Printer Friendly

On Lorca's 'San Rafael (Cordoba)' and some other texts.

The principles of syllable scansion in the Divine Comedy are generally well understood. They have been most authoritatively treated in the Enciclopedia dantesca entries by Gian Luigi Beccaria on dialefe and dieresi, which draw partly on Ciafardini's much earlier articles; more recently they have been treated (but without a predominant focus on Dante) in Pietro Beltrami's and Aldo Menichetti's substantial manuals of Italian metrics, and (rather more questionably) in Remo Fasani's La metrica della 'Divina Commedia'. (1) Like his contemporaries and immediate predecessors, Dante allowed himself a degree of license in this matter that would have been inadmissible by the standards of later versifiers, from the Petrarchists of the sixteenth century onwards. But the degree of license was limited: there is considerable consistency in Dante's metrical practice in respect of syllable divisions both within and across word boundaries. The purpose of this article is to attempt a more exhaustive survey of this practice than earlier discussions have offered, with the help of a number of computer-based routines that have made such work easier, faster, and more accurate than by any manual method. As a result it will be possible to define the nature of Dante's practice, and particularly the degree of regularity that it displays, much more systematically and precisely than has previously been possible.

But there are at least two major qualifications to this claim. First, defining Dante's practice here means defining that practice as represented in Giorgio Petrocchi's edition of the Divine Comedy. (2) The subject of my analysis has already been filtered through Petrocchi's editorial choices, and except on one issue of scansion I do not propose in the present context to call these choices into question: reasonably enough, I hope, given the authority that the edition enjoys. Second, while much of the present project involves the attempt to quantify various aspects of Dante's writing, such quantification rests on interpretative decisions that can sometimes be queried, and inevitably have, in a number of cases, an element of the arbitrary in them. The statistics I provide, particularly in Table 10 at the end of the article, must be read with this caution in mind.

In any case the project of this article is, I trust, computer-aided rather than computer-driven. What is offered here is the necessary first stage of a full study of the rhythm and metre of the Divine Comedy that makes extensive use of computing routines, but does so in order to address long-standing issues of literary scholarship, and has constantly involved the exercise of critical judgement. The analysis of Italian metre does nevertheless lend itself particularly well to the use of computing resources. Because of the near-phonological character of Italian orthography, the great majority of syllable divisions can be identified in simple structural terms, as coming after any group of one or more consonants or after a printed diaeresis mark (the identification of accentual structures, which is the next stage of the study, is unfortunately not nearly so straightforward). While this procedure does not strictly follow the standard rule for Italian, which in most cases makes syllables begin with the consonant that precedes a vowel, it seems perfectly adequate for the purposes of the present enquiry. A simple electronic routine counts syllables on this basis, supplemented by a procedure for counting intervocalic i as a syllable divider (it virtually always acts as such), and by another procedure for registering in a table, and then routinely applying, syllable divisions that regularly occur between adjacent vowels within a given word (as in maestro, for instance, where the a and e are always separate syllables). On occasion, of course, the routine has to be corrected, and also it frequently has to be supplemented through the manual insertion of irregular syllable divisions within the word (diaereses), as well as marks for hiatus (dialefe) between adjacent vowels on either side of a word boundary. Equally important, as well as making the syllable count simpler, the use of various computing routines makes it enormously easier to manipulate the results of the scansion process, as I try to illustrate below.

First, a point of terminology: following common practice I use the terms diaeresis (dieresi) and dialefe to stand for all syllable divisions between adjacent vowels, respectively within and across word boundaries; conversely I use their opposites, synaeresis (sineresi) and sinalefe ('synaloepha' exists in English, but is not much used), to stand for all cases where two or more adjacent vowels are counted as a single syllable. In this looser sense synaeresis and sinalefe are the rule in Italian versification, in that with certain regular exceptions, any two adjacent vowels count as a single syllable, whether or not they are separated by a word boundary or indeed by a syntactic division (though as I have said, Dante's practice, like that of his contemporaries and predecessors, was considerably less regular than that of later generations). The term dieresi can be used more strictly, to stand for syllable divisions within a word where in normal parlance one would not occur, as within a diphthong; its opposite, sineresi, can be used only for cases where the division is removed between two adjacent vowels that would normally be separate syllables (Menichetti, Metrica italiana, pp. 182-83). This stricter usage corresponds roughly to Petrocchi's (and standard) editorial practice, according to which the diaeresis mark is usually printed only where the vowels in question might normally be expected to form a diphthong and a single syllable. But given that it is sometimes difficult even in respect of the present-day language to decide what the standards of normal parlance should be, and all the more so in the case of the language of Dante's day, it seems in the present context that no particular purpose is served by such a terminological restriction.

In what follows I shall use the following symbols: [??] at the end of a word for dialefe and [??] for sinalefe; the standard diaeresis mark (a, e, i, o, u), when it is used by Petrocchi, and the symbol | where there is a diaeresis not marked in his edition; | is also used, for the sake of maximum clarity, to mark the syllable division produced by a semiconsonantal intervocalic i, though this would not usually be considered a case of diaeresis or dialefe. References are given using the numbers 1, 2 and 3 for Inferno, Purgatorio, and Paradiso respectively, followed by the canto and line numbers. All quotations from the Divine Comedy have been numbered sequentially, in the order in which they appear below.

To begin with vowels on either side of word boundaries, there are, by my count (see Table 10 and the related discussion below), 10559 sinalefi and 2499 dialefi in the Divine Comedy. The general rules for the application of dialefe in the poem are simply stated: indeed, more simply than in Beccaria's otherwise excellent encyclopaedia article. It occurs, in the Rime and in the Divine Comedy as in the work of Dante's contemporaries, in the great majority of word combinations where an initial vowel (or h) is preceded either (1) by a polysyllable with a final accented falling diphthong (an accented followed by an unaccented vowel: -ai, -ea) or with a final accented vowel (either short, as in virtu, or long as in finii, or an elided falling diphthong, as in cantere'); or (2) by any monosyllable ending in a vowel except di (3) (conjunction), the definite article, the proclitic pronouns (lo, la, li, le, mi, ti, si, ci, vi, ne), and the negative no. As is the norm in Italian verse, dialefe for Dante is principally determined by the nature of the first of the two words in the combination. A separate rule applies to word-initial semiconsonantal i, which does not strictly fall under the heading of dialefe, as I have said, but needs to be mentioned here: in the thirty-two cases where this follows a final vowel it always produces a syllable division, irrespective of the nature of the preceding word, as in Nuovo | Iason (1. 19. 85). On the other hand, word-initial semiconsonantal u follows the general rules for dialefe; it does not itself produce a syllable division, but a syllable division may be produced by the word that precedes it. Thus we have, for instance, avesse [??] uom (2. 14. 83) and fui [??] uom (1. 27. 67), the first without and the second with dialefe.

(1) A total of 327 cases of dialefe are produced in the Divine Comedy by polysyllables with a final accented vowel or final accented falling diphthong. I find only the following twenty-eight cases where the same class of words produces sinalefe instead: cases (I shall call them) of exceptional sinalefe:
Table 1: Exceptional sinalefe after polysyllables

1.  1 15 22    Cosi[??] adocchiato da cotal famiglia,
2.  1 15 50    rispuos' io lui, <<mi smarri'[??] in una valle,
3.  1 17 55    che dal collo[??] a ciascun pendea[??] una tasca
4.  1 22 72    si che, stracciando, ne porto[??] un lacerto.
5.  1 23 116   consiglio[??] i Farisei che convenia
6.  1 26 56    Ulisse[??]e Diomede[??], e cosi[??] insieme
7.  1 26 136   Noi ci[??] allegrammo[??], e tosto torno[??] in
               pianto;
8.  1 33 15    or ti diro perche[??] i' son tal vicino.
9.  1 33 145   che questi lascio[??] il diavolo[??] in sua vece
10. 2 1 116    che fuggia[??] innanzi, si che di lontano
11. 2 3 58     da man sinistra m' appari[??] una gente
12. 2 3 122    ma la bonta[??] infinita[??] ha si gran braccia,
13. 2 4 124    di te [??] omai; ma dimmi: perche[??]
               assiso
14. 2 5 77     quel da[??] Esti[??] il fe far, che m' avea[??]
               in ira
15. 2 8 96     e drizzo[??] il dito perche 'n la guardasse.
16. 2 8 131    che, perche[??] il capo reo [??] il mondo torca,
17. 2 9 92     ricomincio[??] il cortese portina|io:
18. 2 17 30    che fu[??] al dire [??] e [??] al far cosi[??]
               intero.
19. 2 17 109   e perche[??] intender non si puo diviso,
20. 2 18 61    Or perche[??] a questa[??] ogn' altra si
               raccoglia,
21. 2 19 97    Ed elli[??] a me: <<Perche[??] i nostri diretri
22. 2 28 70    Tre passi ci facea[??] il fiume lontani;
23. 2 28 148   poi [??] a la bella donna torna'
               [??] il viso.
24. 2 33 40    ch' io veggio certamente[??], e pero[??] il narro,
25. 2 33 137   da scrivere [??], i' pur cantere'[??] in
               parte (4)
26. 3 1 7      perche[??] appressando se [??] al suo
               disire,
27. 3 20 137   perche[??] il ben nostro[??] in questo ben s'
               affina,
28. 3 29 51    turbo[??] il suggetto d' i vostri[??] alimenti.


Of course, many of the lines above can be scanned in more than one way. Example 13 could conceivably have a dialefe after perche rather than te; but this would be a much less likely solution, since it would move the accent away from the fourth and sixth syllables, the vast majority of lines in the Divine Comedy having an accent on one or both of these. A similar argument applies in the case of examples 14, 16, 23, and 26; example 18, on the other hand, could easily have a dialefe after cosi rather than e. Further instances could also be added to the list in Table 1, since there are a number of lines in the poem for which more than one scansion is feasible, for instance:

29. 2 10 43   e[??] avea [??] in atto[??] impressa[??]
              esta favella
30. 3 7 32    s' era[??] allungata[??], uni [??] a se[??]
              in persona.


where the dialefe could be moved from avea to e, or from uni to se; an exceptional sinalefe is produced whichever solution is adopted. The list in Table 1 does not include, moreover, thirty cases in this category where Petrocchi has a diaeresis on a final accented diphthong followed by sinalefe, and where for reasons that I shall discuss later I believe that synaeresis followed by dialefe would be a more likely presumption.

If these thirty cases are left on one side, it can be seen that the greater number of exceptional sinalefi in this category of polysyllables comes after final accented single vowels rather than final accented falling diphthongs, of which there are only four instances with sinalefe above (3, 10, 14, and 22), if we exclude cases where the final vowel of the diphthong has been elided (2, 23, 25). The important point for my purposes, at all events, is that these exceptional sinalefi are quite infrequent: a total of 327 dialefi after polysyllables in this category, compared with twenty-eight sinalefi in Table 1 above, gives a ratio of dialefi to sinalefi of just under twelve to one. Beccaria states that dialefe is 'di norma' between two accented vowels ('dialefe', p. 422); in fact, where the first word is a polysyllable, Table 1 shows no cases at all of sinalefe between a final accented vowel or diphthong and an initial accented vowel, if we except the initial vowel of una of examples 3 and 11, the accent of which is so slight in normal parlance as scarcely to count as such.

(2) The monosyllables that regularly produce dialefe in Dante's work are all those ending in a falling diphthong (mio, tuo, and so on), together with those listed in Table 2 (1) (see Appendix). Monosyllables are by far the largest producers of dialefe in the Divine Comedy, with 2133 occurrences: 1555 listed in Table 2 (1), plus 578 ending in a falling diphthong, the last figure including the one case of io with a diaeresis followed by dialefe. (5) As can be seen, the monosyllables listed in Table 2 (1) produce sinalefe rather than dialefe in 169 cases. On the other hand, there are very few cases of sinalefe after a monosyllable ending in a falling diphthong, just as in polysyllables sinalefe is less likely to occur after accented falling diphthongs than after accented single vowels. Sinalefe does not occur, as Beccaria points out ('dialefe', p. 423), after tuo and tua; however it does occur on a few occasions after mio, mia and suo: (6)
Table 2 (2): Exceptional sinalefe after mio, mia, and suo

31. 1 2 20     ch' e' fu de l' alma Roma[??] e di suo[??] impero
32. 1 11 35    nel prossimo si danno[??], e nel suo[??] avere
33. 2 1 2      omai la navicella del mio[??] ingegno,
34. 2 1 81     per lo suo[??] amore[??] adunque[??] a noi ti
               piega.
35. 2 31 87    piu nel suo[??] amor, piu mi si fe nemica.
36. 3 10 59    e si tutto 'l mio[??] amore[??] in lui si mise,
37. 3 13 105   in che lo stral di mia[??] intenzion percuote;
38. 3 20 121   tutto suo[??] amor la giu pose[??] a drittura:


Together with the sinalefe after io discussed below (example 69), these are the only cases I have found of sinalefe after a monosyllable ending in a falling diphthong, excluding the thirty further cases where I shall argue that Petrocchi's diaeresis followed by sinalefe should be replaced by synaeresis followed by dialefe. With this same exclusion, it is clear that there are no instances at all of accented falling diphthongs ending in i followed by sinalefe; the combination always produces dialefe (guardai [??] in alto ...), no doubt for the same reason as intervocalic i virtually always acts as a syllable divider within the word.

On the basis of these figures we arrive at a dialefe/sinalefe ratio for these monosyllables remarkably close to the 12-1 figure given above for polysyllables with final accented vowels or diphthongs, a figure that may therefore be taken as a rough index of the degree of regularity of this aspect of Dante's practice. (7) Can we find any motivation for the occurrences of sinalefe in this category? As we can see, it seems never to occur at all after the majority of the different forms listed in Table 2 (1). However, there are more cases of exceptional sinalefe than of dialefe after che, (8) and a few cases after che, chi, cio, di, e, e, fe, fu, gia, ha, ho, i', la, li, ma, o, piu, qui, se, se, se', si, su, tra, tu, va. (9) The absence of sinalefe after some of the other words in Table 2 (1) may therefore be the product of chance, not of a self-imposed rule. But as in the case of polysyllables with final accented vowels or diphthongs, sinalefe is most unlikely to occur after monosyllables in this group if the first vowel of the second word is accented. I have found only the following cases, for the most part after the adverb li:

39. 1 23 54    sovresso noi; ma non li[??] era sospetto:
40. 1 34 9     al duca mio, che non li[??] era[??] altra grotta.
41. 2 1 62     per lui campare[??]; e non li[??] era[??] altra
               via
42. 2 8 69     lo suo primo perche, che non li[??] e guado,
43. 2 13 7     Ombra non li[??] e ne segno che si pa|ia:
44. 3 7 37     ma per se stessa pur fu[??] ella sbandita
45. 3 16 44    chi [??] ei si fosser e[??] onde
               venner quivi,
46. 3 23 108   piu la spera supprema perche li[??] entre.


As Table 2 (1) shows, li is elsewhere followed by dialefe on four occasions; three of these involve the phrase li[??] era. (10) In

47. 2 20 52   Figliuol fu' [??] io d' un
                beccaio di Parigi:


beccaio might normally be expected to have three syllables, with a consequent sinalefe after fu'. But, as I shall show below, there are occasions in the Divine Comedy when intervocalic i does not act as a syllable divider; if we assume beccaio only has two syllables here, we avoid an exceptional sinalefe after fu', and the phrase accent on io falls on the fourth syllable, in line with Dante's frequent practice, rather than on the third.

Why does dialefe occur at all after these monosyllables? Beccaria (pp. 420-21) associates its use here with its use after accented final vowels or diphthongs in polysyllables, suggesting that it is the tonic status, actual or potential, of most monosyllables that produces the syllable division before the following word. It is indeed the case, on the one hand, that a large proportion of dialefe-producing monosyllables in Dante commonly carry an accent in the phrase, and on the other, that the monosyllables that never produce dialefe are proclitics that never carry an accent in normal circumstances. But there are also a number of monosyllables in the Divine Comedy that produce dialefe with great frequency and that generally do not, and in some cases almost never, carry an accent in the phrase: a, che, che, da, de, e, ma, o, se, si, su, tra, e, da and o account, as Table 2 (1) shows, for a very substantial proportion (768) of the dialefi in the poem; in 462 of these cases the following vowel is accented, in 306 it is not. (11) The more likely explanation, it seems to me, is to do with the fact that the monosyllables that produce dialefe are precisely those that in Tuscan normally produce syntactic doubling in a following consonant, whereas those that never produce dialefe are also those that do not produce syntactic doubling. One can see how the lengthening of a following consonant after these monosyllables could also have encouraged the metrical assumption of a syllable division before a following vowel. (12) Further support for this explanation may be found in the disappearance of dialefe in the sixteenth century, associated as it is with non-Tuscan writers who would not normally articulate double consonants: notably Boiardo, Ariosto, and Tasso. While dialefe is prominently used in Lorenzo de' Medici's poetry, it features scarcely at all in the contemporary Orlando innamorato. One counter-consideration should be noted: in present-day Tuscan, at least, sovra and qualche normally cause syntactic doubling, as well as come, dove, sopra when used as prepositions, (13) but these words do not produce dialefe in the Divine Comedy, with the exception of one sovra in 3. 26. 45 (example 95 in Table 4 below).
Table 4: Exceptional dialefi

62. 1 1 11     tant' era pien di sonno [??] a quel punto
63. 1 1 104    ma sapienza [??], amore[??] e virtute,
64. 1 4 30     d' infanti [??] e di femmine[??] e di viri.
65. 1 4 122    tra ' quai conobbi [??] Ettor ed Enea,
66. 1 6 10     Grandine grossa [??], acqua tinta[??] e neve
67. 1 7 103    L'acqua[??] era bu|ia [??] assai piu che persa;
68. 1 10 119   qua dentro[??] e 'l secondo Federico
69. 1 23 142   E 'l frate[??]: <<Io[??] udi' gia dire[??]
               a Bologna
70. 1 25 85    e quella parte[??] onde prima [??] e preso
71. 1 27 29    ch' io fui d' i monti la [??] intra [??] Orbino
72. 1 33 151   Ahi Genovesi[??], uomini diversi
73. 2 2 46     'In exitu[??] Israel de [??] Aegypto'
74. 2 2 105    qual verso[??] Acheronte non si cala.
75. 2 4 17     venimmo[??] ove quell' anime [??] ad una
76. 2 6 10     Tal era[??] io[??] in quella turba spessa,
77. 2 6 146    legge, moneta[??], officio [??] e costume
78. 2 10 44    'Ecce[??] ancilla Dei', propriamente
79. 2 12 2     m' andava[??] io con quell' anima carca,
80. 2 16 13    m' andava[??] io per l' aere[??] amaro[??] e
               sozzo,
81. 2 20 136   'Gloria[??] in excelsis' tutti 'Deo'
82. 2 23 2     ficcava[??] io si come far suole
83. 2 25 13    tal era[??] io con voglia[??] accesa[??] e
               spenta
84. 2 26 18    rispondi a me che 'n sete[??] e 'n foco[??] ardo
85. 2 26 21    che d' acqua fredda[??] Indo[??] o[??] Etiopo.
86. 2 26 146   que vos guida[??] al som de l' escalina,
87. 2 29 135   ma pari[??] in atto[??] e[??] onesto[??] e
               sodo.
88. 2 33 137   da scrivere[??], i' pur cantere'[??] in parte
89. 3 4 26     pontano[??] igualmente[??]; e pero pria
90. 3 16 107   era gia grande[??], e gia eran tratti
91. 3 16 139   era[??] onorata[??], essa[??] e suoi consorti:
92. 3 17 4     tal era[??] io[??], e tal era sentito
93. 3 22 75    rimasa[??] e per danno de le carte.
94. 3 26 17    Alfa[??] e[??] O[??] e di quanta scrittura
95. 3 26 45    di qui la giu sovra[??] ogne[??] altro bando.
96. 3 27 127   Fede[??] e[??] innocenza son reperte
97. 3 31 47    menava[??] io li[??] occhi per li gradi,
98. 3 31 109   tal era[??] io mirando la vivace
99. 3 33 136   tal era[??] io[??] a quella vista nova:


Beccaria points to certain combinations within the categories above as less likely to produce dialefe than others: in first position che, si che, per che, poi che, and perche, and in second position in or words beginning with the prefix in-, un, il, and i. To begin with first position: there are two cases of sinalefe and none of dialefe after both per che and poi che; perche produces eight sinalefi (listed in Table 1), che three sinalefi and si che two sinalefi, and they also produce the following cases of dialefe:
Table 3 (1): dialefe after perche, che, and si che

48. 1 2 123    perche[??] ardire[??] e franchezza
               non hai,
49. 1 25 48    che[??] io che 'l vidi[??], a pena[??]
               il mi consento.
50. 2 10 41    perche[??] iv' era[??] imaginata quella
51. 2 12 96    perche[??] a poco vento cosi cadi?
52. 2 15 90    perche[??] hai tu cosi verso noi fatto?
53. 2 28 103   Or perche[??] in circuito tutto quanto
54. 2 31 44    del tuo[??] errore[??], e perche
               [??] altra volta,
55. 3 5 127    ma non so chi tu se', ne perche [??] aggi,
56. 3 6 114    perche[??] onore[??] e fama li succeda:
57. 3 8 113    E[??] io: <<Non gia; che[??] impossibil
               veggio
58. 3 8 29     sonava[??] 'Osanna' si, che[??] unque poi
59. 3 10 26    che[??] a se torce tutta la mia cura
60. 3 28 57    che[??] io per me[??] indarno[??] a cio
               contemplo.


Example 58 is admittedly marginal to this argument, because of the comma Petrocchi places after si, and in 53 the dialefe after perche could be replaced by a diaeresis on circuito, especially since perche is followed by in. So far, therefore, Beccaria's suggestion seems to be correct, even if the number of words involved is not very great. The point can be given a broader bearing, however: Table 2 (1) shows that che is the only monosyllable that produces dialefe (47 occurrences) with any frequency while also producing sinalefe (62 occurrences) in an even greater number of cases. Taking all cases of che together, therefore, it does seem to be, along with che and perche, a less strong producer of dialefe than other similar words. It should be noted, though, that there do not seem to be any cases of interrogative che with sinalefe.

Beccaria's suggestion about words in second position (in or words beginning with the prefix in-, un, il, and i) can also be given a broader bearing. Table 3 (2) (see Appendix) breaks down all occurrences of dialefe according to the initial vowel (or the vowel following an initial h) of the second word in the combination (the 'following vowel'), and gives the corresponding figures for occurrences of sinalefe after those categories of words for which dialefe is the norm ('exceptional' sinalefe as discussed above, after polysyllables with accents on a final vowel or final falling diphthong, or after the monosyllables that normally produce dialefe). The Table shows clearly that words that normally produce dialefe are more likely to produce an exceptional sinalefe before an unaccented initial i in the next word than before any other vowel: only 21.77% of normal dialefi come before an i, as opposed to 71.36% of the corresponding exceptional sinalefi; indeed, very few exceptional sinalefi (10) occur before an accented vowel. (14) A look at the lists of occurrences in question (not reproduced here) also shows that the vast majority of sinalefi before i involve Beccaria's group: in or words beginning with the prefix in-, il, and i. Of the list of dialefi before an initial i, on the other hand, relatively few involve the definite articles i or il, rather more involve in or the (unaccented) prefix in-, and most (302 out of 544) involve an accented initial vowel, the pronoun io being the most common. Exceptional sinalefe thus occurs more easily before unaccented i, no doubt for the same reason as synaeresis does, as I shall show: because unaccented i naturally forms a diphthong with a preceding vowel. But even in this combination sinalefe still occurs somewhat less frequently than dialefe after the classes of words I have been considering so far.

On the other hand Beccaria's suggestion does not work so well in relation to un. I have counted forty-eight cases of dialefe before un, and only two cases of exceptional sinalefe: in 1. 13. 57, which I shall argue below should really be a dialefe, and in

61. 1 22 72   si che, stracciando, ne porto [??] un lacerto.


Table 3 (2) shows that the number of occurrences of dialefe before u is vastly greater than the corresponding number of exceptional sinalefi.

Table 3 (3) summarizes the figures given so far for the incidence of normal dialefi and the corresponding figures for exceptional sinalefi after the same groups of words. To these it adds figures for normal sinalefi: that is those occurring after the very short list of monosyllables given at the beginning of this discussion (the preposition di, definite articles and proclitic pronouns) and after all polysyllables ending in an unaccented vowel or diphthong; and it gives the corresponding figure for exceptional dialefi occurring after this last category of polysyllables. As the Table shows, by my count there are only thirty-nine cases in the Divine Comedy of exceptional dialefe, after a polysyllable whose final syllable is unaccented; whereas the number of instances of sinalefe in the same category is 9870. The list of dialefi in this category is given in Table 4.

Examples 73, 78, and 81 are explained by Dante's regular use of dialefe in Latin words, extended to Provencal in 86 and, possibly, Greek in 94. As earlier, some of these lines could be scanned diVerently, exchanging the places of dialefi and sinalefi, but they would still require a dialefe after an unaccented syllable: 63, 64, 66, 67, 70, 75, 77, 84, 85, 87, 89, 91. In the case of 64, 67, 89, however, moving the dialefe would produce unaccented fourth and sixth syllables, thus making the lines even more exceptional. In 69 and 88 my assumption of both an exceptional dialefe and an exceptional sinalefe (after Io and cantere' respectively) is probably preferable, again in order to avoid unaccented fourth and sixth syllables. 71 could have a diaeresis on io (which is quite common in the Divine Comedy, as I shall show) instead of the exceptional dialefe, but this would also lose the accented fourth and sixth syllables; so could 76, 79, 80, 83, 92, 98, 99 ('Tal era[??] i|o ...', 'm' andava[??] i|o ...') but at the cost of an unusual accent on the third and the consequent loss of the phrase break, typical of Dante's hendecasyllable, after an accented fourth or sixth. Similarly, in the following two lines Petrocchi seems to have assumed an exceptional dialefe after sangue and armava respectively, where a diaeresis on io would be a possible alternative (compare examples 148 and 149 below):

100. 1 24 129   ch' io 'l vidi[??] omo di sangue[??]
                e di crucci.
101. 3 24 49    cosi m' armava[??] io d' ogne ragione


My inclination would be to leave the dialefe in example 101, thus preserving the typical phrase-break after an accented sixth, but to replace it with a diaeresis on io in example 100, which in this case would make no difference to the phrase-break (the change is reflected in my statistics). In all but two of these cases of exceptional dialefe the first word is paroxytonic (a parola piana), which supports Beccaria's assertion that dialefe in Dante is prohibited after proparoxytones (parole sdrucciole); the two exceptions, if we wish to keep the accented sixth syllable, are 88 and 89. Table 4 also largely bears out Beccaria's assertion (p. 421) that dialefe is very rare between an unaccented vowel in the first word and an accented vowel in the second (about half the examples belong to this group).

There are by my count 1060 instances of diaeresis in the Divine Comedy. Of these about half (577) are marked by Petrocchi, and the remainder almost all involve vowel combinations that normally constitute two syllables, and where the diaeresis is therefore traditionally not marked; there are also a very few cases where Petrocchi ought to have marked a diaeresis, in my view, but has not done so, as well as those where I shall argue that his diaeresis ought to be replaced. The incidence of diaeresis depends partly on the combinations of vowels involved, and partly on the location of the accent; it is also strongly influenced by etymological considerations: learned words close to their original Latin form frequently preserve the syllable division between two vowels where there is such a division in Latin. I shall deal first with combinations of vowels beginning with a, e, and o, then with those beginning with i and u; the two sets of vowels are best dealt with separately, because of the often semiconsonantal character of i and u and the frequency with which they form diphthongs. The treatment will be broken down further, depending on the presence or absence of an accent in the first vowel of the combination, as follows:

(1) unaccented a, e, or o;

(2) accented a, e, or o;

(3) unaccented i or u;

(4) accented i, or u;

which will be followed by a separate treatment of

(5) intervocalic i.

The numerical breakdown of the diaereses in each of these categories is given in Table 5 (1).
Table 5 (1): Diaereses in the Divine Comedy

(1) unaccented a e o followed by accented vowel     387
    " " " " " unaccented vowel                       98
(2) accented a e o followed by vowel                 82
(3) unaccented i or u followed by accented vowel    325
    " " " " " unaccented vowel                       98
(4) accented i or u followed by vowel                70

TOTAL                                              1060


(1) Unaccented a, e, or o. When this is followed by an accented vowel the combination always counts as two syllables, as in pa|ura, le|one, or po|eta; as a result the diaeresis is not marked in modern critical editions, because it is not a diaeresis in the stricter sense defined above. As far as I can see, in the Divine Comedy there are no exceptions to this rule, which produces almost exactly a third of the 1060 diaereses in the poem. When unaccented a, e, or o is followed by an unaccented vowel there is also in many cases a syllable division between the two. Where the second vowel is an unaccented a, e, or o (78 cases), diaeresis is the norm, and is therefore usually not marked in Petrocchi's edition (Cle|opatra); when the second vowel is unaccented i or u (20 cases), diaeresis is usually marked (deiforme) because it is less common. (15)

In several cases unaccented combinations may acquire the diaeresis by association with cognate forms where the combination has the accent. Thus we have le|on (4), le|one (4) with the accent, and le|oncini (1), le|onessa (1), le|onine (1) without, all with diaeresis; po|eta (25), po|eti (5), and po|etar (1), po|etaro (1), po|etando (4); torne|are (1) and torne|amenti (1); pa|ura (30) and pa|urose (1), spaurato (1: here Petrocchi marks the diaeresis); a|una (1), ra|una (2) and a|unasse (1), ra|unai (1). Latinate terms and names with the same combinations of vowels also often keep the syllable division: Deidamia (1), Deidamia (1), deiforme (1), deita (1), deitade (1), reiterando (1), Eufrates (1), Euno|e (2), Moise(5). (16) Apart from several instances of the diphthong au from the Latin and the articulated prepositions ai, coi, dai, dei, nei, there are only a few words in this group that do not have the diaeresis (I do not include here monosyllables such as dai, fai, hai, mai, sai, stai, tai, vai, where the diphthong is accented, and which will be dealt with below under category (2)). Alongside various forms of creare with the accent on the a, all with diaeresis, we have cre|ature both with diaeresis (7 occurrences, not marked in the edition) and, in the following line, without:

102. 3 7 127   e queste cose pur furon creature;


While there is one instance of beatitudo and 21 of Beatrice, there are 43 cases of Beatrice with synaeresis. There are also Borea (1), geomanti (1), Mainardi (1), sciaurati (1), all with synaeresis, as well as Eurialo (1), Euripide (1), Euripilo (1), Europa (4), where the Greek eu diphthong has been preserved as the Latin au has been elsewhere, but contrast Eufrates (1), Euno|e (2). There are also either one or two cases of geometra with synaeresis, depending on whether the diaeresis falls on geometra or Euclide in

103. 1 4 142   E|uclide geometra[??]e Tolomeo,


The preceding palatal g might explain the absence of the diaeresis in geo-. The only other cases in polysyllables of synaeresis in this combination of vowels are Vedeisi in 2. 22. 112, where the -isi is enclitic (vi si vede), and the Latin diphthong ae in Summae, clementiae, and Aegypto, which always remains a single syllable.

(2) Accented a, e, or o followed by a (necessarily unaccented) vowel: here synaeresis is the overwhelming rule, since the combination forms a (falling) diphthong in normal parlance; as a result cases of diaeresis are generally marked in Petrocchi's edition. The total number of diaereses in this category is eighty-two; the full list of words with the diaeresis marked by Petrocchi is as follows: Anteo (1), avea (9), avean (1), correan (1), dicea (1), discendea (1), discernea (1), dovea (1), E olo (1), Enea (2), Eneida (1), facea (1), facean (2), ficcai (1), giacea (2), hai (2), imprendea (1), intendea (1), Israel (1), levai (2), loico (1), mai (1), parea (1), parean (1), Pegasea (1), percoteansi (1), piacea (1), piangea (3), poi (1), portai (1), potea (1), potean (1), ravvolgea (1), Rea (1), reflettea (1), Rifeo (1), Rodopea (1), sapean (1), solea (1), surgea (1), Taide (1), Tanai (1), Tarpea (1), temea (1), tenea (1), Teseo (2), Tideo (1), vedea (7), vedrai (1), vincea (1). Following his usual practice, Dante also preserves the syllable divisions in passages in Latin: Deus, meus, and cognate forms always have diaeresis, sometimes marked by Petrocchi. Many of the forms in the list above are Latinate, but compare Anfiarao (1) and Vincislao (1), Medea (1), Capaneo (1), Giudeo (1), Giudei (3), Pompeo (1), Romeo (2), Sicheo (1), Tifeo (1), Timbreo (1), Timeo (1), Euro (1), as well as Anteo (1) alongside Anteo (but see the next paragraph), all with synaeresis. This list of forms with synaeresis does not include a number that fall at the end of the line, where a diaeresis would never be marked; the convention is that any unaccented vowel following an accented vowel in the tenth position of the hendecasyllable counts as a notional eleventh syllable. There are also a number of more vernacular forms, almost all of them verb-inflections, which have diaeresis in a few cases in Petrocchi's edition, but mostly do not (for the reason just given, I also give the number of forms in each case that fall at the end of the line): avea (9) and avea (118, 2 at end of line), avean (1) and avean (13), dicea (1) and dicea (29, 1 at end of line), discernea (1) and discernea (1), dovea (1) and dovea (12, 3 at end of line), facea (1) and facea (32, 1 at end of line), facean (2) and facean (7), giacea (2) and giacea (3), hai (2) and hai (56, 5 at end of line), levai (2) and levai (9, 3 at end of line), mai (1) mai (139, 16 at end of line), parea (1) and parea (50, 3 at end of line), parean (1) and parean (11), piangea (3) and piangea (4, 2 at end of line), poi (1) and poi (318, 14 at end of line), portai (1) and portai (3), potea (1) and potea (18, 1 at end of line), potean (1) and potean (4), solea (1) and solea (7), tenea (1) and tenea (10), vedea (7) and vedea (18, 1 at end of line), vedrai (1) and vedrai (32, 4 at end of line), vincea (1) and vincea (1). (17) Further diaereses could be inserted, where Petrocchi has not done so, in

104. 1 10 65   m' ave|an di costui gia letto[??] il nome;
105. 1 19 3    de|on essere spose[??], e voi rapaci


and in Ra|ab in 3.9.116.

I shall again suggest below how, in a number of the words listed above, Petrocchi's diaeresis would best be replaced by a dialefe at the end of the word in question, although the replacement would make no significant difference to the scansion of the line: Anteo, avea (8 out of 9 cases), dicea, discendea, discernea, giacea (2), hai (2), intendea, levai (1 out of 2), mai, piacea, piangea (1 out of 3), poi, portai, potea, ravvolgea, reflettea, solea, tenea, vedea (4 out of 7), vedrai. The group of forms in this category that occur with diaeresis as well as without will therefore, if I am right, be substantially reduced. Given how many more of these forms occur without than with, and given the small numbers altogether of diaereses in this category, this reduced group may be taken as the strongest examples of exceptional diaeresis encountered so far. The group is as follows, excluding, as I said, the cases where I believe Petrocchi's diaeresis should be replaced:
Table 5 (2): Exceptional diaeresis (after accented a, e, o)

106. 1 5 140    l' altro piangea; si che di pietade
107. 1 10 71    ch' io facea dinanzi[??] a la risposta,
108. 1 18 117   che non parea s' era laico[??] o cherco.
109. 1 33 49    Io non piangea, si dentro[??] impetrai:
110. 2 12 28    Vedea Briareo fitto dal telo
111. 2 14 129   facean noi del cammin confidare.
112. 2 26 13    poi verso me, quanto potean farsi,
113. 2 28 147   udito[??] avean l' ultimo costrutto;
114. 2 29 127   e[??] or parean da la bianca tratte,
115. 2 29 147   dintorno[??] al capo non facean brolo,
116. 2 33 22    Si com' io fui, com' io dovea, seco,
117. 3 25 27    ignito si che vincea 'l mio volto.
118. 3 25 38    mi venne[??]; ond' io levai li[??] occhi[??]
                a' monti
119. 3 27 103   Ma[??] ella, che vedea 'l mio disire,
120. 3 29 9     fiso nel punto che m' avea vinto.
121. 3 31 49    Vedea visi[??] a carita suadi,


Example 104 above could also be added to the list. In this list the incidence of diaeresis seems not to depend at all on etymological considerations but simply to be a matter of free variation on Dante's part. I shall consider later whether there is any reason for such a variation.

(3) Unaccented i or u followed by another vowel generally forms a so-called rising diphthong in normal parlance. A substantial proportion of the vowel combinations in this category is made up by the vernacular diphthongs ie and uo from Vulgar Latin open e and o, or by semiconsonantal u plus vowel after g or q. Such combinations never have diaeresis in the Divine Comedy, nor does diaeresis occur after an i deriving from a Latin consonant (for example, piacere from placere), nor, evidently, when the initial i in the combination has a purely graphic function, as in angoscioso where the Latin i has been lost. Where diaeresis does occur it tends to depend, once again, on etymological considerations; it is worth noting that in the Latin passages in the Divine Comedy i and u always have diaeresis before a vowel, except obviously for the u after q. There is a substantial list of Latin names and Latinate words in this category of vowel combinations in which the syllable division of the original form is preserved, as in ardua, abituati, accidia, allevio. In fact, apart from gu- or qu- plus vowel and apart from the Italian diphthong uo from Vulgar Latin open o, all cases of unaccented u in the Divine Comedy have diaeresis when followed by a vowel, all for etymological reasons of one kind or another. (18) There are also a few more vernacular words such as disiar, fiata, and cognate forms, niente, obliando, piorno, Rialto, spiar, Chi|unque, where the diaeresis often reflects consonants lost from the original. (19) Altogether this category of vowel combinations accounts for well over a third (423) of the diaereses in the poem.

The prefix ri- from the Latin re- always produces diaeresis, as in riarse, riarso, riempie, riempion, riesca, with the exception of riede (10), riedi (4), perhaps because of their more strongly vernacular form. In other combinations of unaccented i plus vowel there are more marked inconsistencies, the Latin syllable division being preserved in some occurrences of a given word and not in others. There are both celestial (2) with the diaeresis and celestial (1) without, conversione (1) and conversione (1), distinzione (1) and both distinzione (1) and distinzion (2), furie (1) and furie (1), gloriosa (3) and gloriosa (1),20 grazia (1) and grazia (58; 10 at the end of the line), grazioso (4) and grazioso (1), infamia (1) and infamia (3) (and 'nfamia (1)), invidiosi (1) and invidiosa (1), Marzia (2) and Marzia (1), orazion (1) and orazion (5), orazione (1) (21) and orazione (2), passion (1) and passion (3), pietate (1) and pietate (2), sufficiente (1) and suYciente (1) (as well as the more vernacular fiate (12) and fiate (3)). And there are many other words of an equally Latinate character in which the i remains without diaeresis: condizion (7), condiziona (1), condizione (2), condizioni (3), discrezioni (1), disposizion (2), divozione (1), Dominazioni (1), Domizian (1), dubitazion (1), essalazion (1), fastidiosa (1), fastidiosi (1), Ottavian (1), perfezion (3), Trespiano (1). There seems to be a high degree of free variation here as well. However in cognate forms diaeresis is more likely when the second vowel is accented: gaudio (2) but gaudioso (2); gloria (21; 5 at end of line) but gloriosa and many other cognate forms with the diaeresis (but also gloriosa without, as already noted); grazia (58; 10 at end of line) and grazie (5; 1 at end of line) and only one grazia, but grazioso (4) (but also grazioso (1)); invidia (6) but invidiosi (1) (but also invidiosa (1)); lussuria (4) but lussuriosa (1); studio (4) but studiose (1); sustanzia (1) but sustanzial; varie (1) but variar (1).

Diaereses in this category are generally marked in Petrocchi's edition, but not in the following: I|arba and I|ole (2. 31. 72 and 3. 9. 102; the initial I in the original Latin names is probably vocalic rather than semiconsonantal), adri|ano (3. 21. 123 (adjective)), attribu|isce (3. 4. 45), chi|unque (2. 3. 103), distribu|endo (1. 7. 76), Du|era (1. 32. 116), glori|ai (3. 16. 6), influ|enza (3. 4. 59), intu|assi (3. 9. 81), Lu|igi (2. 20. 50), persu|ade (2. 33. 47), ru|ina and cognate forms (several instances), Samu|el (3. 4. 29), and sobri|a (3. 15. 99). It is also not marked, though it is required in both cases, on Iosu|e in 3. 18. 38 and patri|arca in 3. 22. 70; the same two words have a printed diaeresis in 1. 4. 58, 2. 20. 111, 3. 9. 125, and 3. 11. 121. Using a computer obviously makes it easier to be consistent in such matters.

(4) Accented i, or u followed by another vowel generally produces synaeresis. As with accented a, e, and o followed by another vowel, the number of diaereses produced by this combination is small (well under 100; taken all together, combinations in which the first vowel is accented thus produce only 15% or so of the diaereses in the Divine Comedy). The following should be a full list of the present category, counting for the moment only the diaereses marked by Petrocchi: colui (1), convertian (1), cria (1), cui (1), disia (1), disio (2), fia (1), gratuito (1), io (36), mia (1), mio (3), Niccosia (1), pia (2), pii (1), pio (3), Polimnia (1), sua (4), sue (1), suo (3), svia (1), venian (1), Zodiaco (1). Again, a number of these words occur both with and without diaeresis: colui (1) and colui (108; 4 at the end of a line); cui (1) and cui (157; 8 at end of line); disio (2) and disio (49; 24 at end of line); fia (1) and fia (54); io (36) and io (1091; 28 at end of line); mia (1) and mia (197; 4 at end of line); mio (3) and mio (308; 23 at end of line); sua (4) and sua (289; 2 at end of line); sue (1) and sue (84; 18 at end of line); suo (3) and suo (291; 1 at end of line); venian (1) and venian (4); pia, pii, and pio occur without diaeresis only at the end of line. Gratuito has a diaeresis, circuito does not. (22) In addition to the diaereses in this category marked by Petrocchi, I have suggested above (example 100) a further case where a diaeresis should be inserted on io. Conversely, as with some of the words in group (2) above, I shall also shortly suggest a number of instances where Petrocchi's diaeresis followed by sinalefe might best be replaced by synaeresis followed by dialefe: cria, fia, io (18 cases out of 45), Niccosia, Pio, sua (4), suo (1 out of 3).

Apart from io, which occurs with some frequency with diaeresis, the small group of forms in this category that occur with as well as without the feature gives a second set of cases of exceptional diaeresis, alongside those listed in Table 5 (2) (again I do not include the cases where I believe Petrocchi's diaeresis should be replaced):
Table 6 (1): Exceptional diaereses (after accented i, u)

122. 1 8 57     di tal disio convien che tu goda.
123. 1 14 32    d' India vide sopra 'l suo stuolo
124. 1 14 105   e Roma guarda come suo speglio.
125. 2 3 60     e non pareva, si venian lente.
126. 2 23 132   per cui scosse dianzi[??]ogne pendice
127. 3 3 128    ma quella folgoro nel mio sguardo
128. 3 16 141   le nozze sue per li[??] altrui conforti!
129. 3 23 10    cosi la donna mia stava[??] eretta
130. 3 26 37    Tal vero[??] a l' intelletto mio sterne
131. 3 29 48    nel tuo disio gia son tre[??] ardori.
132. 3 31 53    gia tutta mio sguardo[??] avea compresa,
133. 3 33 58    Qual e colui che sognando vede,


The much discussed instances of diaeresis on io are as follows in Petrocchi's edition:
Table 6 (2): Diaeresis on io

134. 1 2 40     tal mi fec' io 'n quella[??] oscura costa,
135. 1 3 11     vid' io scritte[??] al sommo d' una porta;
136. 1 4 134    quivi vid' io Socrate[??] e Platone,
137. 1 13 25    Cred' io ch' ei credette ch' io credesse
138. 1 21 94    cosi vid' io gia temer li fanti
139. 1 32 1     S' io[??] avessi le rime[??] aspre[??] e
                chiocce,
140. 1 33 113   si ch' io sfoghi 'l duol che 'l cor m' impregna,
141. 2 4 58     Ben s' avvide[??] il po|eta ch' io stava
142. 2 10 19    io stancato[??] e [??] amendue[??] incerti
143. 2 23 2     ficcava[??] io si come far suole
144. 3 5 85     Cosi Beatrice[??] a me com' io scrivo;
145. 3 10 145   cosi vid' io la gloriosa rota
146. 3 14 127   Io m' innamorava tanto quinci,
147. 3 24 56    sembianze femmi perch' io spandessi
148. 3 25 22    cosi vid' io l' un da l' altro grande
149. 3 28 85    cosi fec' io, poi che mi provide
150. 3 31 37    io, che[??] al divino da l' umano,
151. 3 31 47    menava[??] io li[??] occhi per li gradi,


I would add the one case where, I have proposed above, io should be given a diaeresis to avoid an exceptional dialefe:

100. 1 24 129   ch' i|o 'l vidi[??] omo di sangue[??] e di
                crucci>>.


As with the list of exceptional cases in Table 5 (2), here too diaeresis does not seem to depend particularly on etymological considerations. It has been observed by others, including Beccaria, that a number of instances of io with diaeresis could be explained by the fact that it is followed by the so-called impure s (s followed by another consonant), the diaeresis serving to facilitate pronunciation in the same way as impure s is sometimes preceded by the so-called prosthetic i. As can be seen, six of the eighteen cases of io above can be accounted for in this way, and so too can seven of the twelve cases of exceptional diaeresis in the preceding Table 6 (1). I shall consider later whether there is any other way of accounting for the exceptional diaereses in this category.

(5) The treatment of intervocalic i does not generally come under the headings of synaeresis or diaeresis, but the issues are obviously related. In word-internal position it seems always to be semiconsonantal in the Divine Comedy, except in the one case of Isa|ia, where it is accented, with a standard diaeresis on the preceding a; thus it generally acts as a syllable divider, but with a few exceptions. There are 1 migliaio (2 syllables, but miglia|ia with 3 in 2. 22. 36 and 3. 23. 28), 1 primaio (but prima|io in 1. 25. 76 and 2. 29. 145), 1 Uccellatoio, 1 gennaio, 1 Tegghiaio (and one Tegghia|io in 1. 16. 41):

152. 1 6 79     Farinata[??] e 'l Tegghiaio, che fuor si degni,
153. 2 13 22    Quanto di qua per un migliaio si conta,
154. 2 14 66    ne lo stato primaio non si rinselva.
155. 3 15 110   dal vostro[??] Uccellatoio, che, com' e vinto
156. 3 27 142   Ma prima che gennaio tutto si sverni


as well as the already mentioned

47. 2 20 52   Figliuol fu' [??] io d'
              un beccaio di Parigi:


This seems to make another set of cases of free variation on Dante's part. In word-initial position intervocalic i always begins a new syllable where it is semiconsonantal; where it is vocalic, as in io, i' (for io), i (the definite article) and the two Latin names Iole and Iarba, it does not, unless dialefe is produced for other reasons by the end of the preceding word.

I now turn to the cases where Petrocchi has a diaeresis in a final falling diphthong followed by a sinalefe, and where I believe, in line with the practice of Antonio Lanza's recent re-edition of the Comedy, (23) that it is better to assume synaeresis followed by dialefe. There has been some argument among metricists as to the choice in such cases. Most would opt for synaeresis followed by dialefe, but Bausi and Martelli prefer to assume diaeresis followed by a sinalefe for all cases of this kind involving falling diphthongs (io[??] avea rather than io [??] avea: see La metrica italiana, p. 15); Fasani, on the other hand, would see diaeresis followed by sinalefe only where the final falling diphthong ends in an open vowel (a, o, or u) and is followed by an unaccented vowel (io[??] avea, but Io [??] era and guardai [??] in alto) (La metrica della 'Divina Commedia', p. 20). All this of course raises the general question of what a diaeresis or dialefe actually represents in phonetic terms, a question I cannot attempt to answer here, though it does seem rather unlikely that Dante himself would have made a distinction as minute as the one that Fasani proposes. What can be addressed is the question of editorial consistency, and the related question of the consistency of Dante's metrical practice.

As I have already indicated, the words in question are the following: Anteo (1), avea (8), cria (24) (1), dicea (1), discendea (1), discernea (1), fia (1), giacea (2), hai (2), intendea (1), io (18), levai (1), mai (1), Niccosia (1), piacea (1), piangea (1), Pio (1), (25) poi (1), portai (1), potea (1), ravvolgea (1), reflettea (1), solea (1), sua (4), suo (1), tenea (1), vedea (4), vedrai (1). (26) The total number of instances involved is sixty. As Tables 5, 6 (1) and 6 (2) show, some of these words, and others similar to them, occasionally have a diaeresis elsewhere in Petrocchi's edition, though the diaeresis in such cases is exceptional: for instance, avea (1, and 1 avean), io (19), levai (1), piangea (2), Pio (1), suo (2, and 1 sue)), vedea (3). More to the point, Petrocchi himself elsewhere assumes synaeresis with dialefe, rather than diaeresis with sinalefe, in cases of exactly the same kind as those I am considering: for instance, in avea, dicea, discernea, fia, giacea, hai, levai, mai, poi, portai, potea, tenea, vedea, vedrai, (27) not to mention 167 cases of dialefe after io, 17 after sua, 21 after suo. Nor, as Fasani points out, is he consistently following the rule that Fasani himself proposes (La metrica della 'Divina Commedia', p. 20, n. 15): while most of his diaereses with sinalefe conform to this rule, three occur before an unaccented initial vowel, three (in 1. 33. 44, 2. 9. 103, and 3. 31. 100) do not, and several, as the list of words above shows, do not involve a diphthong ending in an open vowel.

The point of arguing this issue in some detail is not to criticize an editorial failure on the part of Petrocchi (who, after all, was working without the benefit of a computer); it is simply to remove the implication of inconsistency on Dante's own part. There is, of course, no explicit evidence of Dante's or his contemporaries' precise views on diaeresis and dialefe, since no discussion of the matter is to be found until it is briefly dealt with in an interpolation in Antonio da Tempo's treatise. (28) Nor, indeed, does it make much diVerence, for most purposes, where the syllable division falls in the cases I have been considering. But there is still no need to assume, as Petrocchi's practice might appear to suggest, that Dante sometimes put the syllable division in one position and sometimes, for no particular reason, in another. The important issue is that final accented falling diphthongs followed by an initial vowel regularly count as two syllables in the Divine Comedy, and it seems reasonable to conclude that they should all be treated in the same way, either as diaereses followed by sinalefe or as synaereses followed by dialefe; my own preference, which I have adopted throughout, is for the latter, as the simpler and more consensual solution. There are inconsistencies in Dante's metrical practice, in short, but my argument is that they are quite limited in number, and more limited than Petrocchi's use of the diaeresis might seem to imply.

To argue for Dante's consistency is not to say that there are always reasons for the exceptions to the rules. To begin with dialefe, Beccaria has argued, following others, that the feature is 'determinata molto spesso dalla pausa del pensiero', that it supports 'lente scansioni narrative' or creates 'scansioni che rilevano distinzioni del pensiero' (pp. 423-24). This certainly helps to explain why some of exceptional dialefi in Table 4 should occur: for instance, in the emphatic pairs or trios of nouns or adjectives in 63, 64, 66, 77, 84, 85, 87, 96, the chiasmus in 90, and the pronominal reduplication in 91. But it is also important to recognize that in a number of the cases considered the exceptional dialefe occurs for no apparent reason, other than as an element of free variation, and/or for convenience of scansion. If, moreover, we consider the incidence of dialefe in the Divine Comedy as a whole, the overall lack of connection between the device and any kind of breaks in language or thought is striking. This is illustrated most forcefully by the relationship between dialefe and sinalefe on the one hand and, on the other, mid-line punctuation marks, all of which, including simple quotation marks, must represent some kind of break in the discourse. As Table 10 shows (see Appendix), there are by my count 10559 instances of sinalefe in the 14233 lines of the Divine Comedy, 2499 instances of dialefe, and 5312 mid-line pauses represented by one or more punctuation marks. (29) Of such pauses, 1441 coincide with a sinalefe, and only 258 coincide with a dialefe, reflecting the fact that the largest number of dialefi (2133) occur after monosyllables, most of which are joined in a single phrase to the word that follows. In this substantial non-coincidence of metrical divisions and content-breaks could be seen another aspect of the autonomia del significante for which Beccaria himself has argued so effectively elsewhere. (30)

One advantage of a computer-based study is the ease with which it can identify lines with an especially high incidence of the features being considered. As Table 10 shows, 3 is the largest number of dialefi in a single line in the Divine Comedy, and it occurs in eighteen lines:
Table 7: lines with 3 dialefi

157. 1 2 99     di te[??], e[??] io[??] a te lo raccomando
158. 1 6 43     E[??] io[??] a lui: <<L' angoscia che tu
                [??] hai
159. 1 6 77     E[??] io[??] a lui[??]: <<Ancor vo' che mi
                'nsegni
160. 1 10 47    a me[??] e[??] a miei primi[??] e[??] a mia
                parte,
161. 1 33 110   grido[??] a noi[??]: <<O[??] anime crudeli
162. 2 3 80     a[??] una[??], a due[??], a tre[??], e l'
                altre stanno
163. 2 4 23     lo duca mio[??], e[??] io[??] appresso, soli
164. 2 8 89     E[??] io[??] a lui[??]: <<A quelle tre facelle
165. 2 9 31     Ivi parea che[??] ella[??] e[??] io[??]
                ardesse;
166. 2 10 63    e[??] al si[??] e[??] al no discordi fensi.
167. 2 17 79    E[??] io[??] attesi[??] un poco, s' io[??]
                udissi
168. 2 22 100   <<Costoro[??] e Persio[??] e[??] io[??] e[??]
                altri[??] assai,
169. 2 24 52    E[??] io[??] a lui[??]: <<I' mi son un che,
                quando
170. 2 27 138   seder ti puoi[??] e puoi[??] andar tra[??]
                elli.
171. 3 9 98     no|iando[??] e[??] a Sicheo[??] e[??] a
                Cre|usa,
172. 3 13 114   e[??] al si[??] e[??] al no che tu non vedi:
173. 3 26 17    Alfa[??] e[??] O[??] e di quanta scrittura
174. 3 33 126   e[??] intendente te[??] ami[??] e[??] arridi!


For the most part it is hard to see any special function or effect in these accumulations of dialefi, many of which result from strongly formulaic expressions (E[??] io[??] a lui[??] ...). I have already suggested that a major source of dialefe, the monosyllables, may be accounted for by the phonetics of Tuscan, and it seems likely that for a Tuscan speaker of Dante's time the syllable divisions in these lines were to be taken for granted. At most they serve to underline other stylistic devices, notably polysyndeton, and may perhaps be said to reinforce the characteristic immediacy, energy, and disruptiveness of Dante's language. If they slow down the pace of the narrative, as Beccaria seems to suggest, they do so noticeably only in conjunction with other features.

It is not wholly clear, however, how this view of the function of dialefe relates to Beccaria's understanding of that of sinalefe, which he believes 'da al verso un andamento meno concitato ma disteso' (p. 423), an interpretation supported, he suggests, by the fact that it is more frequent in Paradiso. As Table 10 shows below, this is not actually the case by my count: there are marginally more sinalefi in Paradiso than in Inferno, but significantly fewer than in Purgatorio. As the Table also shows, there are twenty-seven lines in the Divine Comedy with 4 sinalefi (10 in Inferno, 14 in Purgatorio, 3 in Paradiso), and one line only with 5 (in Inferno):

175. 1 28 68   con li[??] altri[??], innanzi[??] a li[??]
              altri[??] apri la canna,


The passage in question, which concerns Pier da Medicina, and leads up to his speech to Dante, is a rather unusual one: the period begins at the start of the previous terzina, includes an enjambement and mid-line break in the line just quoted ('ristato[??] a riguardar per maraviglia | con li[??] altri[??],') and overflows the end of the second terzina to end in the third syllable of the seventh line:
176. 1 28 70   e disse[??]: <<O tu cui colpa non condanna



Put together with the enjambement and overflow (the vast majority of Dante's periods end at the end of a terzina), the large number of sinalefi in example 175 might be said to contribute, with its effect of crowding, to a sense of instability or disruption well suited to the unpleasant scene the poet is witnessing. But in general the more plausible interpretation is probably to see the device as stylistically neutral. Many of the lines with four sinalefi consist, perhaps predictably, of enumerations or repetitions: (31)
Table 8: lines with 4 sinalefi (selection)

177. 1 6 16     Li[??] occhi[??] ha vermigli, la barba[??]
                unta[??] e[??] atra,
178. 1 8 118    Li[??] occhi[??] a la terra[??] e le ciglia[??]
                avea rase
179. 1 13 13    Ali[??] hanno late[??], e colli[??] e visi[??]
                umani,
180. 1 17 2     che passa[??] i monti[??] e rompe[??] i muri[??]
                e l' armi!
181. 1 32 38    da bocca[??] il freddo[??], e da li[??] occhi[??]
                il cor tristo
182. 2 3 107    biondo[??] era[??] e bello[??] e di gentile[??]
                aspetto,
183. 2 12 82    Di reverenza[??] il viso[??] e li[??] atti[??]
                addorna,
184. 2 12 131   e cerca[??] e truova[??] e quello[??] offcio[??]
                adempie
185. 2 14 109   le donne[??] e' cavalier, li[??] affanni[??] e
                li[??] agi
186. 2 19 85    e volsi li[??] occhi[??] a li[??] occhi[??] al
                segnor mio:
187. 2 28 55    volsesi[??] in su[??] i vermigli[??] e [??]
                in su[??] i gialli
188. 2 30 45    quando[??] ha pa|ura[??] o quando[??] elli[??] e
                [??] affitto,
189. 3 6 44     Romani[??] incontro[??] a Brenno[??],
                incontro[??] a Pirro,
190. 13 18      che l' uno[??] andasse[??] al primo[??] e l'
                altro[??] al poi;


If anything slows down the narrative here, in the way that Beccaria suggests, it is the lexical and/or syntactic structure rather than the accumulation of sinalefi in the line. Nor is it easy to see any stylistic point in the exceptional sinalefi in Tables 1 or 2(2). In this connection the view of Tasso on the subject in the Discorsi del poema eroico may be noted:

Il concorso de le vocali ancora suol producere asprezza o piacevol suono, come in quel verso 'fu consumato, e 'n fiamma amorosa arse'; ed in quelli altri di Dante, ne' quali non s'inghiottono le vocali, ma si fa quasi una apertura ed una voragine: 'poi e Cleopatra lussuriosa.' (32)

For Tasso the accumulation of juxtaposed vowels is a general stylistic ornament, not a source of special effect, regardless of whether it involves sinalefe (in the first quotation, from Petrarch), or dialefe and diaeresis (in the second quotation, from Inferno).

On the whole, therefore, I would tend to be sceptical in assigning a particular effect to either dialefe or sinalefe. It is certainly not necessary for the presence of one or the other of these features to affect one's pronunciation of the word after which it comes; for instance, to pronounce the word perche differently in 2. 4. 124 ['di te [??] omai; ma dimmi: perche[??] assiso') and in 1. 2. 123[ ('perche [??] ardire[??] e franchezza non hai') because it has a dialefe in the second case and a sinalefe in the first. In both cases the last e of perche and the initial a of the following word are enunciated distinctly, but without any break between them; whether one pronounces the second syllable of perche as longer in 48 than in 13 depends on the extent to which one allows metrical expectations to govern one's reading: one can do so, but one does not have to. All in all the only firm generalization one can therefore make about the eVect of the two features is that dialefe gives the possibility of lengthening the vowels it relates to, and sinalefe that of shortening them.

On the subject of the motivation of diaeresis there was an active debate in Dante criticism earlier in this century, (33) which Beccaria continued in his encyclopaedia article: diaeresis, while often unmotivated, often 'suggerisce scansioni lente e ritardi nel ritmo' (p. 434), and/or adds an eVect of solemnity to the verse, as in

191. 1 33 31   Con cagne magre, studiose[??] e conte
192. 3 23 25   Quale ne' plenilunii sereni


He observes, astutely, that while celestial sometimes has a diaeresis, bestial never does. No doubt diaeresis, like dialefe, can slow down the pace of recitation, by giving two syllables to vowel combinations that in some cases would occupy only one. But distinctions must be made between: (a) diaereses that are a normal consequence of vowel structure, in combinations beginning with unaccented a, e and o (category (1) above); (b) those that are the regular marker of a particular lexical register, as in the etymological diaereses that refer back to a word's Latin origins (mostly in category (3) above); and (c) exceptional diaereses, such as I have sought to identify in Tables 5 (2) and 6 (1), to which may be added the diaereses on io in Table 6 (2) (categories (2) and (4) above). Unless reinforced by some other feature, a single diaeresis of type (a) is unlikely to create much eVect, since it is a standard fact of pronunciation. Conversely diaereses of type (b) will always connote a degree of elevation, as in examples 191 and 192 above, the elevation being in most cases phonetically marked by the retention of the Latin vocalic i or u in place of the more vernacular semiconsonantal sounds. The majority of diaereses of type (c), Beccaria suggests, 'si verifica sotto accenti ritmici rilevati, soprattutto di 4a, 6a, spesso di 8a'; indeed, this is the case for most of the exceptional diaereses listed in the Tables 5 (2), 6 (1), and 6 (2), the remainder mostly involving accented second and first syllables. This convergence is virtually to be taken for granted, since the exceptional diaereses all involve accented syllables, and accented syllables are overwhelmingly concentrated in the positions that Beccaria has listed: after tenth, fourth, sixth and eighth, then second and first. Sometimes the exceptional diaereses in these Tables may contribute to an effect of special emphasis appropriate to the dramatic or emotional content of the line, by lengthening the vowels in question:

106. 1 5 140    l' altro piangea; si che di pietade
109. 1 33 49    Io non piangea, si dentro[??] impetrai:
129. 3 23 10    cosi la donna mia stava[??] eretta
133. 3 33 58    Qual e colui che sognando vede, (34)
146. 3 14 127   Io m' innamorava tanto quinci,
150. 3 31 37    io, che[??] al divino da l' umano,


Often, however, they seem to serve no purpose at all other than producing the correct syllable count.

But in the thirty-two lines with two diaereses in the Divine Comedy (there are none with more) there are some more interesting effects:
Table 9: lines with 2 diaereses

193. 1 1 69     manto|ani per patria[??] ambedui.
194. 1 4 58     Abra|am patriarca[??] e David re,
195. 1 4 73     <<O tu ch' onori scienzia[??] e [??] arte,
196. 1 5 63     poi [??] e Cle|opatras lussuriosa.
197. 1 8 19     <<Flegias, Flegias, tu gridi [??] a voto>>,
198. 1 14 32    d' India vide sopra 'l suo stuolo
199. 1 16 27    faceva[??] ai pie continuo viaggio.
200. 1 18 20    di Gerion, trovammoci[??]; e 'l po|eta
201. 1 30 8     la le|onessa[??] e ' le|oncini[??] al varco>>;
202. 1 34 62    disse 'l ma|estro[??], <<e Giuda Scariotto,
203. 2 1 125    so|avemente 'l mio ma|estro pose:
204. 2 2 46     'In exitu[??] Israel de [??] A|egypto'
205. 2 4 58     Ben s' avvide[??] il po|eta ch' io stava
206. 2 4 133    se[??] orazione[??] in prima non m' a|ita
207. 2 11 130   se buona[??] orazion lui non a|ita,
208. 2 12 28    Vedea Briareo fitto dal telo
209. 2 17 91    <<Ne cre|ator ne cre|atura mai>>,
210. 2 21 118   dal mio ma|estro[??], e <<Non aver pa|ura>>,
211. 2 23 11    'Labia mea, Domine' per modo
212. 2 25 121   'Summae De|us clementiae' nel seno
213. 2 28 131   E uno|e si chiama[??], e non adopra
214. 2 32 119   del triunfal ve|iculo[??] una volpe
215. 2 33 127   Ma vedi[??] E uno|e che la diriva:
216. 3 1 20     si come quando Marsia tra|esti
217. 3 1 29     per triunfare[??] o cesare[??] o po|eta,
218. 3 2 19     La concre|ata[??] e perpetua sete
219. 3 4 29     Moise, Samu|el, e quel Giovanni
220. 3 7 1      <<Osanna, sanctus De|us saba|oth,
221. 3 15 29    gratia Dei, sicut tibi cui
222. 3 26 123   fiate, mentre ch' io [??] in terra fu'mi.
223. 3 30 101   o cre|atore[??] a quella cre|atura
224. 3 31 49    Vedea visi[??] a carita suadi,


Views will difffer, evidently, as to the stylistic impact of many of these lines. But I would suggest that while in many cases the double diaeresis seems to serve no particular purpose (notably in the Latin lines 204, 211, 212, 220, 221), its elevating and retardant properties can contribute to a quite striking effect when combined, as they are on a number of occasions, with other stylistic features: assonance in 193, 194, 203, 210; paronomasia in 201 (plus the insane Atamante's striking metaphor), 209, 223; rare words or an unusual metaphor, sometimes reinforced by unusual word-order, in 196 (heightened by the preceding dialefe, as Tasso pointed out, with a reverse symmetry in the sounds of the first two and the second two syllables), 199, 214, 218 (developed further with the diaeresis on deiforme in the following line), and 224. The double diaeresis seems generally less interesting when it occurs in a single word, or the repetition of the same word (197, 213, 215); but scienzia in 195 acquires a special prominence through the association with arte, the sinalefe followed by dialefe, and the inverted word-order, all adding a particular solemnity to the poet's address to Virgil. Altogether the retardant effect of diaeresis works much more strongly when there is something there to linger over.

It will already be apparent from the last part of this discussion that there is a considerable difference in precision between the computer-derived description with which I started, and the stylistic interpretation that followed it. The computer analysis, that is to say, provides a basis or context for the interpretation; it does not provide it with ready answers. There are, of course, other things that the computer analysis can do, notably in the way of systematic comparison between different parts of the Divine Comedy, either different cantiche or different elements within each one: for instance, different lines in the terzina or different syllable positions in the line. More interestingly, it can develop comparisons between Dante and other writers in the terzina form, notably Petrarch and Boccaccio, and even later writers of epic such as Tasso. These are topics, however, for which there is no room in the present article, whose aim has been to produce the systematic starting-point for further exercises of this kind, as well as for the rhythmical analysis referred to at the beginning.
APPENDIX: TABLES

Table 2 (1): Monosyllables producing dialefe (excluding those
ending in diphthongs)

             Dialefe   Sinalefe                Dialefe   Sinalefe

a               15        --      ma                37       6
Be (35)          1        --      me                40      --
bee              1        --      mo                 4      --
che             47        62      ne                24      --
che              4         3      no'                1      --
chi             30         2      o (38)            50       1
cio              5         1      oh                 4      --
cu'             12        --      pie                1      --
da              55        --      piu              121       3
da               1        --      Po (river)         1      --
de (Lat.)        1        --      pro                1      --
de'(deve)        1        --      puo                4      --
dee              3        --      qua               12      --
deh              1        --      qui               24       1
D[i] (36)        1        --      re                 1      --
di              15         1      sa'                1      --
e              559        22      se (conj.)         8       7
e               70        17      se'               11       2
fa               6        --      se                30       2
fe               2         2      si (Lat.)          1      --
fo               1        --      si                47       1
fu              31         9      sii                1      --
fu'             13        --      so                 3      --
gia             34         4      sta                1      --
giu             15        --      sto                2      --
ha              14         4      su                 3       3
ha'              2        --      su                10      --
ho               8         1      te                14      --
I (37)           1        --      tra               11       2
i'              31         1      tre               12      --
la              31         3      tu                50       1
li               4         7      va                 9       1
lu'              1        --      TOTAL           1555     169
Table 3 (2): Vowels following dialefe and exceptional
sinalefe (39)

DIALEFE (all cases)

         Accented   Non-accented   Sub-total   % of total

a          217          721           938        37.54%
e          252          321           573        22.93%
i          302          242           544        21.77%
o          205           83           288        11.52%
u           65           91           156         6.24%
Total      1041        1458          2499       100.00%
EXCEPTIONAL SINALEFE

        Accented   Non-accented   Sub-total   % of total

a          0            42           42         20.39%
e          8             1            9          4.37%
i          0           147          147         71.36%
o          0             2            2          0.97%
u          2             4            6          2.91%
Total     10           196          206        100.00%
Table 3 (3): Total words with dialefe and sinalefe

WORDS NORMALLY PRODUCING DIALEFE             Dialefe   Sinalefe
Monosyllables ending in single vowel
  (see Table 2 (1))                          1555      169
Monosyllables ending in accented diphthong   578       9
Total monosyllables                          2133      178
Polysyllables with final accented vowels
  (including elided diph-thongs)             192       21
Polysyllables with final accented
  diphthongs                                 135       7
Total polysyllables                          32728
Total above                                  2460      206

WORDS NORMALLY PRODUCING SINALEFE            Dialefe   Sinalefe
Monosyllables ('ve, ci, di, la, le, li,
  lo, mi, ne, no, si)                        0         478
Non-oxytonic polysyllables                   39        9870 (40)

GRAND TOTAL                                  2499      10554
Table 10: Sinalefe, dialefe, diaeresis, punctuation, word-breaks
in the Divine Comedy: Position in the line and by cantica

Number        SINALEFE   DIALEFE       DIAERESIS

0                 6220         12005       13204
1                 5811          1975         998
2                 1887           235          31
3                  287            18
4                   27
5                    1
6
7
8
9
10
11
TOTALS           10559          2499        1060
Av. per 1.       0.741         0.175       0.074
of which:

Inferno           3403           841         361
Purgatorio        3729           861         303
Paradiso          3427           797         396

Number of
dialefi at
punctuation        258   (10.324% of
mark:                       dialefi)

Number of
sinalefi at
punctuation        144   (13.647% of
mark:                      sinalefi)

                       WORD-
Number        PAUSES   BREAKS

0               9715
1               3783
2                681       12
3                 50      191
4                  3     1109
5                  1     3080
6                        4419
7                        3483
8                        1468
9                         396
10                         67
11                          8
TOTALS          5312    87394 (41)
Av. per 1.     0.373     6.14
of which:

Inferno         1753    29428
Purgatorio      1926    29306
Paradiso        1633    28660

Number of
dialefi at
punctuation
mark:

Number of
sinalefi at
punctuation
mark:

The left-hand column ('Number') gives the number of occurrences
of each feature in the line. The other columns give the number
of lines in the poem containing that number of occurrences of
each feature. Thus there are 6220 lines with no sinalefi (row 1)
and so on.


(1) 'dialefe', 'dieresi' in Enciclopedia dantesca, 6 vols (Rome: Instituto della Enciclopedia Italiana, 1970-78), II, 420-24 , 432-36; E. Ciafardini, 'Dieresi e sineresi nella Divina Commedia', Rivista d'Italia, 13.1 (1910), 888-919 and 'Dialefe e sinalefe nella Divina Commedia', 17.2 (1914), 456-616; Pietro G. Beltrami, Metrica italiana (Bologna: Il Mulino, 1991), pp. 139-53; Aldo Menichetti, Metrica italiana (Padua: Antenore, 1993), pp. 173-359; Remo Fasani, La metrica della 'Divina Commedia' (Ravenna: Longo, 1992) pp. 19-27. I must here thank Peter Hainsworth and Giulio Lepschy for their generous help with this article, and the British Academy Humanities Research Board for a research leave grant, which enabled me to complete it.

(2) Dante Alighieri, La Commedia secondo l'antica vulgata, ed. by Giorgio Petrocchi, 4 vols (Milan: Mondadori, 1966-67).

(3) Petrocchi always elides di to d' before a vowel, with the one exception of 2 31 72: 'o vero[?] a quel de la terra di[??] I|arba', perhaps because of the initial I, followed by a vowel, of the following word (though, as noted below, this I is probably vocalic rather than semiconsonantal). There is therefore only this one case in the Divine Comedy of di with sinalefe, and there are no cases of di with dialefe.

(4) This line is discussed further below, on the subject of exceptional dialefe.

(5) 1 32 1: 'S' io [diconjunction] avessi le rime[??] aspre[??] e chiocce.'

(6) I do not include the one case in Petrocchi's edition of sinalefe after suo (3 31 12), which I shall argue below should be replaced by synaeresis with dialefe.

(7) 2133 divided by (169+9) = 11.98.

(8) 1 5 78, 1 5 94, 1 6 34, 1 7 53, 1 7 69, 1 8 4, 1 16 18, 1 16 32, 1 16 122, 1 17 90, 1 17 117, 1 18 18, 1 19 79, 1 20 30, 1 23 141, 1 26 123, 1 29 41, 1 29 63, 1 30 46, 1 33 30, 1 33 81, 1 34 31, 2 1 57, 2 2 4, 2 2 26, 2 3 124, 2 4 134, 2 5 110, 2 7 4, 2 7 66, 2 11 71, 2 12 8, 2 12 83, 2 13 150, 2 15 24, 2 18 93, 2 20 21, 2 28 43, 2 29 40, 2 29 88, 2 30 58, 3 2 92, 3 2 101, 3 3 12, 3 3 37, 3 6 39, 3 6 116, 3 10 14, 3 10 66, 3 11 31, 3 11 138, 3 12 26, 3 12 105, 3 14 133, 3 16 55, 3 20 16, 3 23 76, 3 24 148, 3 26 27, 3 27 121, 3 29 139, 3 32 85.

(9) che 1 11 113, 2 13 70, 3 15 61; chi 1 5 107, 1 11 58, 3 16 44; cio 1 16 77; di 2 19 38; e 1 1 90, 1 9 131, 1 16 73, 1 21 92, 1 25 113, 2 7 81, 2 10 43, 2 12 61, 2 12 67, 2 15 7, 2 18 107, 2 20 50, 2 22 86, 2 25 95, 2 27 17, 2 31 46, 3 2 97, 3 5 71, 3 9 133, 3 11 76, 3 23 92; e 1 12 71, 1 12 107, 1 18 1, 1 18 132, 1 26 48, 1 28 141, 1 32 7, 2 11 100, 2 18 72, 2 23 5, 2 25 54, 3 7 108, 3 9 112, 3 14 120, 3 19 99, 3 24 73, 3 30 30; fe 1 4 23, 3 19 102; fu 1 5 54, 1 19 63, 1 24 139, 2 9 137, 2 14 82, 2 28 142, 3 5 49, 3 29 55, 3 7 37; gia 1 10 34, 1 32 103, 2 4 136, 2 29 36; ha 2 6 132, 2 16 81, 2 18 51, 2 28 120; ho 3 10 25; i' 1 4 94; la 1 6 87, 2 9 50, 2 19 66; li 1 23 54, 1 34 9, 2 1 62, 2 8 69, 2 13 7, 3 23 108, 3 24 81; ma 1 10 51, 1 21 47, 2 2 93, 2 6 132, 2 6 134, 3 6 130; o 1 23 74; piu 2 13 98, 3 7 107, 3 8 28; qui 1 29 23; se 2 3 28, 2 4 85, 2 4 133, 2 17 84, 3 6 86, 3 12 81, 3 31 31; se 2 8 94, 3 7 32; se' 3 1 91, 3 22 7; si 1 29 2; su 2 4 26, 2 28 55 (2); tra 1 9 133, 3 18 51; tu 1 33 87; va 2 23 119.

(10) 2 10 33, 2 24 67, 2 27 59, 3 18 71.

(11) The criteria I have used for assigning accents are those in my article, 'Rhythm and Metre in the Divine Comedy', in In amicizia. Essays in Honour of Giulio Lepschy, ed. by Zygmunt G. Baranski and Lino Pertile (special supplement to The Italianist, 17 (1997), 100-116); I have counted as accented for the present purpose all categories of syllables that are treated as potentially accented in this article.

(12) The point is also made by Francesco Bausi and Mario Martelli, La metrica italiana. Teoria e storia (Florence: Casa Editrice Le Lettere, 1993), p. 14, though it seems an unnecessary complication on their part to suppose that the association of these monosyllables with syntactic doubling meant that they were somehow felt to be tonic.

(13) Anna Laura Lepschy and Giulio Lepschy, The Italian Language Today (London: Hutchinson, 1979), pp. 65-67.

(14) The criteria for assigning accents are those referred to in note 11 above.

(15) Inconveniently, from the point of view of the computer-analyst, editors sometimes put the diaeresis mark on the first, and sometimes on the second of the two vowels in question. Petrocchi's practice, which is fairly typical, is as follows: where the first vowel is i it always has the diaeresis mark; the same applies where it is u, except in 2 23 132: 'per cui scosse dianzi ogne pendice' (compare 3 33 58: 'Qual e colui che sognando vede',); where the first vowel is e, it carries the diaeresis, unless the second vowel is i, or unless it is u, where we have Eunoe but Eufrates and leuto; where the first vowel is a or o further inconsistency prevails, with Israel, levai, loico, mai, Maometto, portai, vedrai alongside ficcai, hai, Moise, poi, spaurato, Taide, Tanai. There would be something to be said for putting the diaeresis always on the first, or alternatively always on the second, of the two vowels, regardless of what they may be. The diaeresis is marked on the Latin Israel (2 02 46), ae in Latin normally being a diphthong; in Italian the diaeresis is not marked (Israel in 1 04 59).

(16) The full list in this group (where the second vowel in unaccented) is as follows: a|equar (1), a|unasse (1), amma|estrato (1), beatitudo (1), Beatrice (21), Bugge|a (1), Ca|orsini (1), Cle|opatra (1), Cle|opatras (1), co|agulando (1), cre|ator (1), cre|atore (1), cre|atura (15), cre|ature (7), Deidamia (1), Deidamia (1), deiforme (1), deita (1), deitade (1), E ufrates (1), E uno|e (2), E|uclide (1), empire|o (1), le|oncini (1), le|onessa (1), le|onine (1), Maometto (2), Moise (5), No|arese (1), pa|urose (1), po|esi (1), po|etando (4), po|etar (1), po|etaro (1), ra|unai (1), reiterando (1), sa|ettando (1), sa|ettaron (1), sa|ettava (1), so|avemente (3), so|avita (1), spaurato (1), teodia (1), torne|amenti (1).

(17) Dei and mea only occur with a diaeresis in Latin, and Rea with a diaeresis only as the Biblical name.

(18) With the one exception of the Provencal diphthong in puesc (2 26 141).

(19) The full list of words with the diaeresis marked by Petrocchi is as follows (not including words in Latin): nvetriate, 'nviarci, 'nvieranno, 'nvidiosi, ardua, abituati, accidia, accidioso, aVettuoso, allevio, ambrosia, Anfiarao, Anfione, annual, anterior, anzian, Ariete, aspersion, Assuero, audienza, balbuziendo, Briareo, Caliope, casual, celestial, celestiali, circuir, Ciriatto, Clio, continua, continuo, continuando, continuamente, continuo, conversione, coscienza, Curio, Daniel, Daniello, Diascoride, dieta, Dione, Diogenes, Diomede, Dionisio, diurno, disiando, disiante, disianza, disiar, disiassimo, disiata, disiati, disiato, disiava, disioso, distinzione, disviando, disviate, disviato, division, eYgiata, elezion, elezione, elezioni, Elios, Eliodoro, eresiarche, esperienza, esuriendo, Etiope, Etiopia, Etiopo, Ezechiel, Fialte, fiata, fiate, Flegias, furie, furiosa, Gabriel, Gabriello, Galieno, gaudiose, gaudioso, Gerion, Gerione, giovial, gloriar, gloriarla, gloriosa, gloriose, gloriosi, glorioso, gloriosamente, grazia, Grazian, grazioso, ianua, idioma, Ilion, impetuosa, impetuoso, India, inebriate, inebriava, infamia, intellettual, inviasti, invidiosi, Iosue, Iperione, Iustiniano, Labia, lioncel, lilia, Livio, lussuriosa, Marsia, Marzia, Madian, mandrian, meridian, meridiana, meridiano, mutua, mutui, nescia, niente, Niobe, obediendo, obliando, oblivion, odierno, oppinion, oppinione, orazion, orazione, Oriaco, Oriente, oriental, oriuoli, Ostiense, patria, passion, patriarca, pazienza, perfezione, perpetua, perpetue, perpetui, perpetualemente, pietate, piorno, Pigmalion, plenilunii, potenziata, presuntuoso, presunzion, preziosa, prezioso, progenie, pueril, puerili, puerizia, radial, radiando, razionabile, region, religione, Rialto, riarse, riarso, riaccesa, riarmar, riavesse, riempie, riempion, riesca, riudir, sapienza, Scariotto, scienza, scienzia, Scipion, Scipione, scuriada, settentrion, settentrional, Siestri, Sion, Silvio, spezial, spiar, storiata, straniasse, studiose, suadi, suYciente, suYcienti, sustanzial, sviando, sviati, travio, triangol, triunfa, triunfo, triunfal, triunfale, triunfando, triunfante, triunfanti, triunfar, triunfare, triunfaro, Trivia, Tulio, umiliato, variar, variazion, viaggi, viaggio, viole, violenta, violenti, violenza, virtualmente, vision, visione, Vitaliano.

(20) In 3 10 145: 'cosi vid' io la gloriosa rota'; but there could be a diaeresis on gloriosa rather than on io. All other forms cognate to gloria have the diaeresis on i when the following vowel is accented: gloriai, gloriar, gloriarla, gloriosa, gloriose, gloriosi, glorioso. On the other hand gloria occurs with diaeresis only in Latin: 2 20 136: 'Gloria [??] in excelsis' tutti 'Deo'.

(21) In 2 4 133: 'se[??] orazione[??] in prima non m' a|ita'; but there could be a dialefe after se and no diaeresis on orazione.

(22) See example 53, but also the related discussion.

(23) Dante Alighieri, La Commedia. Nuovo testo critico secondo i piu antichi manoscritti fiorentini, a cura di Antonio Lanza (Anzio: De Rubeis, 1995).

(24) Cria o (3 3 87: 'cio ch' ella cria[??] o che natura face') in Petrocchi becomes cria e in Lanza's edition, but keeps the diaeresis, which in this case seems a reasonable alternative, on the analogy of the many instances of creature (but also one of creature) in the poem; the only other occurrences of cria are in rhyme position, where the diaeresis would be irrelevant.

(25) The line in question is 3 27 44: 'e Sisto[??]e Pio[??] e Calisto[??] e [??] Urbano'. The grounds for replacing the diaeresis on Pio with a dialefe are perhaps less strong that with the other forms we are considering here, since elsewhere pio only occurs without diaeresis at the end of the line. However on the other two occasions where it occurs with diaeresis it is an adjective, and here it is a name; so perhaps here the diaeresis is less likely.

(26) In 1 13 57; 1 13 82; 1 14 26; 1 14 93; 1 15 28; 1 15 53; 1 18 43; 1 21 19; 1 24 127; 1 25 130; 1 25 133; 1 26 135; 1 27 82; 1 29 35; 1 30 23; 1 30 119; 1 31 19; 1 31 90; 1 31 139; 1 33 44; 1 33 130; 1 34 53; 1 34 116; 2 3 18; 2 6 64; 2 8 34; 2 9 82; 2 9 83; 2 9 103; 2 24 77; 2 24 83; 2 25 2; 3 3 87; 3 13 50; 3 16 120; 3 17 2; 3 19 94; 3 19 146; 3 20 114; 3 21 49; 3 24 20; 3 25 82; 3 25 121; 3 26 5; 3 26 90; 3 26 123; 3 27 19; 3 27 30; 3 27 44; 3 27 79; 3 28 40; 3 28 78; 3 28 97; 3 30 93; 3 30 99; 3 31 12; 3 31 58; 3 31 77; 3 31 78; 3 31 100. In all these cases Lanza's edition eliminates the diaeresis, except for 1 14 93 and 1 15 53 (where it is retained, but without a sinalefe, since the following initial vowel is lost), 1 18 43 (where the io is lost altogether), and 3 3 87 (see note 24). In 1 24 127, 1 29 35 and 2 24 83 Lanza removes the diaeresis on io in his text, but omits to include the removal among the list of changes in his introduction.

(27) In 1 1 122; 1 3 127; 1 4 12; 1 4 49; 1 5 63; 1 6 67; 1 9 17; 1 9 17; 1 9 95; 1 11 8; 1 11 17; 1 11 62; 1 13 62; 1 14 22; 1 14 40; 1 14 108; 1 16 106; 1 20 97; 1 25 23; 1 25 100; 1 28 123; 1 29 19; 1 29 68; 1 31 88; 1 31 100; 1 33 9; 1 34 85; 2 1 130; 2 3 110; 2 5 19; 2 9 63; 2 10 43; 2 12 38; 2 12 43; 2 17 35; 2 19 29; 2 19 37; 2 19 49; 2 22 94; 2 25 50; 2 25 97; 2 26 80; 2 27 89; 2 28 148; 2 31 73; 3 3 106; 3 5 32; 3 6 83; 3 6 123; 3 6 136; 3 7 11; 3 8 44; 3 13 107; 3 19 65; 3 20 85; 3 20 107; 3 21 37; 3 22 15; 3 26 136; 3 27 90; 3 30 45.

(28) Antonio da Tempo, Summa Rithimici Vulgaris Dictaminis, ed. by Richard Andrews (Bologna: Commissione per i Testi di Lingua, 1977), pp. 104-06, 128.

(29) In the 2057 hendecasyllables of the Rime there are about 1400 instances of sinalefe, about 200 of dialefe, and over 600 mid-line pauses; 12 dialefi by my count coincide with a punctuation mark, as against more than 10 times as many sinalefi.

(30) Gian Luigi Beccaria, L'autonomia del significante (Turin: Einaudi, 1975).

(31) The other lines with four sinalefi are: 1 9 73; 1 14 13; 1 19 25; 1 22 86; 1 27 31; 2 1 100; 2 17 28; 2 17 111; 2 21 10; 2 22 119; 2 23 22; 2 23 41; 3 6 38.

(32) Discorsi del poema eroico, Book v, in Torquato Tasso, Discorsi dell'arte poetica e del poema eroico, ed. by Luigi Poma (Bari: Laterza, 1964), p. 203. The quotations are from Petrarch, Canzoniere ccciv, 2, and from Inferno, 5 63 ('poi[??] e Cle|opatras lussuriosa').

(33) See, for instance, Mario Casella, 'Studi sul testo della Divina Commedia', Studi danteschi, 8 (1924), 5-85; Salvatore Frascino, 'Suono e pensiero nella poesia dantesca', Giornale storico della letteratura italiana, supp. 24 (1928), 1-138; Santorre Debenedetti, 'Intorno ad alcuni versi di Dante', 87 (1934), 74-99.

(34) Petrocchi here replaces the earlier version 'Qual e colui che somniando vede': this is perhaps a more effective diaeresis because of the rarity of the word, even if less satisfactory for the textual critic.

(35) 3 7 14: the first syllable of Beatrice.

(36) 3 18 78: the letter D.

(37) 3 18 78 again: the letter I.

(38) Both conjunction and vocative.

(39) Includes the sixty forms where I argue Petrocchi's diaeresis with sinalefe should be replaced by synaeresis with dialefe.

(40) See note 39.

(41) 87,394 word-breaks plus 14,233 words at the end of the line makes a total of 101,627 words in the poem.

<ADD> DAVID ROBEY UNIVERSITY OF READING </ADD>
COPYRIGHT 1999 Modern Humanities Research Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Southworth, Eric
Publication:The Modern Language Review
Date:Jan 1, 1999
Words:15644
Previous Article:Counting syllables in the Divine Comedy: A computer analysis.
Next Article:The angel of history and the truth of love: Merce Rodoreda's La Placa Del Diamant.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters