Fluid Intransitivity in Old Finnish

A striking feature of the Finnish language is that a subset of clause subjects behave in various ways like objects. For example, they may be marked with the partitive case in negated clauses, which "normal" subjects never do in Finnish, but objects always do. The subset of subjects in question are those of existential clauses (which broadly correspond to English "there..." clauses, e.g. "there is a book on the table") and possessive clauses (which are similar to existential clauses in Finnish: Finnish does not have a "to have"-verb, and the sentence "I have a book" is expressed like "at me is a book").

In the linked paper, I study whether this phenomenon (which is known as "fluid intransitivity" in recent typological literature) is present in the older Finnish language of the 16th and 17th century. I also want to see whether, if there is variation between subject-like and object-like subjects in clauses that correspond to modern Finnish existential and possessive clauses, I can determine which individual factors (such as, for example, negation) promote subject-like or object-like marking.

For this purpose, I gathered a database of relevant clauses from two key Old Finnish texts: Agricola's 1548 New Testament translation, and the New Testament section of the 1642 first Finnish Bible (Biblia). I used the electronic versions published by the Institute for the Languages of Finland.

The relevant clauses in question are clauses which have an intransitive main verb and an indefinite subject NP. I could not select for existential/possessive clauses in a more direct manner as it is precisely in question whether there is an existential clause type in Old Finnish. The resulting corpus (899 clauses from Agricola, 959 clauses from the Biblia) broadly covers the equivalent of modern Finnish existential clauses.

These clauses were then coded for variables studied (agreement and case-marking) as well as for several factor groups which I postulate may have affected the variables. The resulting code was then subject to a statistical analysis using the Goldvarb computer program. This analysis included a multivariate analysis which assigned a specific "weight" to individual factor groups.

For further clarification of methodology, etc. I refer to the linked paper. The file "coding instructions.txt" includes a brief explanation of the codes used in the database. The four spreadsheet files both include raw data (the clauses in question) and coding of variables and factor groups.

NOTE: The data files have undergone slight corrections and therefore differ slightly from those underlying
the 2016 paper.