On the allomophy of the Finnish genitive and partitive plural:
A perspective from the corpus data

The nominal inflection of the Finnish language is full of allomophic alternations. Most of them is actually not true variation, for in the majority of cases the choice of the ending is perfectly predictable on the basis of inflectional class or on the type of stem formation. However, in the genitive and partitive plural, we find that polysyllabic stems do show systematic variation of (at least) two different endings. We name these context-neutral endings A- and B-ending, respectively:

There are two characteristics in this variation. First, the distribution of the ending is not perfectly free, but there exist several distinct tendencies how the preference goes (Itkonen 1957, Anttila 1995). The preference thus diverges from stem type to type. Secondly, the selectional preference sometimes differs singnificantly between the genitive and partitive plural.

In this study, we try to describe the phenomenon using the quantitative data from the corpus of written Finnish. The ultimate aim of the study is to explicate the domain of the variation on the basis of the distributional evidence from the corpus. We use the electoric corpus of the 1987 Suomen Kuvalehti (sk87), a Finnish weekly magazine.

The database was established by the following procedure: First, I retrieved from the corpus all the word-forms which have the two endings. Subsequently, the polysyllabic stems were extracted and classified into the stem groups. The classification of the stem followed the standard criteria (syllable length, quality of the stem-final vowel, and the length of the stem penult).

In this presetation, we concentrate the second problem mentioned above, namely how the distribtion of the two cases diverges. According to our investigation, only one stem group, the i-stems (e.g. ikoni `icon'), shows the statistically significant deviation pattern between genitive and partitive. More concretely, the corpus data show that in the genitive plural the share of the B-ending is about 60% (A-ending ikone-i-den 41.6%, B-ending ikoni-en 58.4%), while in the partitive plural the persentage of the A-ending surpasses the one of the B-ending (A-ending ikone-i-ta 86.7%, while B-ending ikone-j-a 13.3%).

This peculiar deviation pattern in the i stems has already been reported in the literature. We compare our result with the data of Uosukainen (1969, [2]), which studies the variation using the written material of 19th century, and with Itkonen (1957), which is again an experimental investigation of the preference of the endings. Through the comparison, we find the apparent change of the distribution in the genitive plural, according to which in the last century the genitive plural preferred the B-ending to the A-ending. This means that in Modern Finnish, the tendency of the choice of the endings has been changing so that the genitive permits both types of ending. On the other hand, the distribution of the partitive plural is stative throughout all the data, preferring consistently the A-ending.

Besides the historical trend of the variation, our corpus data clearly shows that the deviation of quantitative profile is led by the particular sub-groups of the i stems. We may specify the central group as the stems which end with either -ri or -li (e.g. lääkäri `doctor', a trisyllabic -ri/-li stem with a short penult): In these stems, the frequency of A, B-endings are almost fifty-fifty in the genitive plural, while in the partitive counterpart almost all the words appear with the A-ending, with the profile of 97%. In other i stems, on the contrary, the picture of the distribution is completely defferent, where both the genitive plural and the partitive plural prefer the B-ending. In the previous literature, the i stems have been treated as a uniform stem group concerning the variation. Our quantitative investigation shows, however, that this expectation is not true.

Unfortunately, why the -ri/-li stems show such a deviation is difficult to explicate at the present stage, and no historical data on the behavior of this stem group is currently available. But because the -ri/-li stem type is most frequent in the i stems, at least it can be argued that the stem type is the core of the i stem type. It is also to be noted that in the present Finnish language the -ri/-li stem type is a special vocabulary group, to which new loan words are successively introduced.

Selected References

Anttila, Arto

1995 Deriving variation from grammar: A study of Finnish genitives. Manuscript. Stanford University.

Itkonen, Terho

1957 "Mellakoihin vai mellakkoihin?" Virittäjä 1957: 259—286.

Uosukainen, Riitta

1969 Matalata ~ matalaa, avannoita ~ avantoja, ratkaisuiden ~ ratkaisujen? Unpublished Suomen kielen sivulaudaturtyö, Helsingin yliopiston Suomen kielen laitos.