Spanish
Combinatorial phonotactics in Spanish derivational morphology
Studied especially in the ethnic names of Castilian Spain.
This study has three main purposes:
A) Establish an inventory of suffixes that Castilians use to derivate the ethnic names in "standard" Spanish. Six suffixes covering 95% of the total: -ense, eño, ano, -ero, -és, -ino, in descending order of frequency; the last 5% are covered by fifty other suffixes, much more rare.
Examples: Almería, almeri-ense, Málaga, malagu-eño, Toledo, toled-ano, Tarazona, tarazon-ero, Córdoba, cordob-és, Bilbao, bilba-ino, Cabrales, cabral-iego.
B) Present the current geographical distribution of these suffixes.
C) Demonstrate with statistical procedures that a fundamental condition of choice among the many possible suffixes consists of the structure of the word root (i.e. the name of the city or town), and the phonetic material that the name is made of.
Three structural distributions are present in the material:
1) -ano / -ino / -és are predominating in the North of the peninsula, -eño / -ero in the South. It is likely that -eño / -ero is steadily advancing northward.
2) Equally pronounced is the distribution of ethnic names corresponding to large cities, medium and small: the share of -ano / -ino / -és increases with the number of inhabitants, and that of -eño / -ero decreases.
The linguistic-historical-political and demographic reasons for these two distributions remain to be explained in detail.
3) The third distribution is exclusively linguistic: a non-repetitive distribution. The foundation of this dissimilation here - as everywhere - is very simple: it is the attempt, involuntary, to avoid non-functional repetition. The tendency not to repeat is influencing the choice between suffixes currently used; for instance, behind a n or ñ in the last part of the word root -ano, -eño and -ino are less frequent, while -ero and -és are more numerous. In the same manner -és is used less with city names ending in s, and -ero less if the cityname has an r near the end. And the differencies are statistical significant.
Despite the presence of the two former distributional structures in the ethnic name material of Spain, the dissimilation or non-repetition can be demonstrated, with statistic means.