Detailed rules for scansion

To be read in conjunction with the Introduction: Rules for scansion in the Rhythm and metre Information page.

(1) The word accent of polysyllables is assumed to be as in the modern standard language, except where we know for certain that it was placed differently (e.g. umíle on occasion, for instance, in Dante). The only polysyllables whose accent is considered lost in the phrase are the indefinite article, the articulated prepositions, and double proclitic pronouns (e.g. gliene).

(2) Contrary to the so-called Malagoli-Camilli principle, but in accordance with more modern thinking, it is assumed that two adjacent accented syllables normally retain their accent within a single phrase as well as across phrase boundaries. This is possible in normal parlance, and is a fortiori possible in the kind of poetic recitation that would normally be used for texts in the present corpus. Within a single phrase there is in most cases some weakening of the first of the two accents, but, with certain exceptions noted in (5) below, it usually retains a sufficient degree of prominence to justify counting it as accented.

(3) However we have followed the rule that there cannot be an uninterrupted sequence of three accented syllables unless there is an interruption or pause between at least two of them. Normally the middle accent is lost in such cases, but there may be semantic or syntactic or other contextual reasons, notably the location of the phrase break, for removing the accent from either the first or the third of the three syllables instead. The assumption is that the accent in such circumstances comes off the least prominent word. If the first or the second accented syllable is followed by a syllable 'lost' through sinalefe, the sequence is considered interrupted and three adjacent accents are allowed; for instance, on the 4th, 5th and 6th in Inferno 7.71, where the 'lost' final vowel of ignoranza allows the accent on è in 5th position to be kept:

quanta ignoranza è quella che v' offende!

But the same does not apply to an intervening syllable 'lost' through synaeresis, on the assumption that an unaccented vowel in sinaeresis interrupts a sequence of accented vowels less than an unaccented vowel in sinalefe; in Inferno 6.98, for instance, only the 4th and the 6th are counted as accented, and sua in the 5th loses its accent, despite its 'lost' final vowel:

ripiglierà sua carne e sua figura,

Three successive accents have also been allowed to stand where there is an intervening pause represented by a punctuation mark, or on the few occasions where a pause is unmistakably required by semantic considerations.

(4) The following monosyllables are considered always to be unaccented, except on the few occasions when they occur before a break, marked by any form of punctuation, and are not immediately preceded by an accented syllable (avvegna che followed by a comma has an accent on che, but sì che followed by a comma has the accent on and not on che):

Bisyllables that lose a syllable through sinalefe do not count as monosyllables in this connection, nor do words where the first syllable is lost through elision and the second is normally accented, for instance 'nver' is counted as accented.

(5) All other monosyllables are normally accented. But in some cases it has been necessary to introduce a further, positional rule that allows certain less prominent words to lose their accent if they are immediately followed by another accented syllable (not separated by a sinalefe) within the same phrase. The words are as follows:

However if one of the words in the categories above is the first of three successive syllables that are potentially accented, and the second is a monosyllable whose accent is weakened through close connection to the word that follows, then the accent can be kept on the first rather than the second: thus non son sola has an accent on non and not on son, which normally would have an accent, and non son paurose has an accent on son and not on non, because in this second case son is not immediately followed by an accented vowel. This last rule has been in practice the most difficult one to apply consistently, but is necessary if we are to follow normal practice in such sequences.

(6) If in a given line a long sequence of unaccented syllables (say more than four) is produced by the above rules, nothing should be done to rectify it. Such sequences are possible in normal parlance, though they may often be avoided; in the present context it would be extremely hard to devise consistent rules for avoiding them, and consistency is an overriding aim of the database.

(7) Secondary syllable or word accents are not included in the analysis not only because they are more difficult to assign consistently, but also, and more importantly, because their contribution to the rhythm of the line is far less significant than that of the principal accents in the word, however weakly these may in some cases be pronounced.

(8) As a general principle, where two alternative scansions are equally possible according to the rules above, if one has an accented 4th or 6th it is to be preferred on grounds of probability, since the vast majority of lines in the texts have an accent there.