• Introduction
• All phonemes
• Initial phonemes
• Phonemes by manner
• Accented vowels
• Technical Method

The most obvious potential use of this section is to allow searching for sound repetitions in the form of alliterations and assonances. Since these figures can be viewed in a variety of ways, three different categories of phonemes are represented:

• all phonemes in the line
• word-initial phonemes
• accented vowels

The first two are additionally grouped by manner of articulation, in order to provide additional scope for the study of the possible expressive functions of sounds.

All these groups are arranged in alphabetical order, to make it easier to search for repetitions. Accented vowels are additionally given in the order in which they appear in the text.

Individual phonemes are represented by their nearest written equivalents, except that

• 1 = C[E] / C[I]
• 2 = G[E] / G[I]
• 3 = GL[I]
• 4 = GN
• 5 = SC[E] / SC[I]
• C includes Q

Note that the order of the numerical codes follows the alphabetical order of the equivalent written forms; to facilitate searching it was essential for single-character representations to be used, and it would be more difficult to remember arbitrarily chosen symbols.

Remember the database search engine is not case-sensitive. See Technical method below for a fuller description of the criteria used for transcription.

Example:

• Enter SSSSSSSS with Match any part in the All phonemes box to see what kind of expressive function there may be in lines with an especially high incidence (8 or more) of the phoneme S.

The phonemes represented are those that occur in the Scanned line field in the database. They thus differ in a small number of cases from those in the original texts, as represented in the Line field: see the relevant sections of the Rhyme and Rhythm and metre Information pages.

Word-initial phonemes use the coding system described in the section above, except that the apostrophe is included for words that begin with one, including the isolated apostrophe occasionally used, for instance, to stand for the preposition a. Proclitics are counted as separate words for this purpose, but enclitics (which have regularly been transcribed as part of the word to which they belong) are not: e.g. m' avea is counted as two words, whereas rife'mi is counted as one. The articulated prepositions are counted as one or two words, depending on the practice of the editor of the text used.

Example:

• Enter MMM with Match any part in the Initial phonemes box, and choose Liberata and Conquistata in the Name of text box, to see how many of Tasso's lines reflect his expressed belief (in the Discorsi del poema eroico) that words beginning with m are particularly suited to the description of tears and weeping.

Where the same phonemes are represented by manner, the coding scheme stands for the different manners of articulation of consonants, and at the same time groups together front and back vowels:

• F = Fricatives: F, S, SC[E] / SC[I], V
• L = Liquids: GL[I], L, R
• N = Nasals: GN, M, N
• P = Plosives: B, C, D, G, P, T
• X = Affricates: C[E] / C[I], G[E] / G[I], X, Z
• A = A
• E = Front vowels: E and I
• O = Back vowels: O and U

Note that the letters representing each category have been chosen because they correspond to phonemes that belong to that category.

This option for searching is included, because grouping phonemes together in this way can produce interesting results for those investigating the expressive functions of sounds. Examples:

• Enter PPPPPPPPPPPP with Match any part in the by manner box after All phonemes to examine lines with a particularly high incidence (12 or more) of plosives.
• Enter OOOOOOOOOOO with Match any part in the same box to examine lines with an exceptionally high incidence (11) of the back vowels O and U.

Accented vowels (as represented in the Scanned line box) are given first in alphabetical order, then, in the in text box, in the order in which they appear in the text. They are not grouped into front and back vowels, because the same result can easily be achieved through the use of wildcards. Example:

• Enter [ou][ou][ou][ou][ou][ou] with Match any part in the Accented vowels box for lines with an exceptionally high incidence (6) of the back vowels O and U in accented position. See Searching Instructions about the use of wildcards here. The results of searches like this should induce some skepticism about the expressive function of these so-called 'dark' vowels.

The in text box is included in order to allow searches for structures such as that in Paradiso 1.1, the effect of which depends significantly on the chiastic repetition of the accented vowels OUUO in

La gloria di colui che tutto move

which prefigures the syntactic chiasmus in the third line of the canto:

in una parte più e meno altrove.

No provision has been made for similar text-order searches on All phonemes and Initial phonemes, since the number of possible structures is too vast.

The transcription from text to phonemic representation can be carried out automatically by applying a set of computer routines directly to the text files, with only one exception: it is necessary to intervene manually in order to ensure that the presence, as opposed to the total absence, of the sound i is correctly marked in the few cases such as that of uscia with an accented i, as opposed to uscio with an accented u and an i that is merely diacritical. For present purposes, however, the same routines have been applied to the texts marked up in terms of accents and syllable-divisions, as described in the Information page on Rhythm, so that any further manual intervention has been unnecessary.

No distinction has been made between the alternative phonemes represented by e, o, s and z, and between vocalic and semivocalic i and u. The program has converted all occurrences of c and g (single and double) and sc before e or i (or followed by an apostrophe followed by another word beginning with e or i), and all cases in whatever position of gl and gn, into the appropriate palatal forms, and the immediately following i has been ignored where its function is purely diacritical. Q has been treated automatically as the same as velar c, and h as a marker for velar c and g before e or i.

The distinction between single and double, or short and long consonants has been retained, the latter counting as two, except that no account has been taken of the fact that gl[i], gn, and sc[i] are always pronounced long in intervocalic position; nor has any account been taken of the Tuscan usage of syntactic doubling, which is not of course normally marked in modern editions of texts. On the other hand allowance has been made for the fact that word-final n before a word-initial b, m, or p is in normal speech changed to m, by regularly effecting the substitution except where a punctuation mark or line-break intervenes.

Opinions will differ, of course, as to whether double consonants should be counted as two, or as one, or indeed as one-and-a-half. There is an element of the arbitrary in any of these solutions; the key feature of the transcription is that it has been applied consistently across the whole corpus of texts.