. Wallace Wallin, 
 
 Unit of measurement, 1/40 of a second. 
 
 Explanation and characterization of record. The subject was given a 
 printed copy of the stanza, and, after a reading, was asked to read it into 
 the phonograph as poetry. The record sounds much like the previous, 
 save that the rhythm is not so regular. No peculiarity is manifest 
 in the relation of the elements of the centroid. The delivery sounds 
 natural ; the rate was rather fast. 
 
 Subject, J. M. T. (Telleen, graduate student.) 
 
 7 4 X 3 6 14 5 10 8 
 Take the joys and bear the sorrows. 
 
 4 3 J 6 7 53 
 3 7 ii 6 6 4 
 
 Shut those eyes, next life will 
 
 5 6 14 6 6 ? 
 
 7 5 ii 12 7 8 
 
 open, stop those ears, next life 
 ? 7 4 33 7 6 
 
 5 7 68 85 
 
 will teach hearing' s office. 
 
 5 J 3 5 
 
 (Browning.) 
 
 Unit of measurement, 1/48 of a second. 
 
 Explanation and characterization of record. The same method of tak- 
 ing the record was used as in the W. C. record. These two sentences 
 are selected from the complete stanza, the first because of its regularity 
 of structure, the second because of its consecutive centroids. 
 
 The rhythm sounds less regular than in the previous record. The syl- 
 lables are distinctly uttered ; the centroids receive a good share of duration 
 emphasis. The movement is rather slower than the subject's ordinary 
 speed of talking. 
 
 Subject ', E. H. T. (Tuttle, 1901, student of phonetics. ) 
 
 Perfect scansion. 
 (#) Without regulative concomitant. 
 
 o 24 o 31 o 20 o 33 
 
 456 12 6 10 3 ii 
 
 And so no force, how ev er great, 
 ? 4 6 17 ? 3 3 ii 
 
 O 20 O 28 O 22 O 27 
 
 3 9 2 *3 4 M 3 8 
 
 Can strain a cord, how e v er fine, 
 
 54 IJ ? I.. 4 10 
 
Researches on the rhythm of speech. 2 7 
 
 o 15 o 18 o 21 o 34 
 2738 68 5 22 
 In to a ho rizon tal line, 
 244 62 5 
 
 O 22 O 21 O 24 O 
 
 3 746374 ii 
 That shall be ab so lute ly straight. 
 
 ? 738356 
 
 With regulative concomitant. 
 
 o 23 o 25 o 23 o 28 
 
 ? 8 4 10 6 10 5 8 
 
 And so no force, how ev er great, 
 55 7 ? 7 12 
 
 O 23 O 24 O 24 23 
 
 4 ii 4 ii 4 8 3 13 
 Can strain a cord, how ev er fine, 
 
 6 7 479 
 
 O 21 O 24 G 23 O 25 
 
 2 5 5 I2 5 9 ir . ? 
 In to a ho ri zon tal line, 
 
 ? 5 5 4 3 3 I0 
 
 O 22 O 22 O 24 O 
 
 That shall be ab so lute ly straight. 
 
 Unit of measurement, 1/48 of a second. 
 
 Explanation and characterization of records. The subject was informed 
 that the distinctive purpose of the experiment was to determine how ac- 
 curately the intervals between the emphatic syllables could be coordinated. 
 He was told to scan the selection with the utmost regularity, first, without 
 and, second, with the aid of the finger beating time upon the table. The 
 delivery was rather slow. A second purpose was to determine the rela- 
 tion which long and short syllables would assume. Of the latter purpose, 
 no intimations were given. 
 
 To the ear, the first specimen sounds like an excellent example of rou- 
 tine scansion ; the second sounds still more artificial and mechanical. 
 
 The two records differ not so much in the relation of the component 
 syllables and silences as in the regularity of the intervals ; hence only the 
 measurements of the syllables and silences of the former, which is free 
 from mechanical regulation, are given in the table of results. 
 
28 /. E. Wallace Wallin, 
 
 Subject, W. L. P. (Phelps, Prof. English Literature.) 
 
 o o o 
 
 3 7 10 9 8 4 12 
 
 But one mor ning the min i ster, 
 
 3 4 ? 6 ? 4 n 
 
 c o o o 
 
 6378 79 
 
 in the chapel, gave out 
 
 5 ? 34 3 5 
 
 GO O O 
 
 6 21 10 7 8 3 ? 
 the hymn num her nine ty nine 
 
 3 35 ? 5 6 2 
 
 Unit of measurement, 1/48 of a second. 
 
 Explanation and characterization of record. The sentence is taken 
 from the body of a prose record. The phrase rhythm is fairly-well pre- 
 served ; the accents are normal, only one extra-long centroid occurring 
 in the sentence. The rendition is distinct and emphatic. 
 
 (Swedish poetry. ) 
 Subject ', O. S. (Sandquist, age 13.) 
 
 16 22 19 25 
 
 o o o o 
 
 5 10 4 8 3 ii 5 5 
 Ack, Herre Jes us hor min rost, 
 
 4 ? 4 5 3 5 9 
 
 o 17 o 180 17 032 
 594837? 9 
 
 Gor dig ett tern pel i mitt brost, 
 332422 4 17 
 
 \ 
 
 o 18 0180 18 o 19 
 376 959 5 I0 
 U ti mitt hjar ta blif och bo, 
 21 3 2 2 ? 3 9 
 
 Ol8o 22 O 15 O 42 
 
 Sa har jag trost och evig ro. 
 
 Unit of measurement, 1/32 of a second. 
 
 Explanation and characterization of the record. The subject was 
 chosen from a number of speakers who appeared on a children's pro- 
 gram, because of the monotone fashion of speaking. The verses were 
 recited from memory. They are end-stopped, with three sectional punc- 
 
.Researches on the rhythm of speech. 29 
 
 tuation marks. The sounds and silences could not be accurately measured 
 in the last line. There was a general tendency to let the voice fall in 
 pitch and weaken toward the close. 
 
 To the ear the stanza has a rhythmic swing characteristic of the jog- 
 trot method of declaiming of many children. The degree of emphasis is 
 moderate, considered especially duration-wise. 
 
 {Persian poetry. ) 
 Subject, K. H. K. (Kazanjian, graduate student, philosophy.) 
 
 3 3 2 23 15 A 6 9 
 1156 7 9 57 14 9 78 
 Ahi min el askh i ve ha la ti hi 
 55 6 ? 4 i 6 ? 25 
 
 16 o 31 2 27 74 
 
 7 3 I0 4 7 7 12 65 5 
 Eh rak i kal bi be hera re ti hi 
 4 25 46? 742 32 
 
 3 3 1 A 2 22 O, 21 | 3 71 
 
 ii 8 6 8 10 5 7 7 ii 6 9 
 
 Mo naz er el ayn e i le ghayri kum 
 443424 5 5 22 
 
 | 3 3 2 2 2 9 O 20 O 
 
 6 6 4 711 7 5 14 845 
 Euk sum i bil la hi ve I yatihi. 
 7 34632 ? 
 
 Unit of measurement, 4 ^ of a second. 
 
 Explanation and characterization of record. The primary object in 
 studying the record was to determine whether or not the accent in Per- 
 sian poetry is entirely lacking, as alleged, or, if present, what is its 
 essence ; and the character of the rhythm. The reciter, a native of Persia, 
 was told to exercise care lest the rendition might be affected by our occi- 
 dental mode of utterance. The verses were recited from memory. 
 
 The movement is fairly rhythmical ; the melody strikes the ear as 
 weird and unfamiliar. On careful listening the centroids were evident to 
 the ear. 
 
 3. Sounds. a. Historical summary. Two chief views have been ad- 
 vanced as to the temporal values of emphatic and unemphatic syllables. 
 According to the one, the relation sustained is that of equality. There 
 is no necessary bond between emphasis and duration. As a matter of 
 
30 /. E. Wallace Wallin, 
 
 contingency or as dependent upon the difficulty of the physiological proc- 
 ess of articulation, either syllable may arrogate to itself supremacy of 
 time value. 
 
 The second theory, which has been held in three forms, posits a rela- 
 tion of long and short between the emphatic and unemphatic syllables. 
 According to the first form the relation is indefinite and variable. In 
 this form the theory is held by GuRNEY, 1 GOOLD, BROWN, LoxzE, 1 BOL- 
 TON 2 and a host of others. For the most part, the relation, while fairly 
 constant, is held to be inexact. LANIER 3 may be taken as a representa- 
 tive of the second form, which maintains that the syllables of speech 
 sustain a proportion of aliquot parts to one another. The proportion is 
 always a simple one, such as i, 2, 3, 4, 5, etc., and never fractional. 
 According to the third form, 4 the ratio sustained is precisely as 1:2. The 
 theory is generally applied to classic verse, which theoretically admits 
 but two time values, long and short, as obtaining among its sounds. 5 
 
 b. The measurement of emphatic and unemphatic syllables. The rel- 
 ative amount of time-occupancy of weak and strong syllables is shown 
 in Tables V. and VI. 
 
 TABLE V. 
 
 Duration of emphatic syllables. 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 I 
 a 
 
 n 
 
 A. R. P. 
 
 2.17 
 
 o. 16 
 
 0.03 
 
 0.18 
 
 13 
 
 W. C. 
 
 3-50 
 
 0.20 
 
 0.05 
 
 0.25 
 
 17 
 
 J. M. T. 
 
 3-o8 
 
 0.17 
 
 0.04 
 
 0.23 
 
 18 
 
 E. H. T. 
 
 3-27 
 
 O.2O 
 
 0.05 
 
 0.25 
 
 16 
 
 W. L. P. 
 
 2. 2O 
 
 0.20 
 
 0.05 
 
 0.25 
 
 ii 
 
 0. S. 
 
 3.06 
 
 0.25 
 
 0.03 
 
 0.12 
 
 12 
 
 K. H. K. 
 
 2-95 
 
 0.18 
 
 0.03 
 
 o.i 6 
 
 16 
 
 Av. 2.89 0.19 0.04 0.20 14.7 
 
 Unit of measurement, I s . d, total duration; a, average duration; /, immediate 
 probable error ; , relative immediate probable error ; , number. 
 
 1 GURNEY, The Power of Sound, 443, London 1880. 
 2 BOLTON, Rhythm, Am. Jour. Psych., 1893 VI 34. 
 
 3 LANIER, The Science of English Verse, 73, New York 1880. 
 
 4 POE, The Rationale of Verse, Works, VI 60, Chicago 1895. Elsewhere he main- 
 tains a contradictory position, 52. 
 
 5 LANIER, The Science of English Verse, 113 New York, 1880. 
 ELLIS, The Quantitative Pronunciation of Latin, 8-20, London 1874. 
 
Researches on the rhythm of speech. 3 1 
 
 TABLE VI. 
 Duration of unemphatic syllables. 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 n 
 
 A. R. P. 
 
 i 15 
 
 O.IO 
 
 0.03 
 
 0.30 
 
 II 
 
 W. C. 
 
 i-5 2 
 
 o. 10 
 
 0.03 
 
 0.30 
 
 15 
 
 J. M. T. 
 
 o-93 
 
 O.I I 
 
 O.O2 
 
 0.18 
 
 8 
 
 E. H. T. 
 
 1.27 
 
 0.08 
 
 0.02 
 
 0.25 
 
 16 
 
 W. L. P. 
 
 1-3 1 
 
 o 13 
 
 O.O4 
 
 0.30 
 
 10 
 
 0. S. 
 
 1.56 
 
 O. 12 
 
 O.O2 
 
 0.16 
 
 12 
 
 K. H. K. 
 
 3-85 
 
 O.I 3 
 
 O.O2 
 
 0.15 
 
 28 
 
 Av. 
 
 i76 5 
 
 0. II 
 
 0.025 
 
 0.23 
 
 15-7 
 
 t of measurement 
 
 , IS ; 
 
 d, total duration ; 
 
 a, average 
 
 duration ; p, 
 
 immediate 
 
 able error; , relative immediate probable error 
 
 (a) Observations and conclusions. The tables of measurements point 
 towards the following conclusions. 
 
 1. The absolute duration of any syllable is variable. The records re- 
 veal that the length of a given long or short syllable is never absolutely 
 fixed : the precise length is different for every time it is spoken. 
 
 2. The average emphatic syllable is invariably longer than the average 
 unemphatic syllable. The average of the former for all the records occu- 
 pies 0.08 sec. more time than the average of the latter. The ratio is as 
 1.7 : i j hence the average long syllable is almost one and three-fourths 
 times as long as the average short syllable. 
 
 3. The ratio sustained between the average unemphatic and emphatic 
 syllable varies with different subjects. The following are the ratios for 
 each of the records, given in the order of the tables : i : i . 6 ; 1:2; 
 1:1.5; I ' 2 -S ') I:I -5j 1:2. 08 and i : 1.38. With the exception of 
 one record, all the average emphatic syllables are at least half again as 
 long as the average unemphatic. The exception is the Persian ; probably 
 the element of duration is a property less prominent in the centroids of 
 Persian poetry. 
 
 The highest ratio is furnished by the routine scansion record. Here 
 the centroid syllable is two and a half times longer than the non-centroidal 
 syllable. Scanned poetry thus appears to furnish the instance where the 
 difference between the length of emphatic and unemphatic syllables is at 
 its greatest. This suggests a conclusion with reference to the method in 
 which classical Roman and Greek poetry was scanned. 
 
 The ratio in the prose record (W. L. P.) is slightly less than the aver- 
 age. Whether this is a valid distinction, as affecting prose and poetry, 
 cannot, however, be maintained on the basis of this one record. 
 
32 /. E. Wallace Wallin, 
 
 4. The ratios existing between individual emphatic or unemphatic 
 syllables of the same or of different records, are neither invariable, nor are 
 they ratios of simple proportion. 1 The possible relations, as may be seen 
 by a glance at the records, and by an examination of the/ column (the 
 column of irregularity) are practically unlimited. Not infrequently the 
 ratio is reversed, so that sporadic unemphatic syllables may be longer 
 than the average of the emphatic syllables. 
 
 These facts incontestably refute the theory of "simple proportion." 
 As an illustration, notice the following time relations in the consecutive 
 syllables of the first line of the specimen of routine scansion, which, 
 rhythmically considered, furnishes a type of poetry at its best : 0.083, 
 0.104, 0.125, 0.250, 0.250, 0.208, 0.062 and 0.229. 
 
 5. The long and short syllables are about equal in regularity. As 
 respects the comparative amount of absolute irregularity, the average 
 long syllable is one and six-tenths times more irregular than the short. 
 The average amount of irregularity for each long syllable is -^-^ ; for 
 each short syllable, -%\-$. In proportion to the length of the span, 
 however, the short syllable exceeds the long in the amount of irregularity 
 by a trifle, namely, -%\-*. 
 
 6. The ratio between the number of emphatic and unemphatic sylla- 
 bles (exclusive of the Persian record) is as 1.2 : i, which is smaller than 
 the ratio between the total duration of the same. In none of the records 
 are there less emphatic than unemphatic syllables, except in the Persian, 
 where the numerical ratio is i : 1.75, and the duration ratio (the relative 
 time-occupancy of all the emphatic and unemphatic syllables) as i . 33 : i . 
 The unemphatic syllables occur, as might be expected from the non- 
 accentual theory of Persian poetry, one and three -fourths times as often 
 as the emphatic. In English, however, the converse often occurs, as may 
 be seen in the J. M. T. record. On the whole, the ratios between the 
 duration and the number of long and short syllables do not coincide. 
 The ratio between the latter may be equal ; the ratio between the former 
 is never that of equality for the total average duration of any record. 
 
 (<) Bearing on the quantitative theory. The most general deduction 
 to be made from this series of measurements is the disparity in the time- 
 span of so-called short and long syllables. It has been shown that the 
 temporal distinction is true to fact. 
 
 All speech is quantitative ; and the distinction, popularly and con- 
 fidently posited, between quantitative and non -quantitative verse, is 
 grounded upon fallacious assumptions. The question of the quantitative 
 character of poetry or prose, is closed. The only crux of contention that 
 
 1 Cf. LANIER, The Science of English Verse, 73, New York 1880. 
 
Researches on the rhythm of speech. 33 
 
 remains affects the character of the laws governing the temporal relation 
 of the components of language. The question contains several phases. 
 
 The first determination has settled the fact that speech as a phenomenal 
 occurrence is conditioned on the time-intuition of sensibility. 
 
 The second question has also been settled, namely that the strong 
 syllable is temporally longer than the weak. 
 
 The third question is partly closed. It has been shown that for Eng- 
 lish, Swedish and Persian the ratio between the individual long and weak 
 syllables is never constant. The general character of the ratio was, for 
 artistically free declamation of English poetry, in the proportion of i : i. 7 ; 
 i : 1.38 for Persian poetry ; i : 2.08 for Swedish poetry ; i : 1.5 for Eng- 
 lish prose; and i : 2.5 for mechanical scansion. 
 
 It has generally been supposed that the ratio in classical verse is a fixed 
 one, in the proportion of i to 2. Supposing the theory true, what justi- 
 fication is there for setting classical verse apart as sui generis ? 
 
 Evidently the distinction between so-called quantitative and non-quanti- 
 tative poetry cannot be validated on the mere ground of proportionality, 
 for our records have furnished examples of English and Swedish poetry in 
 which the proportion is, respectively, as i : 2, and as i : 2.08 ; hence, if 
 an examination of modern verse furnishes no support for the theory, on 
 the basis of the amount of proportion (since it duplicates what the theory 
 assumes as unique), the remaining alternative is to postulate a strict 
 equality between each and every strong syllable and each and every weak 
 syllable ; that is to say, to suppose that the Roman and Greek ear was 
 capable of exactly coordinating every individual emphatic and unemphatic 
 syllable so that, relative to one another, they should sustain the propor- 
 tion of i : 2, and, regarded apart by themselves, a proportion invariably 
 as 2 : 2 in the former, and i : i in the latter case. That such a proportion 
 cannot be effected by the modern ear is indisputable. Even in the case of 
 the strictest metronome scansion, the average inequality is 0.02" for every 
 weak, and 0.05' for every strong syllable. When the modern ear is thus 
 incapable of preserving the strict temporal equality of syllables, even in 
 mechanical scansion, the quantitative theory as limited to classical poetry 
 may, in regard both to extent and invariability of proportion, be con- 
 sidered exploded. 
 
 A third supposition is still possible, namely, that recited Greek and 
 Latin verse, not improbably, was unique in the musical character of the 
 utterance. 1 According to SYMONDS,* the poetry probably was sung or 
 
 1 ELLIS, The Quantitative Pronunciation of Latin, 20, London 1874. 
 
 2 SYMONDS, Sketches and Studies in Southern Europe, II 326, New York 1880. 
 
34 / E. Wallace Walhn, 
 
 chanted ; hence the tempo would have been an adagio, supposing ours to 
 be allegro. This would naturally intensify the appreciation of the dis- 
 parity in the durations of the two syllables, since more time would be 
 afforded to the ear to listen to each syllable, and the quantity of sensation 
 would be augmented. Hence the illusion would arise that the difference 
 between classic and modern scansion of poetry lay in the relation of the 
 weak and strong syllables, in respect of exact regularity and proportion, 
 instead of in the relatively greater amount of time occupied by the syllables 
 in classical scansion as compared with the non-classical, or reading, scan- 
 sion of modern poetry. 
 
 The psychological explanation here advanced of the distinction between 
 quantitative and non-quantitative poetry, as arising from a natural audi- 
 tory illusion, seems quite compatible with the general psychological theory 
 of illusions occurring in normal life. 
 
 (V) Theoretical. How can we best account for the differences of du- 
 ration existing between syllables ? 
 
 The most prevalent theory may be called the physiological theory, 
 which resolves the question into facility or difficulty of enunciation into 
 " lingual evolution." Long syllables are those that are encumbered- by 
 consonants, 1 or those that contain long vowels 2 ; and short syllables are 
 those unencumbered by consonants, or those containing short vowels. 
 Hence the conclusion that there is no quantity 3 in English. 
 
 This theory undoubtedly explains a large percentage of cases. Some 
 syllables, by reason of the vowels and consonants composing them, re- 
 quire a labored adjustment of the vocal apparatus, and hence a retarded 
 utterance. 
 
 On the other hand, as was proved by the experiments of KRAL and 
 MARES/ the addition of consonants does not proportionately or regularly 
 lengthen the syllable ; it often shortens it. Moreover, the theory applies 
 equally well to both weak and strong syllables, and is hopelessly inade- 
 quate to explain the large array of instances where syllables are normally or 
 regularly long, although easy of articulation. It fails to render a satis- 
 factory account of the fact that the average emphatic syllable is invariably 
 longer than the unemphatic. Only the psychological theory can satisfac- 
 torily account for this circumstance. 
 
 1 FOE, The Rationale of Verse. Works, VI 59, Chicago 1895. 
 
 2 GUEST, A History of English Rhythms, 103, London 1882. 
 
 3 GUEST, as before, v. 
 
 *KRAL A MARES, Trvdnihlasek a slabik die objektivne miry, Listy Filologicke, 1893 
 IV 17. 
 
Researches on the rhythm of speech. 35 
 
 According to GURNEY 1 the long syllable occupies more time than the 
 short, because the amount of effort required for its pronunciation is greater 
 than that required for the short syllable. Another theory, partly physio- 
 logical, partly psychological, attributes the prolongation to the fact that 
 when a special intensification is made it is accompanied by a retardation 
 of the action of the voice apparatus. It requires a longer period to over- 
 come the contraction ; hence the sound is prolonged. Against this sup- 
 position it may be urged that the time required to effect the contraction 
 is proportionately shorter ; hence a neutrality of effect. 
 
 The most fruitful theory, we believe, is the centroid theory advanced 
 above. The centroid is a mental center of force, of which duration is an. 
 inseparable element. It is constituted a centroid by reason of the promi- 
 nence, proportionate or unproportionate, which is given by the speaker to- 
 fts properties. It is as natural to intensify the duration factor, as the pitch 
 and loudness factors. 
 
 4. Silences. a. Explanation. By silences is here meant those gaps 
 which in listening to a slowly revolving phonograph cylinder, are per- 
 ceived to exist between the successive syllables of phrase groups. The 
 gaps between phrase groups, or expiration intervals, may be called, in 
 contradistinction to the former, vacant intervals or pauses. It is only the 
 former with which we are now concerned. 
 
 The popular idea, says SWEET, 2 that we make a gap between every syl- 
 lable is erroneous ; the only phonetic divisions justifiable in speech are 
 " those into sounds and syllables and intervening glides." It is patent, 
 however, that the speech organs must consume a fraction of time in the 
 transition from one sound to another, which will be long or short accord- 
 ing to the inherent difficulty of the transition, and the amount of energy ex- 
 pended. A series of such transitions denotes, physiologically, a number 
 of points at which the energy utilized by the organs of speech ( diaphragm, 
 lungs, larynx) may be considered to be at its lowest ebb ; and, physically,, 
 points at which the vibrations of the air are diminished in amplitude and 
 frequency. When the diminution reaches a certain point the transitions, 
 become, psychologically, gaps or silences. The perception of these, how- 
 ever, is a distinct problem. According as the mind is intensely absorbed 
 in the thought element of the utterance, the gaps fail to rise above the 
 threshold of consciousness, and hence it is natural to suppose that an ex- 
 piration group of syllables is psychologically a continuum. The illusion 
 is exposed when attention is directed to the purely structural aspect of 
 
 1 GURNEY, The Power of Sound, 429, London 1880. 
 
 2 SWEET, Primer of Phonetics, 42, Oxford 1890. 
 
36 /. E. Wallace Wallin, 
 
 speech. The rapidity of ordinary utterance is so great as to render the 
 elements of language too subtle for the processes of work-a-day analysis, 
 and recourse must be taken to some means of aiding the power of mental 
 analysis, namely, to magnifying the phenomena; this can be done by 
 listening to a phonograph record, the rate of the rotation of the cylinder 
 of which has been sufficiently reduced. 
 
 In a few instances it was impossible, in measuring, to detect any gaps 
 between the syllables : they seemed to coalesce. In yet other instances, 
 the gaps could be perceived, but were so indistinct as to make it hazard- 
 ous to measure them. These instances are indicated in the records by a ? 
 
 In the table which follows a comparison is made between ante-, inter- 
 and post-centroid silences. Only the average duration is given. 
 
 TABLE VII. 
 
 Duration of silences. 
 
 Subject. .(. _|_ 
 
 A. R. P. 0.103 0.200 0.103 
 
 W. C. 0.108 o.ioo 0.136 
 
 J. M. T. 0.095 0-139 0.104 
 
 E. H. T. 0.080 0.104 
 
 W. L. P. 0.079 0.072 o.in 
 
 O. S. 0.099 0.084 
 
 K. H. K. 0.086 0.078 
 
 Av. 0.092 0.127 o. 102 
 
 Unit of measurement, I s . 
 
 , silence preceding centroid. 
 |- , silence between centroids. 
 -)-, silence following centroid. 
 
 b. Deductions. Owing to the limited number of silences upon which 
 the determinations are based, the following are offered as tentative con- 
 clusions. Later researches may verify or modify them. 
 
 i. The silences that occur between centroids rank highest in the order 
 of length. 
 
 It has been denied by GUEST 1 and affirmed by SWEET and MAYOR, 2 
 that two centroids can follow in an immediate sequence. The silence 
 occurring between the contiguous centroids has been constituted the indis- 
 pensable condition of such a series by some writers. GUEST, e. g., 
 affirms (p. 86) that the condition of the immediate sequence of two 
 accented monosyllables, is the intervention of a " pause. ' ' When the 
 words are in groups, the law affects only the group. 
 
 1 GUEST, A History of English Rhythms, 551, London 1882. 
 
 2 MAYOR, Chapters on English Metre, 29 ff., London 1886. 
 
UNIVERSITY 
 
 OF 
 
 Researches on the rhythm <?/" speech. 37 
 
 The inter-centroid silence has been conceived in the nature of a pause. 
 This is untenable. These silences are essentially the same as those which 
 precede or follow centroids. The difference in duration between the 
 averages is almost negligible 0.02 5 s . 
 
 2. The post-centroid occupies more time than the ante-centroid silence. 
 The converse has been affirmed to be true by writers in phonetics. 
 
 3. The average duration of the silences of speech is about o. 107". 
 
 5. Comparison of the duration of sounds and silences. (#) Observa- 
 tions. The aggregate duration of all the sounds is equal to 31. 82 s ; of 
 all the silences, 13. 35 s . The ratio of quantity of silence to quantity 
 of sound is thus in the proportion of 1:2.38. 
 
 This fact may contain material of importance of a psychological and 
 physiological nature. It has been affirmed that the heart, for example, is 
 alternately working and resting, resting between, working during, the 
 pulsations. Similarly the alternation of sounds and silences suggest a 
 constant alternation of work and rest on the part of the larynx. 
 
 There is this difference between the two cases. The period of work in 
 cardiac action is about equal to, or slightly shorter than, the period of 
 rest. In the action of the vocal cords, the period of work is nearly two 
 and a half times longer than the period of rest. Obviously, this explains 
 why the former mechanism is able to continue at work incessantly, while 
 the latter requires long periods of rest for recuperating its energy. These 
 periods are supplied partly by pauses, and partly by limiting the activity 
 to a certain length of time, varying with each individual. 
 
 The above ratio is exclusive of pauses. In the later comparison of the 
 various intervals of speech, the duration of the expiration intervals will 
 be compared with the duration of the pause intervals. 
 
 None of the general averages of the silences is equal to the average 
 duration of the centroid syllables. The average inter-centroid silence, 
 however, is a trifle longer than the average short syllable. 
 
 (<) Conclusions i. The quantity of the sounds of expiration inter- 
 vals of speech is nearly 2^ times greater than the quantity of the silences. 
 
 2. The average emphatic and unemphatic syllables of speech are both 
 uniformly longer than the average silences. 
 
 V. THE MEASUREMENT OF INTERVALS IN SPEECH. 
 
 A. Explanatory. 
 
 No little confusion has obtained in the meaning ascribed to the term 
 meter. Some writers employ the term to signify exactly regulated rhythm ; ] 
 
 1 ABBOTT AND SEELEY, English Lessons for English People, 146, Boston 1880. 
 
3 8 J. E. Wallace Wallin, 
 
 others, the "foot," 1 " measure," 2 or "bar"; and others, 3 the scheme 
 whereby a fixed number of stresses at fixed points is marked out for the 
 verse, whence we derive the terms dimeter, trimeter, tetrameter, penta- 
 meter and hexameter. According to POE,* meter has to do with the num- 
 ber of feet, rhythm with the character of feet (the arrangement of sylla- 
 bles). Finally, meter has been made synonymous with the line 5 or verse 
 group, whence our "long" and "short" meter. This is the only kind 
 of meter, or rhythm, recognized by Japanese and Persian prosody, the 
 poetry of both of these languages having, theoretically, no accent. 
 
 GUEST 6 provides, perhaps, the most comprehensive definition of the 
 term. He makes it inclusive of three factors, according to the following 
 scheme : 
 
 f i. Elements syllables, verses, staves. 
 
 Meter \ 2. Accidents quantity, accent, modification of sound. 
 [ 3. Law of succession. 
 
 Into a discussion of the propriety of these several views it is not our 
 present purpose to enter. In so far as meter is synonymous with the verse 
 and foot intervals, the two orders of meter will receive due consideration 
 in the appropriate place. 
 
 The word verse is subject to a similar ambiguity. In prosody it is em- 
 ployed in three different senses : ( i ) A number of feet, or succession of 
 words, constituting a metrical line. (2) A group of lines thus com- 
 posed. (3) Metrical composition, as distinguished from prose. 
 
 The words foot, bar and measure are the popular words used to signify 
 the intervals bounded by accented syllables. 
 
 The foot, in poetry, is generally defined as a succession of long and 
 short or accented and unaccented syllables ; the measure, as meter, or a 
 rhythmical period ; and the bar is a variant of the two. Th^ first is ex- 
 clusively applied to poetry ; the second and third are also applied to 
 music. In poetry they are generally employed as synonymous terms. 
 
 To designate the intervals of vacancy in speech the terms pause, rest, 
 stop and silence are indifferently employed. The term rest finds favor in 
 musical terminology ; the word stop is synonymous sometimes with the 
 punctuation mark and sometimes with the making of a change in pitch. 
 
 1 MAYOR, Chapters on English Metre, 5 ff., London 1886. 
 
 2 ARNOLD, A Manual of English Literature, appendix, Boston 1891. 
 3 GuRNEY, The Power of Sound, 425, Lond. 1880 ; CORSON, ARNOLD, as before. 
 4 POE, The Rationale of Verse, Works, VI 56, Chicago 1895. 
 
 S LANIER, The Science of English Verse, 234, New York 1880. 
 6 GUEST, A History of English Rhythms, London 1882. 
 
Researches on the rhythm of speech. 39 
 
 The sound sequence enclosed within two pauses may perhaps be termed 
 a phrase. 
 
 What is the common characteristic of these various groups of facts the 
 verse, foot, pause and phrase sequences ? Obviously, they all agree as par- 
 taking of the same temporal and spatial characteristics. They are phe- 
 nomena of duration. The term interval may be used as a general word to 
 denote any of these sequences. A verse interval is thus a sequence con- 
 sisting of a line of poetry. 
 
 An interval extends between the opening and closing point of a given 
 sequence. Spatially, it is the distance between the first printed object and 
 the last printed object of the sequence. To illustrate : a verse interval is 
 the distance upon the printed page from the first syllable to, and inclu- 
 sive of, the last syllable of the line. Temporally, it is the time which is 
 consumed in speech in passing from the first to the last sound of the 
 sequence. 
 
 In the collection of records to follow, the explanation of the centroid 
 symbols is the same as on p. 15. 
 
 The words enclosed in ( ) are inaudible on the phonograph record ; 
 those in [ ] were supplied by the subject. 
 
 The duration of the interval between any two centroids is indicated by 
 the numeral which is placed midway between them. The measurements 
 of the line and expiration groups are given only in the tables. Pauses 
 marked ? are too indefinite for accurate measurement. 
 
 In the tables, "The immediate probable error" p is determined ac- 
 cording to the formula : 
 
 
 
 k 2 + v* + - 
 \ n i 
 
 where v lt z> 9 , , z/ n represent the differences that exist between each 
 measurement and the average of all the measurements. The immediate 
 probable error furnishes an expression for the amount of irregularity or 
 inequality existing between the components of a series of measurements 
 and the average of the series. 
 
 The ' ' relative immediate probable error' ' furnishes an expression 
 
 of the immediate probable error as a fraction of the average. It will be 
 larger or smaller than the former according as the average is smaller or 
 larger than the unit of measurement. The degree of regularity is ex- 
 pressed by this quantity. 
 
 Both the p and are measures of irregularity on the part of the subject. 
 They are mental quantities. The inaccuracy of the average may be indi- 
 
40 /. E. Wallace Wallin, 
 
 cated by dividing/ by \/n. This gives the " final probable error." It 
 gives the limits within which the average of a similar set of measurements 
 may be expected to vary with a probability of one-half. 
 
 The duration of the average interval may be represented to the im- 
 magination as a line of a definite length, e. g. = i". Every com- 
 ponent of the given series of intervals will be either exactly equal to 
 this line in length, /. e. i 8 , or will deviate from it in one of two di- 
 rections (a) plus (e. g. i. 5o s ) or (b) minus (<?. g. 0.50"). 
 
 The e column gives for each record the lengths of the intervals which 
 deviate the farthest from the average. Thus we get two kinds of ex- 
 tremes, and -f . The former represents the shortest component of the 
 series, the latter the longest. 
 
 The c column gives the character of the difference of the greater of the 
 two extremes from the average. To illustrate graphically : 
 
 Av. 
 
 = 0.50" 
 
 The middle line is the length of the average ; the line above is the 
 longest, and the line below the shortest, single interval in the series. The 
 differences between the average and the -j- line is i 8 ; the difference be- 
 tween the average and the line is 0.50". Hence the character of the 
 greater extreme is +. 
 
 The cipher indicates that the extremes are equal. 
 
 The following records are explained and characterized upon the basis of 
 the impression made upon the ear. They are classified, according to the 
 style of delivery, into four types of poetical scansion. 
 
 The rhythmically free declamation of poetry, or reading scansion, cor- 
 responds to the rhythmically free reading of prose. The one is the 
 basal type of poetical speech ; the other, of prose speech. The former 
 admits of several distinct varieties, the latter of two or three less distinct 
 varieties. 
 
 B. Records. 
 First specimen, J. W. R. (James Whitcomb Riley, poet, lecturer.) 
 
 o 41 1* 2 29 o 30 
 They ain't no style about 'em, 
 
 9 
 
Researches on the rhythm of speech. 41 
 
 3i o 30 2 45 3 67 
 And they're sort o'pale and faded ; 
 
 45 
 o 31 a 42 o 25 o 28 
 
 11 10 
 
 Yit the doorway here, without 'em, 
 
 10 
 
 O 29 3 87 Q 2 22 
 
 Would be lonesomer, and shaded 
 
 J 9 32 2 2 30 . 22 
 
 With a good'eal blacker shadder 
 
 23 28 | 2 28* 50 
 Than the mornin' -glories makes, 
 
 o 3i , 45 o 37 2 40 
 
 10 10 12 
 
 And the sunshine would look sadder 
 
 o 39 2 3 9 8 34 o 
 For their good old-fashion sakes. 
 
 67 
 
 o 2 43 o 37 o 2 28 
 I like 'em 'cause they kind o' 
 
 o 23 o 3 26 o 20 o 75 
 Sort o' make a feller like 'em ; 
 
 3 20 2 l6 20 o 22 o 34 
 And I tell you, when I find a 
 
 o 2 42 o 30 o 2 39 o 70 
 Bunch out whur the sun kin strike 'em, 
 
 3 37 o 
 
 10 
 
 40 o 2 23 
 
 
 It allus sets 
 
 me thinkin' 
 
 
 19 1. 
 
 44 44 
 
 20 
 
 o 43 
 
 18 
 
 O' the ones 
 
 'at used to 
 
 grow, 
 
 o 2 21 
 
 2 23 31 
 
 22 
 
 And peak 
 
 in thro' the 
 
 chinkin' 
 
42 /. E. Wallace Wallin, 
 
 o 23 o 2 25 o 2 25 o 
 O' the cabin, don't you know. 
 ? 65 
 
 2 36 o 26 ^ 64 
 And then I think o' mother, 
 
 40 
 
 Ai8 3 20* 28 o 53 
 
 15 
 
 And how she used to love 'em, 
 
 33 
 
 o 24 3 42 A 53 
 
 When there wuzn't any other, 
 
 40 
 
 o ? 37 o 2 27 o 24 A 78 
 'Less she found 'em up above (em) ! 
 
 53 
 
 o 40 * 3 43 o 35 o 2 58 
 
 23 
 
 And her eyes, afore she shut 'em, 
 
 44 
 
 o 3 o 33 2 46 o 22 
 
 24 
 
 Whispered with a smile, (and) said 
 
 8 
 
 o 34 o 26 2 55 o 2 19 
 We must pick a bunch and put 'em 
 
 24 
 
 o 27 o 2 74 o 32 o 
 
 In her hand when she was dead. 
 37 69 
 
 O 20 O 24 2 II 56 
 
 But, as I was a saying, 
 
 12 
 
 44 0, 32 o 34 
 They ain't no style about 'em 
 8 19 
 
 2 9 2 55* 3 o 2 73 
 
 9 25 12 
 
 Very gaudy or displaying 
 
 53 
 
Researches on the rhythm of speech. 43 
 
 A 34 a 20 2 7 ^ 
 Yet I wouldn't be without 'em, 
 
 40 
 
 o 44 2 33 A 44 2 54 
 'Cause I'm happier in these posies, 
 
 28 
 
 2 22 O 52 O 48 
 
 (And the) hollyhawks and sich, 
 15 39 
 
 O 25 O 21 O 38 | 3 36 
 
 Than the hummin' bird, 'at noses 
 
 15 
 o 26 ., 49 v 
 
 16 
 
 In the roses of the rich. 
 
 (Old Fashioned Roses, Hoosier Dialect, J. W. Riley. ) 
 
 Unit of measurement, -fa of a second. 
 
 The poem was recited from memory. No suggestions were offered. 
 
 The style abounds in the pathetic and the grave. The movement is 
 rather slow, with occasional diminuendos and crescendos and retarded 
 and quickened pace. The melody and swing are peculiarly Rileyan. 
 The rhythm is not continuous. The total effect upon the ear is pleasing. 
 
 The verses are largely end-stopped. Numerous sectional punctuation 
 marks occur. Five stanzas were measured in order to obtain ample 
 material for a study of the regularity of the verse intervals. 
 
 Second specimen, E. W. S. (Scripture.) 
 
 17 o 21 2 31 
 The cities are full of pride, 
 
 20 
 
 2 24 o 20 o 35 
 Challenging each to each 
 29 
 
 0- 2 28 o 23 o 27 
 
 This from her mountain side, 
 
 16 
 
 2 20 2 19 O 
 
 That from her burthened beach. 
 
44 /. E. Wallace Wallin, 
 
 o 2 21 o 29 2 34 
 They count their ships full tale 
 
 15 
 
 o 24 o 25 2 36 
 Their corn and oil and wine, 
 
 25 
 
 o 25 o 20 o 40 
 
 Derrick and loom and bale, 
 
 27 
 
 o 26 o 27 o 42 
 And ramparts gun-flecked line, 
 
 3i 
 
 O 21 O 23 O 32 
 
 City by city they hail : 
 18 
 
 2 19 o 23 A 
 Hast aught to match with mine ? 
 
 O 13 %, 21 O 27 
 
 And the men that breed from them 
 
 12 
 
 015 o 18 o s 41 
 They traffic up and down, 
 
 25 
 
 o 2 24 18 030 
 But cling to their cities hem 
 
 12 
 
 3 23 O 17 2 
 
 As a child to the mother's gown. 
 
 O 25 2 21 041 
 
 When they talk with the strange bands, 
 
 2 9 
 
 27 020 O 42 
 
 Dazed and newly alone ; 
 24 
 
 o 25 o '22 o 43 
 
 When they walk in the stranger lands, 
 
 30 
 
Researches on the rhythm of speech. 45 
 
 2 22 o 23 o 33 
 By roaring streets unknown ; 
 
 24 
 
 2 26 o 26 o 39 
 Blessing her where she stands 
 
 20 
 
 o 2 19 o 19 o 
 For strength above their own. 
 
 A Z 17 o 24 o 28 
 On high to hold her fame 
 9 
 
 | 3 30 o 21 o 39 
 That stands all fame beyond, 
 
 26 
 
 0190 27 040 
 
 By oath to back the same, 
 
 24 
 
 o 2 26 o 190 54 
 Most faithful -foolish -fond ; 
 
 42 
 
 A 26 , 26 o 24 
 
 Making her mere breathed name 
 
 O 19 O 20 O 
 
 Their bond upon their bond. 
 
 
 
 A 3 
 
 19 o 29 2 26 
 So thank I God my birth 
 
 8 7 
 
 o 17 A 2 27 o 32 
 Fell not in aisles aside 
 24 
 
 o 12 o 15 16 3 31 
 Waste headlands of the earth, 
 
 2 2IQ 29 O 30 
 
 Or warring tribes untried 
 16 
 
46 /. E. Wallace Wallin, 
 
 2 15 O 20 2 28 
 
 But that she lent me worth 
 
 2 19 2 21 O 
 
 And gave me right to pride. 
 
 s 2 7 o 26 2 25 
 Surely in toil or fray 
 
 3 J 9 o 2 25 2 35 
 
 Under an alien sky, 
 
 24 
 
 * 3 18 o 19 2 24 
 
 Comfort it is to say : 
 
 ii 
 
 24 O 22 
 
 Of no mean city am I. 
 (Dedication to the City of Bombay, Kipling.) 
 
 Unit of measurement, ^ of a second. 
 
 The first two stanzas were declaimed from memory ; the others were 
 read from the book. The verses are end-stopped. No sectional punc- 
 tuation marks occur. 
 
 The record is typical of the subject, especially as regards emphasis, 
 distinctness of articulation and rapidity of utterance. The point wliich 
 most impresses the ear is the preservation of the rhythmic flow, both ir 
 respect of the recurrence of the centroids and the equality of the verse 
 intervals. The last stanza has a peculiar jog-trot swing ; it will be used in 
 the study of sing-song rhythm. The rhythm of the other stanzas is, in 
 places, so pronounced that the record could have been subsumed under 
 the head of doggerel scansion. The thirty-five lines furnish ample ma- 
 terial for studying the verse interval rhythm. 
 
 Third specimen, A. D. B. (Bissell, musical composer, graduate 
 student. ) 
 
 (A.) 
 
 , 2 3 |s 32 3 8 g 2 36 
 
 Over the chimney the night-wind sang 
 
 7 15 
 
 3 25 o 22 0130130, 77 
 
 And chanted a melody no one knew ; 
 6 56 
 
Researches on the rhythm of speech. 47 
 
 A 
 3 
 
 32 o 15 o 27 o 38 
 And the woman stopped, as her babe she tossed, 
 23 20 
 
 o 18 o 38 , 55 60 
 
 18 12 
 
 And thought of the one she had long since lost, 
 
 18 38 
 
 o 2 65 o 2 34 3 29 o 81 
 
 And said, as her tear-drops back she forced, 
 38 8 55 
 
 a 34 2 40 o 
 
 8 e 
 
 " I hate the wind in the chimney." 
 
 13 15 
 
 (What the Chimney Sang, Bret Harte.) 
 
 (B.) 
 
 o 25 o 2 30 3 30 o 2 27 o 40 
 Though the mills of God grind slowly, 
 17 18 22 
 
 o? 22 o 2 39 | 2 33 o 7 1 
 
 22 
 
 Yet they grind exceeding small ; 
 
 ? 56 
 
 o 28 ^ 3 41 o 29 42 
 
 Though with patience He stands waiting, 
 
 15 ? 27 
 
 o 26 c 2 43 o 29 o 
 With exactness grinds He all. 
 
 20 (Retribution, Longfellow.) 
 
 Unit of measurement, -fa of a second. 
 
 The first stanza was recited from memory ; the second, from the book. 
 The subject was uninstructed. The verses are end-stopped, with sectional 
 pauses. 
 
 The record presents a variety of smooth and rugged movements. The 
 swing in some phrases is almost pendulum like. The centroids are well 
 emphasized. The subject was intent, apparently, on forcibly and artis- 
 tically expressing the thoughts, rather than guarding against interruptions 
 in the flow. The movement is rather slow. 
 
 Fourth specimen, W. W. (Wallin.) 
 
 21 , 44 , 23 26 43 3 19 43 
 The sky is changed ! and such a change ! Oh night, 
 34 35 I 9 
 
48 /. . Wallace Wallin, 
 
 3 5 3 42 17 3 1 o 43 
 And storm and darkness, ye are wondrous strong, 
 21 ii 28 
 
 A 3 l8 o 3 2 2 43 ?e 20 ?7 28 o 30 
 Yet lovely in your strength, as is the light 
 
 33 
 
 * 2 J 4 A 2 2 4 o 45 o 20 o 32 
 
 11 13 
 
 Of a dark eye in woman ! Far along, 
 
 35 J 3 
 
 23 o 35 o 22 o 25 36 
 
 From peak to peak, the rattling crags among 
 18 25 
 
 03 2 7 o 18 * 2 55 o 24 ; 2 i4? 2 36 o 3 
 
 15 10 
 
 Leaps the live thunder ! Not from one lone cloud, 
 30 18 
 
 o 16 2 25 o 18 o 21 o 43 
 But every mountain now hath found a tongue, 
 
 2 3 
 
 9 l6 o 35 o 20 o 20 43 
 
 And Jura answers, through her misty shroud, 
 14 3 2 
 
 o 2 28 21 o 41 o 19 o 33 o 
 
 Back to the joyous Alps, who call to her aloud ! 
 
 22 ? 22 
 
 2 17 o 14 80 8 41 o 2 54 
 
 And this is in the night : Most glorious night ! 
 45 44 
 
 * 2 I2 a II O II O 22 58 % 2 14 O 
 
 3 O 
 
 Thou wert not sent for slumber ! let me be 
 
 37 
 
 18 ,, 28 15*3 36 , 17 53 
 
 34 12 
 
 A sharer in thy fierce and far delight, 
 10 40 
 
 o 2 30 , 28 o 15 ,51 
 
 A portion of the tempest and of Thee ! 
 18 39 
 
Researches on the rhythm of speech. 
 
 49 
 
 A 
 3 
 
 How 
 
 12 
 
 the 
 
 14 
 
 48 
 
 lit lake 
 
 shines, a 
 
 o ? 9 2 IZ o 35 
 phosphoric sea, 
 
 2S 
 
 A 
 
 And 
 
 ii 2 14 14 013 J 3 25 o? 10 o 52 
 the big rain comes dancing to the earth ! 
 
 38 
 
 o, 35 o 27 o 67 30 , 63 
 
 And now again 'tis black, and now, the glee 
 20 15 40 8 23 
 
 19 71 , 47 o 19 o 54 
 
 Of the loud hills shakes with its mountain mirth, 
 5 2 45 
 
 0180 15^348* 20012 A 2 54 o 
 As if they did rejoice o'er a young earth-quakes birth. 
 33 23 
 
 (Childe Harold, Byron.) 
 Unit of measurement, ^ of a second. 
 
 The stanzas were chosen as typical examples of Byronic vigor. The 
 record strikes the ear as aiming more at painting the scene than moving 
 in measured strides. The rhythm is vague. Several glides occur. The 
 centroids are emotional, rather than logical. The elements of pitch, in- 
 tensity and duration are each quite conspicuous. The verses afford an in- 
 teresting study of the sectional and end-line pauses ; several are run-on, 
 with sectional pauses. Several of the larger punctuation marks occur 
 within the line. 
 
 Fourth specimen, J. K. (Kawabe, 
 graduate student, philosophy. ) 
 
 o 16 
 
 o 19 
 
 Kimiga yo wa 
 
 21 
 
 3 I70I5 
 
 Chiyo ni yachiyo ni 
 
 12 A 18 
 Sazare ishe 
 
 02 
 
 2 J 
 
 no 
 7 
 
 3 2 5 O 20 
 
 Iwao to narite 
 
 o 8 o 2 10 o 10 2 
 Koke no musu made. 
 
 (National poem of Japan.) 
 
 36 
 
 45 
 
 40 
 
 48 
 
 Fifth specimen, I. M. (Miyake, 
 graduate student, psychology. ) 
 
 09090 17 
 Kimiga yo wa 
 
 2 22 010^ 
 
 Chiyo ni yachiyo ni 
 
 28 
 
 35 
 
 5 1 
 
 o i 
 Sazare 
 
 A 
 
 .02 
 
 ishi 
 
 no 
 39 
 
 29 o 25 
 Iwao to nari te 
 
 o 15 o ii o ii 2 
 
 Koke no musu made. 
 
 5 2 
 
 Unit of measurement, ^ of a second. 
 
50 J. E. Wallace Wallin, 
 
 The numerals following each verse indicate lengths of the lines. 
 
 Both subjects are natives of Japan. The poem was recited (chanted) 
 and declaimed, strictly following Japanese usage. The above are the 
 records of declamation and not of song. 
 
 The line meter strikes the ear as being perhaps more prominent than 
 the centroid or "foot" meter. Theoretically, Japanese poetry is non- 
 accentual. The centroids, however, were easily located by the experi- 
 menter, and the result was verified by the judgment of J. K. Japanese 
 admit emphasis, as distinguished from accent, but make the distribution 
 wholly arbitrary. It will be seen, however, that both records correspond 
 in regard to the location of the centroids, with the exception of one 
 instance. 
 
 The only meter which Japanese prosody recognizes is the line meter, 
 the above consisting of an alternation of the five-syllable with the seven- 
 syllable type. The five verses constitute one sentence. 
 
 The following is a free English rendering of the poem: "May thy 
 throne last forever, until the sands shall become rock and mosses shall 
 grow thereon." 
 
 Sixth specimen, H. O. (Ohrnstedt, age 12). 
 
 A 3 20 2 3 2 2I 3 8 
 
 I stilla glans han trader fram : 
 8 15 
 
 3 20 o 3 2 I 9 48 
 
 Af spridda far och spada lam 
 
 ii 28 
 
 2 20 O 27 ^ 20 O 38 
 
 Han sig en hjord forsamla vill, 
 12 18 
 
 ^ 3 190 20 o 3 200 55 
 
 Den himmelriket horer till. 
 
 ? 33 
 
 % 3 22 O 20 2 20 O 37 
 
 Och himmel skall forgas och jord, 
 
 *9 
 
 S 20 O IQ 2 20 O 42 
 
 Men ej hans helga dyra ord, 
 
 22 
 
 A 3 22 021% 21 O 
 
 Hans namn af alia tungorsljud 
 
 20 
 
Researches on the rhythm of speech. 51 
 
 39 O 21 20 O 20 O 
 
 Skall kallas Fralsare och Gud. 
 
 Unit of measurement, ^ of a second. 
 
 The record was taken for the purpose of studying sing-song. The 
 rhythm is very marked. 
 
 The end-stopped verses predominate. No intra-line punctuation marks 
 occur. 
 
 Seventh specimen, O. S. (Sandquist, age 13). 
 
 O 2O O 22 O 19 O 26 
 
 O. Jesu Krist att nalkas dig 
 ? ii 
 
 o 16 o 190 39 
 
 Och dig i tro tillbedja, 
 
 18 
 
 O 21 O 21 O 2O O 26 O 26 O 
 
 Det endast kan pa sorgens stig ratt innerlig 
 
 10 
 
 For first stanza and explanation see p. 28. 
 
 This record furnishes an excellent specimen of the doggerel scansion so- 
 common to children. 
 
 Eighth specimen, G. A. A. (Andreen, lecturer ; professor of Scandi- 
 navian languages.) 
 
 2 16 o 16 * , 27 *, 14 031 o 19 3 
 
 Washington was a statesman of the highest order, never a wily 
 
 18 
 
 X 9 02 I 5 IJ 2 4 2 *6 3 20 A 20 02 I0 2 5 
 
 scheming pol i tician. When a cabal was formed to remove him from 
 32 ? 
 
 A 16 * 2 16 47 | 3 15 0,140 16 % 2 20 
 
 the command and elevate Gates, in perfect equipoise of mind and 
 
 27 
 
 o 25 * 2 23 o 59 2 2 5 A l6 o S 2 
 
 temper, he moved not a finger. Neither was he a brilliant orator. 
 
 6 47 33 
 
 o 10 o 28 o 13 o 20 ^ 2 15 c 4 3 2 5 
 
 So that his excellence consisted not in show and glitter, but in nobleness 
 
 24 
 
52 /. E. Wallace Wallin, 
 
 2 48 A 18 OISA, 34 20 o 59 
 
 of heart, solidity of understanding, and tenacity of purpose. No 
 23 J 4 37 
 
 ^ 3 16 o 25 o 15 o 18 29 o, 18 3 35 , 
 wonder then that every American feels as if it were true, that " take 
 16 19 21 
 
 20 O 13 O 32 * 2 9 O 12 O 10 O 15 O 12 O 4! 02 
 
 him for all in all, we shall not look upon his like again," that none 
 18 23 
 
 25 oiio 150 34 o 2 26 o 25 o 42 o 2 31 
 
 do we more love to honor, none so impresses our souls, as he, "the 
 
 23 23 15 
 
 3 18 o 32 2 20 o 38 2 24 2 23 o 
 first in war, the first in peace, the first in the hearts of his countrymen. ' ' 
 17 21 
 
 Unit of measurement, -J-% of a second. 
 
 The address was read from the manuscript. The style was rhetorical ; 
 the utterances rapid and forcible. The subject was asked to make the 
 record a typical address to an audience. 
 
 Tenth specimen, W. L. P. (Phelps, lecturer ; professor, Eng. Lit. ) 
 
 A 12 o 18 2 62 ^ 3 20 oi70 29 o 2 34 
 
 The class of ninety-nine did not hold a very high reputation as 
 
 22 12 
 
 o 70 o 38 o 2 20 27 2 80 37 ^ 3 21 o 
 
 scholars, either for study or for morals. And it was notorious that 
 23 J 3 7o 
 
 21017^2 170 16 021 3 93 , "ft .34 
 
 they paid no attention to the Sunday services ; but one morning, the 
 
 56 
 
 2 41 o 14 2 54 o ii o 24 55 | 2 22 a 20 s 84 
 
 minister, in the chapel, gave out the hymn number ninety-nine ! 
 
 13 34 3 6 74 
 
 o 20 A Q3 16 55 o 2 35 o 25 o 17 ft 48 o 20 
 
 And with one accord, the students of the class of ninety-nine opened 
 3 2 34 
 
 o 40 2 I 9 2 77 o 16 o 17 2 34 
 
 the hymn book, to find the place and what was their amazement, on 
 13 52 
 
Researches on the rhythm of speech. 53 
 
 o 21 * 3 20 , 65 o 20 | 2 i6 70 | 2 25 3 45 
 
 reading the first line, number ninety-nine; great God! what 
 ? 49 53 10 16 
 
 2 43 a 40 O 
 
 worthless worms are we ! 
 
 Unit of measurement, ^ of a second. 
 
 The words were spoken extemporaneously. The movement is, on the 
 whole, slow ; the articulation distinct ; the style colloquial. The sen- 
 tences were punctuated by the speaker. 
 
 Eleventh specimen, C. O. S. (Scoville, clergyman, graduate student, 
 philosophy. ) 
 
 A 
 
 02 17 2 HO 13 O 13 2 20 A 2 23 O 22 
 
 Almighty God, unto whom all hearts are open, 
 
 7 
 2 3 2 o a 25 34 * 28 o ii o 16 A 34 
 
 20 IS 
 
 all desires known, and from whom no secrets are hid ; 
 9 15 8 
 
 44 25 o 2 17 13 o 31 020 o 2 33 
 
 Cleanse the thoughts of our hearts by the inspiration of 
 
 39 8 ' 
 
 o 20 o 35 g 18 ^ 3 26 o 3 36 
 
 thy Holy Spirit, that we may perfectly love thee, and 
 14 27 ? 19 
 
 o 15 o 32 *, 14 o 16 o 45 o 34 
 
 10 9 
 
 worthily magnify thy holy Name ; through Christ our Lord, 
 ii 25 12 
 
 2 33 o 29 o 29 o 34 
 
 10 
 
 Lord, let me know mine end, and the number of my 
 ? 
 
 O 20 O 21 OHO 28 O 28 O 24 O 58 0J 
 
 17 
 
 days : that I may be certified how long I have to live. Behold, 
 7 3 6 
 
 2 3 O 26 O 27 O 30 O 22 2 l8 50 
 
 thou hast made my days as it were a span long : and my 
 7 12 ? 16 
 
54 / E. Wallace Wallin, 
 
 21 O 27 ^ 8 l6 O 20 O 21 46 O 19 O 30 
 
 age is even as nothing in respect of thee ; and verily every man 
 
 27 
 2 5 28 o 19 o 
 
 18 
 
 living is al together vanity. 
 ? 7 
 
 Unity of measurement, -^ of a second. 
 
 Both prayers were read from the manual. The latter is offered at the 
 burial of the dead. The movement is slow, the melody solemn, the 
 intonation monotonous, the style that of devout consecration or solemn 
 importunation, and the rhythm quite manifest in certain phrases. The 
 duration centroids are prominent. 
 
 Test records. The rhythmic feeling has been asserted to be the 
 most general aesthetic endowment of man. 1 Most people undoubtedly 
 distinguish spoken verse from prose largely on the basis of this feel- 
 ing. The external embodiment of verse (aside from the nature of the 
 thought), that is, its verse and stanza-like structure as it appears in 
 print, may be presumed, however, from the law of association, to play no 
 inconsiderable part in lending support to the distinction. How, now, 
 would the judgment of the individual be affected after the props of asso- 
 ciation have been removed ? How would na'ive, unsophisticated thought 
 regard the distinction of prose and poetry into measured and unmeasured 
 language. 
 
 To get at the uncritical, spontaneous judgment of the individual, the 
 following test was devised : 
 
 The subject was conveniently seated before the phonograph, and was 
 given a short passage in print, the contents of which were unknown to 
 him, to be immediately read into the phonograph. He was urged to make 
 his reading natural and representative. 
 
 Two devices were employed. In the first the subject was given a selec- 
 tion of poetry printed in the form of prose ; in the second, a passage of 
 melodious prose printed as a stanza of poetry. In the latter case the sub- 
 ject was allowed to read the stanza through, before speaking into the tube, 
 .and was then told to read it according to his own discretion. 
 
 After the passage had been read the following questions were asked : 
 
 1 . Did the passage, as you read it, sound familiar ? 
 
 2. What did you spontaneously conclude regarding its form? If the 
 question had never been raised, would you have taken it for granted that 
 you read a piece of poetry or prose ? What did you feel it to be ? 
 
 1 LANIER ; GURNEY, as before, 128 ff. ; BOLTON, as before. 
 
Researches on f/ie rhythm of speech. 55 
 
 3. Without engaging in reflective analysis, what do you instinctively 
 feel to be at the basis of your judgment? Why poetry? Why prose? 
 
 4. Did the passage sound smooth or rugged as you were reading, or 
 was it indifferent ? 
 
 5. What did you feel these properties to be due to ? 
 
 6. Did the passage sound melodious to you? If so, what does melody 
 in speech mean to you ? What are its elements ? 
 
 Care was taken lest the subject should begin to unduly reflect upon the 
 questions. Those judgments were rejected where the subject was familiar 
 with the quotations. The questions were submitted to several persons 
 from whom no records were taken. 
 
 The questions are in no wise to be considered as a statistical enquiry. 
 They are only intended to reveal the state of consciousness, or the spon- 
 taneous judgments, of unsuspecting subjects. 
 
 After the answers had been given and the device exposed, the sub- 
 jects of the first experiment were handed a copy in which the poetical 
 quotation was printed in its original form, that is, as poetry. The sub- 
 jects were then asked to familiarize themselves with the contents, and read 
 it as poetry. Thus it was possible to make a comparison of the two ren- 
 derings by a study of the phonograph records. 
 
 A third device consisted in eliminating all the punctuation marks. 
 The subject was not allowed to examine the passage before speaking it 
 into the recording tube. 
 
 ( a ) Records of verse. 
 
 Twelfth specimen. J. M. T. (Telleen, graduate student, English.) 
 (A, without punctuation marks.) 
 
 o 2 16 o 14 o 25 o 25 o 41 028 40 o 
 
 Nothing henceforth man's existence bows [bows] to the mo ni- 
 ii 29 24 20 
 
 8 o 16 o 30 o 25 o 24 o 26 o 9 o 17 2 27 
 tion wait take the joys and bear the sorrows neither with extreme con- 
 
 o 25 o 15 35 o 170 20 2 40 o 20 Q 2 i3 2 3 
 
 cern living here means nescience simply 'tis [the] next life that 
 
 12 24 25 10 
 
 o 15 o 18 9 14 21 2 13 o 39 %. 2 27012 14 27 
 
 helps to learn shut those eyes next life will open stop those ears 
 
 7 8 29 12 13 
 
 02 I2 33 o 19 o 20 o 37 14 13 o 26 o 14 o n 
 next life will teach hearing's office close those lips next life 
 17 ii 23 17 
 
5 6 /. E. Wallace Wallin, 
 
 o 16 p 35 o 35 o 
 will give the power of speech ! 
 
 (B, read as prose. ) 
 
 2 21 15 o 14 O 21 O 25 2 28 Q 7 O 26 
 
 Nothing ! Henceforth man's existence bows to the mo nition 
 8 5 8 
 
 2 32 o 18 2 28 * 2 29 o 30 o 12 o ? 32 o 20 
 wait ! Take the joys and bear the sorrows neither with extreme 
 23 13 
 
 o 39 A 2 io 16 o 18 o 17 o 35 012 o n o 23 
 concern! Living here means nescience simply: 'tis next life that 
 26 25 
 
 O 21 2 41 II 010 * 2 32 O 12 20 O 41 2 36 O 9 
 
 helps to learn. Shut those eyes, next life will open, [those] stop 
 29 20 6 23 
 
 19041 2 20 3 1 015 i 24 o 40 2 8 o 7 2 37 
 those ears, next life will teach hearing's office, close those lips, 
 27 . 27 25 
 
 A 16 v 25 o 23 o 33 o 
 next life will give the power of speech ! 
 
 (C, read as poetry. ) 
 
 2 23 o 52 20 o 20 o 26 2 35 o 23 
 
 Nothing ! Hence forth man's existence bows to the monition 
 10 44 8 
 
 o 43 
 wait ! 
 
 35 
 
 o 17 2 4 2 o 24 g 2 37 c_ 12 p 20 o 25 o 40 
 
 Take the joys and bear the sorrows neither with extreme concern ! 
 
 20 28 34 
 
 18 * 2 26 015 i a 19 o 43 o, 13 Q 13 p 24 o ^20 
 
 Living here means nescience simply: 'tis next life that helps^to 
 ? 30 
 
 o 50 
 learn. 
 42 
 
 8 9 :, J 3 * 5 o 2 130 19 48 2 12 o 17 , 46 o 2 12 o 
 Shut those eyes, next life will open, stop those ears, next life 
 
 34 33 37 
 
 30 22 
 will teach 
 
Researches on the rhythm of speech. 57 
 
 A 16 o 50 2 8 o 2 13 2 37 A 2 12 o 28 o 270 36 
 
 Hearing's office, close those lips, next life will give the power of 
 
 37 28 ? ? 
 
 o 
 speech. 
 
 (Browning.) 
 
 Unit of measurement, -fa of a second. 
 
 Answers to questions. i. No. 2. Prose. 3. Form of print. 4. Rug- 
 ged. 5. The thought; loose rhythm. 6. No. 
 
 Thirteenth specimen. C. O. (Olson, graduate student, English.) 
 
 (A, read as prose. ) 
 
 o 2 32 o 30 o 22 o 29 2 29 o ? 9 o 21 
 
 Nothing ! Henceforth man's existence bows to the mon ition 
 
 18 6 ii 
 
 o 3 39 o 18 24 , 2 23 o 47 o ii o 20 A 21 
 wait ! Take the joys and bear the sorrows neither with extreme con- 
 33 3 1 
 
 A 46 o 14 o 16 o 40 o 26 o 41 12 A 2 13 ^ 25 
 cern ! Living here means nescience simply: 'tis next life that 
 38 27 26 
 
 2 I 9 3902 I2 14 3 29 o 12^ 19 o 39| 2 I2 2 
 helps to learn. Shut those eyes, next life will open, stop those 
 3 1 18 25 
 
 ii 250250 23 o 14 , 19 o 43 2 l6 l6 3 25011 
 ears, next life will teach hearing's office, close those lips, next 
 15 26 15 
 
 , 24 o 12 ^ 2 25 o 
 life will give the power of speech. 
 6 9 
 
 (B, read as poetry. ) 
 
 o a 46 o 16 18 o 21 o 26 o 29 o 21 o 2 41 
 
 Nothing ! Henceforth man's existence bows to the monition wait ! 
 
 30 10 10 32 
 
 o 12 32 19 51 o 2 9 o 17 o 22 o 49 
 
 1 11 
 
 Take the joys and bear the sorrows neither with extreme concern ! 
 X 5 3 2 35 
 
 A 10 ^20 013 j 23 o 34 o ? ii0 2 i2o 20 o 18 
 Living here means nescience simply: 'tis next life that helps to 
 
 22 
 
 o 42 
 learn. 
 
58 /. E. Wallace Wallin, 
 
 3 
 
 3 8 o 9 02 2 3 o 12 2 20 ^ 2 23 2 8 2 12 * 2 24 2 90 
 
 Shut those eyes, next life will open, stop those ears, next life 
 9 l8 9 3 1 
 
 47 o 21 
 will teach 
 ii 
 
 Q 3 16 o 4 1 o 80 9 o 26 ^ 3 12 9 2 21 o 16 2 2 
 
 Hearing's office, close those lips, next life will give the power 
 
 28 18 
 
 31 o 
 of speech ! 
 16 
 
 Unit of measurement, -% of a second. 
 
 Answers to questions. i. No. 2. Dramatic prose. 3. Did not "feel" 
 it to be poetry. 4. Smooth. 5. The movement. 6. Yes. Smoothness 
 and movement. 
 
 Fourteenth specimen. G. F. A. (Abel, graduate student, philosophy. ) 
 
 ( Without punctuation marks.) 
 
 g, 12 * 2 17 2 30 o 3 29 37 3 19 o 17 
 
 Henceforth man's existence bows to the monition wait take the 
 
 14 26 10 
 
 0, 25 2 19 2 39 o 10 o, 24 o 3 24 o 2 io o ? 14 
 
 joys and bear the sorrows neither with extreme concern living 
 
 25 8 ? 
 
 O l8 0?II ? 220 52 * 2 14 3 31 O 17 41 2 
 
 here means nescience simply 'tis next life that helps to learn shut 
 7 33 l6 3i 
 
 i2 2 10* 22o 2 10 3 15 o 36 o 9 13 37 o 2 12 o 19 o 10 
 those eyes next life will open stop those ears next life will teach 
 8 25 24 
 
 3 X 5 o 39 * 12 2 15 330 120 18 o 17 o 29 
 
 hearing's office close those lips next life will give the power of 
 26 26 8 
 
 o 
 speech ! 
 
 Unit of measurement, ^ of a second. 
 
 Answer to questions. i. No. 2. Prose. 3. Arrangement of words ; 
 form; lack of rhyme. 4. First part, rough; second part, smooth. 5. 
 Lack of meaning and punctuation marks. 6. Yes and no, as in 4. 
 Rhythm, if anything. 
 
Researches on the rhythm of speech. 59 
 
 Fifteenth specimen. W. C. (Churchill, graduate student, philosophy.) 
 (A, read as prose. ) 
 
 o 18 2 23 09 o ? 15 23 o 16 -0 2 15 2 17 o 
 Oh yet we trust that somehow good will be the final goal of ill, 
 5 
 
 49 o 2 23 o 24 o 21 o 53 o 19 3 25 2 21 
 
 to pangs of nature, sins of will, defects of doubt, and taints of 
 39 I2 45 *9 
 
 o 56 2 21 o 30 21 o 57 on 014 2 
 
 blood; that nothing walks with aimless feet; that no one life 
 
 5 l8 49 9 
 
 24 o 15 o 40 o 22 o 19 o ii o 49 2 
 
 shall be destroyed or cast as rubbish to the void, when God hath 
 3 1 4i 
 
 24 O l8 O 22 O 
 
 made the pile complete. 
 
 ( B, read as poetry . ) 
 
 o 20 2 28 o 9 3 
 Oh yet we trust that somehow good 
 
 13 ? 9 
 
 23 o 19 2 18 2 210 40 
 Will be the final goal of ill, 
 
 37 
 
 o 23 , 27 o 24 o 45 
 
 To pangs of nature, sins of will, 
 
 15 37 
 
 O 23 O 4 6 O 22 O 6l 
 
 Defects of doubt, and taints of blood ; 
 34 48 
 
 A 2 18 o 23 o 20 o 48 
 That nothing walks with aimless feet ; 
 
 43 
 
 2 10 o 13 3 27 o 19 o 36 
 That no one life shall be destroyed, 
 ii 33 
 
 0270 22 014 048 
 
 Or cast as rubbish to the void, 
 ? 34 
 
 o 26 o 23 | s 41 o 
 When God hath made the pile complete. 
 
 16 
 (Tennyson. ) 
 
6o /. E. Wallace Wallin, 
 
 Unit of measurement, ^ of a second. 
 
 Sixteenth specimen. B. S. G. (Gowen, college senior. ) 
 
 ( Without punctuation marks. ) 
 
 25 o 22 ^28 ,2 J 7 2 29 o 14 o 16 41 o 
 Oh yet we trust that somehow good will be the final goal of ill 
 
 7 12 25 5 
 
 22 o 24 o 39 o 60 c 30 o 18 o 2 28 o 23 o 
 
 to pangs of nature sins of will defects of doubt and taints of blood 
 26 35 19 
 
 32 O 15 2 4 2 19 O 350 8 02 130 23 
 
 that nothing walks with aimless feet that no one life shall be 
 20 25 18 5 
 
 16 o 40 o 22 o 19 o ii o 49 2 24 o 18 
 
 destroyed or cast as rubbish to the void when God hath made the 
 21 41 
 
 O 22 O 
 
 pile complete. 
 
 Unit of measurement, -fa of a second. 
 
 Answers to questions. i. No. 2. Prose. 3. " Felt " no rhythm. 4. 
 Quite smooth. 5. Smooth succession of words ; occasional rhythm. 6. 
 In parts. Fitness of language to thought, and character of thought. 
 
 Seventeenth specimen. S. (Smith, laboratory mechanic. ) 
 
 (A, read as prose. ) 
 
 o 19 o 25 o 24 o 17 o 18 o 19 o 17 o 
 
 Oh yet we trust that somehow good will be the final goal of ill, 
 
 5 42 
 
 50 o 24 o 33 c 20 o 33 o 21 o 32 o 19 
 
 to pangs of nature, sins of will, defects of doubt, and taints of 
 18 25 23 
 
 o 37 o 18 30 o 23 o 50 on 2 120 
 
 blood ; that nothing walks with aimless feet ; that no one life 
 
 3 2 39 
 
 21 o 32 53 o 25 o 29 o 66 o 16 
 
 shall be destroyed or cast as rubbish to the void, when God hath 
 19 30 I2 53 
 
 O 20 O 23 O 
 
 made the pile complete. 
 
 (B, read as poetry.} 
 
 o 42 o 25 o 51 o 33 
 
 Oh yet we trust that somehow good 
 18 27 
 
Researches on the rhythm of speech. 61 
 
 o 20 o 21 o 34 o 58 
 Will be the final goal of ill, 
 
 45 
 
 o 26 o 45 o 20 o 34 
 
 To pangs of nature, sins of will, 
 
 29 27 
 
 o 22 o 40 o 20 o 59 
 
 Defects of doubt, and taints of blood ; 
 
 2 5 43 
 
 o 19 o 29 o 22 47 
 
 That nothing walks with aimless feet ; 
 
 ? 29 
 
 o 10029 o 25 o 18 o 36 
 
 That no one life shall be destroyed, 
 
 7 22 
 
 o 23 o 430 ? Z 7 o 49 
 Or cast as rubbish to the void, 
 
 23 3 2 
 
 o 29 o 19 o 24 o 
 When God hath made the pile complete. 
 
 7 
 
 Unit of measurement, ^ of a second. 
 
 Answers to questions, i. No. 2. Prose. 3. Felt little rhythm. 4. 
 Not very smooth. 5. Form of expression. 6. Yes. Rhythm and vowel 
 sounds. 
 
 Two other persons replied to the questions, but no records of their 
 recitations were made. (B) i. No. 2. Poetry. 3. Rhythm, rhyme 
 and sentiment. 4. Smooth. 5. Metrical perfection, uniform length of 
 words, avoidance of sibilants. 6. Yes. Meter, onomatopoetic words, 
 o sounds, polished expression of sentiment. (C) i. No. 2. Prose. 
 3. Form of print. 4. Smooth. 5. Rhythm and rhyme. 6. No. 
 
 The above two specimens of poetry are taken from the two foremost 
 poets of the century just closed, the one standing for the highest met- 
 rical perfection, the other emphasizing the content rather than the formal 
 investiture of thought. Each instance contains one case of run-on 
 verses. The Browning selection, however, contains by far the greater 
 number of intra-line punctuation marks. 
 
 To the ear the rhythm of the prose and poetry renderings appears to 
 be about equally balanced. Appeal was made to several ears. On the 
 
6i /. E. Wallace Wallin, 
 
 basis of the rhythm the attempt to distinguish the prose from the poetry 
 in the readings did not prove very successful. 
 
 The movement in the records without the punctuation marks appears 
 the most rugged. In these records the pauses strike the ear as if placed 
 with diffidence and hesitancy. 
 
 (<) Records of prose. 
 Eighteenth specimen. S. I. (Isakson, stenographer.) 
 
 2 19 o 26 o 13 , 20 o 47 
 
 To the poor Christian that sits bound in the galley ; 
 
 28 
 
 , 20 o 27 o 17 o 32 o 14 o 39 
 
 To despairful widows, pensive prisoners, and deposed kings ; 
 
 13 12 29 
 
 o 21 2 28 o 30 o 15 o 50 
 
 To them whose fortune runs back and whose spirits mutiny 
 
 7 21 
 
 o 19 2 23 o 29 o 14 o 17 o 
 
 Unto such death is a redeemer, and the grave a place for rest. 
 9 10 
 
 (Essay on Death, Bacon.) 
 Unit of measurement, -fa of a second. 
 
 Answers to questions. i. No. 2. Poetry. 3. Form of print, easy 
 movement, and beautiful sentiment. 4. Smooth. 5. Meter. 6. No. 
 Nineteenth specimen. O. E. S. (Staaf, graduate student, Latin.) 
 
 2 ii o 28 013 o 70 ioo 38 
 To the poor Christian that sits bound in the galley ; 
 
 21 
 
 o 16 o 26 o 190 38 o 18 o 31 
 
 To despairful widows, pensive prisoners, and deposed kings ; 
 
 12 14 20 
 
 2 29 o o 021 o 14 o 17 o 43 
 
 To them whose fortune runs back and whose spirits mutiny 
 
 8 , 15 
 
 2 21 O 25 O 29 O 15 O 15 O 
 
 Unto such death is a redeemer, and the grave a place for rest. 
 14 10 
 
 Units of measurement, -fa of a second. 
 
 Answers to questions. i. No. 2. Poetry. 3. Meter, outward form. 
 4. Neither. 6. No. 
 
Researches on the rhythm of speech. 63 
 
 Answers to questions were obtained in several cases without taking 
 phonograph records. (D) i. No. 2. Poetry. 3. Periodic recur- 
 rence of accents. 4. Smooth. 5. Rhythm. 6. No. (E) i. No. 2. 
 Doubtful. First two lines poetry ; last two lines upset judgment. 3. 
 Rhythm of first two lines (poetry) ; prosaic thought (prose). 4. Feel- 
 ing of smoothness gradually vanishing. 5. Correspondence of main 
 stress-points in first two lines (3 each). 6. Yes. Rhythm. (F) i. 
 No. 2. Poetry. 3. Verse-like structure, inversion, a certain rhythm. 
 6. Yes. Don't know. (G) i. No. 2. Poetry. 3. Mechanical form 
 and poetic sentiment. 4. Smooth. 5. ?. 6. Yes. (H), poet i. No. 
 
 2. Rhythmical prose. 3. No meter. 4. Smooth prose. 5. Swinging 
 movement. 6. Yes. Graceful expression of thought, alliteration and 
 interior rhymes. (I) i. No. 2. Poetry. 3. Meter and poetic language, 
 especially the second line. 4. Not altogether smooth. 5. Smooth flow. 
 6. Yes. Pleasing sounds. (J) i. No. 2. Both. Lines 2 and 4, poetry; 
 lines i and 3, prose. 3. Rhythm and lack of rhythm. 4. Lines 2 and 4, 
 smooth. 5. Rhythm. 6. Yes. ? 
 
 The second specimen of rhythmical prose arranged as a stanza of 
 poetry is found on p. 26. The questions elicited the following replies : 
 (K) i. No. 2. Poetry. 3. Rhythm. 4. Smooth. 5. Regular recur- 
 rence of accents. 6. Vague feeling of melody. (L) i. No. 2. Poetry. 
 
 3. Regularly recurring accents (primarily), rhyme, printed form. 4. 
 Fairly smooth. 5. Rhythm. 6. No. (M) i. No. 2. Poetry. 3. 
 Rhythm and rhyme. 4. Smooth. 5. Uniform number of syllables to 
 line. 6. Yes. Rhyme. (N) i. No. 2. Prose printed as verse. 3. 
 Subject-matter prosaic. 4. First part smoother. 5. Rhythm and 
 rhyme. 6. No. 
 
 The judgments were derived from persons who, according to admission, 
 possessed more or less poetic appreciation. Nearly all were admirers of 
 music and the majority possessed some knowledge of its rudiments. The 
 answers throw some light on the varied manifestation of the instinctive or 
 native appreciation of rhythm, smoothness and melody in speech. The 
 connotation of the term melody of speech is in most people's minds 
 shifting and vague. This is suggested by the answers and by a cursory 
 glance at the popular discussions of the subject. The word answers, in 
 most people's minds, to no fixed conception. To most subjects it is 
 interchangeable with rhythm (motion); to some, with vowel sounds, 
 rhyme and alliteration (tone-color); to others, graceful and fit expression 
 of thought ; and yet to others the term has no signification. 
 
 The connotation of the term smoothness of speech is more fixed than 
 that of melody. This quality was felt to be due to the quality, first, of 
 
64 /. E. Wallace Wallin, 
 
 the movement, and second, of the sound. In some cases no definite 
 notion corresponded to the term ; in others, no smoothness was felt. In 
 some of the latter cases, as was noticed by the experimenter, the reading 
 was rugged. 
 
 The success of the deception varied. Three subjects declared the 
 verses of Tennyson and Browning to be prose. The reasons given were 
 "lack of rhythm," "form of print," "indefinite." Only one subject 
 pronounced the verses of Browning, and two the verses of Tennyson, 
 poetry. All the judgments were based on the feeling of rhythm. Six 
 subjects called the prose sentence of Bacon poetry. The reasons 
 given were the outward form, sentiment, poetic language, inversion of 
 words, or rhythm. One subject was doubtful, one pronounced the sen- 
 tence half-and-half and one (a poet) pronounced it, correctly, rhythm- 
 ical prose. The specimen of mathematical prose was declared to be 
 poetry by three subjects on the basis of rhythm, rhyme, or the form of 
 print ; and to be prose written as verse by one on the basis of the prosaic 
 subject matter. 
 
 Some of the subjects, who felt no rhythm in the lines, pronounced the 
 passages smooth, others feeling no smoothness pronounced the lines 
 rhythmical. 
 
 The following corollaries, to be held tentatively, are deducible from a 
 comparison of the judgments : 
 
 (#) Poetry is spontaneously and instinctively distinguished from prose 
 most largely upon the basis of a rhythmic affection of sensibility. All 
 other differences, such as inversion, sentiment, rhyme, etc., are subordinate. 
 
 (/) For unsuspecting and nai've thought, the arrangement of the words 
 into verses and stanzas serves as an important prop for differentiating the 
 rhythms of prose and poetry. The removal of this support tends in the 
 majority of cases to convert the rhythm of poetry into the rhythm of prose, 
 and vice versa. 
 
 (<r) The rhythmic instinct is an affection of sensibility, varying to some 
 extent with individuals. It has two aspects motor and sensory. The 
 test concerns only the former the rhythm felt by the subject in uttering 
 and not in hearing uttered a series of words. As a motor phenomenon 
 the instinctive perception of the rhythmical time sense is subject to varia- 
 tions in the immediacy of response and in the universality of manifestation. 
 This is most obviously dependent upon the degree of development of the 
 power of coordinating time-units. 
 
 We begin now the detailed examination of the quantitative determina- 
 tion of the various groups of recurrences. The verse intervals will most 
 conveniently be dealt with first. 
 
Researches on the rhythm of speech. 65 
 
 C. Verse intervals. 
 
 The term verse is used in this discussion as synonymous with a line of 
 poetry. The verse interval signifies the interval of sound which consti- 
 tutes a run-on or end-stopped line of poetry. It is the time required in 
 speaking to pass from the first to the last sound of the line. 
 
 The measurements are made from the beginning of the first to the end 
 of the last sound of the verse. The pauses at the end of the verses are 
 not included in the measurements. Hence the determinations yield a 
 measure of the length and inequality of the verses only. 
 
 TABLE VIII. 
 
 ( Poetry. ) 
 
 Subject. 
 
 a 
 
 P 
 
 I 
 
 a 
 
 e c 
 
 n 
 
 J. W. R. 
 
 2.20 
 
 0.36 
 
 0.15 
 
 + 3-18 
 -1.56 + 
 
 32 
 
 E. W. S. 
 
 I-I5 
 
 0.09 
 
 0.07 
 
 H- I -39 
 0.89 
 
 36 
 
 A. D. B. A 
 
 2.41 
 
 0.30 
 
 O.I 2 
 
 f C\ I 
 
 6 
 
 
 
 
 
 1.91 
 
 
 W. W. 
 
 2.95 
 
 o-45 
 
 0.15 
 
 + 4.04 , 
 -I.8 7 
 
 18 
 
 A. R. P. 
 
 1-63 
 
 0.21 
 
 0.12 
 
 4- 1-92 4 
 1-35 
 
 4 
 
 H. 0. 
 
 2.05 
 
 O.I2 
 
 0.05 
 
 + 2.25 
 -1.67 
 
 8 
 
 O. S. 
 
 1.87 
 
 0. 12 
 
 O.O6 
 
 + 2.12 
 
 7 
 
 
 
 
 
 1-53 
 
 
 Av. 
 
 2.05 
 
 0.23 
 
 O. IO 
 
 
 
 TABLE IX. 
 ( Pvetry read as prose. ) 
 
 Subject. 
 
 a 
 
 P 
 
 
 
 a 
 
 c n 
 
 'J. M. T. 
 
 4.14 
 
 0.50 
 
 0.12 
 
 + 5 
 
 C. O. 
 
 3-92 
 
 0.30 
 
 O.O7 
 
 + 5 
 
 W. C. 
 
 1.74 
 
 0.13 
 
 O.O7 
 
 8 
 
 S. 
 
 1.66 
 
 0.12 
 
 0.07 
 
 8 
 
 Av. 
 
 2.86 
 
 0.26 
 
 0.08 
 
 
 TABLE X. 
 ( Poetry read as poetry. ) 
 
 ' 
 
 Subject. 
 
 J. M. T. 
 
 4.62 
 
 o-55 
 
 O.I I 
 
 + 5 
 
 C. O. 
 
 3-84 
 
 0.24 
 
 0.06 
 
 + 5 
 
 W. C. 
 
 1.92 
 
 0.23 
 
 0. 12 
 
 8 
 
 S. 
 
 2-39 
 
 0.26 
 
 O. IO 
 
 + 8 
 
 Av. 
 
 3-18 
 
 0.32 
 
 O.O9 
 
 
66 J. E. Wallace Wallin, 
 
 TABLE XI. 
 ( Prose read as poetry. ) 
 
 Subject. a p en 
 
 S. I. 2.85 0.27 0.09 -f- 4 
 O. E. S. 3.53 0.31 0.08 4 
 
 Av. 3.16 0.29 0.08 
 
 TAPLE XII. 
 {Summary of TABLES VIII., ix., and x., poetry. ) 
 
 Table. a -b 
 
 
 
 
 a 
 
 VIII. 
 
 2.05 
 
 0.23 
 
 o. 10 
 
 IX. 
 
 2.86 
 
 0.26 
 
 0.08 
 
 X. 
 
 3.18 
 
 0.32 
 
 0.09 
 
 Av. 
 
 2.69 
 
 0.27 
 
 0.09 
 
 Unit of measurement, I s . -, relative immediate probable error. 
 
 a, average duration. 
 
 ,. e. extreme intervals. 
 
 p, immediate probable error. 
 
 c, character of greater extreme. 
 
 These observations are based upon a study of 163 verses, various as 
 to length and structure. 
 
 Extreme intervals. The longest interval among the in verses is 5.01", 
 4th verse, record of J. M. T.; the shortest o.8 9 s , Table VIII. , E. W. S.; 
 a range of 4. 1 2". This is applicable to verse intervals, similarly composed, 
 in general, presupposing that the figures fairly well represent the extremes. 
 
 Character of extreme intervals. Column c of Tables VIII. , IX. and X. 
 furnishes the following data bearing upon the nature of the largest extreme 
 deviation among individual verses : 60 per cent, are extremes of excess, 
 40 per cent, of deficiency, and o per cent, of equality. The tendency 
 is thus to cause the longest individual deviation to vary more than the 
 shortest from the length of the average. 
 
 Average duration. (i) For the series. The average for the series of 
 163 intervals is 2.69'. For the series of records of prose read as poetry 
 it is 3.16". 
 
 (2) For the sets. The shortest average of any set of records is 2.05", 
 Table VIII. , various types of scansion; the longest, 3.18", Table X., 
 poetry read as poetry, making a range of i . 1 3" for the average verse in- 
 tervals of the different sets of records. 
 
 The average in the set of records where poetry was read as poetry is 
 0.32" longer than in the records where poetry was read as prose. 
 
 The average in the set of records where prose was read as poetry is 
 
Researches on the rhythm of speech. 67 
 
 o.o4 s longer than the average of Tables IX. and X. (test records), and 
 o.ii 8 longer than in the set of records of Table VIII. (poetry). 
 
 (3) For the records. The longest average for any record is 4.62*, J. 
 M. T., Table X.; the shortest, 1.15", E. W. S., Table VIII.; a range of 
 3.47". These figures presumably represent the time limits within which 
 the average verse interval, similarly composed, of any reciter will fall. 
 
 The range for different records appears to be nearly three times larger 
 than the range for different sets of records ; and the range of the indi- 
 vidual verses is about 1.29 times the range existing between the average 
 intervals of different records. 
 
 Relative inequality. (i) For the series. The inequality for the series 
 of poetry is 9 % of the average duration of the intervals, and for prose, 
 8 %. 
 
 (2) For the sets. The highest percentage is 10 %, Table VIII., types 
 of scanned poetry ; the lowest, 8 % , poetry read as prose, and prose read 
 as poetry; a range of only 2 %. 
 
 (3) For the records. The highest percentage among the individual 
 records is 15 % , W. W. and J. W. R ; the lowest, 5 <f c . This gives a 
 range at 10 % for different records. Every subject was slightly more 
 regular in reading the poetry as prose than as poetry. The highest differ- 
 ence between the regularity of the two records of the same subject is 0.05 
 (W. C.) ; the lowest, .01 (J. M. T. and C. O.). 
 
 Importance attaching to the duration of the verse interval. The verse 
 is the longest interval with which these measurements are concerned ; 
 hence the relevancy of the following remarks. 
 
 The interest that attaches to the length of the verse-group arises 
 chiefly from a consideration of the span of the rhythmic time sense. 
 How long a period is consciousness capable of rhythmizing ? How short 
 or how long must the unit be to enable the mind to coordinate it with 
 other units of the same kind ? 
 
 GURNEY 1 holds that the most highly developed rhythmic sense can 
 take in no more than a "small group"; and SiEVERS 2 and SCHMIDT/ 
 equally vague, "strictly limit" the length of the sentence so that the 
 time occupied in reading it may constitute a rhythmical whole. BOLTON * 
 concludes from experiments on subjective rhythmisation that there is a 
 correspondence between the average time of a rhythmical unit, no matter 
 
 1 GURNEY, The Power of Sound, 127, London 1880. 
 
 2 SIEVERS, Grundztige der Phonetik, 4. AufL, 218, Leipzig 1893. 
 
 3 SCHMIDT, Introduction to the Rhythmic and Metric of the Classical Languages, 
 79ff., Boston 1878. 
 
 4 BOLTON, Rhythm, Am. Jour. Psych., VI 145-238. 
 
68 /. E. Wallace Wallin, 
 
 what the form, and the normal wave of attention. The rate of speed 
 determined the form, or the number of the components (8, 6, 4, or 2), 
 of the group. No matter what the number, within certain limits, of the 
 impressions, the duration of the group remained relatively the same, 
 about i s . He gives 1.5" as the upper limit. 
 
 It seemed that the natural period of attention exercised a regulative 
 influence upon, and definitely limited the span of, the groups. 
 
 LANGE 1 found that the periodicities of the maxima and minima of 
 attention for acoustic sensations varied from 3.5 to 4.0"; for optic, from 
 3.0 to 3. 4 s ; and for electrical cutaneous, from 2.5 to 3.0". 
 
 STEVENS 2 posits a time interval between 0.53" and 0.87% which is 
 capable of being reproduced with the greatest accuracy. A plus interval 
 results in a plus error; and a minus in a minus error. JOHNSON, 3 how- 
 ever, experimenting on practice and habit, concludes that there is no 
 " indifference point " from which variations do not occur. 
 
 DiETZE 4 reports that the subject of his experiments in rhythm was in- 
 capable of grouping the impressions after the period had reached 4.25'. 
 
 Other experiments have given the following results : for the MASSON 
 disc, an average period of 5 3.5" (with average variation of 0.3"), and 
 of about 5 or 6 s6 (with possible variation from 6 to 27 s7 ), the oscillation 
 having been found most rapid 8 with a certain difference between the 
 grayness of the disc and the ring ; for voluntary effort, 9 a period of three- 
 fourths of a second, attended by a large fluctuation and recovery ; and 
 for subjective rhythmisation of sounds, the "lower limit" 10 of the inter- 
 vals has been placed at less than 2 s ; others give a limit of less than 0.4 s111 
 and o.5 s12 , and of more than i.o s . 13 
 
 1 LANGE, Beitrdge zur Theorie der sinnlichen Aufnierksarnkeit und der activen ap- 
 perception, Philos. Studien, 1888 IV 404. 
 
 2 STEVENS, On the time sense, Mind, 1886 X 394. 
 
 3 JOHNSON, Researches in practice and habit, Stud. Yale Psych. Lab., 1898 VI 91. 
 
 4 DIETZE, Untersuchungen tiber den Umfang Jes Bewusstseins bei regelmdssig aufein- 
 anderfolgenden Schalleindriicken, Philos. Studien, 1885 II 383. 
 
 5 PACE, Zur Frage der Schwankungen der Aufmerksamkeit nach Versuchen mit Mas- 
 son 'sc hen Scheiben, Philos. Studien, 1893 VIII 394. 
 
 5 TiTCHENER, A Primer of Psychology, 89, New York 1898. 
 7 TiTCHENER, An Outline of Psychology, 141, New York 1896. 
 
 8 MARBE, Die Schwankungen der Gesichtstmpjindungen, Philos. Studien, 1893 VIII 
 615. 
 
 9 SCRIPTURE, The New Psychology, 125, New York 1897. 
 
 1 SMITH, Philos. Studien, 1900 XVI 282. 
 
 ^MEUMANN, Philos. Studien, 1894 X 302. 
 
 12 MARTius, Philos. Studien, 1891 VI 196. 
 
 13 TiTCHENER, Experimental Psychology, I, ii 340, New York 1901. 
 
Researches on the rhythm of speech. 69 
 
 These figures indicate : ( i ) That, while the differences in the results 
 of experiments may be partly explained by differences and inconstancy 
 in the amount of the stimuli and in the physical and mental conditions 
 of the observers (due to fatigue, inattention, irrelevant movements, sen- 
 sations and memory images, passive or active attitude, etc.), the natural 
 period of attention will differ for the different senses and for different 
 individuals. The period for the more purely sensory (receptive) waves 
 may also differ from the more purely motor (voluntary) waves ; and those 
 of involuntary subjective grouping from those of voluntary grouping. 
 
 (2) That the span of the rhythmic sense is definitely limited for each 
 individual, though we do not yet know its precise upper or lower limits, 
 nor how it differs in voluntary and involuntary subjective rhythmisation, 
 in its sensory and motor aspects, nor its dependence on the psycho- 
 physical conditions and age of the individual. 
 
 (3) That the regularity of intervals will be greater for those that fall 
 within the span than for those that do not. This will be illustrated 
 presently. 
 
 Rhythmical signification of irregularity of intervals. The definition of 
 rhythm entails a limiting conception, as regards both the span and regu- 
 ularity of intervals. What does the concept of rhythm require ? Abso- 
 lute or relative conformity to a standard ? How great a variation may 
 intervals admit of without becoming unrhythmical, or non-periodic ? 
 
 A series of intervals of an average duration of 4.62 s (J.M.T. ) mani- 
 festly will admit of more absolute variation than one of 1.15" (E.W.S. ). 
 Hence the amount of variation expressed as a percentage of the average 
 duration (the relative immediate probable error) yields the index of 
 regularity. How small must this be to preserve the rhythmical unity of 
 intervals ? 
 
 This question emphasizes the distinction between hap-hazard recur- 
 rences and strict periodicities between coordinated (that is, rhythmical) 
 sequences, and uncoordinated (that is, unrhythmical) sequences. It is a 
 mistake to subsume all recurrences 1 of maxima and minima, all repeti- 
 tious undulations and oscillations, under the concept of rhythm, and 
 treat literary, social, geological, et aL, phenomena as rhythmical. Prior 
 to legitimately making such sweeping subsumptions a measure or limit to 
 the irregularity for any series of occurrences must first be adopted. This 
 involves the adoption of a criterion of irregularity. 
 
 It has been established, by facts admitting of profuse illustration,* that 
 things come and go in waves and periods ; that all motion is oscillatory 
 
 1 SPENCER, First Principles, X, The Rhythm of Motion, New York 1883. 
 2 FisKE, Outlines of Cosmic Philosophy, II. 
 
70 J. E. Wallace Wallin, 
 
 and not uniform, either in the sense of continuing at precisely the same 
 rate of motion, or invariably in the same direction. But undulations that 
 are not well coordinated periodicities are mere recurrences. 
 
 The limit of irregularity cannot be fixed a priori. It may be deter- 
 mined by taking an easily rhythmisable series of intervals and increasing 
 the inequality until the subject is incapable of rhythmising them. This 
 practical test is based upon the fact that the feeling of rhythm is non- 
 existent for me, until it becomes an affection of my sensibility. The only 
 rhythm which can become an object of direct experience is that of a suc- 
 cession of psychoses. Hence the ultimate criterion is to be found in the 
 sensibility of beings rhythmically constituted. 
 
 In the present investigation the limits may be approximately deter- 
 mined in units of time by ascertaining the average amount of inequality of 
 those records the intervals of which the ear is incapable of coordinating. 
 
 Each record of Table VIII. was listened to a number of times, and the 
 attempt made to coordinate the sequences of impressions representing the 
 verses. The reproducing speed was relatively the same as the recording 
 speed. Appeal was then made, where the rhythm was obscure, to the 
 motor aspect of the rhythmical sense. 
 
 With reference to the more purely motor aspect of the attempted group- 
 ing, it was difficult to impart a very distinct unity to the verses of the J. 
 W. R. and A. D. B. records, unless the lines were repeated rapidly and 
 the terminal pauses were considerably prolonged. The rhythm then be- 
 came marked. The verses of the W. W. record with this device, were 
 capable of being only vaguely coordinated and grasped as totalities. 
 These verses contain twelve syllables. 
 
 Listening to the records on the phonograph, it was difficult to co- 
 ordinate, and grasp as rhythmical wholes, several of the verses of the 
 J. W. R., and most of the W. W. record. In the A. D. B. records it 
 was possible to feel a certain verse rhythm ; in the A. R. P. record, it 
 was more distinct ; in the three remaining records it was marked, espe- 
 cially in the H. O. and in the last stanza of the E. W. S. record. In 
 general, other things being equal, the shorter verses were more unitary 
 and rhythmical. 
 
 Bearing upon the limit of inequality and span of the rhythmic sensibility. 
 These results suggest that when the deviation rises to a fraction of 1 5 % 
 of the length of the whole interval, the mind finds it hard to rhythmise 
 the impressions ; when the fraction is 1 2 % the rhythm is still a little 
 vague ; and when it is 7 % the intervals are easily coordinated. When 
 the fraction is smaller, of course, the rhythm becomes increasingly distinct. 
 
 As regards units of time, the rhythm of the verse intervals was almost 
 
Researches on the rhythm of speech. 71 
 
 destroyed when the deviation reached a third of a second ; it was some- 
 what disturbed when it reached a fifth of a second ; and when it was about 
 one-tenth of a second, it was well preserved. 
 
 From the measurements of the irregularity of the verse intervals, it is 
 difficult to determine the maximum length of an interval that can be 
 synthesized into a rhythmical whole. The longest span where the rhythm 
 was felt, though vaguely, was 2.41"; except in the Persian record, where 
 the line meter rhythm was very distinctly felt and the intervals averaged 
 2.45 s . Here the absolute irregularity and relative irregularity reached the 
 low figures of o.o6 s and 0.02 respectively. That the Persian line group 
 should manifest a relatively high degree of regularity was to be expected 
 apriori from the Persian theory of poetry. In Japanese poetry, which 
 theoretically follows the same law with respect to its meter, the irregularity 
 is 0.06" for J. K. (av. length of verses = o.87 8 )and 0.09" for I. M. (av.= 
 0.95"), and the fraction is 0.07 and 0.09 respectively. While, therefore, 
 the verse meter of both these languages conforms with the requirements, as 
 respects degree of regularity, of rhythm, the records show that English 
 poetry, which theoretically does not recognize the verse meter, except as 
 secondary, affords specimens which surpass the Japanese in regularity and 
 which, if we include the record of mechanical scansion without regulative 
 concomitant where the figures are o.o6 s and 0.03, are practically equal to 
 the Persian. No other intervals in the records of the oriental languages 
 are equal to the regularity of the verse intervals. (In the Persian record 
 the expiration and the verse intervals coincide. ) In English the simple 
 centroid intervals are frequently as regular. 
 
 The long limit of the rhythmic sense for verse intervals, themselves 
 containing minor intervals (rhythms), probably does not generally ex- 
 ceed two and a half seconds. Perhaps we may say that those of less 
 than one and three-fourths of a second in length (with inter-verse pauses 
 of average length) give for most subjects the best results rhythmically 
 (1.67* = average length of verses in the most rhythmical records, viz., 
 E. W. S., H. O., O. S., A. R. P., J. K., I. M. and K. H. K.). This 
 figure is considerably above MEUMANN'S, MARTIUS'S and TITCHENER'S 
 results, and slightly below SMITH'S, for subjective accentuation ; consid- 
 erably above SCRIPTURE'S results for volitions, and below PACE'S, LANGE'S 
 and TITCHENER'S results for the fluctuation of sensations. Hence the 
 results of this experiment may apply only to verse intervals. This span 
 should not be identified offhand with the spans (a) between the oscilla- 
 tions of attention for sensations, memory-images, illusions and simple 
 voluntary efforts, and (<) between the accents in voluntary and involun- 
 tary subjective rhythmisation. 
 
72 J. E. Wallace Wallin, 
 
 Results. The verse interval rarely exceeds 5" in length, nor does it fall 
 much below 0.89" (a range of a little over 4 s ). 
 
 About 60 per cent, of the verse intervals that deviate most from the 
 length of the average are extremes of excess. The balance are extremes 
 of deficiency. 
 
 The average duration of a verse interval is about 2.69 s . ^ varies 
 according to the character of the line and will lie for a given number of 
 reciters, somewhere between approximately 2.05" and 4.62", a range of 
 
 3-47 s - 
 
 In units of time the verse intervals of a given number of recited verses 
 will vary from the length of the average by about 0.27". 
 
 The range of irregularity existing between a given number of recited 
 verses of a given number of subjects may be approximately 0.46* (from 
 o.o9 9 to -55 8 )- 
 
 As a fraction of the average, the irregularity of a given number of 
 recited verses of a given number of subjects will be about 9%; and the 
 range about 10% (from 5% to 15%). 
 
 The regularity is never precisely the same for different records of the 
 same person. The difference will rise to about 5%. 
 
 Whenever the amount of irregularity exceeds 1 5 per cent, of the average 
 duration, the mind finds it hard to rhythmise the series. The rhythm is 
 obscure. Apparently when the percentage does not rise above 10%, the 
 rhythm is fairly well preserved. 
 
 If we accept, provisionally, 10 per cent, of deviation as the limit of 
 allowable inequality in good rhythm, 59 per cent, of the records of verse 
 intervals are rhythmical and 41 per cent, are not. 
 
 The nature of the unity of the verse interval. Assuming that a 
 spoken line of poetry does not exceed the limits of the grasp of the 
 rhythmic sense, and that it is x rhythmically coordinated with other lines, 
 it will stand forth as a unity in consciousness. Every verse is so far forth 
 an absolute unity, whether its components are homogeneous or not, /. e . , 
 made up entirely of sounds, or predominantly of sounds and partly of 
 pauses. Psychologically a rhythmical line is a mental synthesis, regard- 
 less of its structure. 
 
 As a physical event, however, to constitute a unity the elements of 
 speech must make an uninterrupted continuum. The unity of a verse is 
 coterminous with homogeneity of process. The sounds constitute one 
 order of homogeneity ; the pauses, another. Whenever the two inter- 
 mingle the spoken verse physically becomes heterogeneous, i. e., non- 
 unitary. Hence to be a physical, or physiological, unity, a verse must 
 contain no sectional pauses. Such a verse unity, however, would, if 
 
F 
 Researches on the rhythm of 
 
 frequently repeated, become thread-bare and monotonous, although it 
 would result in the best quality*of rhythm. 
 
 Examples of verse intervals, that are physically homogeneous unities, 
 maybe seen by referring to the I. M., J. K., K. H. K., E. W. S. and 
 H. O. records of V., B, above. 
 
 A physically homogeneous unity will coincide with a psychological 
 unity when its limits are so coordinated as to coincide with the rhythmic 
 beats of the latter. 
 
 The principles of the rhythmical coordination of verse intervals. The 
 principle according to which the records were differentiated as rhyth- 
 mical and non-rhythmical, was that of time-coordination. The inter- 
 vals formed a series of rhythmical intervals only provided the law of 
 rhythm, as regards amount of inequality and length of span was observed. 
 This principle, however, was found incapable of comprehending a com- 
 plete explanation of the totality of the experience. To account fully for 
 all cases of verse rhythm, it is necessary to invoke another principle, 
 that of emphasis. This is best exemplified in sing-song. 
 
 6. Sing-song scansion. The table of measurements justifies the belief 
 that the verses of the O. S. record (reading scansion) should give rise to 
 the most distinct feeling of rhythmical unities. This is counter to the 
 judgment of the ear, however, which found it easiest to grasp and retain 
 in consciousness as rhythmical unities the first three verses, respectively, 
 of the two stanzas of the H. O. record, and the last stanza of the E. W. 
 S. record. These appeared completely unified sequences, not because the 
 coordination impressed the ear as perfect, but because the principle of 
 emphasis rendered the unity distinct in consciousness. 
 
 This unifying principle is more or less operative in all the records, 
 but in the record of sing-song it follows a unique law, which may be 
 expressed as an exact correspondence line for line, of certain centroids, 
 as regards distribution and degree of intensity. When the major cen- 
 troids are correlated vertically, that is, placed on the numerically cor- 
 respondent syllables of each line the type of scansion differs from read- 
 ing, doggerel and routine scansion, which at their best, merely attain a 
 maximum regularity, in its strict observance of this law of emphasis. This 
 species of scansion we may designate sing-song. It is this which pro- 
 duces the most vivid impression of verse unities. No verse rhythm can, 
 psychologically considered, be compared with the sing-song. 
 
 To illustrate : in the E. W. S. record (last stanza), the beginning of 
 every line received a specially strong accent upon the corresponding syl- 
 lables. The accentuation was uniformly of the third degree, except in the 
 last line. The stress of the last syllables of the verses, was uniformly sub- 
 
74 / E. Wallace Wallin, 
 
 ordinated to the stress of the first syllables. The intervening syllables, 
 with one exception, were subordinate to those at the end. The predomi- 
 nant elements of the centroids were intensity for those at the close, and 
 pitch and intensity for those at the beginning of the lines. 
 
 In the H. O. record the same general law is exhibited. The second 
 degree centroid was placed, however, on the last syllable but one. This 
 did not affect the sing-song swing. This record also differed in having 
 a sectional pause between the two chief centroids. In both, however, the 
 pauses at the end of the verses served to support the unity of the lines. 
 
 Conclusions. These are suggested by the foregoing. Sing-song scan- 
 sion is peculiar in being largely a matter of verse interval rhythm. 
 
 This peculiarity consists (a) in paying a strict regard to the temporal 
 coordination of the intervals and (b) in observing a law of distribution 
 and subordination of centroids. In respect to distribution, the law seems 
 to require two strong centroids to the verse, placed on the corresponding 
 syllables of each line, the one near the beginning, the other near the close. 
 Hence in the distribution of the sing-song centroids, there is a balanced 
 line arrangement. 
 
 In respect to subordination, there is a like correlated arrangement, line 
 for line. But in each line, the centroids are unevenly balanced. It 
 seems to open with a very strong explosion of the cords in which the ele- 
 ment of pitch is prominent, is followed by a weak one and ends up with a 
 fairly strong explosion in which the element of pitch is less prominent. 
 For different speakers the distribution of the second and third degree 
 centroids may be reversed. 
 
 It has been customary to explain the peculiarity of sing-song scansion 
 as merely a phenomenon of exact rhythm. 1 To fully account for it as an 
 occurrence in consciousness, however, a twofold principle needs be in- 
 voked. 
 
 Logical or reading scansion and doggerel and routine scansion observe 
 no such principle, and the rhythm of the centroid intervals (cf. V., F., 
 below) in reading scansion is almost invariably, and in doggerel scansion 
 generally, less regular. Routine has the presumption of greater regularity 
 than either doggerel or sing-song scansion, as may be seen by a com- 
 parison of the records. 
 
 Sing-song scansion might figuratively be called pendulum -swing scan- 
 sion, where the height of the two swings differ somewhat, and routine or 
 mechanical scansion, metronome scansion, where the strength or reach of 
 the two swings is equal. In the latter kind, where the motive is merely 
 to rigorously coordinate the distances between centroids, the function of 
 
 1 ELLIS, The Quantitative Pronunciation of Latin, London 1874. 
 
Researches on the rhythm of speech. 75 
 
 the sectional and terminal pauses becomes compensatory. They must be 
 strictly limited. 
 
 Doggerel scansion is the most nearly like reading scansion, which is 
 rhythmically free. The latter differs from the former in that it pays more 
 heed to the logic or thought contained in the verses. When the logical 
 and metrical accents coincide it tends to become doggerel. The two often 
 coincide, as is instanced by the first six stanzas of the E. W. S., and the 
 stanza of the A. R. P. records. 
 
 D. Expiration {or sound) intervals. 
 
 i. Explanation. SWEET'S statement that the only divisions which are 
 actually made in language are those into ' ' breath groups, ' ' each group 
 forming a closed system, in which, contrary to general opinion, there is 
 no pause, as it stands, vague and unqualified, must be rejected. It does, 
 however, express an important truth and may, with proper limitations, be 
 accepted. 
 
 It is obvious that speech, as a physiological process, cannot continue 
 indefinitely so as to form a perfect continuum. The vocal cords are in- 
 capacitated from functioning in an unbroken continuity, partly because 
 such activity is antagonistic to the necessary process of alternate waste 
 and repair, partly because the capacity of the air supply in the lungs is 
 definitely limited for each individual, and partly because the specific func- 
 tion of speech, the expression of thought, could not be thus subserved. 
 Hence the need of making divisions into sounds and pauses is inescapable. 
 
 In the sense, therefore, of a succession of physical or physiological 
 processes in time, the only speech divisions are those into filled and 
 vacant groups. These represent periods of expiration and inhalation. 
 While physiologically they are indisputably the only divisions of speech, 
 psychologically, other divisions obtain. 
 
 It is both true and false that within each expiration group no gaps 
 occur. In the sense in which the word silence is used on p. 35 above, it 
 is false. It has been shown that practically every sound is separated from 
 every other by a gap, which, however, is generally so minute as to be 
 imperceptible by the naked ear. The average duration is only o.io". 
 Hence, if by pauses is not meant the necessary silences which intervene 
 between separate sounds, but those gaps which, in the main purposeful, 
 separate groups of words by silences other than those involved in the 
 mere production of a series of sounds, it is true that the intervals in 
 question constitute a closed system. 
 
 By expiration interval, then, is indicated a sequence of sounds limited 
 by pauses. The following tables give the results of the measurement of 
 
76 J. E. Wallace Wallin, 
 
 these intervals. The measurements extend through the entire length of 
 the sequences. Only the lengths of the different sound intervals are 
 given : not the sound intervals and pauses taken together. Every sound 
 interval is invariably convertible into an expiration interval, since the 
 outgoing of breath is the indispensable condition of producing vocal 
 sounds. The pause intervals, contrariwise, are not invariably convertible 
 into inhalation intervals. It requires less time to inhale a quantity of air 
 than it takes to consume it in speaking. And oftentimes the quantity in- 
 haled is so oppressive or excessive as to require exhaling during the pause 
 interval, as is frequently seen in persons of a nervous disposition. 
 
 2. Tables of measurements of expiration intervals. 
 
 TABLE XIII. 
 (Poetry.} 
 
 Subject. 
 
 a 
 
 P 
 
 /- 
 a 
 
 e 
 
 c n 
 
 J. W. R. 
 
 2.60 
 
 0.98 
 
 0-37 
 
 + 4-77 
 0.48 
 
 + 27 
 
 E. W. S. 
 
 1.18 
 
 0.22 
 
 0.18 
 
 + 2.31 
 
 0.41 
 
 4- 35 
 
 A. D. B. A 
 
 0-75 
 
 0.28 
 
 o-37 
 
 + 1-73 
 0.14 
 
 4- i5 
 
 A. D. B. B 
 
 1. 00 
 
 0-37 
 
 0-37 
 
 + 2.16 
 0.25 
 
 4- 8 
 
 W. W. 
 
 i. 06 
 
 0-37 
 
 0-34 
 
 -, 2.62 
 o 18 
 
 4- 38 
 
 A. R. P. 
 
 1.26 
 
 0.45 
 
 0-35 
 
 4-1.90 
 
 O.I2 
 
 5 
 
 J. K. 
 
 2.15 
 
 0-35 
 
 o. 16 
 
 
 O 2 
 
 I. M. 
 
 2-39 
 
 o-33 
 
 0.13 
 
 
 2 
 
 K. H. K. 
 
 2-45 
 
 0.06 
 
 O.O2 
 
 + 2.54 
 
 4 
 
 
 
 
 
 2.31 
 
 
 H. 0. 
 
 1.41 
 
 0.30 
 
 O.2I 
 
 + 2. 02 
 
 0.65 
 
 ii 
 
 0. S. 
 
 1.87 
 
 O.I2 
 
 0.06 
 
 + 2.12 
 
 T CO 
 
 7 
 
 
 
 
 
 I -53 
 
 
 E. H. T. A 
 
 1.72 
 
 0.87 
 
 0.50 
 
 4-3-45 
 0-54 
 
 4- 4 
 
 E. H. T.B 
 
 1.68 
 
 O.I 3 
 
 o 07 
 
 + 2.00 
 -1.56 
 
 4- 4 
 
 Av. 
 
 1.65 
 
 0-37 
 
 0.24 
 
 
 - 
 
 
 
 TABLE 
 
 XIV. 
 
 
 
 ( Prose. ) 
 
 Subject. 
 
 a 
 
 P 
 
 I 
 
 a 
 
 e 
 
 c n 
 
 G. A. A. 
 
 1-37 
 
 0.42 
 
 0.30 
 
 + 3.00 
 0.32 
 
 + 21 
 
 W. L. P. 
 
 1.25 
 
 0.19 
 
 0.15 
 
 4-1-90 
 
 0.12 
 
 + 20 
 
 C. 0. S. 
 
 1.50 
 
 0.72 
 
 0.48 
 
 4-3-42 
 0.15 
 
 4- 21 
 
 Av. 
 
 i-37 
 
 o.44 
 
 0. 3 I 
 
 
 
Researches on the rhythm of speech. 
 
 77 
 
 TABLE XV. 
 ( Poetry read as prose. ) 
 
 Subject. a 
 
 C. O. 0.93 
 
 W. C. 1.09 
 
 s. 
 
 Av. 
 
 1. 12 
 
 1.04 
 
 0.36 
 
 0.37 
 
 0-39 
 oTs7 
 
 0.38 
 o-33 
 0-34 
 o^35 
 
 Subject. a 
 
 C. O. 0.85 
 W. C. 0.90 
 
 TABLE XVI. 
 ( Poetry read as poetry. ) 
 
 P 
 
 S. 
 Av. 
 
 1. 08 
 
 o-35 
 0.32 
 0.50 
 o-39 
 
 a 
 0.41 
 
 o-35 
 0.40 
 0.40 
 
 +3-00 
 
 + 
 
 Subject. 
 
 G. F. A. 
 
 B. S. G. 
 
 Av. 
 
 Subject. 
 
 S. I. 
 
 O. E. S. 
 
 Av. 
 
 TABLE XVII. 
 
 {Poetry read as prose without puuctuation marks.) 
 P 
 
 1. 01 
 
 0.90 
 o-95 
 
 / 
 0-37 
 0.30 
 
 -33 
 
 0.36 TO 
 
 -33 *; ; 
 
 0.34 
 
 a 
 
 .1.07 
 1.40 
 1.23 
 
 TABLE XVIII. 
 ( Prose read as poetry. ) 
 / 
 
 0-34 
 0.66 
 0.50 
 
 a 
 
 0.31 
 0.47 
 o-39 
 
 17 
 
 12 
 
 12 
 
 n 
 16 
 16 
 
 TABLE XIX. 
 ( Table of summaries of expiration intervals of speech. ) 
 
 Tables. a p 
 
 XIII. 1.65 0.37 0.24 
 
 XIV. 1.37 0.44 0.31 
 XV. 1.04 0.37 0.35 
 
78 /. E. Wallace Wallin, 
 
 XVI. 
 
 0.94 
 
 o-39 
 
 0.40 
 
 XVII. 
 
 o-95 
 
 0-33 
 
 0-34 
 
 XVIII. 
 
 I 23 
 
 0.50 
 
 o.39 
 
 Av. 
 
 1.19 
 
 0.40 
 
 0-34 
 
 Unit of measurement, I'; a, average duration; /, immediate probable error; 
 , relative immediate probable error ; e, extreme intervals ; c, character of greater 
 extreme. 
 
 3. Observations. The study is based on 362 measurements. 
 
 Extreme intervals. Of the 362 intervals the longest is 4.77*; the 
 shortest, 0.12"; a range of 4.65 s . 
 
 Character of greater extreme. 72 per cent, of the intervals which devi- 
 ate the farthest from the average are extremes of excess, 19 per cent, of 
 deficiency, and 8 per cent, are equal. The tendency in about three- 
 fourths of the cases is to lengthen unduly, rather than shorten, the expira- 
 tion intervals. 
 
 Average duration (i) For the series. The average for the 362 inter- 
 vals is 1. 19 3 . 
 
 (2) For the sets. The longest average of any set of records is 1.65". 
 Table XIII. , poetry; the shortest, 0.94", Table XVI., poetry read as 
 poetry. This gives a range of o. 7 I s . 
 
 The average of the set of prose records, Table XIV., is 1.3 7 s . This is 
 o.28 3 less than for the corresponding set of records of poetry, Table XIII. 
 
 The set for poetry read as prose without punctuation marks, Table 
 XVIII. , affords the second shortest interval, 0.95". It is o.oi s longer than 
 for poetry read as poetry. The number of intervals is about the same in 
 both tables. 
 
 The reading of the poetical passage as prose, Table XV. , test records, 
 resulted in lengthening the interval. The difference is o.io 8 . This is in 
 disagreement with the result noted above, that the expiration interval in 
 reading poetry is longer than in reading prose. The discrepancy may be 
 explained by the circumstance that the passage when read as prose was 
 unfamiliar to the subject. This would entail a retarded utterance. 
 
 The intervals of the test records where prose was read as poetry, Table 
 XVIII. , are longer than in any of the other three test records. 
 
 (3) For the records. The highest average duration of any single record 
 is 2.60", J. W. R., Table XIII. , reading scansion; the lowest, 0.75"; 
 Table XIII., A. D. B. A , poetry ; a range of 1.85*. The range for the 
 records is 2.60 times larger than for the sets ; and the range for the indi- 
 vidual intervals is 2.51 times larger than for the records. 
 
 The greatest difference between the averages of two different records of 
 the same subject is o.i9 s ; the smallest, 0.04" ; a range of o. 15*. 
 
Researches on the rhythm of speech. 79 
 
 Regularity ( i ) For the series. The average inequality for the series 
 of records is 0.40". This is 34% of length of the average. 
 
 (2) For the sets. The highest degree of irregularity is 40%, Table 
 XVI., poetry read as poetry; the lowest, 24%, Table XIIL, different 
 types of scansion of poetry ; a range of 1 6 % 
 
 In units of time, the greatest irregularity is 0.50" (Table XVIII.) ; the 
 lowest, o.33 s (Table XVII.) ; a range of o.iy 3 . 
 
 The irregularity was 5% smaller in the records where poetry was 
 read as prose than in the records of the same passages read as poetry. 
 The regularity for poetry read as prose without punctuation marks is 
 greater than either ; it is unexpectedly high. 
 
 The regularity for the records of scanned poetry is noticeably greater 
 than for any other set. 
 
 ( 3 ) For the records. The greatest absolute irregularity of any individual 
 record, J. W. R., Table XIIL, poetry, is 0.98", or over one-third of the 
 average length of the intervals ; the lowest, ignoring the Persian record 
 where the verse and expiration groups are coincident, is o. 12 s , O. S., Table 
 XIIL, routine scansion ; a range of o.86 s . This is five times the range 
 existing between the sets of records. 
 
 The highest amount of relative irregularity is 50% ; the lowest 6% ; a 
 range of 44 % . This range is 2 ^ times the range existing between the 
 sets. 
 
 In the selections of the test read as prose and poetry, the amount of 
 irregularity is higher, both absolutely and relatively, for the poetic ren- 
 dering. This is at variance with Tables XIIL and XIV. A like discrep- 
 ancy was noted above between the set of test records and those of prose 
 and poetry. The discrepancy calls for explanation. May we assume the 
 existence of a special talent for reading poetry, distinct from the ability 
 to properly read prose ? Some persons discourse eloquently on scientific 
 themes, but make a botch of reciting poetry. 
 
 The greatest difference in the different records recited by the same 
 person is 6% ; the smallest, 2% ; a range of 4%, which is exceedingly 
 small. 
 
 4. Results. The length of the expiration interval of ordinary speech 
 may be presumed to lie somewhere within the limits of about o. 12 s 
 and 4. 7 7 s . In impassioned oratory the upper limit will evidently be 
 overreached ; the lower limit is relatively fixed (average unemphatic 
 syllable = o.n 8 ). 
 
 About 72 per cent, of the expiration intervals of speech which 
 deviate the farthest from the normal or average interval are extremes of 
 excess ; about 19 per cent., of deficiency ; and about 8 per cent, are equal. 
 
8o /. E. Wallace Wallin, 
 
 The length of the average expiration interval of speech will be approxi- 
 mately ii s . 
 
 The difference in the length of the average expiration intervals of or- 
 dinary speech of different speakers is limited to about i.85 s . 
 
 As between the length of the average expiration interval of prose and 
 poetry, the evidence suggests that the interval is slightly longer in the 
 declamation of poetry. The explanation suggested is, (a) that the in- 
 tervals in poetry tend to become unities coterminous with the verses (the 
 average verse interval is about 2.26 times longer than the average expira- 
 tion interval) ; and (b) that the more lyrical and impetuous emotions of 
 poetic diction result in prolonging the intervals. The latter circumstance 
 should obtain equally in impassioned prose. 
 
 The average length for the same person reciting the same lines on dif- 
 ferent occasions is not the same. The difference varies roughly from 0.04* 
 (smallest difference) to 0.19' (largest difference). 
 
 The inequality in the length of successive expiration intervals will be 
 equal to about one-third of the length of the average. 
 
 In units of time, the range in the inequality of the expiration inter- 
 vals of different speakers will be about o.86 s (from o.i2 s to 0.98*). 
 
 As a fraction of the average, the range of the inequality for different 
 speakers will be about 44% (from 6% to 50%). 
 
 The regularity for the same person reciting the same passages on dif- 
 ferent occasions is not the same. It varies from something like 2% 
 to 6%. 
 
 As between prose and poetry, the evidence suggests that the regularity 
 is slightly higher for the latter. 
 
 The elimination of punctuation marks does not seem to increase the 
 irregularity. 
 
 Accepting provisionally 10 per cent, as the criterion, the coordination 
 of the lengths of expiration intervals of speech is rhythmical in 12 per 
 cent, of the records. 
 
 5. Interpretation of results. What is the psychological import of the 
 above results? With what laws of mental precedure can, for example, 
 the brevity of the expiration group be correlated ? If we accept as true the 
 doctrine of the periodic character of attention, 1 w r e find here a fitting illus- 
 tration of its applicability to speech. Attention, subjectively considered, 
 is detention in consciousness of sensations and other psychical phenomena 
 that, through special fixation, have been exalted to a high degree of inten- 
 sity, vividness and completeness. But this process of fixating the mind is 
 of the nature of a wave or pulsation : it consists in an alternation of max- 
 
 1 LADD, Psychology, Descriptive and Explanatory, 71 fl, New York 1896. 
 
Researches on the rhythm of speech. 8 1 
 
 imal and minimal efforts. The period of these attentive, efforts has been 
 variously given (cf. VI. , C, 3, above). The average sound group easily 
 falls within the extremes. Manifestly, were the laws of attention disre- 
 garded in the expiration intervals of speech, the result would be a special 
 strain upon the power of attention, resulting in a sacrifice of interest, in 
 fatigue and diminished mental grasp. The laws of mental activity seem 
 to demand that the words in speaking be grouped into short unities that 
 agree, in the main, with the ' ' unity of consciousness ' ' ; and that frequent, 
 though brief, pauses be made to enable the mind to easily grasp and syn- 
 thesize this manifold of sensation. 
 
 All this is important as affects the doctrine of conservation of psychical 
 not speaking now of physiological energy. The mind conserves its 
 energy by dividing its work into brief, but frequent and forceful, efforts, 
 and by introducing brief and frequent, rather than long and infrequent, 
 rests. This demand is unconsciously fulfilled in spoken language. The 
 frequent pauses afford momentary opportunities for rest for both speaker 
 and listener. Hence the possibility of listening attentively to a long dis- 
 course. . A speaker who easily tires himself and his audience, and fails to 
 rise to the full measure of his possibilities, will find here, it would seem, 
 a suggestion. 
 
 It appears from our tables that the average duration of the intervals is 
 less than our average wave of attention (i ^ s ). This suggests an explana- 
 tion of the phenomenon that the extremes of excess are vastly more 
 numerous than the extremes of deficiency. On this supposition, we should 
 expect to find, with an average that is rather in excess of the wave of 
 attention, a greater number of minus extremes. The theory is suggestive, 
 and is offered as an explanatory hypothesis that seems plausible and con- 
 sistent with known laws of mental functioning. 
 
 The high degree of irregularity exhibited in the expiration intervals 
 makes it evident that a supplementary principle is needed to complete the 
 theory advanced above, namely the principle of change or variety. 
 While the laws of mental behavior require the duration to be limited 
 they do not require invariable uniformity of length within the limits. 
 Such uniformity would result in intolerable monotony, and would be an- 
 tagonistic of the essential rhythm of speech, the centroid rhythm. Hence 
 the high degree of inequality displayed in the expiration intervals may 
 conduce to restive and pleasing variety. 
 
 The numerical relation of syllables to the normal and maximal expira- 
 tion intervals is shown in VI., 5, below. 
 
82 /. E. Wallace Wallin, 
 
 E. Vacant intervals. 
 
 i. Explanation. A vacant interval in speech is the sequence which 
 intervenes between the adjacent termini of two consecutive expiration 
 intervals. The interval is entirely void of sound. In contra distinction 
 to the gaps which separate different components of a sound or expiration 
 interval, these vacancies separate different sound intervals themselves. 
 They may be called pauses, every variety of which occurring in uttered 
 language is included in the term. In music the terms rest and stop are 
 most frequently used. 
 
 The measurements extend from the end of the last or vanishing sound 
 of the one interval, to the beginning of the first or appearing sound of the 
 following interval. Only the length of successive pauses are compared. 
 The rhythmic character of the alternation of vacant intervals and expira- 
 tion intervals (roughly the rhythmic action of the lungs in speaking) 
 can be obtained by taking both as unities of a series, finding the sum of 
 the measurements of both, and the probable error, according to the for- 
 mula on p. 39. 
 
 2. Measurements of vacant intervals. 
 TABLE XX. 
 
 (Poetry.} 
 
 4- 25 
 
 + 28 
 
 4- 14 
 
 + 7 
 
 4- 37 
 
 f 4 
 
 + 3 
 
 -f 10 
 
 4- 7 
 3 
 
 4- 3 
 
 Subject. 
 
 a 
 
 P 
 
 t 
 
 t 
 
 
 
 
 a 
 
 
 J. W. R. 
 
 0.63 
 
 0.19 
 
 0.30 
 
 + I.IO 
 
 0.18 
 
 E. W. S. 
 
 0-43 
 
 O.I I 
 
 0.25 
 
 + 0.87 
 0.14 
 
 A. D. B. A 
 
 0.48 
 
 0.23 
 
 0.47 
 
 -f- o. 16 
 
 0.12 
 
 A. D. B.B 
 
 0.52 
 
 0.19 
 
 0.36 
 
 f 1.16 
 0.31 
 
 W. W. 
 
 0-57 
 
 0.15 
 
 0.26 
 
 -f 1. 06 
 0.16 
 
 A. R. P. 
 
 0.46 
 
 0.17 
 
 0.36 
 
 + 0.82 
 
 0.20 
 
 K. H. K. 
 
 0.54 
 
 O.O6 
 
 O.I I 
 
 -|-o.66 
 0.45 
 
 H. 0. 
 
 0.46 
 
 0.13 
 
 0.28 
 
 -l-o.So 
 
 0.20 
 
 O. S. 
 
 0.44 
 
 0.09 
 
 0.20 
 
 H-0.62 
 0.28 
 
 E. H. T. A 
 
 0.21 
 
 0.01 
 
 O.O4 
 
 
 E. H. T.B 
 
 0.21 
 
 0.017 
 
 0.08 
 
 4-0.25 
 
 0.18 
 
 Av. 
 
 0-45 
 
 0.12 
 
 0.24 
 
 
Researches on the rhythm of speech. 
 
 TABLE XXI. 
 
 (Prose.) 
 
 Subject. 
 
 a p e c 
 a 
 
 n 
 
 G. A. A. 
 
 0.57 0.14 0.24 -f 
 
 20 
 
 C. 0. S. 
 
 0.40 0.16 0.40 +-97 i 
 0.17 
 
 19 
 
 W. L. P. 
 
 07_ 2 0*8 038 +1% + 
 
 10 
 
 Av. 
 
 0.56 O.I9 O.34 
 
 
 
 TABLE XXII. 
 
 
 
 (Poetry read as prose. ) 
 
 
 Subject. 
 
 p t 
 
 n 
 
 J. M. T. 
 
 0.39 0.12 0.30 l^io 
 
 H 
 
 C. 0. 
 
 0.43 0.13 0.30 -L. 
 
 16 
 
 W. C. 
 
 0.71 0.28 0.39 I-2 5 
 
 II 
 
 S. 
 
 0.67 O.22 0.32 -|- 
 
 II 
 
 Av. 
 
 0.55 0.18 0.32 
 
 
 
 TABLE XXIII. 
 
 
 
 (Poetry read as poetry. ) 
 
 
 Subject. 
 
 fl 
 
 a p e c 
 a 
 
 n 
 
 J. M. T. 
 
 0.62 0.15 0.24 
 
 14 
 
 C. 0. 
 
 0.44 0.13 0.29 +73 4- 
 
 16 
 
 W. C. 
 
 0.68 0.22 0.32 + 1 ' 20 4- 
 
 12 
 
 
 O.22 
 
 
 S. 
 
 0.62 0.18 0.29 4-I-I2 4. 
 
 14 
 
 Av. 
 
 0.59 0.17 0.28 
 
 
 
 TABLE XXIV. 
 
 
 (Poetry read as prose without punctuation marks.) 
 
 Subject. 
 
 P 
 a p e c 
 a 
 
 n 
 
 J. M. T. 
 
 0.35 o.io 0.28 4- 
 
 17 
 
 G. F. A. 
 
 0.47 0.15 0.31 iJJ + 
 
 15 
 
 B. S. G. 
 
 0.44 0.15 0.34 ;*7 + 
 
 15 
 
 Av. 
 
 0.42 0.13 0.31 
 
 
84 /. E. Wallace Wallin, 
 
 TABLE XXV. 
 ( Prose read as poetry. ) 
 
 Subject. 
 
 a 
 
 p 
 
 e 
 
 
 
 a 
 
 
 S. I. 
 
 0.40 
 
 0.14 0.35 
 
 ^17 
 
 0. E. S. 
 
 0-43 
 
 0.09 O.2O 
 
 f^ si * 
 
 
 
 
 0.25 
 
 Av. 
 
 0.41 
 
 o.i i 0.27 
 
 
 
 
 TABLE XXVI. 
 
 
 . 
 
 
 {Summaries.} 
 
 
 
 Tables. 
 
 a p 
 
 t 
 a 
 
 
 XX. 
 
 0.45 0.12 
 
 0.24 
 
 
 XXI. 
 
 0.56 0.19 
 
 0-34 
 
 
 XXII. 
 
 0.55 0.18 
 
 0.32 
 
 > 
 
 XXXIII. 
 
 0.59 0.17 
 
 0.28 
 
 
 XXIV. 
 
 0.42 0.13 
 
 0.31 
 
 
 XXV. 
 
 0.41 o.n 
 
 0.27 
 
 Av. 0.44 0.15 0.29 
 
 Unit of measurement, I s ; a, average duration ; p, immediate probable error ; 
 
 P 
 
 , relative immediate probable error ; e, extreme intervals ; c, character of greater 
 
 extreme ; n, number of intervals. 
 
 3. Observations. The tables afford a study of 238 pauses. 
 
 Extreme intervals. The W. L. P. record, colloquial prose, Table 
 XXI., contains the longest individual interval, of the entire series, 1.45"; 
 The J. M. T. record, Table XXII., poetry read as prose, the shortest, 
 o.io 8 ; a span of 1.35". 
 
 The second longest is 1.3 2 s , poetry read as prose. Where poetry was 
 read as prose without punctuation marks, the length of the longest inter- 
 val does not exceed 0.87". Several intervals occupy 0.12". 
 
 Doubtlessly the longest extreme, especially in conversation, is frequently 
 exceeded. These measurements do not include stops, such, for example, 
 as are made in lecturing to students taking notes, in collecting one's wits, 
 in deliberating, etc. 
 
 Character of greater extreme. 88 per cent, of the extremes are +, 
 12 per cent. , and o per cent. o. (The first are extremes the difference 
 between which and the average is greater than the difference between the 
 average and the shortest extreme ; the second are extremes the difference 
 between which and the average is greater than the difference between the 
 average and the longest extremes ; in the third case the extremes of excess 
 and deficiency are equal. ) 
 
 Average duration. (i) For the series. The average length of the 
 238 pauses, Table XXVI., is o.44 s . 
 
Researches on the rhythm of speech. 85 
 
 (2) For the sets. The highest average is 0.59% Table XXIII. , poetry 
 read as poetry, which is followed by 0.56*, Table XXL, prose ; the low- 
 est, 0.41", Table XXV., prose read as poetry; a range of o.i8 s . The 
 average for the set read without punctuation marks is o.42 s . 
 
 The average for the table of poetry is o.n 8 less than for the table of 
 prose, whereas in Table XXIII., poetry read as poetry, it is o.o4 s greater 
 than in Table XXII. , poetry read as prose. An examination of the four 
 tables suggests that the discrepancy might be due to the range in the 
 character of the records of Tables XX. and XXI. The former contains 
 the shortest average interval, in the record of routine scansion ; the latter, 
 the longest, in the record of colloquial prose. Eliminating these extremes, 
 the averages will very nearly coincide. 
 
 The duration of the pauses in the set of records of the passage read 
 without punctuation marks is o.o8 s shorter than for the pauses of the same 
 passages reads as prose and poetry with the punctuation marks. The 
 average number of pauses for the former set, however, is 15.6, for the 
 latter, 13.5, a difference of 2.1. The highest number in any one record 
 of the sets, 17, is contained in the Browning passage without the marks; 
 the lowest number, 1 1 , in the Tennyson passage read as prose with the 
 marks. In the records of the same passage (Tennyson) without the 
 marks the number is 15. 
 
 (3) For the records. The highest average for any individual record is 
 o.ya 8 , W. P. L., Table XXI., prose, which is followed by 0.71', W. C., 
 Table XXII., poetry read as prose, and o.68 s ,W. C., Table XXIII. , poetry 
 read as prose ; the lowest is 0.21% E. H. T.A and B., Table XX., routine 
 scansion, which is followed by 0.35% J. M. T., Table XXIV., poetry read 
 as prose without punctuation marks; a range of o.5i s . The range for 
 the records is 2.73 times the range for the sets ; and the range for the in- 
 dividual intervals is 2.64 times the range for the average intervals of 
 the records. In the passages that were read by the same subject on two 
 different occasions the highest difference is 0.23" (J. M. T. ); the lowest, 
 o.oi 3 (C. O.); a range of 0.22'. 
 
 Absolute irregularity. (i) For the series. The irregularity in the co- 
 ordination of the duration of the intervals of the series, Table XXVI., is 
 o.is 8 . 
 
 (2) For the sets. The highest amount is 0.19% Table XXI., prose; 
 the lowest, o.n 3 ; Table XXV., prose read as poetry; a range of 
 o.o8 s . 
 
 The average for Table XXIII., poetry read as poetry, is o.oi 8 smaller 
 than for Table XXII., the same passages read as prose. Table XXIV., 
 where the punctuation marks were left out, is more regular than either. 
 
86 /. E. Wallace Wallin, 
 
 The poetry records, Table XX., are 0.07* more regular than the prose 
 records, Table XXL 
 
 (3) For the records. The highest is 0.28", W. L. P., Table XXL, 
 prose, and W. C., Table XXII. , poetry read as prose ; the lowest, o.oi 8 , 
 Table XX., E. H. T.A, routine scansion ; a range of 0.27'. This is ap- 
 proximately three and one -half times the range for the sets. 
 
 Relative irregularity. (i) For the series. The average is 29%. 
 Table XXVI. 
 
 (2) For the sets. Table XXL, prose, contains the highest percentage 
 of irregularity, 34%; Table XX., different types of scansion, the lowest, 
 24% ; a range of 10%. 
 
 The inequality for the set where poetry was read as poetry, Table 
 XXIII. , is 4% smaller than for the sets where poetry was read as prose. 
 The reading of the passages without the punctuation marks is i % more 
 regular than the former. The pauses in Table XX., poetry, are 10% 
 more regular than in Table XXL , prose. 
 
 (3) For the records. The largest amount of irregularity in any record 
 is 47%, A. D. B., Table XX.; the smallest, 4% ; making the extent of 
 the range 43%. This is 4.30 times the range existing between the dif- 
 ferent sets of speech pauses. 
 
 The largest difference in the regularity of the pauses of the different 
 records of the same person is 7%, W. C., the lowest, i%, C. O.; a 
 range of 6%. 
 
 The irregularity in the Tennyson stanzas of Tables XXII. and XIII. , is 
 greater than in the Browning stanza. In Table XXII. the difference is 
 5%, and in Table XXIII. , 4%. This result appears anomalous. 
 Metrically the verses of Tennyson are superior to the verses of Browning, 
 both as regards the alternation of centroids and the scarcity of sectional 
 pauses. The explanation, probably, is that the subjects W. C. and S. are 
 more irregular in their pauses than J. M. T. and C. O. 
 
 4. Results. The range of duration for the pauses of speech was ap- 
 proximately 1.35" (from o.io s to 1.45"). This should hold for ordinary 
 pauses. 
 
 Of the greater extreme deviations of pauses from the length of the 
 average pause, approximately 88 % are extremes of excess, 12 % extremes 
 of deficiency, and o % equal extremes. 
 
 The duration of an average pause in speech is approximately o.44 s . 
 
 As between prose and poetry, the difference in the average duration is 
 inconsiderable. This was apriori to be expected. The amount of rest 
 required in both cases is the same. The pause is a physiological and 
 psychological resting period. 
 
Researches on the rhythm of speech. 87 
 
 The range in the duration of the average pauses of a given number of 
 reciters will perhaps be about y^ (from o.2i s to o-72 8 ). 
 
 The elimination of punctuation marks slightly increases the number, 
 and decreases the average duration of pauses. 
 
 The duration of the average pause of the speech of the same person 
 on different occasions is never exactly the same. It will vary ordinarily 
 from about o.oi s (smallest difference) to 0.23" (largest difference). 
 
 The longest and shortest averages are found, apparently, in colloquial 
 prose and routine scansion, respectively. 
 
 The average irregularity in the length of pauses in speech is equal to 
 approximately 0.15% or about 30% of the average duration. 
 
 In respect to units of time, the amount of irregularity in the pauses 
 of different speakers will vary within the limits of about o.oi* and 0.28' 
 (a range of 0.27*). 
 
 In respect to the length of the average, the irregularity of the pauses 
 of a given number of persons will be limited to a range of about 43% 
 (from 4% to 47%). 
 
 The coordination of the lengths of pauses is slightly more exact in 
 poetry than in prose. 
 
 The elimination of punctuation marks, apparently, does not tend to 
 increase the inequality of the coordinaton. 
 
 The regularity is most exact in routine scansion. 
 
 With a provisional criterion of 10 per cent, of permissible inequality, 
 approximately 8 per cent, of the records of pauses of speech are rhythmi- 
 cally coordinated in length. 
 
 5. Distribution of the pauses in poetry. Explanation. A sectional 
 pause is a vacant interval that occurs within a verse ; a terminal pause, 
 one that occurs at the end of a verse that is, between two verses of 
 poetry. The former is an intra-verse, the latter an inter-verse, pause. 
 
 The term sectional pause has been applied to the pauses of such poetry 
 as contains intra-verse punctuation marks. A given verse may frequently 
 contain one or more, or none of these. By some writers the term is lim- 
 ited to what is technically designated the ' ' caesura, ' ' which is defined 
 as a mid-verse pause occurring within a foot. In the sense in which the 
 word is here employed it denotes any pause, given in the records, occur- 
 ring within a verse. 
 
 The phrase " end-stopped line " is employed to denote a verse which 
 is separated from the following verse by a punctuation mark. A ' ' run-on 
 line ' ' is one that is not thus separated. In the run-on verses the punctu- 
 ation mark is generally shifted to the first part of the following verse, 
 where the pause should logically occur. The records present a variety 
 of both kinds of verses. 
 
88 /. E. Wallace Wallin, 
 
 The term terminal pause, in preference to the uncouth term end-line 
 pause, will be employed to signify any pause that separates verses, 
 whether run-on or end-stopped. 
 
 We shall tabulate the measurements under these two heads. 
 
 Measurements of 
 
 (a) Sectional pauses. 
 
 TABLE XXVII. 
 
 {Poetry. ) 
 
 A 
 Subject. dap 
 
 
 J 
 
 
 
 j. 
 
 a 
 
 1 
 
 
 
 J 
 
 . W. R. 
 
 3-40 
 
 0.48 
 
 0.16 
 
 0-33 
 
 7 
 
 7 
 
 E. W. S. 
 
 0.16 
 
 0.16 
 
 
 
 i 
 
 o 
 
 A. D. B. 
 
 4-37 
 
 0-33 
 
 O.IO 
 
 0.30 
 
 13 
 
 4 
 
 
 W. W. 
 
 12. 6l 
 
 0-54 
 
 0.15 
 
 0.27 
 
 23 
 
 14 
 
 A. R. P. 
 
 0.20 
 
 0.20 
 
 
 
 i 
 
 
 
 
 H. 0. 
 
 0.77 
 
 0.25 
 
 0.03 
 
 0.12 
 
 3 
 
 o 
 
 
 0. S. 
 
 
 
 
 
 
 
 2 
 
 Av. 
 
 7s8 
 
 0.32 
 
 O.II 
 
 0.25 
 
 Ts 
 
 3-8 
 
 TABLE XXVIII. 
 
 (Poetry read as prose. ) 
 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 [*, 
 
 *, 
 
 
 J. M. T. 
 
 3-90 
 
 0.35 
 
 0.12 
 
 0-34 
 
 ii 
 
 8 
 
 
 C. 0. 
 
 4-85 
 
 0-37 
 
 O.II 
 
 0.29 
 
 13 
 
 8 
 
 
 W. C. 
 
 1.57 
 
 0.31 
 
 O.II 
 
 0-35 
 
 5 
 
 2 
 
 
 S. 
 
 1.92 
 
 0.38 
 
 O.II 
 
 0.28 
 
 5 
 
 2 
 
 
 Av. 
 
 3.06 
 
 o-35 
 
 O.II 
 
 0.31 
 
 s.s 
 
 5 
 
 
 
 
 TABLE 
 
 XXIX. 
 
 
 
 
 ( Poetry read as poetry. ) 
 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 , 
 
 n 2 
 
 
 J. M. T. 
 
 6.22 
 
 0-37 
 
 0.15 
 
 0.40 
 
 ii 
 
 8 
 
 
 C. 0. 
 
 5.16 
 
 o-39 
 
 0.12 
 
 0.30 
 
 13 
 
 8 
 
 
 W. C. 
 
 2.22 
 
 0.44 
 
 0.15 
 
 0.34 
 
 5 
 
 2 
 
 
 S. 
 
 3.80 
 
 0.47 
 
 0.13 
 
 0.27 
 
 8 
 
 2 
 
 
 Av. 
 
 4-35 
 
 0.41 
 
 013 
 
 0.32 
 
 9.2 
 
 5 
 
 
 
 
 TABLE 
 
 XXX. 
 
 
 
 
 ( Prose read as poetry. ) 
 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 n \ 
 
 ", 
 
 
 S. I. 
 
 1.27 
 
 0.25 
 
 0.03 
 
 0.12 
 
 5 
 
 3 
 
 
 0. E. S. 
 
 1.81 
 
 0.36 
 
 0.05 
 
 0.13 
 
 5 
 
 3 
 
 
 Av. 
 
 1-54 
 
 0.30 
 
 0.04 
 
 0.12 
 
 5 
 
 3 
 
Researches on the rhythm of speech. 89 
 
 TABLE XXXI. 
 (Poetry read as prose without punctuation marks. ) 
 
 Sztbject. d a p n 
 
 J. M. T. 
 
 5-45 
 
 0.38 
 
 0.09 
 
 0.23 
 
 14 
 
 G. F. A. 
 
 6.15 
 
 0.47 
 
 0.1$ 
 
 0.31 
 
 13 
 
 B. S. G. 
 
 4.00 
 
 0.50 
 
 1.19 
 
 0.38 
 
 8 
 
 Av. 
 
 5-20 
 
 o-45 
 
 0.14 
 
 0.30 
 
 U. 
 
 TABLE XXXII. 
 (Summaries of sectional pauses of poetry. ) 
 
 Table. dap , 
 
 XXVII. 
 
 3-58 
 
 0.32 
 
 O.II 
 
 0.25 
 
 6.8 
 
 3.8 
 
 XXVIII. 
 
 3.06 
 
 0-35 
 
 O.II 
 
 0.31 
 
 8-5 
 
 5- 
 
 XXIX. 
 
 4-35 
 
 0.41 
 
 0.13 
 
 0.32 
 
 9-2 
 
 5- 
 
 XXX. 
 
 1-54 
 
 0.30 
 
 0.04 
 
 0.12 
 
 5- 
 
 3- 
 
 Av. 
 
 3-13 
 
 0-34 
 
 0.09 
 
 0.25 
 
 7.37 
 
 4.2 
 
 Unit of measurement, I", d, total duration; a, average duration; /, immediate 
 
 )bable error ; -> relativ 
 a 
 
 ber of punctuation marks. 
 
 probable error; -, relative immediate probable error; n lt number of pauses; n v num- 
 
 (b) Terminal pauses. 
 
 TABLE XXXIII. 
 
 (Poetry.} 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 ft, 
 
 2 
 
 J. W. R. 
 
 1 6. 60 
 
 0.77 
 
 0.25 
 
 0.32 
 
 21 
 
 20 
 
 E. W. S. 
 
 12. OO 
 
 0.44 
 
 O.II 
 
 0.25 
 
 27 
 
 22 
 
 A. D. B. 
 
 6.02 
 
 o-75 
 
 0.24 
 
 0.32 
 
 8 
 
 5 
 
 W. W. 
 
 8.64 
 
 0.61 
 
 0.13 
 
 0.21 
 
 14 
 
 13 
 
 H. 0. 
 
 3.87 
 
 o.55 
 
 O.IO 
 
 0.18 
 
 7 
 
 5 
 
 0. S. 
 
 2.93 
 
 0.42 
 
 0.09 
 
 0.21 
 
 7 
 
 5 
 
 A. R. P. 
 
 1.62 
 
 0-54 
 
 0.16 
 
 O.29 
 
 3 
 
 2 
 
 Av. 
 
 7.38 
 
 0.58 
 
 0.15 
 
 0.25 
 
 12.4 
 
 xo.: 
 
 TABLE XXXIV. 
 ( Poetry read as prose. ) 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 *, 
 
 , 
 
 J. M. T. 
 
 1.62 
 
 0.54 
 
 0.06 
 
 O.II 
 
 3 
 
 3 
 
 C. 0. 
 
 2.12 
 
 0.70 
 
 0.05 
 
 0.07 
 
 3 
 
 3 
 
 W. C. 
 
 6-37 
 
 1. 06 
 
 O.IO 
 
 0.09 
 
 6 
 
 6 
 
 S. 
 
 5-52 
 
 0.92 
 
 0.16 
 
 0.17 
 
 6 
 
 6 
 
 Av. 
 
 3-90 
 
 0.80 
 
 0.09 
 
 O.II 
 
 4-5 
 
 4- 
 
9 o /. E. Wallace Wallin, 
 
 TABLE XXXV. 
 ( Poetry read as poetry. ) 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 n \ 
 
 *, 
 
 J. M. T. 
 
 2.31 
 
 o.77 
 
 0.05 
 
 0.06 
 
 3 
 
 3 
 
 C. O. 
 
 2.25 
 
 0.56 
 
 0.15 
 
 0.26 
 
 4 
 
 3 
 
 W. C. 
 
 6.O2 
 
 0.86 
 
 O.2O 
 
 0.23 
 
 7 
 
 6 
 
 S. 
 
 4.95 
 
 0.82 
 
 0.12 
 
 0.14 
 
 6 
 
 6 
 
 Av. 3.88 0.75 0.13 0.17 5 4.5 
 
 TABLE XXXVI. 
 (Prose read as poetry. ) 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 ! 
 
 , 
 
 S. I. 
 
 1-95 
 
 0.65 
 
 0.07 
 
 O.IO 
 
 3 
 
 3 
 
 0. E. S. 
 
 1-75 
 
 0.58 
 
 0.06 
 
 O.IO 
 
 3 
 
 3 
 
 Av. 1.85 0.61 0.06 o.io 3 3 
 
 TABLE XXXVII. 
 (Poetry read as prose without punctuation marks.} 
 
 Subject. d a p n^ n 2 
 
 J. M. T. 0.62 0.20 0.03 0.15 3 3 
 
 G. F. A. 1.02 0.51 0.24 0.47 2 3 
 
 B. S. G. 2.62 0.37 o.io 0.27 7 6 
 
 Av. 1.42 0.36 0.12 0.29 4 4 
 
 TABLE XXXVIII. 
 (Summaries of terminal pauses of poetry.} 
 
 Table. da- p ^ n, n 
 
 
 
 
 
 a 
 
 
 
 XXXIII. 
 
 7.38 
 
 0.58 
 
 0.15 
 
 0.25 
 
 12.4 
 
 10.2 
 
 XXXIV. 
 
 3-90 
 
 0.80 
 
 0.09 
 
 O.II 
 
 4-5 
 
 4-5 
 
 XXXV. 
 
 3.88 
 
 0-75 
 
 0.13 
 
 0.17 
 
 5 
 
 4-5 
 
 XXXVI. 
 
 1.85 
 
 0.61 
 
 0.06 
 
 O.IO 
 
 3 
 
 3 
 
 Av. 4.25 0.68 o.io 0.15 6.22 5.55 
 Unity of measurement, I s . d, total duration ; a, average duration ; /, immediate 
 
 probable error; , relative immediate probable error; n v number of pauses ; 2 , number 
 of punctuation marks. 
 
 Comparison of tables. The study is based on 129 sectional pauses 
 and 73 sectional punctuation marks ; and on 122 terminal pauses and 114 
 terminal punctuation marks. This is exclusive of Tables XXXI. and 
 XXXVII. , poetry read as prose without punctuation marks. 
 
 Numerical relation of pauses to punctuation marks. It is a trite obser- 
 
Researches on the rhythm of speech. 91 
 
 vation, enforced by an examination of the records, that punctuation marks 
 and pauses do not invariably coincide, and that pauses frequently occur 
 irrespective of any visible symbol of punctuation. Furthermore, the 
 tables warrant the assertion that the number of pauses, both sectional and 
 terminal, exceeds the number of punctuation marks. That is to say, 
 while, as the records show, the number of punctuation marks exceeds the 
 number of correlative punctuation-mark pauses, the number of non-punc- 
 tuation-mark pauses is so large that, for poetry in general, the pauses out- 
 number, quite considerably, the marks. 
 
 The proportion of the number of sectional pauses to sectional punctua- 
 tion marks is as 5.37 : 4.2 ; of terminal pauses to terminal punctuation 
 marks, as 6.22 : 5.55. In other words, the pauses occurring within the 
 verses are 1.75 times more frequent than the corresponding punctuation 
 marks; and those following, 1.12 times more frequent than the corre- 
 sponding punctuation marks. 
 
 The highest number of sectional pauses occurring in any record is 23, 
 W. W., Table XXVII. , where there are fourteen punctuation marks for 
 eighteen verses of Byron's Childe Harold. In several records the sectional 
 pause, as well as punctuation mark, is entirely lacking. 
 
 The proportion of the number of terminal punctuation marks to sec- 
 tional punctuation marks is as 5.55 : 4.2, or as 1.33 : i. 
 
 While the punctuation mark is thus, for a given variety of verses, gener- 
 ally largely of inter-verse occurrence, in certain kinds of verse structure 
 it is exclusively a matter of inter-verse occurrence ; yet in other cases the 
 sectional punctuation marks are predominant. (As instances, see W. W. 
 and C. O. records for the latter, and E. W. S., A. R. P. and H. O. records 
 for the former). 
 
 The number of pauses occurring within the verse is proportioned to the 
 number occurring at the end of the verse as 7. 37: 6. 22, or as 1.18 : i. 
 
 The proportion between the number of sectional and terminal pauses is 
 thus smaller than the proportion between the number of sectional and 
 terminal punctuation marks. Whereas the number of sectional punctua- 
 tion marks is smaller than the number of terminal punctuation marks, the 
 number of sectional pauses is greater than the number of terminal pauses. 
 These facts may be interpreted as signifying that a certain artificiality 
 obtains in the distribution of the punctuation marks in verse. The tend- 
 ency is to unwarrantably eliminate the marks from the interior of the line. 
 In reading, the deficiency of punctuation marks is restored by introducing 
 extra sectional pauses. The tendency of the former should be to preserve 
 the rhythm and obscure the thought, of the latter, to disturb the rhythm 
 and elucidate the thought. Both principles find application in the records 
 cited above. 
 
9 2 /. E. Wallace Wallin, 
 
 The number of sectional pauses in Table XXVIII. is smaller than in 
 the corresponding table (XXIX.), where the rendering was purposely 
 poetical. This, possibly, explains why the latter does not sound more 
 distinctly poetical than the former. Where the selections were read 
 without punctuation marks, the number of sectional pauses, relatively to 
 the number of punctuation marks in the corresponding records of Tables 
 XXVIII. and XXIX., was increased. The number of terminal pauses, 
 however, remains, relatively to the number of punctuation marks, the same 
 as in the prose and poetry rendering of the same passage. 
 
 Relative total duration. The average aggregate duration, Table XXXII. , 
 of the sectional pauses sustains the proportion to the terminal pauses 
 (Table XXXVIII.) of 3.13": 4.25% or as 1:1.35. In respect of number 
 of pauses the proportion is reversed namely as 1.18:1. That is, while 
 the number of sectional pauses is approximately 20% higher than the 
 number of terminal pauses, the duration of the sectional pauses is ap- 
 proximately 33% less than the duration of the terminal pauses. The 
 difference in duration is thus larger than the difference in number. 
 
 The most notable exception is afforded by Tables XXXI. and XXVII. , 
 where the proportion of the duration of the sectional pauses to the 
 terminal pauses is as 3.66: i. The selection from Bryon, and the passages 
 from Browning and Tennyson, also differ in this regard, the former 
 (Browning) having the greater mass of silence within, the other (Ten- 
 nyson) without, the verses. This suggests important considerations with 
 reference to the metrical quality of the verses of the two types of poetry. 
 The relatively greater number and aggregate duration of the sectional 
 pauses tends to jeopardize the unity of the verse, and to cause to coalesce 
 rather than segregate as distinct wholes, the successive verses. When the 
 bulk of silence is thrown between, instead of within, the verses, they 
 must needs stand forth in consciousness as unified groups of presentation 
 complexes. 
 
 Relative average duration. The duration of the average sectional 
 pause is proportioned to that of the terminal as o.34 s :o.68 s , or as 
 1:2. In other words, the average terminal pause of a series of struc- 
 turally different verses was precisely twice as long as the average sectional 
 pause. 
 
 . As has been suggested, this subordination of length of the sectional 
 to the terminal pause, is probably the most potent principle of the segre- 
 gation of verses into distinct groups. 
 
 . The records in which the verse intervals most pronouncedly appear as 
 distinct wholes are those in which the terminal pauses are predominant in 
 the number of occurrences and in the quantity of duration (cf. E. W. S., 
 
Researches on the rhythm of speech. 93 
 
 O. S., A. R. P., H. O., and E. H. T. records). In the records in 
 which this principle of subordination is violated, the unity of the verse 
 interval is perceptibly disturbed (cf. W. W. record). When the prin- 
 ciple of subordinating the sectional to the terminal pauses is reversed, 
 the verse loses its character as a distinct unitary group. This is instanced 
 in all the records where the punctuation marks were left out. 
 
 The observance of this principle largely gives the character of poetry to 
 the passages that were read as prose. These passages with doubtful rhythm, 
 sound more like poetry than prose. In fact, from this point of view, 
 they are rather more poetical than the corresponding readings of the selec- 
 tions as poetry. Examination of the two series of records shows the fol- 
 lowing : the average sectional pause is o. o6 s shorter in the table (XXVIII. ) 
 for poetry read as prose than in the table (XXIX. ) for poetry read as poetry ; 
 the terminal pause, o. 05* longer. Naturally, we would expect the sectional 
 pause to be shorter in the distinctly poetical renderings. 
 
 The observance of the rule of subordinating the length of the sectional 
 to terminal pauses in the prose rendering of the set of test record, sug- 
 gests that the structural (organic) and the logical pause arrangement in 
 the verses coincide. Where the selection was printed as prose, no sug- 
 gestions from the verse arrangement of the words could tend to make the 
 terminal pauses longer. They were prolonged, evidently, because the 
 subject felt that at those points in the sequence of words a long pause 
 should naturally because representing to the hearer a large stop or tran- 
 sition in the thought be made. To preserve the integrity of the verse, 
 the larger transitions in thought should be so arranged as to coincide with 
 the terminal pauses. In the records of W. W. and J. M. T. this rule is 
 not observed. The result is that the verses are split up and lose their 
 unity, to retain which they would have to be scanned in such a fashion as 
 to involve a sacrifice of thought. As between the two, the average 
 reader prefers sacrificing rhythmical perfection to impairing the distinct 
 articulation of the thought. 
 
 The introduction of the " caesura " illustrates the same principle. The 
 difference between the length of the caesuras and the terminal pauses in 
 the records is less than the difference between the other sectional pauses 
 and the terminal pauses. The tendency of the caesura is to split the verse 
 up into two unities. The caesural and the terminal punctuation marks 
 should both coincide with naturally large thought transitions, to preserve 
 the unity of the verse interval. 
 
 The subordination of pauses in verse finds its parallel in prose. The 
 pauses between sentences are longer than those contained within sentences. 
 The average for all kinds of pauses in prose (Table XXL) is 0.56'. The 
 
94 / E. Wallace Wallin, 
 
 average for the pauses at the end of sentences is i.2i s . (The records 
 afford only seven instances.) The relation is as 1:2.16. 
 
 The difference between the two is, that the terminal pauses of poetry 
 are relatively equidistant. In prose the sentence may be of a great variety 
 of lengths ; hence the terminal pause in prose is not nearly so periodically 
 recurrent as in poetry. This constitutes the essential difference, which is 
 so easily heard, between the pauses of prose and poetry. 
 
 All the records are in complete agreement with the principle of pause 
 subordination, except two of the records of Table XXXI., prose without 
 punctuation marks, where the average of the sectional pauses is o. 4 5 s . The 
 pauses of these records, according to the judgment of the ear, are placed 
 with hesitancy and diffidence. 
 
 Regularity. The absolute regularity of the two series differs only by 
 o.oi s . The relative inequality of the terminal pauses is 15%; of the 
 sectional 25%. The terminal pauses are one and two-thirds times as reg- 
 ular as the sectional. 
 
 The regularity is higher in the prose reading of the test than in the 
 corresponding reading of the passages as poetry. 
 
 The lowest inequality for any set of records of sectional pauses is 12%, 
 Table XXX., prose read as poetry, the highest, 32%, Table XXIX., 
 poetry read as poetry ; a range of 20%. 
 
 The lowest inequality for the terminal pauses is 10%, Table XXXVI., 
 prose read as poetry; the highest 29%, Table XXXVII. , poetry read as 
 prose without punctuation marks ; a range of 19%. 
 
 The lowest degree of irregularity for any record of sectional pauses is 
 12%, records of H. O. and S. I; the highest, 40%, J. M. T. ; a range 
 of 28%. 
 
 For terminal pauses the lowest is 7%, C. O. ; the largest 47%, G. F. 
 A., the record without the punctuation marks ; a range of 40%. 
 
 The range for the individual records of sectional pauses is 8% larger 
 than the range for the sets ; and, of the terminal pauses, 21% larger. 
 
 Results 
 
 The number of pauses of all kinds in recited poetry is generally greater 
 than the number of punctuation marks. 
 
 The number of punctuation -mark pauses is frequently if not generally 
 slightly less than the number of punctuation marks, in recited poetry. 
 
 The proportion of the number of pauses to punctuation marks varies 
 according as they are sectional or terminal. In respect to the sec- 
 tional pauses, there were approximately 7 for every 4 punctuation marks ; 
 and for the terminal pauses, approximately 10 for every 9 punctuation 
 marks. 
 
Researches on the rhythm of speech. 95 
 
 The terminal punctuation marks for a varied collection of verses seem to 
 be about ip$ times the frequency of the sectional punctuation marks. 
 
 The frequency of sectional pauses, for a variety of verses, is approx- 
 imately i^ times that of terminal pauses. Most punctuation marks, for a 
 variety of verses, seem to be terminal ; most pauses, sectional. 
 
 The combined length of the terminal pauses is approximately ife times 
 that of the sectional pauses. In some verses, however, the aggregate 
 length of the latter may be approximately 3^3 times that of the former. 
 
 The average terminal pause in a variety of verses is about twice as long 
 as the average sectional pause. 
 
 The subordination of sectional to terminal pauses helps largely to unify 
 the verses into distinct wholes. 
 
 The terminal pauses of poetry occur, with a high degree of regularity, 
 at relatively equidistant points ; those of prose, at indefinite distances. 
 The length of sentences in prose admits of nominal coordination ; of 
 verses in poetry, of a high degree of coordination. The latter may, 
 and often do, become rhythmical ; the former never or rarely do. 
 
 Removing punctuation marks reverses the rule of subordination : the 
 terminal pauses become subordinate to the sectional. The latter are 
 increased in both length and frequency when compared with records of 
 the same selections with the marks ; the former remain relatively the same. 
 This reversal tends to destroy the unity of verse intervals. 
 
 The irregularity of the terminal pauses of poetry is about 15% of the 
 average duration; of the sectional, 25%. The former are about ij/z 
 times more regular than the latter. This suggests that sectional pauses 
 are a more important disturbing factor of the rhythm of verses than 
 terminal. 
 
 The irregularity of the sectional pauses for different persons ranges 
 from about 12% to 40% (28%) ; of the terminal pauses, from about 
 7% to 47% (40%) of the average. 
 
 Accepting 10% as the limit of inequality for rhythmical units, none 
 of the records of sectional pauses, and only 25% of terminal pauses, are 
 rhythmically coordinated in length. 
 
 6. The time value of punctuation marks. 
 
 Two divergent views have obtained : Complete lawlessness reigns 
 among the different punctuation marks with respect to their relative 
 time value ; and, they may be arranged in a scale of values, whether or 
 not the temporal and logical values correspond. 
 
 The tables are arranged in two groups : punctuation-mark pauses, and 
 non-punctuation-mark pauses. The former correspond, in the records, 
 
9 6 /. E. Wallace Wallin, 
 
 with the symbols which occur between the same words in print. These 
 symbols were directly perceived, perhaps vaguely, as the selections were 
 read ; and were pictured perhaps, consciously or unconsciously, to the 
 imagination, when the words were recited without looking at the printed 
 page. 
 
 The term non-punctuation-mark pauses, however uncouth, indicates 
 pauses in the records between the words of which there are no punctua- 
 tion marks upon the printed page. These pauses are of two kinds : 
 those purely accidental, and those conducive to emphasis. The latter 
 may be emotional or logical. These add clearness and emphasis to the 
 thought. This is not true of the fortuitous ; they tend rather to ob- 
 cure. In the table the division into logical and emotional is disre- 
 garded. A punctuation mark may frequently be inserted where they oc- 
 cur. The division into accidental and emphasis has been made by the 
 judgment of the experimenter's ear, on the basis of the context or the 
 mode of delivery. Terminal non-punctuation-mark pauses have gener- 
 ally been regarded as emphasis pauses. 
 
 i 
 
 Measurements of punctuation-mark pauses. 
 TABLE XXXIX. 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 t 
 
 a 
 
 n 
 
 J. W. R. 
 
 1.79 
 
 0.89 
 
 0.03 
 
 0.03 
 
 2 
 
 E. W. S. 
 
 2.52 
 
 0.63 
 
 O.I I 
 
 0.17 
 
 4 
 
 A. D. B. 
 
 2-33 
 
 I.I6 
 
 0.00 
 
 0.00 
 
 2 
 
 A. R. P. 
 
 1.20 
 
 0.60 
 
 0.21 
 
 o-35 
 
 2 
 
 G. A. A. 
 
 0-57 
 
 0.57 
 
 
 
 I 
 
 W. L. P. 
 
 1.16 
 
 1.16 
 
 
 
 I 
 
 C. O. S. 
 
 2.27 
 
 0-75 
 
 0. 12 
 
 0.16 
 
 3 
 
 W. C. 
 
 2.27 
 
 1.13 
 
 0.05 
 
 0.04 
 
 2 
 
 S. 
 
 1.80 
 
 0.90 
 
 0.16 
 
 0.17 
 
 3 
 
 S. I. 
 
 1.42 
 
 0.71 
 
 0.00 
 
 0.00 
 
 2 
 
 0. E. S. 
 
 1.28 
 
 0.64 
 
 o.oo 
 
 0.00 
 
 2 
 
 W. C. A 
 
 2-47 
 
 1.23 
 
 0.00 
 
 o.oo 
 
 2 
 
 S.A 
 
 1.77 
 
 0.88 
 
 0.08 
 
 0.09 
 
 2 
 
 Total 
 
 22.85 
 
 
 
 
 28 
 
 Av. 0.86 0.07 0.09 
 
Researches on the rhythm of speech. 07 
 
 TABLE XL. 
 
 Subject. 
 
 d 
 
 a 
 
 W. L. P. 
 
 1-45 
 
 1-45 
 
 G. A. A. 
 
 3-72 
 
 0-93 
 
 C. O. S. 
 
 0.90 
 
 0.90 
 
 O. S. 
 
 0.62 
 
 0.62 
 
 H. 0. 
 
 0.82 
 
 0.82 
 
 C. 0. 
 
 0.62 
 
 0.62 
 
 J. M. T. 
 
 0.87 
 
 0.87 
 
 C. O. A 
 
 0.64 
 
 0.64 
 
 Total 
 
 9.64 
 
 
 Av. 
 
 
 0.85 
 
 TABLE XLI. 
 
 Or) 
 
 TABLE XLII. 
 
 CO 
 
 I 
 
 a 
 o.io 
 
 
 ii 
 
 O.IO 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 L 
 
 n 
 
 
 
 
 
 a 
 
 
 W. W. 
 
 1.66 
 
 0.83 
 
 o.oo 
 
 o.oo 
 
 2 
 
 W. L. P. 
 
 1. 12 
 
 1. 12 
 
 
 
 1 
 
 J. M. T. 
 
 i-45 
 
 0.72 
 
 0.03 
 
 0.04 
 
 2 
 
 C. 0. 
 
 1. 06 
 
 o-53 
 
 o.oo 
 
 0.00 
 
 2 
 
 Total 
 
 5-29 
 
 
 
 
 7 
 
 Av. 
 
 
 0.80 
 
 
 
 
 Subjtct. dap 
 
 J. W. R. i.io i.io I 
 
 W. W. 5.37 0.76 0.05 0.06 7 
 
 W. L. P. 1.87 0.93 0.58 0.62 2 
 
 J. M. T. 1.64 0.54 0.19 0.35 3 
 
 C. O..B 2. 02 0.67 0.03 0.04 3 
 
 C. O. 4 1.85 0.61 0.14 0.22 3 
 
 Total 13-85 19 
 
 Av. 0.77 0.19 0.26 
 
9 8 
 
 /. E. Wallace Wallin, 
 TABLE XLIII. 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 t 
 
 a 
 
 n 
 
 J. W. R. 
 
 10.29 
 
 0.68 
 
 0.19 
 
 0.28 
 
 15 
 
 E. W. S. 
 
 5-43 
 
 0.49 
 
 0.02 
 
 0.04 
 
 II 
 
 A. D. B. 
 
 4.64 
 
 0.66 
 
 o. 16 
 
 0.24 
 
 7 
 
 W. W. 
 
 7.56 
 
 0.47 
 
 0. 12 
 
 0.25 
 
 16 
 
 H. 0. 
 
 1.47 
 
 0.49 
 
 0.05 
 
 O.IO 
 
 3 
 
 O. S. 
 
 1.81 
 
 0-45 
 
 O.O9 
 
 O.2O 
 
 4 
 
 G. A. A. 
 
 6. 20 
 
 0.48 
 
 0.09 
 
 0.18 
 
 13 
 
 W. L. P. 
 
 3-77 
 
 0-75 
 
 0.21 
 
 0.28 
 
 5 
 
 C. O. S. 
 
 1.70 
 
 0.42 
 
 0.13 
 
 0.30 
 
 4 
 
 J. M. T. 
 
 2.06 
 
 0.68 
 
 0.06 
 
 0.08 
 
 3 
 
 C. O. 
 
 1. 12 
 
 0.28 
 
 O.O7 
 
 0.25 
 
 4 
 
 W. C. 
 
 4-75 
 
 0.79 
 
 0.13 
 
 0.16 
 
 6 
 
 s. 
 
 4-50 
 
 0-75 
 
 0.13 
 
 0.17 
 
 6 
 
 S. I. 
 
 0.87 
 
 0.29 
 
 O.O3 
 
 O.IO 
 
 3 
 
 0. E. S. 
 
 1. 12 
 
 0-39 
 
 O.02 
 
 0.05 
 
 3 
 
 C. O.A 
 
 2.70 
 
 0-45 
 
 O.O9 
 
 O.2O 
 
 6 
 
 W. C. A 
 
 4.67 
 
 0.77 
 
 0.21 
 
 0.27 
 
 6 
 
 S.A 
 
 4-77 
 
 Q.79 
 
 0.21 
 
 0.26 
 
 6 
 
 Total 
 
 69.43 
 
 
 
 
 12! 
 
 Av. 
 
 
 0.56 
 
 O.I I 
 
 0.18 
 
 
 
 
 TABLE 
 
 XLIV. 
 
 
 
 Subject. 
 
 d 
 
 a p 
 
 . 
 
 a 
 
 E. W. S. 
 
 1-75 
 
 0.43 0.09 
 
 0.20 4 
 
 J. M. T. 
 
 0-57 
 
 0-57 
 
 I 
 
 S. I. 
 
 0.52 
 
 0.52 
 
 I 
 
 0. E. S. 
 
 0.46 
 
 0.46 
 
 I 
 
 C. 0. 
 
 0.66 
 
 0.66 
 
 I 
 
 Total 
 
 3^96 
 
 
 8 
 
 Av. 
 
 
 0.53 0.09 
 
 O.2O 
 
 
 
 TABLE XLV. 
 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 t 
 
 a 
 
 n 
 
 E. W. S. 
 
 0.60 
 
 0.30 
 
 0.06 
 
 0.20 
 
 2 
 
 C. 0. S. 
 
 0.57 
 
 0.28 
 
 0.15 
 
 0-53 
 
 2 
 
 H. 0. 
 
 o-37 
 
 0-37 
 
 
 
 
 W. L. P. 
 
 1. 10 
 
 1. 10 
 
 
 
 
 C. O.B 
 
 0-45 
 
 0.45 
 
 
 
 
 J. M. T. 
 
 0.62 
 
 0.62 
 
 
 
 
 C. 0. 
 
 0.56 
 
 0.56 
 
 
 
 
 Total 
 
 4.27 
 
 
 
 
 9 
 
 Av. 
 
 
 0.52 
 
 O.IO 
 
 0.36 
 
 
Researches on the rhythm of speech. 
 
 99 
 
 TABLE XL VI. 
 
 ( Summaries 
 
 for punctuation-mark 
 
 pauses) 
 
 
 Table. 
 
 d 
 
 a 
 
 P 
 
 t 
 
 a 
 
 n 
 
 XXXIX. (;) 
 
 22.85 
 
 0.86 
 
 0.07 
 
 0.09 
 
 27 
 
 XL. (.) 
 
 9.64 
 
 0.85 
 
 O.IO 
 
 O.IO 
 
 II 
 
 XLI. (,-) 
 
 5.29 
 
 0.80 
 
 
 
 7 
 
 XLII. (!) 
 
 13.85 
 
 0.77 
 
 0.19 
 
 0.26 
 
 19 
 
 XLIII. (,) 
 
 69.43 
 
 0.56 
 
 O.I I 
 
 0.18 
 
 121 
 
 XLIV. ( ) 
 
 3.96 
 
 0-53 
 
 0.09 
 
 O.2O 
 
 8 
 
 XLV. (:) 
 
 4.27 
 
 0.52 
 
 O.IO 
 
 0.36 
 
 9 
 
 Total 
 
 129.29 
 
 
 
 
 202 
 
 Av. 
 
 
 0.70 
 
 O.I I 
 
 0.19 
 
 
 Unit of measurement, I s . 
 
 d, duration. 
 
 a, average duration. 
 
 /, immediate probable error. 
 
 , relative immediate probable error. 
 
 n, number of punctuation marks. 
 
 Measurements of non-punctuation-mark pauses. 
 
 TABLE XLVII. 
 {Emphasis} 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 i 
 
 a 
 
 n 
 
 J. W. R. 
 
 2.14 
 
 0.42 
 
 0.15 
 
 0-35 
 
 5 
 
 E. W. S. 
 
 i. ii 
 
 0.27 
 
 0.07 
 
 0.25 
 
 4 
 
 A. D. B. 
 
 2.45 
 
 0.27 
 
 O.IO 
 
 o.37 
 
 9 
 
 W. W. 
 
 4.07 
 
 6.50 
 
 0.17 
 
 0-34 
 
 8 
 
 A. R. P. 
 
 0.42 
 
 0.42 
 
 
 
 i 
 
 O. S. 
 
 0-34 
 
 0-34 
 
 
 
 i 
 
 H. 0. 
 
 1-95 
 
 0-39 
 
 0.13 
 
 0-33 
 
 5 
 
 W. L. P. 
 
 2.68 
 
 0.44 
 
 0.15 
 
 o-34 
 
 6 
 
 G. A. A. 
 
 0.87 
 
 0-43 
 
 0.03 
 
 007 
 
 2 
 
 C. O. S. 
 
 2.02 
 
 0.25 
 
 0.04 
 
 o. 16 
 
 8 
 
 J. M. T. 
 
 0.58 
 
 0.29 
 
 0. II 
 
 0-37 
 
 2 
 
 C. 0. 
 
 0-95 
 
 0.23 
 
 0.05 
 
 0.21 
 
 4 
 
 W. C. 
 
 0.90 
 
 0.30 
 
 0.05 
 
 0.16 
 
 3 
 
 S. 
 
 1.42 
 
 0.47 
 
 0.16 
 
 0-34 
 
 3 
 
 S. I. 
 
 0.60 
 
 0.30 
 
 0.02 
 
 0.06 
 
 2 
 
 O. E. S. 
 
 o-43 
 
 0-43 
 
 
 
 I 
 
 Total 
 
 22.93 
 
 
 
 
 64" 
 
 Av. 
 
 
 0.36 
 
 0.09 
 
 0.25 
 
 
100 
 
 /. E. Wallace Wallin, 
 
 TABLE XLVIII. 
 (Accidental) 
 
 Subject. 
 
 d 
 
 a 
 
 P 
 
 
 
 a 
 
 n 
 
 J. W. R. 
 
 0.47 
 
 0.23 
 
 O.O6 
 
 0.26 
 
 2 
 
 E. W. 3. 
 
 0.31 
 
 0.15 
 
 0.00 
 
 o.oo 
 
 2 
 
 A. D. B. 
 
 o-93 
 
 0.31 
 
 0.04 
 
 0.12 
 
 3 
 
 W. W. 
 
 0-95 
 
 0.47 
 
 o.oo 
 
 00 
 
 2 
 
 A. R. P. 
 
 0.20 
 
 0.20 
 
 
 
 I 
 
 0. S. 
 
 0.31 
 
 0. 3 I 
 
 
 
 I 
 
 W. L. P. 
 
 0-54 
 
 0.27 
 
 0.00 
 
 o.oo 
 
 2 
 
 C. O. S. 
 
 0.2 7 
 
 O.27 
 
 
 
 I 
 
 J. M. T. 
 
 O.gi 
 
 0.91 
 
 
 
 I 
 
 C. 0. 
 
 1. 00 
 
 0.50 
 
 0.14 
 
 0.28 
 
 2 
 
 W. C. 
 
 0.32 
 
 0.32 
 
 
 
 I 
 
 S. 
 
 1.02 
 
 0-34 
 
 0.15 
 
 0.44 
 
 3 
 
 O. E. S. 
 
 0.25 
 
 0.25 
 
 
 
 i 
 
 Total 
 
 7.48 
 
 
 
 
 22 
 
 Av. 
 
 
 0-34 
 
 0.05 
 
 0.15 
 
 
 TABLE XLIX. 
 (Summaries of non-punctuation-mark pauses'} 
 
 Table. d 
 
 XLVII. (Emphasis) 22.93 
 
 XLVIII. (Accidental) 7.48 
 
 Total 30.41 
 
 Av. 
 
 Unit of measurement, I s . 
 #, average duration. 
 
 , relative immediate probable error. 
 
 a 
 
 0.36 
 0-34 
 
 P 
 
 0.09 
 0.05 
 
 i 
 
 a 
 0.25 
 
 0.15 
 
 n 
 64 
 
 22 
 
 0.35 O.O7 O.2O 
 
 d, total duration. 
 
 /, immediate probable error. 
 
 , number of pauses. 
 
 Results. Two hundred and eighty-eight pauses were measured, 202 
 punctuation -mark and 86 non-punctuation-mark. Unfortunately several 
 of the punctuation-marks occurred so infrequently as to render some of 
 the results entirely provisional. Far more measurements are needed. 
 
 In respect to the frequency of pauses occurring in speech, approxi- 
 mately 75 per cent, are punctuation-mark pauses. The balance are non- 
 punctuation-mark pauses. In the records, 60% of the former, were 
 commas, 13% semicolons, 9% exclamation-marks, 5% periods, 4% 
 colons, 3% dashes and 3% comma-dashes (,-). The records of poetry, 
 although more numerous than those of prose, contain : of semicolons 
 
Researches on the rhythm of speech. 101 
 
 66%, periods 45%, comma-dashes 86%, exclamation-marks 87%, com- 
 mas 72%, colons 62%, and dashes 75%. This may possibly indicate 
 that the periods are more abundant in prose, and the exclamation-marks 
 and comma-dashes in poetry. 
 
 About 75% of the non-punctuation-mark pauses are emphasis pauses. 
 These tend to contribute strength and perspicuity to, and the accidental 
 pauses to detract these from, the thought of the sentences. The ac- 
 cidental pauses, as was to be expected in rational, coherent speech, are 
 very rare (about 7% of all pauses). 
 
 Apparently, over 40% of the pauses of speech are comma pauses. 
 The relative frequency of the different punctuation marks differs some- 
 what for poetry and prose, and for different styles of writing. 
 
 The punctuation-mark pauses in the records occupy about 80 % of the 
 aggregate duration of all kinds of pauses. Of the former, commas 
 occupy 53%; semicolons, 17%; exclamation-marks, 10% ; periods, 
 7%; comma-dashes, 4%; colons, 3%; and dashes 3%. The time 
 subordination corresponds with that of frequency, but the percentages do 
 not exactly correspond. 
 
 The time value of the average punctuation-mark pause is about twice 
 that of the non-punctuation-mark pause (former in records =0.70'; 
 latter =o. 3 5 s ). 
 
 The shortest average punctuation-mark pause (:) is seemingly longer 
 than the longest average non-punctuation-mark pause. The difference 
 was about o.i6 s . Only nine of the former were measured. It may be 
 somewhat uncertain. 
 
 The longest average punctuation-mark pause is considerably longer 
 than the shortest average punctuation -mark pause [about 0.34*; from 
 0.52* (:) to o.86 s (;)] 
 
 The average accidental and emphasis pauses appear to be about equal 
 in length (range, o.o2 s ). 
 
 The duration of an average punctuation-mark pause of spoken poetry 
 and prose is about o.7o 3 . 
 
 Punctuation marks differ in respect to the time which they occupy in 
 speech, both absolutely and relatively. Generally, no two consecutive 
 punctuation marks, of the same or different kind, are exactly coordinated 
 in length. The longest and shortest averages of the tables are as fol- 
 lows : semicolons, 1.23% and 0.57" (range, o.66 s ) ; periods (Table XL.), 
 i.45 s an d o.62 s (range, 0.83") ; comma-dashes (Table XLL), 1.12* and 
 o.53 8 ( ran g e > -59 8 ) ') exclamation-maiks (Table XLII.), i.io 8 and 0.54" 
 (range, 0.56') ; commas (Table XLIII.), o. 79 s and o. 28 s (range, 0.51") ; 
 dashes (Table XLIV. ), o. 66 s and 0.43* (range, o.23 s ) ; and colon pauses 
 
102 /. E. Wallace Wallin, 
 
 i. io s and o.37 s (range/ 0.7 3 s ). The range for the three punctuation- 
 mark pauses of greatest frequency follows the subsumption according to 
 the average duration : it is greatest for the semicolon, and least for the 
 comma, pauses. The difference in the ranges of these is 0.15*. 
 
 The average lengths of different punctuation marks obey a rule of 
 subordination, more or less indefinite in details. In the main, we can 
 arrange them in three groups: (i) the semicolon and period ; (2) the 
 exclamation-point and comma-dash; and (3) the comma, dash and 
 colon. The interrogation point presumably belongs to one of the first 
 two. The approximate value of the first group is o.86 s , of the second, 
 0.78% and the third, 0.5 3 s . No fixed line of demarkation exists be- 
 tween the groups. The difference between (i) and (2) (it is possible 
 to combine them into one) appears less marked than between (2) and (3). 
 
 The regularity of punctuation and non-punctuation-mark pauses is 
 about equal. The average amount of irregularity is approximately o. io s , 
 or 19% of the average duration. 
 
 Apparently, commas are more irregular than semicolons. Exclama- 
 tion-marks are doubtful, owing to extreme irregularity of W. L. P. 
 The frequency of the other punctuation-marks is insufficient to justify 
 any conclusions. 
 
 The range of irregularity for the punctuation -mark pauses of a given 
 number of speakers is, in units of time, about o.ip 8 (fromo.o2 s to o.2i s ); 
 and, as a fraction of the average, about 50% (from 3% to 53%). This 
 is slightly larger than for non-punctuation mark pauses (0.15% or 38%). 
 
 Accepting tentatively the 10% criterion, 43% of the semicolons, 40% 
 of the exclamation-marks, 28% of the commas, 15% of the emphasis 
 pauses and o% of the accidental pauses, are rhythmically coordinated in 
 length. The figures for the comma and emphasis pauses are the most 
 trustworthy. 
 
 The concept of the punctuation mark. Upon the printed page it is a 
 symbol, like any other character of type, meaningless apart from an 
 interpreting mind. Through experience, it acquires a definite connota- 
 tion just like any of the letters. 
 
 Every character upon the page has a twofold significance for speech : 
 it represents a physiological and a psychological process. The latter is 
 conditioned upon a presentation of sense ; and is translated into a defi- 
 nite motor innervation as soon as it is apperceived. When the characters 
 are letters, in isolation or in combination, the apperception of a letter or 
 word expresses itself in a definite adjustment of the larynx, giving rise 
 to a definite sound. 
 
 Similarly, punctuation marks are meaningless presentations until 
 
Researches on the rhythm of speech. 103 
 
 apperceived. They acquire a definite meaning according to the way the 
 contours differ. The significance of any cognate characters depends 
 upon such distinguishing traits of contour. A . and a ;, just as an s and 
 a c, are connected with distinct apperceptions. 
 
 As interpreted presentations (apperceptions), punctuation marks 
 denote breaks or transitions in the continuity of the thought and differ- 
 ences in the quality or character of the pause. A period denotes the 
 termination of a single wave (pulse or oscillation) of thought ; a semi- 
 colon, a ripple in the wave without being a break ; and the comma, a 
 minor ripple. The two latter are a species of subordinate waves com- 
 prehended in the unity of the whole wave. There are several such 
 species of thought-waves, e. g., those of declaration, interrogation, excla- 
 mation. In the main, the different punctuation marks have as fixed a 
 signification for thought as do the letters of the alphabet. 
 
 Finally, as motor resultants they denote physiological processes differing 
 according as the marks and the character of the waves of thought differ. 
 The physiological differences are threefold : (i) As to modulation. The 
 pitch of the voice varies according to the character of the sign. This is a 
 subject for special research. ( 2 ) As to timbre. This is specially noticeable 
 in the bracket and parenthesis (or in the parenthetical commas), question- 
 mark and exclamation-mark. This subject also merits special investiga- 
 tion. (3) As to the pause. This is perhaps the most obvious difference. 
 Punctuation marks, physiologically considered, are pauses, the average 
 length of which varies for the different marks, as has been shown 
 above. 
 
 It is largely upon the basis of these three characteristics, that a listener 
 can punctuate the speech of a lecturer, who is observing in his speech, 
 evidently unconsciously, the laws of the physiology of the punctua- 
 tion mark. 
 
 Hence a given punctuation mark is on the one hand a visible symbol, 
 directly perceived by the eye, or represented to the imagination how- 
 ever vaguely which signifies to the reader a turn or break in the move- 
 ment of thought ; and, on the other, an auditory image which, as a 
 moment in the physiological process of speech, is distinguished by 
 changes in duration of pause, in pitch and quality of sound. In these 
 respects it is a mental item, representing quantitative and qualitative 
 peculiarities. 
 
 F. O roid intervals. 
 
 By a centroid interval is .iderstood the interim, spatial upon the 
 page to the eye, temporal i' -peech to the ear, which stretches between 
 two successive centroids, i ., from centroid a, to centroid b, from cen- 
 
104 / E - Wallace Wallin, 
 
 troid b to centroid c, and so on. Hence every centroid interval in- 
 cludes one centroid. 
 
 According to BRUCKE the centroid interval lies between two "Arsen- 
 gipfeln' ' ; according to MINOR, it begins with the syllable standing in 
 arsis ; and according to MEYER/ when the beginning of the arsis coin- 
 cides with the beginning of a syllable, and the end of the thesis with 
 the end of a syllable, the limits of the foot and syllable coincide. 
 MEUMANN 2 and BOLTON 3 find experimentally that accented sounds oc- 
 cupy the first place in the interval. Our measurements extend as 
 nearly as possible between two successive centroids. 
 
 In music the centroid interval is popularly designated a bar ; in poetry, 
 a foot or a measure ; and in prose it has no distinctive name. 
 
 The centroids of speech and music are, as centroids, one and the 
 same. The intervals between them are also as intervals, identical. All 
 centroids can be subsumed under the laws of centroid composition ; all 
 the intervals, under those of centroid-interval composition. 
 
 Hence the term centroid interval is applicable to all sorts of human 
 utterance poetry, prose and music. No justification exists for drawing 
 hard and fast lines between either the centroids or centroid intervals of 
 these, although slight differences may, and do, obtain between the regu- 
 larity of the intervals and the relation of the elements in the centroids 
 of poetry, prose and music. 
 
 The composition of centroid intervals. Only two kinds of materials can 
 enter into their composition sounds and silences, or syllables and 
 pauses. A centroid interval may be composed purely of a sound- 
 plenum, though not of a pause-plenum. It must always contain a certain 
 quantum of sound. 
 
 Centroid intervals consisting of sounds, or syllables, only may be 
 called sound- centroid intervals ; those of sounds and silences, or syllables 
 and pauses, composite- centroid intervals. 
 
 Sound-centroid intervals may be divided into the following varieties : 
 (i) \-sy liable intervals. Here the centroid syllables are directly con- 
 tiguous; no unemphatic syllable or pause intervenes between them. 
 Hence the interval consists simply of one centroidal syllable. The 
 existence of such intervals has been both denied* and affirmed. 5 The 
 
 1 MEYER, Beitrdge zur deutschen Metrik, Neuere Sprachen, 1898 VI 136-138. 
 
 2 MEUMANN, (Jntersuchungen zur Psychologic und Aesthetik des Rhythmus, Philos. 
 Stud., 1894 X 303. 
 
 3 BOLTON, Rhythm, Am. Jour. Psych., 1893 VI 222. 
 
 4 ABBOTT AND SEELEY, English Lessons for English People, 154, Boston 1880. 
 5 MAYOR, Chapters on English Metre, III, London 1886. 
 
Researches on the rhythm of speech. 105 
 
 records furnish numerous instances of this type. (2) 2-syllable, con- 
 taining an intervening syllable, or one centroidal and one non-centroidal 
 syllable. (3) 3- syllable, composed of one centroidal and two non-cen- 
 troidal syllables. (4) \-syllable, consisting of one centroidal and three 
 non-centroidal syllables. (5) ^-syllable centroid intervals, consisting 
 of one centroidal and four non-centroidal syllables. These are exceed- 
 ingly rare ; the number in the records is too small for purposes of study. 
 GURNEY * holds that no centroid interval can be composed of more than 
 one centroidal and four non-centroidal syllables; and MAYOR, 2 that even 
 three consecutive syllables without a metrical accent is impossible. 
 
 The composite-centroid intervals may be divided as follows: (i) 
 \-pause-i-syllable intervals, consisting of one pause in addition to the 
 centroidal syllable. (2) \-pause-2-syllable, consisting of one pause, one 
 centroidal and one non-centroidal syllable, the^ pause preceding or follow- 
 ing the unemphatic syllable. (3) \-pause-$- syllable, composed of one 
 pause, one centroidal and two non-centroidal syllables. (4) \-pausc- 4- 
 syllable, composed of one pause, one centroidal and three non-centroidal 
 syllables. (5) \-pausc-$-sy liable, containing one pause, one centroidal 
 and four non-centroidal syllables. 
 
 The records furnish instances, too scant for purposes of study, of other 
 modes of composition, such as 2 -pause- 2 -syllable, and 2 -pause- 1 -syllable 
 centroid intervals. 
 
 The lengths of each variety of the two groups of centroid intervals were 
 measured. 
 
 Results. The study is based on 69 i-syllable, 238 2-syllable, 47 
 3-syllable and 12 4-syllable intervals. 
 
 The average duration of the sound-centroid intervals of speech is about 
 o.5i s . This is about the same as Martius' period for subjective rhyth- 
 misation (0.50*). The average for the i -syllable intervals was o.32 s ; for 
 the 2-syllable, 0.44'; for the 3-syllable, 0.62"; for the 4-syllable, o.69 s . 
 
 Any i -syllable interval occurring in speech may be presumed to fall 
 within the limits of about 0.14" and o-56 s (range, o.42 s ); any 2-syllable, 
 within o.i8 s and 0.79" (range, o.6i s ); and any 3-syllable, within o.37 s 
 and o.97 s (range, o.6o s ). (For the 4-syllable interval the measurements 
 are insufficient). Any one of the different varieties may be presumed to 
 fall within 0.14" and 0.97" (range, 0.83"). 
 
 The range for the averages of different records is about as follows : For 
 i -syllable intervals, o.i8 s (from 0.2 5 s to 0.43'); for 2-syllable, 0.17* 
 (from 0.36' to o.53 s ); f r 3-syllable, 0.40" (from o.49 s to 0.89"); and 
 
 1 GURNEY, The Power of Sound, 433, London 1880. 
 
 2 MAYOR, Chapters on English Metre, III, London 1886. 
 
io6 /. E. Wallace Wallin, 
 
 Measurements of sound-centroid intervals. 
 TABLE L. 
 
 i-syllable sound-centroid intervals. 
 
 Subject. 
 
 a 
 
 P 
 
 ^ 
 a 
 
 e c 
 
 n 
 
 J M. T. 
 
 0.28 
 
 0.05 
 
 0.17 
 
 + 0.45 -f 
 
 14 
 
 
 
 
 
 0.14 
 
 
 C. O.B 
 
 0.25 
 
 0.05 
 
 O.2O 
 
 + 0-43 + 
 
 14 
 
 
 
 
 
 0.16 
 
 
 W. W. 
 
 0.31 
 
 0.05 
 
 0.16 
 
 -f 0.47 4- 
 
 17 
 
 
 
 
 
 0.18 
 
 
 E. W. S. 
 
 0.40 
 
 0.03 
 
 0.07 
 
 
 
 2 
 
 A. 1). B. A 
 
 0.27 
 
 
 
 
 2 
 
 G. A. A. 
 
 0.30 
 
 0.05 
 
 0.18 
 
 + 0.47 + 
 
 5 
 
 
 
 
 
 O.22 
 
 
 W. L. P. 
 
 0.29 
 
 0.07 
 
 0.24 
 
 + 0. 4 I 4- 
 
 3 
 
 
 
 
 
 0.22 
 
 
 C. O. S. 
 
 0.36 
 
 0.08 
 
 O.22 
 
 -f- 0.50 
 
 5 
 
 
 
 
 
 0.20 
 
 
 S. I. 
 
 0.38 
 
 0.05 
 
 0.13 
 
 + 0.47 -f 
 
 3 
 
 
 
 
 
 0.32 
 
 
 O. E. S. 
 
 0.43 
 
 0.05 
 
 O. II 
 
 + 0.56 4- 
 
 4 
 
 
 
 
 
 0-34 
 
 
 Av. 
 
 0.32 
 
 0.05 
 
 0.16 
 
 
 
 
 
 
 TABLE LI. 
 
 
 
 2-syllable sound-centroid 
 
 intervals. 
 
 
 Subject. 
 
 a 
 
 P 
 
 I 
 
 a 
 
 e c 
 
 n 
 
 J. M. T. 
 
 0.46 
 
 0.08 
 
 0.17 
 
 + 0.75 4- 
 
 17 
 
 
 
 
 
 0.25 
 
 
 C. O. 
 
 0.36 
 
 0.06 
 
 0.16 
 
 + 0.54 o 
 
 H 
 
 
 
 
 
 0.18 
 
 
 S. I. 
 
 0-43 
 
 0.04 
 
 0.09 
 
 + -5 2 + 
 
 6 
 
 
 
 
 
 0-35 
 
 
 O. E. S. 
 
 0.48 
 
 0.09 
 
 0.18 
 
 -f 0.65 
 
 8 
 
 
 
 
 
 0.21 
 
 
 W. W. 
 
 0.42 
 
 0.07 
 
 0.16 
 
 + 0.66 4- 
 
 42 
 
 
 
 
 
 0.20 
 
 
 A. D. B. A 
 
 0-53 
 
 O.IO 
 
 0.18 
 
 4-0.79 4- 
 
 6 
 
 
 
 
 
 0.31 
 
 
 E. W. S. 
 
 0.44 
 
 
 
 
 60 
 
 G. A. A. 
 
 0.41 
 
 
 
 
 33 
 
 W. L. P. 
 
 0.41 
 
 
 
 
 26 
 
 C. O. S. 
 
 0.52 
 
 
 
 
 26 
 
 Av. 
 
 0.44 
 
 0.07 
 
 0.15 
 
 
 
Researches on the rhythm of speech. 
 
 107 
 
 TABLE LII. 
 
 3 -syllable sound- centroid intervals. 
 
 Subject. 
 
 a 
 
 P 
 
 i 
 
 a 
 
 e 
 
 c n 
 
 W. W. 
 
 0.72 
 
 0.08 
 
 O.II 
 
 + 0.97 
 
 + 5 
 
 
 
 
 
 0.58 
 
 
 E. W. S. 
 
 0.49 
 
 0.04 
 
 0.08 
 
 + 0.56 
 
 12 
 
 
 
 
 
 0.37 
 
 
 A. D. B. A 
 
 0.56 
 
 0.14 
 
 0.25 
 
 + 0.79 
 
 + 3 
 
 
 
 
 
 0.37 
 
 
 G. A. A. 
 
 0.51 
 
 0.05 
 
 0.09 
 
 + 0.62 
 
 10 
 
 
 
 
 
 0.37 
 
 
 W. L. P. 
 
 0.60 
 
 0.08 
 
 0.13 
 
 + 0.70 
 
 4 
 
 
 
 
 
 0.41 
 
 
 c. o. s. 
 
 0.63 
 
 0.08 
 
 0.12 
 
 + 0.75 
 
 8 
 
 
 
 
 
 0-37 
 
 
 S. I. 
 
 0.61 
 
 0.06 
 
 0.09 
 
 + 0.70 
 
 3 
 
 
 
 
 
 0.50 
 
 
 O. E. S. 
 
 0.89 
 
 
 
 
 2 
 
 Av. 
 
 0.62 
 
 0.07 
 
 0.12 
 
 
 
 TABLE LIII. 
 
 
 
 ^.-syllable 
 
 sound- centroid intervals. 
 
 Subject. 
 
 a 
 
 P 
 
 i 
 
 a 
 
 e 
 
 c n 
 
 J. M. T. 
 
 0.72 
 
 
 
 
 I 
 
 C. 0. 
 
 0.60 
 
 
 
 
 I 
 
 G. A. A. 
 
 0.65 
 
 O.O2 
 
 0.03 
 
 + 0.70 
 
 + 3 
 
 
 
 
 
 0.62 
 
 
 W. L. P. 
 
 0.68 
 
 0.08 
 
 O.II 
 
 + 0.77 
 
 3 
 
 
 
 
 
 0-54 
 
 
 C. 0. S. 
 
 0.83 
 
 
 
 
 2 
 
 S. I. 
 
 0.62 
 
 
 
 
 I 
 
 0. E. S. 
 
 0.78 
 
 
 
 
 I 
 
 Av. 
 
 0.69 
 
 0.05 
 
 0.07 
 
 
 
 Unit of measurement, I s . 
 
 a, average duration. 
 
 p, immediate probable error. 
 
 , relative immediate probable error. 
 
 e, extreme intervals. 
 c , character of greater extreme. 
 n, number of intervals. 
 
io8 /. E. Wallace Wallin, 
 
 Measurements of composite- centroid intervals. 
 TABLE LIV. 
 
 i -pause- i-sy liable comp^site-centroid intervals. 
 
 Subject. 
 
 a 
 
 P 
 
 I 
 a 
 
 e c 
 
 n 
 
 E. W. S. 
 
 0.74 
 
 O.II 
 
 0.14 
 
 + I.I2 + 
 
 II 
 
 
 
 
 
 0.52 
 
 
 A. D. B. A 
 
 0.66 
 
 
 
 
 I 
 
 W. W. 
 
 0.98 
 
 0.14 
 
 0.14 
 
 + i-47 + 
 
 II 
 
 
 
 
 
 0.72 
 
 
 J. M. T.c 
 
 0.92 
 
 0.08 
 
 0.08 
 
 + 1.04 
 
 5 
 
 
 
 
 
 0.77- 
 
 
 C. O..JB 
 
 0.64 
 
 0.16 
 
 0.25 
 
 +1.02 + 
 
 6 
 
 
 
 
 
 -0.43 
 
 
 W. L. P. 
 
 1-34 
 
 0.19 
 
 0.14 
 
 + 1-75 + 
 
 ,5 
 
 
 
 
 
 1. 00 
 
 
 C. 0. S. 
 
 o.75 
 
 0.32 
 
 0.42 
 
 
 2 
 
 S. I. 
 
 0.47 
 
 
 
 
 I 
 
 O. E. S. 
 
 0.65 
 
 
 
 
 I 
 
 Av. 
 
 0.68 
 
 0.16 
 
 0.19 
 
 ' 
 
 
 TABLE LV. 
 
 
 i -pause- 
 
 2- syllable 
 
 composite- centroid intervals. 
 
 Subject. 
 
 a 
 
 P 
 
 i 
 
 a 
 
 e c 
 
 n 
 
 E. W. S. 
 
 0.71 
 
 0.08 
 
 O.II 
 
 + 0.89 
 
 13 
 
 
 
 
 
 0.50 
 
 
 A. D. B. 
 
 1.07 
 
 0.28 
 
 0.26 
 
 + 1.68 + 
 
 7 
 
 
 
 
 
 0.70 
 
 
 W. W. 
 
 0.91 
 
 0.17 
 
 0.18 
 
 + 1.66 + 
 
 25 
 
 
 
 
 
 0.56 
 
 
 J.M.T.O 
 
 o-79 
 
 0.15 
 
 0.19 
 
 + 1.04 
 
 7 
 
 
 
 
 
 0.47 
 
 
 C. 0. 
 
 0-75 
 
 0.13 
 
 0.17 
 
 + 1.06 
 
 10 
 
 
 
 
 
 0.43 
 
 
 G. A. A. 
 
 o-95 
 
 0.17 
 
 0.17 
 
 + 1-47 + 
 
 H 
 
 
 
 
 
 0.62 
 
 
 W. L. P. 
 
 1.26 
 
 0.20 
 
 0.15 
 
 + 1.66 + 
 
 7 
 
 
 
 
 
 0.68 
 
 
 C. O. S. 
 
 0.78 
 
 O.2I 
 
 0.26 
 
 + I-45 -f 
 
 ii 
 
 
 
 
 
 0.47 
 
 
 S. I. 
 
 0.82 
 
 
 
 
 2 
 
 O. E. S. 
 
 0.90 
 
 
 
 
 3 
 
 Av. 0.89 0.17 o. 18 
 
Researches on the rhythm of speech. 
 
 109 
 
 TABLE LVI. 
 
 i-pause-j-syllable composite-centroid intervals. 
 
 Subject. 
 
 a 
 
 / e c 
 
 n 
 
 E. W. S. 
 
 0-75 
 
 
 2 
 
 A. D. E. A 
 
 0.82 
 
 0.25 0.30 +1.35 + 
 
 4 
 
 
 
 0-45 
 
 
 W. W. 
 
 1.31 
 
 
 I 
 
 G. A. A. 
 
 0.86 
 
 0.34 0.39 +1-47 + 
 
 3 
 
 
 
 0.50 
 
 
 W. L. P. 
 
 0.90 
 
 0.15 o. 16 +1.29 + 
 
 5 
 
 
 
 0.70 
 
 
 C. O. S. 
 
 0.92 
 
 0.14 0.15 H- 1.25 -}- 
 
 4 
 
 
 
 0.77 
 
 
 S. I. 
 
 0-75 
 
 
 i 
 
 Av. 
 
 0.91 
 
 0.22 0.25 
 
 
 
 
 TABLE LVII. 
 
 
 
 i -pa use -4- 
 
 syllable composite-centroid intervals. 
 
 
 Subject. 
 
 a 
 
 > P a 
 
 n 
 
 G. A. A. 
 
 1.05 
 
 0.15 0.14 -|- 1.30 -{- 
 
 3 
 
 
 
 0.85 
 
 
 W. L. P. 
 
 i-93 
 
 
 i 
 
 S. I. 
 
 -95 
 
 0.21 0.22 
 
 2 
 
 O. E. S. 
 
 1.04 
 
 0.13 0.12 
 
 2 
 
 Av. 
 
 1.24 
 
 O.I 6 O.l6 
 
 
 
 
 TABLE LVIII. 
 
 
 I -pause-j- syllable composite-centroid intervals. 
 
 
 Subject. 
 
 P- 
 a p n 
 a 
 
 
 
 S. L 
 
 I. O2 O.2I O.2O 2 
 
 
 
 O. E. S. 
 
 1.26 0.07 005 2 
 
 
 
 Av. 
 
 I.I4 O.I4 O.I2 
 
 
 Unit of measurement, I s . 
 
 a, average duration. 
 
 p, immediate probable error. 
 
 e, extreme intervals. 
 c, character of greater extreme. 
 n, number of intervals. 
 
 , relative immediate probable error. 
 
no /. E. Wallace Wallin, 
 
 for 4-syllable intervals, 0.23" (from o.6o s to 0.83'). The range appears 
 to be. highest for the average 3- syllable intervals. 
 
 The length of the average sound-centroid interval is invariably pro- 
 portioned to the number of syllables of which it is composed. This is 
 true alike of the reading scansion of poetry and prose. Single intervals 
 frequently depart from this rule. 
 
 The addition of a syllable to sound-centroid intervals increases their 
 average duration about 50%. The addition to the 3 -syllable interval 
 appears to result in the smallest, and to the 2 -syllable, in the largest 
 increment. (Average 2-syllable, 37% longer than i-syllable; 3-syl- 
 lable, 40 ffo longer than 2-syllable; 4-syllable, 11% longer than 3 -syl- 
 lable.) 
 
 The difference between the lengths of the different intervals appears 
 to be very slightly larger in poetry than in prose. 
 
 No sound-centroid intervals are of precisely the same length. The 
 average inequality of the four kinds is about o.o6 s , or a trifle over 12% 
 of the duration. Apparently it is very slightly smaller for the longer 
 than the shorter intervals. 
 
 The extreme deviations of the i -syllable (75% = -}-, u% = , 
 n% = o) and 2-syllable (66% = -j-, 16% = , 16% = o) intervals 
 are predominantly extremes of excess. For the others (for 3 -syllable, 
 28% = -f, 71% = ) the predominant extremes are liable to be those 
 of deficiency. Of all the varieties, 58 % were extremes of excess, 33 % of 
 deficiency, and 9 % were equal. 
 
 The 2-syllable interval is the predominant sound-centroid interval of 
 speech. This is true alike of prose and dactylic and trochaic verse. 
 (For poetry and prose, 65% = 2-syllable ; 18%, i syllable ; 13%, 
 3-syllable; 3%, 4-syllable). The 5-syllable and 4-syllable are the 
 most infrequent. They seem to be less frequent in poetry than in 
 prose. 
 
 According to the criterion of 10% of irregularity, n% of the 
 i -syllable, 16% of the 2-syllable and 43% of the 3-syllable sound-cen- 
 troid intervals were rhythmically coordinated in length. The large per- 
 centage for the, 3-syllable may indicate that centroid intervals of about 
 o.65 s are m o st easily rhythmized. For all varieties of sound-centroid in- 
 tervals the figure is about 25%. 
 
 The study of the composite intervals is based on the measurement of 
 43 i -pause -i -syllable, 99 i-pause-2 -syllable, 20 i -pauses-syllable, 8 
 i -pause-4-syllable and 4 i -pause- 5 -syllable intervals. 
 
 The average duration of the five varieties of these intervals in speech 
 was about 0.9 7 s . 
 
Researches on the rhythm of speech. in 
 
 It seems to be larger in proportion to the number of syllables of which 
 the interval is composed. This rule meets with many variations in the 
 individual records. 
 
 The difference between the averages of all the i -pause- 1- syllable and 
 the i -pause -2 -syllable intervals was o.2i s ; between the latter and the i- 
 pause-3-syllable, 0.02". 
 
 Any single one of the five varieties occurring in ordinary speech may 
 be presumed to fall within the limits of 0.43" and i-93 s (a range of 
 i.5o 8 ). The range for the different varieties decreased progressively 
 as the number of components increased. [Extremes for i -pause- 1- 
 syllable=i.75 s and o.43 s (range, i-32 s ) ; for i-pause 2-syllable=i.66 s 
 and o.43 s ( ran g e > I - 2 3 s )j f r i-pause-3-syllable= 1.47* and 0.45" 
 (range, i.o2 s ) ; for i-pause-4-syllable = 1.93* and 0.85" (range, 
 i.o8 s ).] 
 
 The longest and shortest averages were: for the i -pause- 1 -syllable 
 interval, 1.34" and 0.47" (range, 0.87") ; for the i -pause- 2 -syllable, 
 i.26 s and 0.71" (range, 0.55") ; for the i-pause-3 -sellable, i.3i s and 
 0.75 (range, 0.56"). The different records of the same person, as well 
 as those of different persons, gave a wide range of differences for the 
 lengths of the different averages. 
 
 Approximately 80% of the extremes of the i -pause- 1 -syllable, and 
 62% of the i -pause- 2 -syllable intervals were extremes of excess. The 
 balance were deficiency extremes. Of all the different varieties, 
 about 70% were extremes of excess, 20% of deficiency and 10% of 
 equality. 
 
 No intervals are of equal duration. The average inequality for the 
 five different modes of composition was about 0.17", or 18% of the 
 duration. 
 
 The most frequent mode of composition, both in prose and in poetry, 
 was the i-pause-2 -syllable. It was about 8% more frequent in the records 
 of poetry than in those of prose (52% of all modes). 
 
 The i -pause-5 -syllable, followed by the i-pause-4-syllable, was the most 
 infrequent mode of collocation. Both were less frequent in poetry than in 
 prose. (In records of prose, the former = 4%, the latter 11% ; in those 
 of poetry both = o%.) 
 
 On the basis of the 10% standard, 16% of the i -pause -i -syllable, 
 6% of the i -pause -2 -syllable and 9% of all composite-centroid intervals 
 are rhythmically coordinated. 
 
 Comparison of sound centroid and composite-centroid intervals. The 
 average composite-centroid interval of speech occupies about 1.90 times 
 more time than the sound-centroid interval. 
 
H2 /. E, Wallace Wallin, 
 
 The extremes of both are predominantly extremes of excess, the per- 
 centage for the composite being about 1.20 times higher than for the 
 sound-centroid intervals. 
 
 The range for a given composite interval of speech is likely to be about 
 i. 80 times larger than that for a given sound-centroid interval. 
 
 The latter is relatively one and a half times more regular than the 
 former. 
 
 The difference in the length of the several modes of composition of 
 the two kinds of 'intervals is more pronounced, and less variable, in the 
 case of the sound-centroid intervals. 
 
 In both, the 2-syllable mode of composition (over 50% of all modes) 
 occurs most frequently and the 5- and 4-syllable combination least fre- 
 quently. 
 
 (#) The distribution of the syllables of centroid intervals. Modes of 
 distribution. In the i -syllable interval, the centroid extends throughout 
 its entire length. This interval is customarily called a spondee. The 
 definition of this as " a foot consisting of two equally accented syllables," 
 is based upon the misconception that a centroid interval consists of more 
 than one centroid. When the components of a series of intervals are 
 equally strong (are real centroids as the theory supposes) each cen- 
 troid constitutes a so-called foot : there is no reason for including two 
 centroids in the interval, rather than three or four, or all in a spondaic 
 verse. The spondee of prosody is thus based upon a misconception and 
 a certain arbitrariness. The spondee, to mean anything, should signify 
 a i -syllable centroid interval; and as signifying this, it may be a con- 
 venient term. It is arbitrary and unwarranted to include two centroid 
 intervals in one interval. 
 
 The i -syllable interval is limited to short sequences. In long ones the 
 tendency is to so emphasize certain of the centroids as to make the others 
 subordinate. 
 
 The 2-syllable interval admits of two situations : (i) a non-centroidal 
 followed by a centroidal syllable (the iambus, - -). Here the centroid 
 closes the interval, unlike the next; (2) a centroidal followed by a 
 non-centroidal syllable (the trochee, - -). 
 
 The components of the 3-syllable interval admit of this distribution : 
 (i) two non-centroidal syllables followed by one centroidal (the ana- 
 pest, ~ - -) ; (2) one centroidal followed by two non-centroidal syl- 
 lables (the dactyl, - ~ ~) ; (3) one non-centroidal, one centroidal and 
 one non-centroidal syllable (the amphibrach or amphiambus, ~ -). 
 
 In the 4-syllable interval this distribution is possible: (i) one cen- 
 troidal followed by three non-centroidal syllables; (2) three non-cen- 
 
Researches on the rhythm of speech. i i 3 
 
 troidal syllables followed by one centroidal. This interval appears to be 
 generally the opening interval of a sequence. No names have been 
 applied to these two modes ; (3) one centroidal, two non-centroidal, fol- 
 lowed by one centroidal syllable, (the choriamb, - - -, in ancient pros- 
 ody). This mode is spurious, if the centroidal syllables are genuine 
 centroids. 
 
 Bearing on the conceptions and nomenclature of prosody. Our treatment 
 of the centroid and centroid interval suggests two general lines of criticism. 
 
 First, as long ago pointed out by SHELLEY/ the distinction between 
 " measured" and "unmeasured" speech is unscientific. Both prose 
 and poetry contain centroids and centroid intervals, following the same 
 laws of composition and distribution. It is inadmissible to restrict the 
 application of "measure" or "foot" to poetry alone. Centroid 
 intervals pervade all uttered language. 
 
 Second, as pointed out by GURNEY, the terms "foot," "measure" 
 or "bar" are inapt. "The arrangement of the foot is a mere matter 
 of the eye. Nothing can prevent ictus from being ictus." The funda- 
 mental concept of the centroid interval in speech is that of a unitary 
 whole, juxtaposed to, and coalescent with, other similar wholes, requir- 
 ing a portion of time for the voice to pass over its sounds. These 
 intervals between force centers may be called centroid-intervals, bars, 
 feet, or measures. But, unless interrupted by pauses, no divisions or 
 "bars " exist between them. Nor are centroid-intervals to be conceived 
 as a succession of bars of invariable length, nor as "feet" of similar 
 length, nor as successive quanta of time conforming to an invariable 
 " measure. ' ' Moreover, these terms, besides being customarily restricted 
 to poetry, are applied with different meanings to classical and modern 
 
 poetry. 3 "In English poetry the names of feet denote groups of 
 
 accented and non -accented syllables, without reference to quantity" 
 (ABBOTT and SEELEY). On the whole, the term centroid interval seems 
 preferable. 
 
 Third, a convenient nomenclature for the different modes of distribu- 
 tion of the centroidal and non-centroidal syllables of the intervals is desira- 
 ble. The old terminology of prosody is perhaps associated with mis- 
 conceptions, as already indicated with reference to the spondee and 
 choriamb. 
 
 Fourth, there is no physical distinction between the several types of 
 the different modes of distribution. There may be a mental, or felt, 
 
 SHELLEY, A Defense of Poetry, Works, VII 6, London 1880. 
 
 2 GURNEY, The Power of Sound, 426, London 1880. 
 
 3 ELLIS, The Quantitative Pronunciation of Latin, 5, London i8?4. 
 
ii4 / E. Wallace Wallin, 
 
 difference between the iambus and trochee, and the anapest and dactyl ; 
 but physical measurements of them always extend from centroid to cen- 
 troid, independently of type differences. The modes of distribution are, 
 upon the whole, the same in prose and poetry. Prose, however, observes 
 no systematic arrangement of the different kinds of intervals throughout 
 the sentences in this respect. In poetry, the verses of a given stanza 
 are supposed to consist of intervals of the same mode of distribution, or, 
 if of different modes, to follow an orderly arrangement. 
 
 The succession or recurrence of centroid intervals. As recurrences 
 within sequences, they may be divided into two classes. In the complex 
 centroid intervals are included all the intervals occurring between the 
 first and the last centroid of a given record, irrespective of the composi- 
 tion of the intervals. It includes sound-centroid and composite-centroid 
 intervals. 
 
 Simple-centroid intervals include all those occurring within expiration 
 groups, or all the varieties of sound-centroid intervals. The latter were 
 studied, not as successive members of rhythmical sequences, but as in- 
 tervals, whether in isolation or succession, different in composition and 
 time- value. 
 
 Each expiration group is an uninterrupted sequence of simple-centroid 
 intervals. A pause destroys this continuity. 
 
 A comparison of these two intervals, the one including pauses, the other 
 not, will determine the differentia of language rhythm, the function of the 
 pause, etc. 
 
 Measurements of complex-centroid intervals. 
 TABLE LIX. 
 
 ( Reading scansion, English poetry ) 
 
 + "5 
 
 -f 102 
 
 + 24 
 
 4- 16 
 
 -f 101 
 
 + 15 
 
 Subject. 
 
 a 
 
 P 
 
 r_ 
 
 a 
 
 e 
 
 J. W. R 
 
 0.74 
 
 0.29 
 
 0.38 
 
 1.81 
 
 
 
 
 
 0-33 
 
 E. W. S. 
 
 0.52 
 
 o. 10 
 
 0.19 
 
 -f- 1. 12 
 
 
 
 
 
 0.25 
 
 A. D. B. A 
 
 0-75 
 
 0.25 
 
 0-33 
 
 + 1.68 
 
 
 
 
 
 -0.27 
 
 A. D. B..B 
 
 0.72 
 
 0.16 
 
 O. 22 
 
 + 1-47 
 
 
 
 
 
 0.45 
 
 W. W. 
 
 O.6o 
 
 0.21 
 
 0.35 
 
 -f 1.66 
 
 
 
 
 
 -0.18 
 
 A. R. P. 
 
 0.51 
 
 0.15 
 
 0.29 
 
 -f 1.20 
 
 
 
 
 
 0.27 
 
 Av. 
 
 0.64 
 
 O.I9 
 
 0.29 
 
 
Researches on the rhythm of speech. 115 
 
 TABLE LX. 
 
 {Reading scansion, Japanese poetry} 
 
 Subject. 
 
 a 
 
 , 
 a 
 
 e c 
 
 n 
 
 J. K. 
 
 0.32 
 
 0.06 0.18 
 
 + 0-52 + 
 0.16 
 
 13 
 
 I. M. 
 
 -39 
 
 o. 16 0.42 
 
 o.i 8 
 
 14 
 
 K. H. K. 
 
 0.72 
 
 0.27 0.37 
 
 + 1-54 + 
 
 15 
 
 Av. 
 
 
 
 0.31 
 
 
 047 
 
 0.16 0.32 
 
 
 
 
 
 TABLE LXI. 
 
 
 
 {Sing-song and doggerel scansion) 
 
 Subject. 
 H. 0. 
 
 a 
 0.65 
 
 ' 1 
 
 0.16 0.24 
 
 e c 
 
 + 1-37 + 
 
 n 
 31 
 
 0. S. 
 
 0.68 
 
 0.13 0.19 
 
 0.47 
 
 28 
 
 Av. 
 
 oT66 
 
 
 0.47 
 
 
 
 0.14 0.21 
 
 
 
 
 TABLE LXII. 
 
 
 
 
 
 {Routine scansion) 
 
 
 
 Subject. 
 
 a 
 
 p t 
 
 e c 
 
 n 
 
 E. H. T. A 
 E. H. T.s 
 
 Av. 
 
 0.50 
 
 0.48 
 
 0.07 0.14 
 
 O.O2 O.O4 
 
 + 0.70 
 
 0-31 + 
 + 0.58 
 
 0-43 + 
 
 15 
 
 15 
 
 0.49 
 
 O.O4 O.O9 
 
 
 
 
 TABLE LXIII. 
 
 
 
 (Summary of complex-centroid intervals of poetry) 
 
 
 Table. 
 
 a p 
 
 L 
 
 a 
 
 
 
 LIX. 
 LX. 
 LXI. 
 LXII. 
 
 0.64 0.19 
 0.47 0.16 
 0.66 0.14 
 0.49 0.04 
 
 0.29 
 0.32 
 
 0.21 
 O.O9 
 
 
 
 Av. 
 
 0-55 0.13 
 
 O.22 
 
 
n6 
 
 /. E. Wallace Wallin, 
 
 
 
 TABLE LXIV. 
 
 
 
 
 (Prose] 
 
 
 Subject. 
 
 a 
 
 p I 
 
 n 
 
 G. A. A. 
 
 0-55 
 
 0.19 0.32 4i-47 4 
 
 71 
 
 
 
 O.22 
 
 
 W. L. P. 
 
 0.70 
 
 0.28 0.40 4 1.93 4 
 
 35 
 
 
 
 0.23 
 
 
 C. 0. S. 
 
 0.63 
 
 0.17 0.26 4 J -45 4 
 
 61 
 
 
 
 0.20 
 
 
 Av. 
 
 0.62 
 
 0.21 0.32 
 
 
 
 
 TABLE LXV. 
 
 
 
 
 (Poetry read as prose] 
 
 
 Subject. 
 
 a 
 
 f 1 
 
 n 
 
 J. M. T. 
 
 0.46 
 
 0.13 0.27 40.85 
 
 48 
 
 
 
 0.14 4- 
 
 
 C. O. 
 
 0.48 
 
 0.14 0.29 4 1.02 
 
 47 
 
 
 
 0.18 4 
 
 
 W. C. 
 
 0.62 
 
 0.22 0.35 4 J -42 
 
 33 
 
 
 
 0.22 4 
 
 
 S. 
 
 0.67 
 
 0.21 0.31 4 1.65 
 
 3 1 
 
 
 
 0.27 4- 
 
 
 Av. 
 
 0.56 
 
 0.17 0.30 
 
 
 
 
 TABLE LXVI. 
 
 
 (Poetry read as prose ivithout punctuation marks] 
 
 Subject. 
 
 a 
 
 p ?- e c 
 a 
 
 n 
 
 J. M. T. 
 
 0.45 
 
 0.12 0.26 40.85 
 
 19 
 
 
 
 0.16 4 
 
 
 G. F. A. 
 
 0.51 
 
 0.16 0.31 4 1. 12 
 
 44 
 
 
 
 0.24 
 
 
 B. S. G. 
 
 0.66 
 
 0.19 0.28 4 1.5 
 
 32 
 
 
 
 o. 20 4~ 
 
 
 Av. 
 
 o-54 
 
 0.16 0.28 
 
 
 
 
 TABLE LXVII. 
 
 
 
 ( Poetry 
 
 read as poetry, Browning verses'] 
 
 
 Subject. 
 
 a 
 
 p e- c 
 
 n 
 
 J. M. T. 
 
 0.52 
 
 0.18 0.34 4- 1.09 
 
 48 
 
 
 
 0.16 4- 
 
 
 C. 0. 
 
 0.45 
 
 0.16 0.35 4 i. 06 
 
 47 
 
 
 
 0.16 4 
 
 
 Av. 
 
 0.48 
 
 0.17 0.34 
 
 
Researches on the rhythm of speech. 117 
 
 TABLE LXVIII. 
 
 {Poetry read as poetry ', Tennyson verses} 
 
 t 
 
 a 
 0.30 -j- 1.52 33 
 
 O.22 -j- 
 
 0.27 -f 1.47 32 
 
 0.25 4- 
 0.28 
 
 TABLE LXIX. 
 
 {Summary of complex-centroid intervals of verses of Browning and Tennyson read as 
 
 poetry) 
 
 Subject. 
 
 a 
 
 / 
 
 W. C. 
 
 0.66 
 
 O.2O 
 
 s. 
 
 0.77 
 
 0.21 
 
 Av. 
 
 0.71 
 
 0.20 
 
 Table. 
 
 a 
 
 P 
 
 I 
 
 a 
 
 LXVII. 
 LXVIII. 
 
 0.48 
 0.71 
 
 0.17 
 
 0.20 
 
 0.34 
 0.28 
 
 Av. 
 
 0-59 
 
 0.18 
 
 0.31 
 
 TABLE LXX. 
 
 {Prose read as poetry} 
 
 Subject. 
 
 a 
 
 > 7 
 
 e 
 
 c 
 
 n 
 
 S. I. 
 
 0.61 
 
 O.I7 0.27 
 
 + 1-25 
 
 
 21 
 
 
 
 
 0.32 
 
 + 
 
 
 O. E. S. 
 
 [0.69 
 
 0.20 0.29 
 
 + I-37 
 
 
 23 
 
 
 
 
 0.21 
 
 
 
 
 Av. 
 
 0.65 
 
 0.18 0.28 
 
 
 
 
 TABLE LXXI. 
 
 
 {Summary 
 
 of complex-centroid 
 
 intervals 
 
 speech-) 
 
 
 
 Table. 
 
 a 
 
 P 
 
 I 
 a 
 
 
 
 LXIII. 
 
 0-55 
 
 0.13 
 
 O.22 
 
 
 
 LXIV. 
 
 0.62 
 
 0.21 
 
 0.32 
 
 
 
 LXV. 
 
 0.56 
 
 0.17 
 
 0.30 
 
 
 
 LXV1. 
 
 0.54 
 
 0.16 
 
 0.28 
 
 
 
 LXIX. 
 
 0-59 
 
 0.18 
 
 0.31 
 
 
 
 LXX. 
 
 0.65 
 
 0.18 
 
 0.28 
 
 
 
 Av. 
 
 0.58 
 
 0.17 
 
 0.28 
 
 
 Unit of measurement, I s . e, extreme intervals. 
 
 a, average duration. c, character of greater extreme. 
 
 /, immediate probable error. n, number of intervals. 
 
 , relative immediate probable error. 
 a 
 
 Results. The number of intervals measured was 1054. 
 
n8 /. E. Wallace Wallin, 
 
 The duration of a given complex-centroid interval in speech will prob- 
 ably fall somewhere within the limits of 0.14' (= shortest in series, J. 
 M. T., Table LXV.), and 1.93* (= longest in series, W. L. P., Table 
 LXIV. ) , a range of about i y. The range for the average interval of 
 different records is about % of this (from o.32 s , J. K., Table LX. to 
 0.75% A. D. B. A , Table LIX.), and for the averages of different sets of 
 records less than | (from 0.47% Table LX. to 0.71% Table LXVIIL). 
 The highest range between the averages of the selections read twice by 
 the same person was o.io s (S.); the lowest, 0.03" (C. O.). 
 
 The extreme deviations of these intervals from the average were always 
 of the nature of excesses. In the measurements 100 %were plus extremes. 
 This may be due more to the pauses than the brevity of the average. 
 The former enables the range to be extended almost indefinitely, while 
 it can be only slightly abbreviated. The extremes for the different sets, 
 may be obtained in the e and c columns of the tables. 
 
 The length of the average complex-centroid interval is a little over half 
 a second (0.58"). 
 
 The averages in prose (0.59") and poetry (0.57") are practically equal 
 in length. The tendency, just barely perceptible, is to make the inter- 
 vals of prose the longer. This conclusion may only hold for such inter- 
 vals as were measured, which were prevailingly 2 -syllable sound- and 
 i -pause- 2 -syllable composite-intervals. 
 
 The longest average in any record was 0.07* shorter in prose (0.70", 
 W. L. P.), than in poetry (S., Table LXVIIL); and the shortest 
 average was o.io s shorter in the reading scansion of English poetry (C. 
 O., Table LXVII.) than in prose (0.55% G. A. A.). 
 
 The intervals composed regularly of two syllables (0.71" = average for 
 set of Tennyson verses; o.66 s , same for sing-song and doggerel scan- 
 sion) are longer than those composed of a mixture of one, two, three, 
 four and five syllables, provided the number of i -syllable intervals is 
 relatively larger than the number of 3-, 4- and 5 -syllable intervals. 
 
 The average inequality of the lengths of complex-centroid intervals in 
 all varieties of speech was about o. 1 7," or 28 % of the length of the average. 
 
 The regularity in a given set of records of a variety of verses of 
 poetry of the predominantly 2 -syllable type, scanned according to the 
 fourt ypes of scansion, was about 8 / higher than that for the intervals in 
 a given number and variety of sentences of prose, uttered in various 
 ways. The irregularity for the former was 22% (Tables LIX., LXL, 
 LXIL, LXIX.) ; of the latter, 30% (Tables LXIV., LXX.). 
 
 The regularity in a given set of records of English poetry, the inter- 
 vals of which are mostly of the 2 -syllable pattern, and the scansion of 
 
Researches on the rhythm of speech. 119 
 
 which is rhythmically free (reading scansion, Table LIX. ) was about 3% 
 higher than that in a corresponding set of records of English prose, the 
 reading of which is rhythmically free (32%, Table LXIV. ). When 
 the scansion or reading is natural, artistic and rhythmically free, the 
 complex-centroid intervals are only slightly more regular in poetry than 
 in prose. 
 
 The most regular coordination of the intervals in prose (e. g., C. O. 
 S., Table LXIV. ; and Table LXX.) was superior to the coordination in 
 many records of poetry. 
 
 The most regular coordination in poetry occurs in routine scansion ; 
 it is never perfect. The lowest percentage in seventeen records was 4% 
 (Table LXII. ; E. H. T. ). Beating the time with the finger in- 
 creased the regularity about three and a half times. 
 
 The intervals in- the most regular scansion of poety were about six and 
 one-half times more regular than in the most regular reading of prose. 
 Excluding the record of routine scansion with regulative concomitant, 
 this is reduced to one and six-sevenths. 
 
 The coordination in doggerel and in the most regular specimen of' 
 reading scansion, was about equally good (19%, O. S., Table LXI., and 
 E. W. S., Table LIX.). In routine scansion it was about twice as reg- 
 ular as this (9%, Table LXII.). 
 
 In sing-song scansion it was slightly poorer than in doggerel scansion, 
 but better than in reading scansion in general. Reading scansion repre- 
 sents the highest degree of irregularity of any form of scansion. 
 
 The test records of the verses of Tennyson and Browning indicate 
 that the highest regularity in reading scansion occurs when the intervals 
 consist of a uniform mode of composition or the same number of 
 syllables (cf. Tables LXVII. and LXVIIL). The difference in the regu- 
 larity of the renderings of the tests is small. The absence of the punc- 
 tuation marks did not decrease it. 
 
 The range of irregularity for the records of a given number of per- 
 sons will be about 0.27" (from 0.02", routine scantion, Table LXII. to 
 0.29% reading scansion, J W. R., Table LIX.), or 38% (0.04 to 0.42, 
 reading scansion, Japanese poetry, I. M., Table LX. ). The range for 
 the averages of the sets was 25% (from 9%, routine scansion, to 34%, 
 verses of Browning read as poetry, Table LXVII. ) ; and for the same 
 persons repeating the same selections 7 % (from i% to 8%, in both cases, 
 J. M. T.). 
 
 With a standard of 10% of inequality, the complex-centroid inter- 
 vals are rhythmically coordinated in about 6% of the records of poetry 
 and in none of those of prose. With 15% as the standard none of the 
 
120 J. E. Wallace Wallin. 
 
 latter and 12% of the former satisfy the demands of rhythm (this is 
 exclusive of the test records read as prose, both with and without punc- 
 tuation marks). 
 
 The above items for the individual records (test records, records of 
 foreign languages, etc. ) may be obtained in the tables. 
 
 Measurements of simple- centroid intervals. 
 TABLE LXXII. 
 
 {Reading scansion , English poetry} 
 
 Subject. 
 
 a 
 
 P 
 
 f_ 
 
 a 
 
 e c 
 
 n 
 
 J. W. R. 
 
 0.63 
 
 O.I I 
 
 0.17 
 
 -f- I 12 -j- 
 
 91 
 
 
 
 
 
 0-33 
 
 
 E. W. S. 
 
 0.44 
 
 0.05 
 
 O.I I 
 
 -f 0.62 
 
 74 
 
 
 
 
 
 0.25 
 
 
 A. D. E. A 
 
 0.47 
 
 0.12 
 
 0.25 
 
 + 0.79 + 
 
 ii 
 
 
 
 
 
 0.27 
 
 
 A. D. B.5 
 
 O.6o 
 
 O.O6 
 
 O. IO 
 
 + 0.81 + 
 
 9 
 
 
 
 
 
 0.27 
 
 
 W. W. 
 
 0.41 
 
 0.10 
 
 0.24 
 
 + 0.97 + 
 
 64 
 
 
 
 
 
 0.18 
 
 
 A. R. P. 
 
 0-43 
 
 O.O6 
 
 o 14 
 
 + 0.57 
 
 ii 
 
 
 
 
 
 0.27 
 
 
 Av. 
 
 0.50 
 
 0.08 
 
 0.17 
 
 
 
 
 
 TABLE 
 
 LXXIII 
 
 . 
 
 
 
 ( Reading 
 
 scansion, Japanese and 
 
 Persian poetry} 
 
 
 Subject. 
 
 a 
 
 P 
 
 i 
 
 a 
 
 e c 
 
 n 
 
 J. K. 
 
 0.32 
 
 0.06 
 
 0.18 
 
 + 0.52 + 
 
 12 
 
 
 
 
 
 o.i 6 
 
 
 I. M. 
 
 0-34 
 
 o. 10 
 
 0.29 
 
 -f 0.60 -f- 
 
 13 
 
 
 
 
 
 0.18 
 
 
 K. H. K. 
 
 0-53 
 
 0.08 
 
 0.15 
 
 -f 0.68 ' 
 
 12 
 
 
 
 
 
 -0.31 
 
 
 Av. 
 
 0-39 
 
 0.08 
 
 O.2O 
 
 
 
 
 
 TABLE 
 
 LXXIV 
 
 
 
 (Sing-song and 
 
 dogger al scansion} 
 
 Subject. 
 
 a 
 
 P 
 
 a 
 
 e c 
 
 n 
 
 H. 0. 
 
 0.50 
 
 0.04 
 
 0.08 
 
 + 0.55 + 
 
 21 
 
 
 
 
 
 0.47 
 
 
 0. S. 
 
 0-59 
 
 0.04 
 
 0.06 
 
 + 0.78 + 
 
 21 
 
 
 
 
 
 0.47 
 
 
 Av. 
 
 0-54 
 
 0.04 
 
 0.07 
 
 
 
Researches on the rhythm of speech. 121 
 
 
 
 TABLE LXXV. 
 
 
 
 
 ( Routine scansion } 
 
 
 Subject 
 
 a 
 
 a 
 
 n 
 
 E. H. T. A 
 
 0.47 
 
 0.06 o.i 2 -j--7 + 
 
 12 
 
 
 
 0.31 
 
 
 E. H. T.j, 
 
 0.48 
 
 o.oi 0.03 +0.52 
 
 12 
 
 
 
 0.43 
 
 
 Av. 
 
 0.47 
 
 0.03 0.07 
 
 
 
 
 TABLE LXXVI. 
 
 
 
 ( Summa ry 
 
 of simple-centroid intervals of poetry} 
 
 
 ' 
 
 Table. 
 
 a p 
 
 
 
 LXXII. 
 
 0.50 0.08 0.17 
 
 
 
 LXXIII. 
 
 0.39 O.O8 O.2O 
 
 
 
 LXXIV. 
 
 0.54 O.O4 O.O7 
 
 
 
 LXXV. 
 
 o 47 0.03 0.07 
 
 
 
 Av. 
 
 0.47 0.06 0.13 
 
 
 
 
 TABLE LXXVII. 
 
 
 
 
 (Prose] 
 
 
 Subject. 
 
 a 
 
 , l 
 
 n 
 
 G. A. A. 
 
 0-44 
 
 O.IO O.22 -f-O.7O -f 
 
 51 
 
 
 
 0.20 
 
 
 W. L. P. 
 
 0.44 
 
 0.12 0.27 +0.83 + 
 
 36 
 
 
 
 0,23 
 
 
 C. O. S. 
 
 o 54 
 
 O. II O.2O -f O.8o 
 
 42 
 
 
 
 0.20 
 
 
 Av. 
 
 047 
 
 o. ii 0.23 
 
 
 
 
 TABLE LXXVIII. 
 
 
 
 
 ( Poetry read as prose} 
 
 
 Subject. 
 
 a 
 
 t ~ 
 
 n 
 
 ]. M. T. 
 
 0-37 
 
 o.io 0.27 -f-a68 _(- 
 
 34- 
 
 
 
 0.14 
 
 
 C. 0. 
 
 0.36 
 
 0.08 0.22 -j- 0.62 4- 
 
 31 
 
 
 
 0.18 
 
 
 W. C. 
 
 0.44 
 
 0.07 0.15 + 0.60 
 
 22 
 
 
 
 0. 22 
 
 
 S. 
 
 0.50 
 
 0.06 0.12 +0-75 + 
 
 20 
 
 
 
 0.27 
 
 
 Av. 
 
 0.41 
 
 0.08 0.19 
 
 
I 22 
 
 /. E. Wallace Wallin, 
 TABLE LXXIX. 
 
 ( Poetry read as prose without punctuation murks} 
 
 Subject. 
 
 a 
 
 P 
 
 jr 
 a 
 
 e c 
 
 n 
 
 G. F. A. 
 
 0.38 
 
 0.08 
 
 0.21 
 
 + 0.60 + 
 
 29 
 
 
 
 
 
 0.24 
 
 
 B. S. G. 
 
 0.49 
 
 O.I I 
 
 0.22 
 
 + 0.75 
 
 17 
 
 
 
 
 
 O.2O 
 
 
 J. M. T. 
 
 0.36 
 
 O.IO 
 
 0.27 
 
 + 0.72 + 
 
 32 
 
 
 
 
 
 o.i 6 
 
 
 Av. 
 
 0.41 
 
 0.09 
 
 0.23 
 
 
 
 
 
 TABLE 
 
 LXXX 
 
 
 
 
 ( Poetry 
 
 read as poetry, verses 
 
 of Browning} 
 
 
 Subject. 
 
 a 
 
 P 
 
 I 
 
 a 
 
 e c 
 
 n 
 
 J. M. T. 
 
 0.38 
 
 0.09 
 
 0.23 
 
 + 0.75 + 
 
 48 
 
 
 
 
 
 0.16 
 
 
 C. 0. 
 
 0.30 
 
 0.06 
 
 O.2O 
 
 + 0.50 -f 
 
 31 
 
 
 
 
 
 0.16 
 
 
 Av. 
 
 0.34 
 
 0.07 
 
 0.21 
 
 
 
 
 
 TABLE 
 
 LXXXI. 
 
 
 ( Poetry 
 
 read as poetry, verses 
 
 of Tennyson} 
 
 
 Subject. 
 
 a 
 
 P 
 
 I 
 
 a 
 
 e c 
 
 n 
 
 W. C. 
 
 0.48 
 
 0.09 
 
 0.18 
 
 + 0.65 
 
 21 
 
 
 
 
 
 0.22 
 
 
 S. 
 
 0-55 
 
 0.09 
 
 0.16 
 
 + 0.85 o 
 
 18 
 
 
 
 
 
 0.25 
 
 
 Av. 
 
 0.09 
 
 0.17 
 
 TABLE LXXXII. 
 
 {Summary of simple-centroid intervals of verses of Browning and Tennyson, read as 
 
 poetry} 
 
 Table. a p ~ 
 
 a 
 
 LXXX. 0.34 0.07 0.21 
 
 LXXXI. 0.51 0.09 0.17 
 
 Av. 0.42 0.08 0.19 
 
Researches on the rhythm of speech. 123 
 
 TABLE LXXXIII. 
 
 (Prose read as poetry} 
 Subject. a p e en 
 
 S. I. 0.46 0.07 0.15 +0.70 -f- 13 
 
 0.32 
 O. E. S. 0.53 0.13 0.24 +0.90 -f- 15 
 
 0.21 
 
 Av. 0.49 o. 10 0.19 
 
 TABLE LXXXIV. 
 
 (Summary of simple-centroid intervals of speech} 
 
 Table. 
 
 a 
 
 P 
 
 f 
 
 LXXVI. 
 
 0.47 
 
 0.06 
 
 0.13 
 
 LXXVII. 
 
 0.47 
 
 0. II 
 
 0.23 
 
 LXXVIII. 
 
 0.41 
 
 0.08 
 
 0.19 
 
 LXXIX. 
 
 0.41 
 
 0.09 
 
 0.23 
 
 LXXXII. 
 
 0.42 
 
 0.08 
 
 0.19 
 
 LXXXIII. 
 
 0.49 
 
 O.IO 
 
 0.19 
 
 Av. 
 
 0.44 
 
 0.08 
 
 0.19 
 
 Unit of measurement, I s . *>, extreme intervals. 
 
 a, average duration. c, character of greater extreme. 
 
 p, immediate probable error. , number of intervals. 
 
 , relative immediate probable error. 
 Results. Eight hundred and twenty-three measurements were made. 
 
 The duration of a given simple-centroid interval of speech will pre- 
 sumably lie somewhere between 0.14* (J. M. T., Table LXXVIII.; and 
 i.i2 s (J. W. R., Table LXXIL), a range of about i second. The range 
 for the averages of a given series of records will be about y$ of this (from 
 o. 3 o s , C. O., Table LXXX., to o.6 3 s J. W. R., Table LXXIL); and 
 for those of different sets, about \ (from 0.34", verses of Browning read 
 as poetry, to 0.54", sing-song and doggerel scansion). The highest 
 range for the selections repeated by the same person was o.o6 s (C. O.); 
 the lowest, o.oi s (J. M. T.). For the complex intervals J. M. T.'s 
 range was the highest ; here it is the lowest. He seems most irregular in 
 his pauses. 
 
 The longest average for any set of prose records was 0.49" (Table 
 LXXXIII.); for poetry, 0.54* (Table LXXIV., sing-song and doggerel 
 scansion). The shortest for poetry was 0.34" (Table LXXX., Browning 
 verses) . The longest average for any record was o. 54 s for prose (prayer, 
 
124 / E> Wallace Wallin, 
 
 C. O. S., Table LXXVIL), and o.6 3 s for poetry ( J. W. R., Table 
 LXXIL); the shortest for prose, 0.44* (G. A. A., Table LXXVIL), and 
 for poetry, 0.30* (C. O., Table LXXX.). The range seems larger in 
 poetry. 
 
 In 69% of the records the largest deviations from the average were 
 those of excess; in 27%, of deficiency; and about 3% were equal. 
 
 The average duration of these intervals for different kinds of speech is 
 somewhat below half a second (o.44 s ). This corresponds closely with 
 MEUMANN'S period (o.4o s ) for purely subjective rhythmisation. 
 
 The averages for a variety of sentences of English prose (0.48") and a 
 variety of verses of English poetry (0.46') are practically equal. The 
 tendency, barely perceptible, is towards a longer interval in prose. The 
 intervals in the poetry measured were predominantly 2 -syllable. 
 
 The average is longer in verses consisting uniformly of 2 -syllable inter- 
 vals, than in verses consisting of a haphazard alternation of i-, 2-, 3-, 4-, 
 and 5 -syllable intervals, provided that the number of i -syllable is higher 
 than the number of 3-, 4-, and 5 -syllable intervals. The average for 
 the verses of Tennyson was 0.51"; for those of Browning, 0.34* 
 (Table LXXXIL). The latter contained numerous i-syllable inter- 
 vals. So did those of Byron where the average was o.4i s (W. W., Table 
 LXXIL). 
 
 The inequality in the lengths of these intervals in various kinds of 
 speech is about o.o8 s , or 19^ of the average length. 
 
 The- regularity for a given series of records of a variety of poetical 
 verses of the predominantly 2 -syllable type of intervals, scanned accord- 
 ing to the four types of scansion will be about 8% higher than that for 
 records of a variety of prose sentences spoken in various ways (21%, 
 av. irregularity of Tables LXXVIL and LXXXIII. ) . This is exclusive of 
 the Japanese and Persian records. For records of English verses com- 
 posed mainly of the 2 -syllable pattern, the scansion of which is rhyth- 
 mically free, it will probably be about 6% higher" than for records of 
 English prose, the rendering of which is rhythmically free (irregularity 
 2 3%, Table LXXVIL). 
 
 The most regular coordinations in prose (15%, S. L; 20%, C. O. S.) 
 may often be higher than some coordinations in poetry (cf. A. D. B. A , 
 W. W., J. M. T., C. O., etc.). 
 
 The intervals in the most regular type of scansion of poetry (E. H. 
 T. B , Table LXXV. ) were about five times more regular than in the most 
 regular reading of prose (15%, S. L, Table LXXXIII. , melodious prose). 
 Excluding the record of routine scansion with regulative concomitant, 
 the regularity was two and a half times greater. 
 
Researches on the rhythm of speech. 125 
 
 The most regular coordination of the intervals in poetry is never per- 
 fect. The lowest observed percentage of irregularity in seventeen rec- 
 ords was 3%. 
 
 The highest regularity occurred in the type of routine scansion 
 accompanied by beating time. This type represents the acme of regu- 
 larity in the accentual or centroid rhythm of speech. 
 
 The coordination in sing-song (irregularity, 8%) and doggerel scan- 
 sion (irregularity, 6%) may be approximately the same. Both are fully 
 twice as irregular as mechanically regulated (routine) scansion. 
 
 The most regular coordination in reading scansion (10% of irregu- 
 larity, A. D. B. 5 ) is slightly inferior to the coordination in sing-song 
 scansion. The regularity in reading scansion is the most inexact of all 
 the types of scansion of poetry. It is most exact, other things equal, 
 when the intervals are composed quite uniformly of the same number of 
 syllables (contrast Tennyson's verses with Browning's and Byron's). The 
 regularity in the reading of the selections as prose and poetry was the 
 same. When the punctuation marks were eliminated, it was only slightly 
 increased. 
 
 The range in the irregularity of the intervals in various kinds of speech 
 may be about o. n s (from o.oi s , routine scansion, to 0.12", A. D. 
 B.^, poetry, and W. L. P., prose), or 26% (from 3%, routine scansion, 
 to 29%, Japanese poetry). The range for the different sets was 16% 
 (from 7%, Tables LXXIV. and LXXV., to 23%, Tables LXXVII. and 
 LXXIX. ), or about 2/3 of the former (the individual records) ; and 
 for the repeated readings by the same person, 2% (from 2%, C. O., to 
 4%, J. M. T. andS.). 
 
 According to the 10% standard, 23% of the records of poetry are 
 rhythmical, exclusive of records of poetry read as prose, both with and 
 without punctuation marks. With a standard of 15%, 46% of the rec- 
 ords of poetry, and 20% of the records of prose (record of melodious 
 prose), are rhythmical. Even in the latter case, more than one-half of 
 a fairly comprehensive collection of speech records was non-rhythmical. 
 
 Comparison of complex- and simple-centroid intervals. 
 
 The range for the former was approximately 1.80 times that for the 
 latter. 
 
 The general character of the extreme deviations was the same for both. 
 For the complex, however, they were uniformly, for the simple chiefly 
 excess extremes. 
 
 The average for the complex was 0.14% or 1.31 times, longer than for 
 the simple intervals. 
 
126 /. E. Wallace Wallin, 
 
 The averages of both were very slightly longer in prose than in 
 poetry. This is probably due (i) to the composition of the intervals. 
 In the prose records the longer intervals (3-, 4-, and 5 -syllable) were 
 relatively more numerous. (2) The tendency in poetry, especially in 
 routine and doggerel scansion, to rhythmize the pause ; this limits its 
 length. 
 
 Both intervals had the longest averages in verses consisting of the uni- 
 form 2 -syllable type. 
 
 The coordination was about i^ times more regular for the simple than 
 for the complex. 
 
 The difference was less between the regularity of the simple inter- 
 vals in poetry and prose, than between the regularity of the complex in 
 poetry and prose. 
 
 For the reading scansion of English poetry and prose, the difference 
 was twice as large for the complex as for the simple. 
 
 In the most regular type of scansion, the regularity was slightly higher 
 for the simple than the complex. The beating of time increased the 
 regularity of the former most. Relatively to other types of scanning, 
 however, it greatly increased the regularity of the complex intervals. 
 
 The range of irregularity for a variety of speech of different persons 
 was about i^ times larger (1.46) for the complex than for the simple 
 intervals. 
 
 With a provisional standard of 15%, the simple intervals were rhyth- 
 mically coordinated in about four times as many records of poetry as 
 were the complex. 
 
 General deduction. The most perfect manifestation of centroid rhythm 
 in speech never reaches the zero point of irregularity. This point is 
 most nearly approached in simple-centroid intervals of uniform compo- 
 sition, mechanically scanned. 
 
 Bearing on the rhythmic function of the pause. The facts reached re- 
 garding the duration and regularity of sound and composite, and single 
 and complex intervals necessitate a revision of the prevalent notion of 
 the pause as compensatory, 1 as the rhythmical equivalent for a missing 
 syllable or a foot. POE 2 so regards the caesura. We have seen that the 
 introduction of the pause uniformly disturbs the rhythm of the centroids. 
 The records without exception show this. The coordination of the 
 complex intervals was almost without exception so irregular as to defy 
 rhythmisation. In only two records does the inequality fall to 
 
 LANJER, The Science of English Verse 189, New York 1880. 
 GUEST, A History of English Rhythms, 77, London 1882. 
 2 POE, The Rationale of Verse, Works, VI 89, Chicago 1895. 
 
Researches on the rhythm of speech. 127 
 
 while for the simple intervals, there are eight such records of poetry and 
 one of prose. The pause constitutes the one rhythmical ; the other, in the 
 main, non-rhythmical. It creates two alternating, though interpenetrat- 
 ing, sequences within the unity of the discourse ; the one consisting of 
 expiration intervals, being rhythmical, the one consisting of expiration 
 and vacant intervals, being non-rhythmical. 
 
 We saw that the terminal pauses, which were longer and more regular 
 than the sectional, unified the verses. The tables for the sectional 
 pauses and the simple and complex intervals show that when the number 
 of sectional pauses is large the coordination of the intervals is poor. 
 When they occur quite regularly in long verses these are split up into 
 two. Most of those reading the Browning verses written as prose who 
 made regular sectional pauses, when asked to write the sentences as 
 verses, doubled the number. Here the pauses segregated tlje verses into 
 two parts. Both the terminal and sectional pauses may thus become 
 rhythmically recurrent, provided they recur at fairly regular intervals and 
 are of fairly regular length. 
 
 Rhythmically, pauses check the continuity of the centroid rhythm of 
 speech ; they limit the length of centroid sequences. This makes cen- 
 troid rhythm essentially discontinuous t an alternately interrupted and 
 recommencing flow, an inter-pause rhythm. Mechanical rhythm, e. g. 
 clock ticks, consists of continuous sequences of coordinated intervals. 
 
 Pauses also assist in imparting a felt unity to sequences of several cen- 
 troid intervals, provided they follow a certain law, whence arises a 
 secondary rhythm, that of vacant and expiration intervals. Thus we get 
 the verse interval rhythm. 
 
 Psychologically, the pause serves to avert the monotony of long 
 sequences of regularly recurrent stimuli. The length of these will affect 
 the quality of the rhythm. Too long or too short series are objectionable. 
 This may differ for different kinds of scansion and rhythmisation. In 
 subjective rhythmisation a continuance of clicks for 45 seconds has been 
 found favorable, for 70 seconds disgusting. 1 The pause thus contributes 
 change, variety, relaxation, and also a complex feeling of rhythm due to 
 the involution or interpenetration of the two orders of rhythm. 
 
 Bearing on some phases of speech rhythm. (a) The elements of rhythm. 
 Speech rhythm has been defined as a 'Maw of succession " ( GUEST); 
 a ''principle of proportion introduced into language, " " inferior meter ' ' 
 (SEELEY and ABBOTT); "periodic stimulation of sounds or of a small 
 group of sounds" (GURNEY); "the succession and involution of unities, 
 that is, unities within unities, applicable to feet, verses and stanzas ' ' 
 
 1 TITCHENER, Experimental Psychology, I n 340, New York 1901. 
 
128 J. E. Wallace Wallin, 
 
 (CORSON); "a system of accentuation " (HEGEL); "a succession of 
 tones in various length and shortness, following in a certain time form ' ' 
 (LOBE); "the coordination of speech sounds in time and duration" 
 (LANIER). In these definitions are involved two theories. 
 
 (i) The "time theory " holding that the element of time is funda- 
 mental, and emphasis subordinate. The sounds of a rhythmical sequence 
 must, above all things, be coordinated, periodical. LANIER ] ("simple 
 time relations "), GuRNEv 2 ("a fixed scheme of recurrence"), BOL- 
 TON 3 (regular recurrence the most important element; "accentuated 
 sounds form a secondary rhythm out of the primary"), HAUPTMANN/ 
 WESTPHAL/ LOBE/ HERBART/ LOTZE/ SCHOPENHAUER,* and ED- 
 WARDS/ may be cited as representatives. 
 
 ( 2 ) The " accentnalist theory, ' ' making emphasis the essence, and time 
 a secondary element. Some of the exponents are GUEST 6 ("accent the 
 sole principle," " rhythm of accent independent of quantity "), HEGEL * 
 ("rhythm a system of accentuation"), GUMMERE/ and KosTLiN 9 
 ("change of accents of the tones"). 
 
 It is hard to pin the writers down to any one statement, but the posi- 
 tions seem to be as indicated. When accent is made primary the ele- 
 ment of time becomes, at best, merely "a regulative principle or a prin- 
 ciple of embellishment." The opposite rigorist holds that no number 
 of accents, nor manner of variation of the elements, can render sounds 
 rhythmical unless they are strictly periodic. 
 
 The centroid theory of speech rhythm emphasizes the truth contained 
 in both of these theories. No recurrence of centroids can be rhythmized 
 unless the length and regularity of the intervals fulfil the requirements of 
 the rhythmical time-sense. This may differ slightly for different individ- 
 uals. It may allow of slight grades of rhythm, bad, good, excellent. At 
 a certain point, however, the rhythmic perception or feeling ceases 
 entirely. 
 
 The criterion of regularity may also differ for the different kinds of 
 rhythms, sensory and motor, speech, music, walking, dancing, etc. For 
 routine scansion, with the attention focused on the regularity, the crite- 
 
 1 LANIER, The Science of English Verse, III., New York 1880. 
 2 GURNEY, The Power of Sound, 439, London 1880. 
 3 BoLTON, Rhythm, Am. Jour. Psych., 1893 VI. 
 
 4 MEUMANN, Untersuchnnsen zur Psychologie iind Aesthetik des Rhythinus, Philos. 
 Stud., 1894 X 250. 
 
 5 GUEST, A History of English Rhythm, 108, London 1882. 
 
 6 GUEST, as before, 108-110. 
 
 7 HEGEL, Aesthetic, trans, of Kedney, 257. 
 
 8 GUMMERE, Handbook of Poetics, 137, Boston 1885. 
 9 MEUMANN, Philos. Stud., 1894 X 250. 
 
Researches on the rhythm of speech. 129 
 
 rion may be higher than for reading scansion, where the best expression 
 of thought is essential. A variation (-f- and ) of g- 1 ^ of empty intervals 
 between clicks 4.27* in length was always correctly noticed in the exper- 
 iments of HALL and JASTROW ;* one of yi-g-, nearly always. 
 
 These requirements, however, give only subjective or objective period- 
 icity. Grouping may be temporally perfect or periodic (Lanier's primary 
 rhythm), and still not be psychologically rhythmical. To constitute 
 perceived rhythms, some member of the groups must be accentuated, at 
 least subjectively. Walking may be periodic ; it is not normally rhyth- 
 mical. It becomes rhythmical when one step is made more energetic, 
 or when attention is given to it, or a sound is uttered or heard (e. g., a 
 drum) as it descends. 
 
 EBBINGHAUS 2 found that the coordination of the lengths of syllables 
 spoken loud was impossible without differences of accent : no rhythmis- 
 ing of syllables mechanically uttered was possible. In repeating the 
 numerals i-io regularly, I could feel no rhythm until some numbers 
 were subordinated, i. e., formed into centroid intervals. Then grouping 
 soon became inevitable. A trained vocalist sang the vowel 6 into the 
 phonograph without interruption for nearly 4o s . He was told to make 
 it perfectly smooth, without modification in intensity and pitch. The 
 resulting sound was fairly smooth at the beginning; after about io s 
 oscillations in the intensity were noticeable. Speech as a motor phe- 
 nomenon is necessarily centroidal. 
 
 Emphasis is unavoidably imposed upon certain members of a sound 
 sequence. Numerous experiments 3 show that these are always formed 
 into groups because of subjective emphasis. 
 
 The form of these groups will depend upon representative factors aris- 
 ing from the experience of the subject ; they will conform to some repre- 
 sented rhythm (counting, beating time, clock beats, pendulum swings, 
 puffs of the engine, etc.). 
 
 In BOLTON'S* experiments the normal, easiest or psychologically prior 
 form was the 2 -group (first syllable accented, trochee) and 4-group. The 
 3-group (dactylic form easiest) and 2 -f 3-group were readily suggested, 
 but not the 5 -group. HALL and JASTROW 5 found that in counting a 
 series of sounds or clicks 3- and 4-groups must be farther apart than 
 
 1 HALL AND JASTROW, Studies of rhythm, Mind, 1886 XI 61. 
 
 2 MEUMANN, Philos. Stud., 1894 X 423, 424, 
 3 BOLTON, Am. Jour. Psych., VI 145-238. 
 
 EBBINGHAUS and DIETZE, in MEUMANN, Philos. Stud. , 1894 X 423. 
 
 4 HOLTON, as before, 212, 216, 222. 
 
 5 HALL and JASTROW, Studies of Rhythm, Mind, 1886 XI 58, 61. 
 
130 /. E. Wallace Wallin. 
 
 2-groups. Twenty-four to forty clicks per second were easily distin- 
 guished by the average ear [limit of ability to discriminate sounds = T -|"2 
 (HELMHOLTZ) or -~\- (EXNER) of a second]. Counting, however, in- 
 volved naming of the clicks. The number-names are not of uniform 
 ease and length. This influenced the grouping. 
 
 Thus the centroid, or the element of emphasis, is always present in 
 speech. Consciousness is centroidal. Perceived rhythm is a highly 
 regular alternation of intense and lesser intense states of conscious- 
 ness. 
 
 (b) The nature of rhythm. It is a series of mental events involving 
 affection, intellection and conation. The role of the first has been 
 stated by WUNDT * (feeling of strain and expectation, of unity, and a cer- 
 tain degree of emotion), by SMITH 2 (rhythm a progressive emotion with 
 coordinated motor discharges), by MEUMANN 3 (aesthetic feelings) and by 
 LIPPS* (associative aesthetic feelings). The latter are probably such as 
 proportion, symmetry, unity in variety, harmony, euphony, beauty, sub- 
 limity. The emotions may be those of different degrees of pleasantness, 
 ecstasy, joy, vivacity, erethic diathesis, orgasm, love, hate, sweetness, 
 well-feeling, etc.; and perhaps disagreeableness monotony, insipidity, 
 sadness, discomfort, disgust, etc. The sense feelings may include 
 smoothness, assonance, dissonance, intense or weak, long or short, 
 fast or slow, excitations, freshness or tiredness, muscular activity or 
 innervation, tingling, heightened temperature and blood pressure, 
 etc. 
 
 MEUMANN 5 (rhythm a perception) emphasizes the role played by intel- 
 lection. This is best illustrated by the change of grouping produced by 
 ideation (association factor). This factor will largely determine the 
 character of the affectional response of the individual. 
 
 The present researches emphasize the factor of conation : without sub- 
 jective emphasis, the centroid, no rhythm. Intellective perception alone 
 may perceive periodicities, but not rhythm ; for this conative perception, 
 or intermittent attention, aroused and sustained by feeling, is necessary. 
 Felt motor innervations are fundamental to rhythmic perception. They 
 need not necessarily be movements or vaso-motor discharges, but motor 
 impulses (probably only mental beats). Rhythm is a peculiarly felt 
 motor perception of movement in time. 
 
 1 WUNDT, Outlines of Psychology (Eng. tr.), 167, 176, Leipzig 1897. 
 
 2 SMITH, Philos. Stud., 1900 XVI 291. 
 3 MEUMANN, Philos. Stud., 1894 X 264. 
 
 * LIPPS, Aesthetische Einfuhlung, Zeit. fur Psychol. u. Physiol., CXXII 441. 
 5 MEUMANN, as above, 272, 284. 
 
Researches on the rhythm of speech. 131 
 
 VI. COMPARISON OF INTERVALS IN SPEECH. 
 
 Tables of average number of syllables per centroid and expiration interval, 
 and the interval of the unit of measurement ( I s ). 
 
 TABLE LXXXV. 
 
 ( Poetry"] 
 
 Subject. 
 
 c 
 
 S 
 
 e 
 
 E. W. S. 
 
 2.21 
 
 4.22 
 
 6.42 
 
 J. W. R. 
 
 2.06 
 
 2.75 
 
 8.77 
 
 A. D. B. 
 
 2.15 
 
 2.90 
 
 3-73 
 
 W. W. 
 
 1.74 
 
 3.01 
 
 4.89 
 
 A. R. P. 
 
 2.00 
 
 3-73 
 
 6.00 
 
 H. 0. 
 
 2.09 
 
 3-i6 
 
 5.81 
 
 O. S. 
 
 2.10 
 
 3.10 
 
 7.80 
 
 J. M. T. 
 
 I. 5 6 
 
 2-95 
 
 
 C. O.s 
 
 1-56 
 
 3-40 
 
 4.41 
 
 W. C.B 
 
 1.90 
 
 2 90 
 
 4.26 
 
 S..B 
 
 2.OO 
 
 2. 5 6 
 
 4.26 
 
 G. F. A. 
 
 1.6 5 
 
 3-13 
 
 4.56 
 
 B. S. G. 
 
 2.00 
 
 2.99 
 
 4.00 
 
 Av. 
 
 1.92 
 
 3-14 
 
 5-41 
 
 TABLE LXXXVI. 
 
 ( Prose) 
 
 Subject. 
 
 c 
 
 S 
 
 e 
 
 G. A. A. 
 
 2-53 
 
 4.48 
 
 4.48 
 
 W. L. P. 
 
 2.20 
 
 3-14 
 
 6.05 
 
 C. 0. S. 
 
 2.24 
 
 3-53 
 
 6-57 
 
 S. I. 
 
 2.7J 
 
 4-38 
 
 6-33 
 
 O. E. S. 
 
 2-47 
 
 3-54 
 
 6-33 
 
 Av. 2.43 3.81 5.95 
 
 c, complex centroid intervals. 
 
 s, intervals of i . 
 
 e, expiration intervals. 
 
 Results. The average number of syllables for each complex centroid 
 interval of all varieties of speech was 2.17. 
 
 For the records of poetry it was 1.92 ; for those of prose, 2.43, the 
 latter being a trifle over 25% larger than the former. For verses com- 
 posed predominantly of the 3 -syllable type of intervals, this relation 
 would probably be reversed. Since there is no orderly arrangement of 
 interval pattern in prose, the variation, while it may be constant and 
 sometimes large, will in the average be small (cf. records of Table 
 LXXXVI.). The range between the highest and lowest average for 
 
1 32 J. E. Wallace Wallin, 
 
 prose was 0.51 syllable; for poetry, 0.65 syllable. The range for the 
 individual intervals was from i to 5 syllables. A centroid interval in 
 speech may thus range from a i -syllable to a 5 -syllable. 
 
 The average number of syllables per second of time for all kinds of 
 speech was 3.47 ; for the records of poetry, 3.14; for those of prose, 
 3.81, or about 20^ more than for poetry. 
 
 The range for the averages of prose was 1.34 syllable per second 
 (from 3.14 to 4.48) ; for poetry, 1.66 syllable (from 2.56 to 4.22), or 
 about 23% larger than the range for prose. 
 
 On the basis of the general average, the number of syllables uttered 
 per minute in speaking is approximately 208. This involves an equal 
 number of changes in the action of the vocal cords. The most rapid 
 contractions in aspirating t and k has been found to never exceed 6 
 double or 12 single contractions per second. 1 
 
 The average number of syllables per expiration interval for all kinds 
 of speech was 5.68 ; for poetry, 5.41 ; for prose 5.95, or 10^ more than 
 for poetry. 
 
 The range for the averages of prose was 1.85 syllable; of poetry, 
 5.04, or nearly two and three-fourths times longer than the former. 
 
 This range applies only to the averages of ordinary expiration inter- 
 vals, and not to those of maximum expiration intervals. The following 
 test was made : The subject, after having inhaled as much air as the 
 lungs would comfortably hold (attempting to make each inhalation 
 equal) read as many of a variety of short and long words as was possible 
 without replenishing the original supply of air. The test was stopped 
 when fortuitous inhalations occurred. The tendency was to increase the 
 rate as the context became familiar and practice set in. All the utter- 
 ances were considerably more rapid than those in ordinary speech. 
 Sometimes this increased the number of syllables ; frequently it produced 
 a more rapid expenditure of air, and decreased the number. Several 
 relatively slow readings contained the largest number. 
 
 These were the average number of syllables for seven or eight tests each 
 of ii subjects: 122, 147, 148, 153, 163, 165, 187, 190, 230, 257, and 276 
 The general average is 182.5. The highest average came from a trained 
 vocalist. Most subjects gained up to a certain point and then fell off. 
 The lowest percentage of gain was 19 syllables; the highest 119. The 
 range is 154 syllables (from 122 to 276). 
 
 Compared with the average normal interval, the average maximum 
 contains approximately 32 times as many syllables. The range of a breath 
 group may thus exte'nd from i to about 175 syllables. The normal is 
 
 1 HALL AND JASTROW, Studies of rhythm, Mind, 1886 XI 59. 
 
Researches on the rhythm of speech. 133 
 
 the stock-in-trade of ordinary speech. The maximum is perhaps never 
 reached in speech. The nearest approach is found in impassioned 
 speech. 
 
 Comparison of intervals. The average interval of the second was 0.60 
 syllable longer than the complex centroid interval, and o 63 syllable 
 shorter than the expiration interval. For the prose records, it was 0.56 
 syllable longer than the centroid interval, and the same shorter than the 
 expiration interval. For poetry it was 0.63 syllable longer than the 
 centroid, and 0.72 syllable shorter than the expiration interval. 
 
 The expiration intervals of poetry were 2.81 syllables longer than the 
 centroid intervals of poetry and prose respectively. 
 
 The range between the longest and shortest averages of the tables was : 
 for the centroid interval, 0.58 syllable; for the second, 1.45; and for 
 the expiration, 5.04. The average range for the second was 2.50 sylla- 
 bles longer than for the centroid, and 4.75 syllables shorter than for the 
 expiration interval. 
 
 The difference between the range of syllables for the second and cen- 
 troid interval was over 4 times the difference between the average number 
 of syllables for each ; and the same for the second and expiration inter- 
 val was 7.5 times the difference between the averages for each. 
 
 Character of the greater- extreme. This was for all the intervals of 
 speech in the vast majority of cases of the nature of excess. In no case 
 does the percentage fall below 59%. 
 
 The extremes were relatively least frequent in the verse intervals, and 
 most frequent in the complex, vacant, expiration and simple centroid in- 
 tervals in the order named. The converse order represents the grade of 
 frequency of the deficiency extremes. 
 
 Only in the expiration and simple centroid intervals were there any 
 equal extremes, the percentage in both cases being very small (6% 
 and 3%). 
 
 May we say that the longer the average of the interval the smaller, and 
 the shorter the larger, is the percentage of excess extremes ? The average 
 vacant and simple centroid intervals are equal in length, yet differ in the 
 percentage of plus extremes. The complex are longer, yet have no 
 minus extremes : all are plus. The simple intervals are considerably 
 shorter than the expiration and complex, yet have a higher percentage 
 of minus extremes than these more than double as many as the vacant 
 intervals. 
 
 Hence, other things being equal, the extremes of deficiency are rela- 
 tively most infrequent and those of excess most frequent, in intervals con- 
 taining pauses. 
 

 134 /. E. Wallace Wallin, 
 
 Range of duration for single intervals. Taking the range for the simple 
 centroid (0.98') as the norm, the range for the vacant was 1.37 longer, 
 for the complex 1.82, for the verse 4.20, and for the expiration inter- 
 vals 4.72. The difference between the longest (4.65") and the shortest 
 range is 3.57 s . 
 
 The range for the average of the complex intervals is 1.30 times that 
 of the simple (=0.33") ; for the vacant 1.54 times ; for the expiration 
 5.60 times; and for the verse 10.5 times. The longest range (3. 47 s ) is 
 3.14 s longer than the shortest. The order for the individual intervals 
 and the averages is seen to differ. Both fall into two groups, the one a 
 small, the other a large, range. 
 
 Average duration. Taking the average of the simple centroid as the 
 norm (o.44 s ), the average for the vacant was of the same length ; for the 
 complex, 1.31 times longer; for the composite, 2.20 times longer; for 
 the expiration, 2.70 times longer; and for the verse intervals, 6.11 
 times longer. 
 
 With the sum of the averages as the unit, the verse interval consti- 
 tutes 42%, the expiration 19%, the composite 15%, the complex 9%, 
 and the vacant and simple centroid 7% each, of this unit. The aver- 
 age for the verse intervals is about 2^ times longer than for the expira- 
 tion. 
 
 The proportion between the averages of the expiration and vacant in- 
 tervals, or between the quantity of sound and silence in speech, was as 
 2.71 to i. Apparently about 73% of the time consumed in speaking is 
 utilized in sound production and in exhalation, and 27% in pausing. 
 Most pauses probably represent an inhalation. 
 
 The proportion of sound to silence was slightly larger for poetry than 
 for prose. The difference is negligible. The largest percentage in 
 English poetry for the expiration intervals was 79% (J. W. R. record) ; 
 the smallest, 57% (W. C.). 
 
 The average for the composite intervals is nearly two and a fourth 
 times larger than that for the vacant and simple. Hence the sounds in 
 the composite intervals occupy more time than the silence. 
 
 All the above relations of the intervals in the various records, may be 
 obtained in the tables. 
 
 Range of irregularity. In units of time, with the range of the simple 
 intervals (o. i i s ) as the norm, that of the complex and vacant is each 2.45 
 times larger, of the verse 4.18 times and of the expiration 7.81 times. 
 As a fraction of the average duration the range for the simple (26%) is 
 2^ times that for the verse; and for the complex 1.46, for the vacant 
 1.65 and for the expiration 1.69 times that for the simple intervals. 
 
Researches on the rhythm of sp 
 
 Relative regularity. The average for the verse intervals is 2.11 as reg- 
 ular as for the simple ( 19% ); for the latter 1.47 times as regular as for the 
 complex, 1.52 as regular as for the vacant, and 1.78 as regular as for the 
 expiration intervals. 
 
 Percentage of records of rhythmically coordinated intervals. Applying 
 a 10% standard of irregularity, it is as follows: For verse, 59% ; simple 
 centroid, 14% ; expiration, 12% ; vacant, 8%, and complex centroid, 
 4%. The manner of coordination differs for the intervals. The ex- 
 piration and vacant were picked out wherever they occurred in the 
 records. The one intervenes between the other ; the tables do not 
 include the measurements of the two in one series. The verse intervals 
 are likewise separated by terminal pauses. These are not included in 
 calculating their rhythm. 
 
 Bearing on metrical problems. If meter requires a relatively perfect 
 coordination of intervals, what conditions are essential to fulfil this re- 
 quirement ? What type of reading and of interval most frequently and 
 fully satisfies the demand ? 
 
 Types of meter. We have seen that there are two types of speech 
 intervals which, on the basis of the criterion of regularity, may with 
 scientific precision be called rhythmical, viz., the verse and simple 
 centroid intervals. 
 
 As affects the simple centroid intervals, the measurements showed that 
 the criterion of metrical perfection was most highly approached in 
 poetry. The conditions for the highest regularity in poetry were : ( i ) 
 the distribution throughout the verses of centroid intervals consisting of 
 the same number of components; and (2) the scanning of the verses 
 according to the routine type of scansion. 
 
 As affects the phrase intervals, the measurements showed that the cri- 
 terion of metrical perfection was -approached only in the verse intervals 
 of poetry, the number of whose centroids was more limited and regular 
 than in the sentences of prose. The conditions for securing the high- 
 est regularity, other things being equal, were : (i) limiting the duration 
 of the intervals; and (2) subordinating the sectional to the terminal 
 pauses, and the sectional to the terminal centroids, of the intervals. 
 
 Meter par excellence choice of types. Which of the two types is prior, 
 primary or fundamental ? The question is complicated. Neither type 
 exists in isolation ; both rhythms interpenetrate. A good quality of 
 verse- interval rhythm will largely depend on, or perhaps demand, 
 a smooth flow of centroids. It may also be associated with pause 
 rhythms, sectional and terminal. Remembering these cautions, which 
 type has, on the whole, the stronger claim ? 
 
136 /. E. Wallace Wallin, 
 
 Hints from the measurements. These indicated that the range of 
 irregularity was about 2^ times larger, the average irregularity over 2 
 times larger, and the number of records of unrhythmical intervals about 4 
 times more for the simple intervals. In only one record was the varia- 
 tion of the intervals less for the simple centroid. From the point of 
 view of regularity, experiment uniformly emphasizes the importance of 
 the verse interval type. 
 
 This type, however, to become a rhythmic unity requires a sort of 
 "verse" centroid, as distinguished from the "foot" centroids, at the 
 beginning or the end. The complicated relations of the verse and foot 
 centroids give rise to a hegemony of centroids. The verse as a rhyth- 
 mical whole might be changed by eliminating the sectional and especially 
 the terminal centroids. It could remain a regular, yet not rhythmical, 
 group. 
 
 Hints from the test. The grouping of the sounds into visible verses 
 produced a delusion : spontaneous, unsuspecting thought did not dis- 
 tinguish verse sequences from sentence sequences. With words arranged 
 like prose sentences the verses were pronounced non-rhythmical sentences 
 of prose. With words arranged like verses the prose sentences were 
 declared rhythmical, and to be poetry. 
 
 In these judgments the smoothness or rhythm of the word flow was a 
 factor. The determining factor was the form of print, or a felt verse 
 meter. Some who read the Tennyson verses as prose pronounced them 
 rugged but still poetry, because of a felt verse interval rhythm. The 
 appeal to unsuspecting thought also shows the importance of this form 
 of rhythm. 
 
 Hints from the poet. The following questions, bearing also on other 
 aspects of the metrical problem, were submitted. 
 
 Questionnaire on the art of versification in respect to meter. 
 
 Describe the attitude which you assume, in the act of poetical production, toward the 
 versification of the poem that you are about to write. 
 
 (I. ) (a) Do you determine in advance the particular kind of meter which you intend 
 to follow? (b) Do you decide that a given poem shall consist of a certain kind and 
 number of measures dactyls, spondees, iambi, anapests, etc. to the line? (c) If so, 
 do you go over your verses to see that they invariably follow the metrical system adopted ? 
 (d) Or are there forces unconsciously operative preventing any deviation from a chosen 
 norm? 
 
 Or do you predetermine the number of emphatic or accented syllables, i. e., stress- 
 points, irrespective of the number of unaccented or weak syllables, for the verses of your 
 poem ? 
 
 (II. ) Or, on the other hand, is the measure in your poetry a subordinate factor? 
 
 Does it sustain a relation to special moDds and emotions and to the character of the 
 
OF 
 
 Researches on the rhythm of speech. 137 
 
 thought which you would express ? Do you become so inspired by the thought to be 
 expressed that the form never for a moment rises consciously before you ? Your foot- 
 scheme would thus be the result of a species of unconscious cerebration, and you would 
 become aware of its character not before, but either in the very act of poetical creative- 
 ness, or upon subsequent examination of your completed product. 
 
 (III.) Do you feel a demand for meter? Whence does it spring ? 
 
 (IV. ) If you employ both methods, what relation do they sustain in reference to (a) 
 frequency; (b) ease or naturalness, and (c) results, metrical and poetical? 
 
 The meter of Christabel, it may here be noted, was founded on the 
 principle of "counting in each line the accents and not the syllables." l 
 Each verse contains four accents ; the number of syllables may vary from 
 seven to twelve. Tennyson's lyric, "The Poet's Mind," may be con- 
 sulted. 
 
 Replies. 
 (English) 
 
 Ella Wheeler Wilcox : I cannot fully analyze my methods. They are not always the 
 same. The idea comes usually first, a sentence, a word, or a thought only like a point 
 of light. I may carry it about for weeks the germ of the poem shapeless and unde- 
 fined. When I take time to sit down and focus my mind upon it, the poem comes. 
 
 Sometimes I begin two or three shapes before I fix upon the style of verse ; oftener I 
 begin the poem as it remains, the lines falling into rhythm themselves. 
 
 Again, I decide upon a sonnet form, the moment the idea comes. One of my best 
 sonnets ("A Minor Chord ") came that way. I met two nurses on the street and the 
 line "The wistful unkissed mouths of nurses" came to me. I knew it was a sonnet 
 line, and days afterward sat down and built the sonnet to fit that line. 
 
 In talking with a very excitable man one day about single tax (that is, listening while 
 he talked), he said : " No question is ever settled until it is settled right." I said, " Oh 
 there is a poem in that," and at once sat down and wrote it it has been used in two 
 political campaigns by the Bryan orators. 
 
 Sometimes without any idea, a rhythm, a wordless song, goes ringing through my head. 
 I walk to it I work to it and it pursues me until I find an idea to put into the measure. 
 
 At other times, without idea or measure, I settle down, knowing something will come. 
 - : 2 My own poetic mental habit is lyrical. As nearly as I can re- 
 member, each poem, or theme, or motif (as one would say in music) occurs to me almost 
 simultaneously in both thought and form a poetic phrase (made up of words in a certain 
 poetic accent and diction) shapes itself in the mind. I do not realize at the moment 
 what the meter is. I may, or may not, realize what the stanza or complete poem-form is 
 to be. I think that most of my lyrics have occurred to me " on the road " when mov- 
 ing about, going back and forth to my office, travelling, sometimes when I am reading. 
 Sometimes a line or two will rest in my mind for years, and add other lines spontaneously ; 
 it may be in distant scenes. For instance, for years I had in my mind these lines, 
 descriptive of the glow above New York as seen at night from the Staten Island ferry : 
 
 Lies like a lily white 
 
 On the black pool of night. 
 
 COLERIDGE, Poetical Works, I 188, Aldine Edition. 
 2 The writer wishes his name reserved. 
 
138 /. E. Wallace Wallin, 
 
 Years afterward, while at Stratford-on-Avon for the first time, the whole poem came to 
 me in a shape that I had never had in mind at all, so far as I know. It was not at all 
 affected in form by the memory of Shakespeare verse forms, as might have been natural 
 in the circumstances. 
 
 AT NIGHT. 
 
 The sky is dark, and dark the bay below 
 Save where the midnight city's pallid glow 
 
 Lies like a lily white 
 
 On the black pool of night. 
 
 O rushing steamer, hurry on thy way 
 Across the swirling Kills and gusty bay, 
 
 To where the eddying tide 
 
 Strikes hard the city's side ! 
 
 For there, between the river and the sea, 
 Beneath that glow, the lily's heart to me, 
 
 A sleeping mother mild, 
 
 And by her breast a child ! 
 
 My busy life has given me little time to build poems ; hence their brevity and subjec- 
 tive quality. They start, with few exceptions, from a personal experience and emotion. 
 At times in my mind a line or verse seems to sing itself (not an actual tune but a series 
 of poetic accents) ; hardly anything at first except a sense of verbal music without words, 
 stirred by the delight, perhaps, in something startlingly beautiful in nature. 
 
 In the volume "In Palestine" you will find a poem, "How to the Singer 
 Comes the Song," which seems to be itself an answer to your question. After a few 
 phrases or lines have started into being something seems to say, this is to be a poem in 
 such or such a conventional form blank verse, four lines with rhymes, six, a sonnet, 
 etc., or else in some new form altogether, with stanzas irregular or unconventional in 
 themselves but made regular by following succeeding stanzas all of the same form, or 
 else with the regular irregularity of the dithyrambic form. After the form is thus estab- 
 lished everything is bent to moulding the poem according to the conventional or sought- 
 out, or accidental pattern. I find the line is in such a form well then, it must be 
 true to that form unless with some variation that cannot be charged to carelessness, but 
 to deliberate intention or deliberate adoption of an accidentally pleasing form. At one 
 time virtually the beginning of my poetical writing I was very much stirred by the 
 Italian sonnets, by Shakespeare's, Milton's, etc., and my poetic impulses naturally flowed 
 into sonnet form ; the seed phrase in the mind was apt to be inevitably sonnet-wise. 
 I have written four in a day. I have no doubt that while Dante was writing his triology 
 his poetical thought fell naturally into the terza rima. 
 
 I think the initial thought is generally a line ; rhymes usually follow with lightning 
 speed. I believe this is because my mind like the minds of all moderns is full of the 
 music of rhymes. If I had been an ancient Hebrew, or a verse-maker of some other 
 primitive race in the early days, the music in my mind would doubtless have been rhyme, 
 less would have fallen with apparent spontaneity into the forms prevalent in the lan- 
 guage of the time. 
 
 : J (I.) I do not determine in advance the stanza forms to be used 
 
 in any particular poem, unless a thought or feeling occurs to me which seems fitted for 
 
 1 The writer wishes his name reserved. 
 
Researches on the rhythm of speech. 139 
 
 one of the more elaborate, recognized stanza forms, like the sonnet, rondeau, Spenserian 
 stanza, etc. In that case I enter upon the construction of it consciously. I do go over 
 my verses to see that there are the right number of lines, that they rhyme in the right 
 order, and I do not commonly count the number of syllables or accents in a line in 
 composing isometric verse. The form once adopted say the ten-syllable iambic my 
 ear keeps count of the syllables half- consciously in the act of composition. When the 
 length of the line varies, however, sometimes I find that I have made a mistake have 
 got a line in that is a foot too short or too long and so have to correct. 
 
 (II.) The adoption of a line as the unit of a verse-composition, is commonly the 
 result of unconscious cerebration, I think ; or rather, perhaps of unconscious imitation, 
 my memory being full of verses of all kinds. It is with some half-remembered measure 
 that a thought or feeling in my mind tends to associate itself, giving possibly a line or 
 two lines which emerge definitely into consciousness. Then the end- word in this line 
 or lines, suggests a rhyme- word and so a stanza or couplet is formed usually in the 
 middle of the poem that is to be. From this, by a conscious effort of construction, I 
 build up the piece, before and after, on the norm thus established. 
 
 (III.) I do feel a demand for meter, but could not say whence it springs without 
 theorizing about it. Certain emotions, not at the moment they are experienced, but after- 
 wards, when recalled, seem to crave metrical expression. A few words come to me 
 metrically arranged, and that is the nucleus of a poem if I carry the thing any further 
 by a definite effort of will. If not, they simply drop out of memory in a day or two. I 
 have sometimes dreamed a line or two. 
 
 (IV.) I have not noticed the relative frequency. The two methods run together and 
 are confused in any actual piece of verse- composition. 
 
 James Riley : (I.) I know nothing of the anatomy (I will call it) of poetry. My 
 own experience after nineteen years, writing my first poem at the age of thirty- three, is 
 that the form strengthens and meets unconsciously the requirements of the poem, in 
 accordance to the inner growth. 
 
 (II.) This question answers all I know of poetry. It seems to me the poet should 
 write his moods as they come to him, and later complete or reject the work according to 
 its worth. In my dialect poems in the New England and Irish dialects I have always 
 known what I was going to write, to a great extent before writing a line. They have 
 generally been of actual occurrences in my life. On the other hand and it has always 
 been puzzling to me why it should be so I have written a great number of stories in 
 prose, and in them I can never find myself able to insert a single actual incident in my 
 experience. 
 
 To show how erratic is the poetic mood as it comes to me sometimes, out of which if 
 worked out comes a poem, I mention a poem written a year ago and which appeared in 
 the " New England Magazine " of last May. I remember the words and thought that 
 led to the opening lines of the poem : 
 
 " After the book is written, after its page is read, 
 After the soul has brought and left what to the age is wed," 
 
 What connection it has with the poem I am unable to tell. The lines were to me a 
 music that led to other music, and allowing the second feeling to express itself, I began 
 striking the words on the typewriter " over and over and over." I did it with no con- 
 nection of any thought but that should in some way involve and be of life. In this way 
 the music came first. My first writing beginning of the poem was with the thought 
 
1 4 o /. E. Wallace Wallin, 
 
 that every stanza should begin as the first, and I can see now that the thought would 
 have been higher, if the refrain had so continued. 
 
 N. M. Nelson : In my few metrical compositions, 1 have never stopped to think 
 whether I used dactyls or spondees, or counted the syllables until I had finished. I 
 simply let my thoughts assume the form suited to their expression. When in such a 
 mood, the metrical part is secondary to thought, but may still be in strict conformity with 
 poetical usage. 
 
 There is certainly a necessity for meter. This may be because we unconsciously imi- 
 tate poetical composition, or the mind may require meter in certain moods to express 
 inspired thought. 
 
 E. L. Fox (Yale '02): (I.) (a) Before beginning to write, yes; but not before the 
 spirit and general idea of the poem are clearly defined in my mind, (b) No. (c) No. 
 (d) Yes. The spirit of the poem generally swings itself into a suitable meter, so that 
 choice of a meter is usually unnecessary (when writing " to order" on a subject I am 
 not enthusiastic about, this is not true). Sometimes I hit upon a meter I like; it sticks 
 in my mind and suddenly occurs to me as just the thing for a subject I have been think- 
 ing over for a week or so. 
 
 (II. ) The measure is distinctly paramount. It sustains a close relation to the charac- 
 ter of the thought. The last sentence applies perfectly to my experience. 
 
 (III. ) Yes ; from the nature of the theme and the enthusiasm which it excites. 
 
 (IV 7 .) I employ but one method. 
 
 Paul F. Gilbert (Yale '01) : My ear tests the quality of meter. Minor defects are 
 corrected by mechanical examination. Forces unconsciously operative prevent with some 
 reliability any eccentricity. The number of accented syllables is predetermined, subject 
 to modification, with little attention to specific forms of verse, the number being settled 
 by their capability of expressing the nucleus of the poem and other favorite sentiments 
 which may be afterwards suggested. 
 
 My moods and emotions are emphasized by the employment of a suitable verse form, 
 but this form is not arbitrary. 
 
 The expression of spontaneous thought often runs into the channels of least resist- 
 ance usually some elementary form of verse. The demand for meter arises from a 
 natural desire for repetition. Repetition is agreeable, the abrupt disagreeable. 
 
 The practice of determining in advance the meter and the number of accented syllables 
 tends to produce artistic results ; but I find the spontaneous expression more satisfactory. 
 The imitation of established verse forms leads to artificiality. 
 
 (Norwegian) 
 
 Wilhelm Petersen : (I.) (a) Sometimes I do determine the meter in advance, e. g., 
 when I write new words to some old melody, either from choice, or by the request of 
 others. In such cases the song known to me in that meter has a certain influence on 
 what I write, perhaps through mood-association. 
 
 (b) Strictly speaking, I do not think I ever determined in advance (except as above 
 stated) any precise meter, as dactyl, spondee iambus, anapest, etc. I do think, how- 
 ever, that after my thought has assumed a certain rhythm (the original poetic form, an 
 emotion- product) I develop it in consciousness into a more regular meter, partly by choice 
 influenced by my acquaintance with prosody and literature, partly as it may happen to 
 fall in with the completed "motif" that is to form the basis of my poem. 
 
 (c) That depends. If I desire to make a "finished product," I follow the laws of 
 prosody and rhetoric in detail. 
 
Researches on the rhythm of speech. 141 
 
 (d) No, I hardly think there are. My opinion is that a person working under the 
 influence of the soul-forces in a subconscious state, simply follows the rhythmic fall of his 
 thought as accentuated by the underlying mood- waves, and the suitable words fall into 
 the place demanded either by a musical force in subconsciousness, or are called forth simul- 
 taneously with the thought. The result is : accented- tone syllables and stress-points. 
 Meter and rhyme are secondary results due to conscious effort toward art- form. 
 
 (II.) When the writer is unconscious of imitation and uninfluenced by mood-associa- 
 tion, "measure" in poetry must be subordinate in degree and secondary in time, and is 
 rather an art-effect than a factor. As the "motif" in original inspired production is 
 " emotion of soul fused into thought " eine Regung der Seele durch Gesinnung bewutz 
 und im Gedanken fliezend gemacht the desire for the materialization of which demands 
 from the soul a conscious effort in language production, and as, furthermore, language 
 follows the most rigid laws, just as lines do in painting and sculpture, I cannot quite see 
 how there can exist any direct relation between the creative and conceiving mood and the 
 outward art-form that the complete poetical thought finally assumes. That some relation 
 may exist, it were rash to deny. The natural steps in the development of a poetical 
 thought would then, it seems to me, be as follows : Underlying emotion, or inspiration, 
 moving towards and shaping itself into a "motif" that rises through subconscious effort 
 and determines "rhythm" or "tone-accent," and flowing into word-expression, follow- 
 ing in the form it finally assumes, the laws of prosody as to meter, and of rhetoric as to 
 choice of words. 
 
 (III. ) Yes. It springs partly, at least, from the nature of the language used. How- 
 ever, language itself has been influenced by this demand, indicating that the demand for 
 meter may follow from the first the steps in the above scheme of development, although 
 not consciously sought for till the last. 
 
 (IV.) I think I have written (a) mostly without predetermined measure, (b) far 
 more easily and naturally then, and (c) with better results, metrically and poetically. 
 
 (Swedish) 
 
 Jakob Bonggren : While sitting, walking, or, though rarely, lying down, some 
 thought or, oftener, picture flashes into my mind, which I conclude will make a good 
 poem. Sometimes I hear a meter or some line of poetry in connection with it, and that 
 is the nucleus of the poem about to be begun. 
 
 ( I. ) If the mind only begets a thought or picture, I try to find a meter ; this is usually 
 an easy matter. I think I can hear what is the proper meter, so I seldom have difficulty 
 in selecting a form. I very seldom, if ever, decide from the beginning the number of 
 lines in the poem ; it grows until it is full-grown. If I have one verse, it is easy to get 
 all the following or preceding ones molded in the same form. I read through the poem, 
 correcting and changing the words here and there, but very seldom the meter. 
 
 If I have a thought as a nucleus, I sometimes study out some of the strongest points 
 and stress- words, select a meter, and work out the poem. Poems worked out in this way 
 (from pictures or thoughts) I have found to be the best. 
 
 (II.) Whenever I am called upon to write a poem on a certain subject I slowly and 
 carefully work it out ; select a suitable meter and think out all I want to include in the 
 poem. After handling my subject for a while, I sometimes feel warmed up, and then 
 everything is easy. There seems to be an unconscious cerebration working along with 
 conscious thought and will, when something worthy the name of a poem is created. The 
 "poems " written to order I have found to be generally poor, unless I have warmed up 
 in the act of composing. 
 
 (IV.) Answers as above (I.) and (II.). 
 
142 J. E. Wallace Wallin. 
 
 As bearing on the present question, these statements indicate that both 
 types of language rhythm, varying more or less with different composers, 
 are felt, and make a demand upon the mind that conceives poetical 
 thoughts and embodies itself in metrical language. Witness the terms : 
 "wordless song," "poetic accents," "verbal music, without words," 
 "measure paramount ;" and " poetic phrase," " sonnet form, " "verse 
 form," "verse line singing itself," music of the lines, the line as the 
 "unit of verse-composition," as the "initial thought," the "seed 
 phrase," feeling that the line must be " true to its form," and " measure 
 secondary. ' ' 
 
 Deduction. Three lines of evidence thus uniformly emphasize the im- 
 portance of the verse interval type of meter in speech rhythm. These 
 intervals may approximate better than the centroid intervals the larger 
 fluctuations of attention in speaking, which alone may possess the power 
 to genuinely attune and cadence the soul. May we therefore regard 
 them as the primary or chief type of the rhythm of speech ? 
 
 It is a pleasant duty to acknowledge my obligations to the subjects, to 
 those answering the questionnaire, and to the editor of the Studies. 
 
NOTE. 
 
 The price of all volumes of these Studies is fixed at $i each. For cash 
 with order the price is 75 cents, provided no correspondence or other 
 clerical labor is involved. 
 
"" 
 
 JAH 1 1 
 
YC 45446'