July 2002

Shell Corner: The Soundex System in C and Shell

Hosted by Ed Schaefer

This month, Mendel Cooper (thegrendel@theriver.com) submits a bash script called soundex.ss. The Soundex system is a hashing algorithm for surnames. It catalogs variants of a given surname, such as Smith, Smyth, and Smythe, and encodes them according to the way they are pronounced. Following Cooper's presentation of soundex.ss, I present my surname tests, a non-GNU tr command class problem, and a Perl programming nugget.

Cooper explains:

soundex.ss is an error-minimizing code, as a misspelled name will often have the same or nearly the same Soundex representation as the correct version. This feature tends to mitigate the inevitable errors of workers who listen to and transcribe many names per work shift. This also, to some extent, corrects for typographical drift in name spelling over the years, which can be very useful when tracing one's ancestors (as an example, Smathers --> Smothers --> Smithers ... same Soundex code).

Soundex has been in use for over 80 years -- since well before the adoption of modern information handling techniques -- and can simplify filing procedures if used creatively. The National Archives, the Census Bureau, and some state governmental agencies use Soundex. It is also used for genealogical research -- by hobbyists and by adoptees attempting to find their birth parents.