• Introduction: Rules for scansion
• Scanned line
• Rhythm / Odd / Even
• Word Accents
• Irregular lines
• Technical method
• Detailed rules for scansion (Information page)

This section of the database has presented the greatest challenge in terms of scholarly method. The data presented here derives from edited versions of the texts marked up in such a way as to allow correct scansions to be generated automatically. The edited versions are presented in the Scanned line field, from which the other fields in the Rhythm section are in turn derived through a set of automatic procedures. The process of editing and marking up the texts, as described in the Technical method section below, has been greatly facilitated through the use of computational methods, but has also required a substantial degree of manual intervention. It has further entailed the formulation of a systematic set of rules for the assignment of accents to the verse line, in order to ensure the maximum degree of consistency in the resultant scansions.

The rules formulated for the present purpose are predominantly linguistic rather than metrical. This results partly from the fact that the development of the present section of the database started with the Divina Commedia, and it is by no means clear that Dante always followed all the rules for the Italian hendecasyllable that later became canonical: while we know from the De vulgari eloquentia that Dante followed the standard definition of a hendecasyllable as a verse line whose final accent falls on the tenth, we do not know whether or not he consciously accepted the later requirement of an additional major accent on the fourth or the sixth followed by a phrase boundary and often a pause. Almost all Dante's hendecasyllable, but not quite all, have an accent on the fourth or sixth, and in the majority of these, but by no means always, this accent is followed by a phrase boundary.

The one fundamental metrical rule followed in this database is therefore that of the final accent on the tenth. All the lines in the database have been scanned on this principle, and in a very few cases adjusted in order to conform to it: see Irregular lines below.  The first stage of the markup process involved the insertion of syllable divisions where needed in order to produce such a result. For the 16th-century texts in our corpus the number of such insertions is quite limited, in accordance with the practice that became standard in Italian versification, of treating almost all adjacent vowels as parts of a single syllable; but the earlier texts, particularly Dante's, required the extensive insertion of diaeresis and dialefe marks (syllable divisions between vowels across word boundaries) in as consistent a fashion as possible: this can be tracked in the database using a search suggested in Scanned line section below.

In the next stage accents were  marked on the basis of purely linguistic rules, following the practice of normal modern parlance; these rules are described in detail in a separate information page on Detailed rules for scansion. Metrical rules have been allowed to override these linguistic rules only in a very few cases at the end of the line, where for the sake of the rhyme it has been necessary to make an unaccented tenth syllable accented or an accented eleventh syllable unaccented; for instance the ne la which rhymes with vela in Purgatorio 17.55:

«Questo è divino spirito, che ne la

or the non ci ha which rhymes with oncia in Inferno 30.87:

e men d' un mezzo di traverso non ci ha.

These instances can all be tracked using  one of the search methods illustrated in the Rhyme Information page.

Following the practice of modern parlance means that the scansion produced by these rules represents a potential performance of the texts, and in fact the rules were derived from a series of actual performances of the Divine Comedy by a group of Italian actors. There are however some important qualifications. First, no distinction is made between degrees of accent, in contrast to traditional Italian metrical analysis, which limits itself to the primary accents in a line; such a distinction would have been virtually impossible to apply consistently. Second, no allowance is made for the variety which characterizes actual parlance. The syllables marked as accented are all typically pronounced as such in normal parlance, but not always: depending particularly on the speed of enunciation, especially in the case of monosyllables, the accent is often lost. If the scanned line is the representation of a potential performance, therefore, it is a performance of an unusually slow, emphatic and monotonous kind: that is the price to be paid for consistency.

See the more detailed information page on Detailed rules for scansion, and the fuller discussion of the procedures adopted in Robey, Sound and Structure in the ‘Divine Comedy', and (with Annalisa Cipollone and Paola Nasti) 'Rhythm and Metre in Renaissance Narrative Poetry' (details are given under Publications).

The line in this box is an edited version of the text marked up in terms of syllable division and accent, as described above. Its purpose is (1) to explain how the line has been scanned, and (2) to create opportunities for metre-related searching. All punctuation and diacritics have been removed from this field to make the searching easier. The following coding system has been used:

• Accented vowels are followed by the number of the syllable to which they belong. To improve legibility, the number 10 has been replaced by 0. All other vowels are counted as unaccented.
• ^ stands for sinalefe, | for dialefe (i.e. for, respectively, the absence and the presence of a metrical syllable division between adjacent vowels on either side of a word-boundary).
• + marks a syllable division between adjacent vowels within the word. Sometimes this matches a diaeresis mark in the edition used (as displayed in the Line box), but very often it does not, since editors only mark such divisions in a limited number of cases. Otherwise all adjacent vowels within the word have been counted as parts of a single syllable.
• Semivocalic i is also preceded by a + sign on all the occasions (the vast majority) where it marks the beginning of a metrical syllable after a preceding vowel, in either word-internal or word-initial position.

Otherwise the text in this box is the same as that in the Line box, except in the few cases listed under Irregular lines below, and where changes have been made to the Rhyme word field in the Conquistata, as detailed in the Rhyme and rhyme words information page.

Examples:

• Search for a particular word in a particular metrical position; for instance, enter  co6rpo with Match any part in the Scanned line box and select Divina Commedia in the Name of text box, to find possible examples of Contini's echi di Dante in Dante: there are 10 instances in the Comedy of corpo with the accent in the sixth syllable.
• Enter | in the Scanned line box with Match any part and, one after another, select different texts in the Name of text box in order to track the incidence of lines with dialefe in the different texts in the databases.
• Enter |%|%| in the Scanned line box with Match any part to find lines with an exceptionally high incidence (3) of dialefe, ^%^%^%^%^%^ for lines with an exceptionally high incidence (6) of sinalefe. These searches make use of the Wildcard %.
• Enter [0-9]^ _[0-9] in the Scanned line box, again using Wildcards, to find all instances where two successive accented syllables are separated by a sinalefe: none in Dante and Petrarch, very few in Tasso, more in Ariosto and more still in Pulci and Boiardo.

See Introduction: Rules for scansion above for the criteria that guided the scansion process.

In the Rhythm box accented syllables, as marked in the Scanned line field, are represented by the syllable number, and unaccented syllables by hyphens. The 10th syllable is represented by 0, and is by definition always present; syllables after the 10th are not represented in this field, but they are represented in the Word Accents field. Thus ---4--7--0 represents all lines with accents only on the 4th, 7th and 10th.

The Odd and Even boxes give respectively the number of accented odd and even syllables in the line, including the 10th.

Examples:

• Enter the following in the Rhythm box, matching either whole fields or any part : ___-_-___ (3 underscores + hyphen + 1 underscore + hyphen + 3 underscores). This will find all non-canonical lines, without an accent on either the 4th or the 6th syllable.
• Enter 4_- or -_6 in the Rhythm box with Match any part to find all lines with, respectively, an accent on the 4th but not the 6th or an accent on the 6th and not the 4th, in order to gain an indication of the balance between a minore and a maiore hendecasyllables.
• Select 5 accents in the Even box, O accents in the Odd box, 8a rima in the Metre box, and then, one after the other, Line  1, Line 2, Line 3, etc., in the Line position box: you will find that in the ottava rima texts the largest number of lines with accents on all the even syllables and none on the odd ones is in Line 8.
• Enter 5 in the Odd box to find all lines with an exceptionally high incidence of accents on odd syllables.

Word boundaries are represented by the forward slash /, unaccented syllables by the hyphen -, and accented syllables by the syllable number. As in the Scanned line field, 10 is represented by 0. Words without vowels (l', ch', n', etc.) have been omitted.

While word boundaries may often not be audible in spoken Italian, they clearly play a part in listeners' and readers' perception of the rhythm of a line. The impact of Inferno 5.142

E caddi come corpo morto cade [/-/2-/4-/6-/8-/0-/]

depends in part on the fact that each accent begins a new two-syllable word. The Word accents box allows structures such as this to be searched for.

Examples (some using Wildcards:

• Enter 0-- with Match any part in the Word accents box to find all lines with proparoxytonic rhymes (rime sdruccioleI).
• Enter 7--/0-/ with Match any part in the Word accents box and select both the Liberata and the Conquistata in the Name of text box, to find examples in Tasso of the line-endings in a 'dactyl' followed by a 'spondee' that he praises in the Discorsi del poema eroico.
• Enter /-/2-/-/-6/-/8-/0-/ in the Word accents box and select the Divina Commedia in the Name of text box, to see how many lines in the Divine Comedy have the same word-accent structure as Inferno 1.1: one of them is Paradiso 1.1.
• Enter [13579]/[13579] with Match any part in the Word accents box, and follow it by a second search for [24680]/[24680], to find the few instances in which two accented words belong to the same metrical syllable as a result of sinalefe: for instance Gerusalemme liberata 20.64.7, where or is in the same metrical syllable as the second syllable of Così.
          Così or paventa ed or desia che tocchi [/-2/2/-4-/-/6/-8/-/0-/]

Splitting the search in this way seems rather complicated, but it is necessary in order to exclude the many lines with successive accented syllables on either side of a word boundary. Finding two accented words in the same syllable is of course a consequence of strict adherence to the rules followed here; whether in a normal reading both syllables would be accented is a matter for debate, but even if one of the two is only potentially accented it is quite unusual to find such a combination in a single syllable.

Following one of the search examples suggested above, under Rhythm, will reveal a good many hendecasyllables in the database that are represented as non-canonical, in the sense of having no accent on either the fourth or the sixth. In many cases this is due to strict adherence to the principles of scansion that have been used, which generally means that in normal parlance there would be no accents on the words in those positions. That does not mean, of course, that accents could not be placed on the fourth or sixth in most such cases: merely that to do so generally involves stretching the rules of normal parlance, as often happens in verse.

A few lines in the editions used for the database, particularly in the Gerusalemme conquistata, are not capable, as they stand, of scansion as hendecasyllables in the minimal sense of having the final accent on the 10th. In order to make them comparable with the rest of the database contents, they have been adjusted in the Scanned line box (but not the Line box) so as to have a final accent on the 10th, and thus to assume hendecasyllable form. Five of these lines are in our edition of the Morgante (3.80.3; 12.6.7; 18.137.3; 27.150.2; 27.275.6), and in all but the second instance we have solved the problem by following the versions in Domenico De Robertis's edition, Morgante e lettere (Florence: Sansoni, 1984). In the second instance we have had to do more violence to the text, by removing the final vowel of miserere in 'peccavi Domine, miserere mei'. Four further instances are in our edition of the Orlando innamorato (1.5.28.7, 2.4.65.3, 2.7.5.6, 2.27.15.6), where we have solved the metrical problem by following the more recent edition by Antonia Tissoni Benvenuti and Cristina Montagnani; for details of this, and for more information on our versions of both Pulci and Boiardo, see Morgante and Orlando innamorato: the digital texts.

For the same reason, for the Scanned line box we have had to implement a number of emendations to the notoriously inadequate Bonfigli text of the Gerusalemme conquistata. The lines in question, with our corrections in square brackets, are as follows:

In instances 2, 3, 4, 6, 9, 10, 11 our corrections correspond to the matching lines in the Liberata; in the other instances there is no exact match in the earlier poem. In all these cases the text has been left in its original form in the Line box and text files.

The analysis of Italian metre lends itself well to computer-based methods. 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. 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 is perfectly adequate for present purposes. 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 it also has frequently to be supplemented through the manual insertion of syllable divisions within the word (diaereses), as well as marks for hiatus (dialefe) between adjacent vowels on either side of a word boundary.

The processing for accent-marking is only partly automatic, and relies very substantially on manual intervention in the first of its two stages. This uses a computer program as a prompt and aid: it proposes a metrically scanned version of each line of the text, applying the above system of syllable divisions, and assigning accents to words of more than one syllable through a simple set of algorithms. The algorithms produce the correct linguistic accent in some 80% of cases; where the accent produced is wrong, the adjustment has to be made by hand the first time it is needed, but is then registered by the program in a table, and routinely applied from then on. Similarly the assignment of accents to monosyllables is recorded in a table and then routinely applied, in accordance with the rules for scansion referred to above. Further manual intervention is also needed for the resolution of cases of ambiguity and for the application of the positional criteria included in the rules for scansion. The outcome of the the process is the electronic version of the text in the Scanned line box, that is sufficiently marked up in terms of accent and syllable divisions to allow correct scansions, according to the rules described, to be produced entirely automatically, and then used for the Rhythm, Odd, Even, and Word accents fields.