[eu-fo-nì-a]: a program to automatically compute euphonic phenomena in the Italian language

The Italian language includes a series of euphonic phenomena used to avoid cacophony or difficulties in pronunciation; the letters involved are d, i, and r.
In the first case, the addition of d concerns the preposition ad, and two conjunctions: ed and the archaic od¹. It was also formerly present in the following cases: ned, sed, and ched. While once widely employed, the current recommendation is to use it only when there are two identical vowels². However, there are some exceptions, such as depending on the letter after the first vowel (if it is d or t), if the foreign aspirated h precedes a, e, or o, or even if ed, ad, or od come before an aside³. In addition, a few accepted cases do not follow the general rules (e.g. ad ogni morte di papa, ad esempio, ad ogni buon conto or ho incontrato Luigi e Enzo)⁴.
The prosthetic i consists in the addition of an i at the beginning of a word in case it begins with an s impurum and is preceded by a word ending in a consonant (e.g. per iscoprire)⁵. Today, it is an extensively obsolete linguistic device⁶ (except for per iscritto, which is still common)⁷.
Finally, the archaic euphonic r occurs with the addition of an r to the preposition su if followed by a word starting with u (e.g. sur un tavolino)⁸.
Given that the rules of euphony are strongly dependent on the tastes of an era, we would expect they change consistently and that, for example, it would be possible to select this parameter, among others, to chronologically collocate a literary work whose author is unknown. Therefore, I developed a Python program to automatically compute the number of times the above-mentioned euphonic phenomena occur.
Furthermore, it is possible to produce a CSV (Comma-Separated Values) output that can be easily imported into Excel or R to carry out further analyses. Importantly, the output is not a mere table of frequencies; rather, the file contains the text of every collocation and its frequency. As such, it is possible to double-check the results and search for potential significant patterns. After this initial phase, data can be sorted and further analysed, employing other programs or visualisation tools as needed.
The next step is to create an adequate corpus containing literary works (in TXT format) spanning 100 years (from the mid-18th to the mid-19th century), allowing the investigation of texts from synchronic and diachronic perspectives.
Once the data are gathered and analysed, we will understand if some or all rules are consistent or if they change significantly according to single authors, genres, or even works. Based on the results, the program will be further perfected to differentiate euphonic phenomena, taking into consideration the identified parameters. 
This feature will be extremely helpful for researchers interested in performing stylistic analyses. Furthermore, progressively expanding the corpus will help identify a linguistic phenomenon that is rarely considered and trace how its use changed through time and authors.

¹ Cf. Treccani (2010, p. 1650)
² Cf. Migliorini and Folena (1957, p. 25)
³ Cf. Treccani (2012, pp. 238-239)
⁴ Cf. Treccani (2010, pp. 1650-1651)
⁵ Cf. Malagoli (1912, p. 156)
⁶ In Malagoli (1912) it is already underlined that modern writers tended to avoid it, especially with proper names.
⁷ Cf. D’Achille (2011, p. 223)
⁸ Cf. Malagoli (1912, p. 157)