Eric Hoekstra en Henk Wolf
The Distinctness Principle.
Abstract
The simple perfect tense of the Frisian verb 'wêze' (to be) can be conjugated both with 'hawwe' (have) and with 'wêze' (be). Thus we find "Ik haw west" (lit. I have been) alongside "Ik bin west" (lit. I am been). However, the option exists only if the auxiliary of the perfect is a simple present or a simple past. If the auxiliary of the perfect is an infinitive, as in "Soe er siik west hawwe / * wêze" (lit. Should he ill been have / * be), 'wêze' is hardly acceptable. We investigate the choice of 'wêze' and 'hawwe' as perfect auxiliary of 'west' (been) in Modern Frisian, Middle Frisian, Old Frisian and Dutch. Our analysis of the facts provides support for Ackema's (2001) interpretation of the OCP. The OCP forbids adjacency of two elements which are identical in some sense to be made precise. We provide our own formulation of this principle, and refer to it, for ease of reference, as the Principle of Distinctivity (PD). In our view, the Principle of Distinctivity requires two adjacent words to differ as much as possible with respect to syntactic category (syntactic features) or with respect to outward shape. Independent evidence for the PD comes from a ban on sequences of identical complementisers, and from a ban on sequences of identical adverbs. We then go on to briefly discuss choice of perfect auxiliary for the equivalent of 'been' in other languages, and argue that the facts provide further support for the PD.
1. Introduction [note 1]
This article deals with the selection of the auxiliary of the perfect in Westerlauwers Frisian by the verb 'wêze' (to be). Westerlauwers Frisian (henceforth Frisian) is spoken in the province of Fryslân, which is situated in the north of The Netherlands. It is spoken by approximately 200,000 persons with varying degrees of fluency, who are all fluent in Dutch as well.
It is shown that superficial syntactic properties may affect speakers' grammaticality judgments. In order to account for the relevant facts (to be illustrated below), we postulate the existence of the so-called principle of distinctivity. This principle puts restrictions on the occurrence of identical, or even similar, function words.
It has been reported in the literature (Popkema 1984) that the perfect participle 'west' (been) may be formed both with 'hawwe' (to have) and 'wêze' (to be). This phenomenon is illustrated below for main and embedded clauses:
(1a) Hy hat siik west.
He has ill been
(1b) Hy is siik west.
He is ill been
“He has been ill.”
(2a) Om't er siik west hat.
Because he ill been has
(2b) Om't er siik west is.
Because he ill been is
“He has been ill.”
Both (a) forms and (b) forms are found in written Frisian, see section 2. To compare, Dutch exclusively features 'zijn' (to be) as auxiliary of the perfect for 'west' (been). Note, however, that some dialects of Dutch, just like Frisian, do feature 'hebben' (to have) alongside 'zijn' (to be) as auxiliary of the perfect for to be (see Van Haeringen 1962). The following questions will be discussed in this article.
(3a) Is there a correlating difference between Dutch and Frisian that helps us understand the different behaviour with respect to choice of auxiliary of the perfect?
(3b) Is there a correlating difference between the Dutch dialects that are like Frisian with respect to auliary selection and those which are not?
(3c) What is the historical background of the variation in the choice of the perfect auxiliary in Frisian?
(3d) Does the correlating difference between Dutch and Frisian, if any, helps us to understand choice of selection of the auxiliary of the perfect of the verb to be in other languages?
Question (3a) will be seen to involve aspects of phonological and syntactic form. The general idea is that the perfect participle of to be can be more easily formed with forms of to be, just in case they are more distinctive. To illustrate, 'west hawwe' is more acceptable than 'west wêze', as the members of the former pair are more distinctive (distinguishable) than the members of the latter pair. These ideas will be worked out in more detail below. It will be shown that the facts of auxiliary selection in Old and Middle Frisian support our proposal. In addition, it will be shown that facts from Standard Dutch and its dialects provide further support for the proposed analysis. A brief investigation of auxiliary selection in other languages concludes our article.
2. BE's choice of auxiliary selection in written Frisian.
2.1. Introduction.
This section deals with written Frisian of the twentieth century. Our instrument of research is the linguistic corpus commonly referred as "Linguistic Corpus Modern Frisian" (LCMF). This corpus contains about 24 million words from various written sources. These sources involve such different genres as newspaper articles, radio news transcripts, poetry, high literature, fiction, text books, plays, and scientic articles. We will restrict ourselves to embedded clauses, in which the auxiliary of the perfect is right-adjacent to the perfect participle. [note 2]
2.2. LC New Frisian
Consider the following table indicating to what extent either to have or to be is used as auxiliary of the perfect for to be.
(4) Table of choice of auxiliary of the perfect for 'west' (been).
Infinitive: West wêze: 2 / West ha(wwe): > 100Present 1sg: West bin: 3 / West ha(w): > 100
Present 2sg: West bist: 0 / West hast: > 60
Present 3sg: West is: 22 / West hat: > 100
Present 123pl: West binne: 7 / West ha(wwe): > 100
Past 2sg: West wiest: 0 / West hiest: 19
Past 13sg: West wie: 20 / West hie: > 100
Past 123pl: West wiene(n): 1 / West hiene(n): > 100
The sign ">" is used for "at least"; in those cases a large amount of occurences is involved, but we abstract away from its exact nature. What we see is that the LC New Frisian almost exclusively features to have as auxiliary of the perfect. Needless to say, the type of Frisian found in the LC cannot be equated with spoken Frisian. The LC consists mostly of Frisian that has been published in the form of a book. It has regularly been corrected by a corrector. Thus it is more formal than spoken Frisian or the type of Frisian that is written in the Internet.
2.3. Spoken Frisian and informal written Frisian.
Our intuition tells us that spoken Frisian features a much more frequent use of the auxiliary to be. This is corroborated by a brief search in the LC Spoken Frisian. The literature confirms our intuitions in this respect (Popkema 1984). In other words, the use of to have as auxiliary of the perfect is a feature more characteristic of written Frisian than of spoken Frisian. [note 3], [note 4].
In order to investigate informal written Frisian, we consult the Internet with the help of the search engine Google. The internet is a more informal medium than the LC New Frisian. Most texts on the internet have not been corrected by a corrector. Given our previous finding that informal language use features more usage of to be as auxiliary of the perfect, we expect that this tendency is also reflected on the (informal) Frisian internet. Thus we expect to encounter more use of to be as auxiliary of the perfect on the (less formal) Frisian internet than in the (more formal) LC New Frisian.
We must first establish the rough size of the Frisian internet. In order to do so, we select a number of fairly neutral words of which we know the frequency of occurrence in the LC New Frisian, of which we know the size (24 million words). We then go on to investigate how often these words occur on the Frisian internet. Given that the relative frequency of these words will be the same in both corpora, we can then determine the size of the Frisian internet. We did this on 15 april 2004. We chose the following words, as they are fairly neutral and therefore may occur both in formal and informal genres:
(5) Table of 5 neutral words for determining the size of the Frisian internet.
Word / In LCNF / In the internet / Ratio / Size of Frisian internetfoar ´for´ / 150.000 / 12.500 / 1/12 / 25 x 1/12 = 2 milj
oan ´to´ / 170.000 / 7000 / 1/25 / 25 x 1/25 = 1 milj
omheech ´upwards´ / 5000 / 800 / 1/7 / 25 x 1/7 = 3,5 m
mear ´more´ / 60.000 / 4000 / 1/15 / 25 x 1/15 = 1,6 m
tsjin ´against´ / 40.000 / 2000 / 1/20 / 25 x 1/20 = 1 m
We thus take the size of the Frisian internet to be about 2 million words. The LC New Frisian is about 12 x as large as the Frisian internet. [Note 5] Even though the Frisian internet is much smaller, it features in absolute numbers a larger usage of to be as auxiliary of the perfect for to be. This is shown in the following table, where it must be born in mind that the LC New Frisian is about 10 times as large as the Frisian internet.
(6) Comparison Frisian internet with LC New Frisian for 'wêze' (be) as auxiliary of the perfect for 'west' (been).
Paradigmatic combination / Frisian Internet versus LC New Frisian /(± 2 million words versus 24 million words)
Infinitive: west wêze / 8 versus 2
Present 1sg: west bin / 2 versus 3
Present 2sg: west bist / 2 versus o [note 6]
Present 123 pl: west binne / 20 versus 7
Past 13sg: west wie / 24 versus 20
Past 2sg: west bist / 2 versus 0
Past 123pl: west wiene(n) [note 7] / 10 versus 1
These facts make clear that to be is used more often as auxiliary of the perfect in the Frisian internet than in the LC New Frisian. This is in keeping with the hypothesis that to be is more often used as auxiliary of the perfect in less formal Frisian. Conversely, to have is used as auxiliary of the perfect more often in the LC New Frisian than on the internet. For example, 'west hast' (has been, present 2sg) is only found 2 times on the internet, but 63 times in the LC New Frisian.
2.4. Spoken Frisian.
The Corpus Spoken Frisian is a corpus that is still in progress. About 50 hours of speech have been transcribed, and about 250 hours of speech have been collected. In order to investigate informal language, we must focus on spontaneous conversation, not on, for example, lectures. We selected a number of transcriptions reporting spontaneous conversation for investigation. As the research interface is still under construction, it could not yet be used. We thus improvised by loading the transcriptions in the program Textpad, which allows of extensive search possibilities. In our texts, we encountered 29 occurrences of the participle 'west' (been). It turned out that the perfectum had been formed with to have in 15 examples, and with to be in 14 examples. Both categories included present tense and past tense. Both categories also featured an example with an infinitive. It is somewhat surprising that spoken Frisian features less use of to be than the Frisian internet. We do not have an explanation for this. Note though that spoken Frisian does feature more use of to be than the LC Modern Frisian, as expected.
2.5. Conclusion.
The LC New Frisian hardly features the use of to be as auxiliary of the perfect for to be. There is clear-cut difference between LC New Frisian, on the one hand, and the internet and the Corpus Spoken Frisian, on the other hand. It would seem that standardisation (prescription of to have as auxiliary of the perfect for to be) has only been succesful in so far as formal written Frisian is concerned. Otherwise, to be is also regularly used. Surprisingly, to be was more often present on the Frisian internet than in the Corpus Spoken Frisian. We shall now investigate choice of auxiliary of the perfect for to be in Old Frisian and Middle Frisian.
3. Middle Frisian and Old Frisian.
Speakers of Frisian are always bilinguals who also hear and speak Dutch. The Frisian use of to be as auxiliary of the perfect for to be is supported by Dutch which also features to be as auxiliary for to be. Nevertheless, the rise of to be in Frisian originally seems to be an autonomous development, in view of the fact that this use of to be is already on the rise in Middle Frisian. In the case of Middle Frisian syntactic interference from Dutch can hardly be postulated. Let us summarise the relevant facts from Old and Middle Frisian, which have been thouroughly reported in Johnston (1994).
Early Old Frisian exclusively features to have as auxiliary of the perfect for ´wessen´, the Old Frisian form of the participle of to be. Synthetic perfects, however, are less frequent than in later stages of the language.
The use of to be as auxiliary of the perfect is first found around 1450 in texts from Westerlauwers Frisian, corresponding roughly to the present-day province of Fryslân. It is not found in Oosterlauwers Frisia, corresponding nowadays to the present-day province of Groningen (situated in The Netherlands) and adjacent provinces situated in Germany (formerly referred to as East Frisia). In the neighbouring languages Middle Dutch and Middle German, to be is found earlier than in Frisian.
When to be makes its first appearances as auxiliary of the perfect participle of to be, the participle itself changes from 'wessen' to 'west'. It is not clear why this change took place. An effect of the change is that the participle of to be is less similar to the infinitival forms ('wêze', 'wêzen').
To be functions as auxiliary for to be only if it is a simple past or a simple present, but not as infinitive. As Johnston (1994:17) puts it: (translated from Frisian): “It appears to be the case that the combination of the infinitive 'wesa' or 'wezze' with its perfect participle (hence something like 'wessen wesa' or 'west wezze') was judged to be ungrammatical.”
4. Dutch and its dialects
Johnston cites De Rooij (1986:306) for making a similar observation concerning present-day Dutch and its dialects. De Rooij observes that some dialects of Dutch feature the infinitive 'wezen' (to be) instead of the Standard Dutch form 'zijn' (to be). This is illustrated below: