metaphonebr

The goal of metaphonebr is to simplify brazilian names phonetically using a custom metaphoneBR algorithm that preserves ending vowels, created for aiding in dataset pairing in the absence of unambiguous keys.

Installation

The package is in the process of submission to CRAN. When it is accepted, the stable version can be installed with:

install.packages("metaphonebr")

You can install the development version of metaphonebr from GitHub with :

# install.packages("remotes")
remotes::install_github("ipeadata-lab/metaphonebr")

Example

This is a basic example which shows how to use the main function:

example_names <- c("João da Silva", "Maria", "Marya",
                    "Helena", "Elena", "Philippe", "Filipe", "Xavier", "Chavier")
phonetic_codes <- metaphonebr::metaphonebr(example_names)
print(data.frame(original = example_names, metaphonebr = phonetic_codes))

The metaphoneBR phonetic encoding algorithm proceeds as follows:

1. Initial Cleanup & Preparation:
   • Remove all diacritics (e.g., “João” becomes “Joao”).
   • Convert the entire string to uppercase (e.g., “Joao” becomes “JOAO”).
   • Remove all characters that are not uppercase letters (A-Z) or spaces.
   • Ensure single spaces between words and trim leading/trailing whitespace.
2. Silent Letter Removal:
   • Remove a silent ‘H’ if it appears at the beginning of any word (e.g., “Helena” becomes “Elena”).
3. Digraph Simplification (Sound Grouping):
   • LH is replaced by 1 (representing a palatal lateral approximant, like in “Filha” -> “FI1A”).
   • NH is replaced by 3 (representing a palatal nasal, like in “Manhã” -> “MA3A”).
   • CH is replaced by X (representing the /ʃ/ sound, like in “Chico” -> “XICO”).
   • SH is replaced by X (for foreign names with /ʃ/ sound, like in “Shirley” -> “XIRLEY”).
   • SCH is replaced by X (approximating /ʃ/ or /sk/, like in “Schmidt” -> “XMIT”).
   • PH is replaced by F (like in “Philip” -> “FILIP”).
   • SC followed by E or I becomes S (like in “SCENA” -> “SENA”).
   • SC followed by A, O, or U becomes SK (like in “ESCOVA” -> “ESKOVA”).
   • QU or QÜ followed by E or I becomes K (e.g., “QUEIJO” -> “KEIJO”).
   • GU or GÜ followed by E or I becomes G (the U is silent, e.g., “GUERRA” -> “GERRA”).
   • Any remaining QU becomes K (e.g., “QUANTO” -> “KANTO”).
4. Similar Consonant Simplification:
   • Ç is replaced by S.
   • C followed by E or I is replaced by S (like in “CELSO” -> “SELSO”).
   • Any other C (not part of an already transformed digraph like CH or SC) is replaced by K (like in “CARLOS” -> “KARLOS”).
   • G followed by E or I is replaced by J (like in “GELO” -> “JELO”; GUE/GUI already handled).
   • Any remaining Q (that wasn’t part of QU) is replaced by K.
   • W is replaced by V (common Brazilian Portuguese pronunciation, e.g., “WALTER” -> “VALTER”).
   • Y is replaced by I (e.g., “YARA” -> “IARA”).
   • Z is replaced by S (e.g., “ZEBRA” -> “SEBRA”).
   • X preceded by S has the X removed (e.g., “EXCELENTE” -> “ESELENTE”, to avoid a double /s/ representation from SKS).
5. Terminal Nasal Sound Simplification:
   • A word-final N is replaced by M (e.g., “JOAQUIN” -> “JOAQUIM”).
   • A word-final AO is replaced by OM (e.g., “JOÃO” -> “JOOM”).
   • A word-final ÃES is replaced by AES (e.g., “MÃES” -> “MAES”).
6. Duplicate Vowel Removal:
   • Sequences of identical adjacent vowels are reduced to a single vowel (e.g., “AARAO” -> “ARAO”).
7. Final Cleanup (Duplicate Letters & Spaces):
   • Sequences of identical adjacent letters (except if they are part of the special codes 1 for LH or 3 for NH) are reduced to a single letter (e.g., “CARRO” might become “CARO”, “LESSA” becomes “LESA”. Note: This rule simplifies sounds like ‘RR’ and ‘SS’ to their single counterparts, which is a common Metaphone-style simplification).
   • Ensure single spaces between any remaining words and trim leading/trailing whitespace again.

The resulting code is an attempt to represent the phonetic signature of the name in a simplified, standardized way for a Brazilian Portuguese context. In particular, by construction it preserves ending vowels since they imply generally gender information in Brazilian Names (ex.: ADRIANO and ADRIANA).

Nota

metaphonebr is developed by a team of researchers at Instituto de Pesquisa Econômica Aplicada (Ipea).