Principales grupos funcionales de la biota del suelo

The first phonological descriptions of SLN since Tervoort (1953) were made in the context of a dictionary project, the so-called KOMVA project, which was started in the early 1980s (Stroombergen & Schermer 1985, KOMVA 1988, 1989, 1993, NSDSK 1988, Schermer 1990). More than 15,000 signs were recorded in total, from five different geographical areas in the Netherlands. Each area was marked by the presence of a school for the Deaf, and a relatively high percentage of Deaf people in the areas around those schools. In order to describe these signs for the dictionaries, a notation system was constructed (NSDSK 1988). This system was “phonetically based” (Schermer 1990: 28), in the sense that it did not involve a thorough phonological analysis of the language. However, it explicitly builds on phonological analyses of American Sign Language (Stokoe 1960, Friedman 1976, Battison 1978, Klima & Bellugi 1979, Mandel 1981), and in constructing the notation system, observations about minimal pairs in SLN are used as well. Distinctions that are adopted from ASL research include those between selected and non-selected fingers (Boyes Braem 1981, Mandel 1981) and presence vs. absence of an aperture relation between the thumb and the fingers. An overview of the notation system is presented in Appendix C.

In the context of the KOMVA data collection, several papers were published analyzing the handshape parameter (Harder 1989, Harder & Schermer 1986, Schermer, Harder & Bos 1987, Schermer 1990). Frequency distributions for the different handshapes are presented in these works, but only for the most frequent handshapes. It turns out that the seven most frequent handshapes occur in the strong hand in 65% of all signs. Distributions of handshapes were then compared for the strong hand vs. the weak hand and for the different regions.46 _{It was found that in}

46 _{In two-handed signs in which only one hand moves, the strong hand is the moving hand and the weak} hand is the non-moving hand. The strong hand typically is also the preference hand for other tasks, and also the hand that is typically used for one-handed signs. However, the hands may shift roles for

the Eindhoven region only 46% of all signs feature one of these seven most frequent handshapes; the reason for this lower percentage is that in this region many signs are initialized. In initialized signs, the handshape marks the first letter of the Dutch word most commonly associated with the sign, by means of the hand alphabet (Janssen 1986). Although six of the seven most frequent handshapes are also used in the Dutch hand alphabet, the use of the remaining eighteen fingerspelling handshapes in signs in the Eindhoven area is responsible for the different frequency distribution in that area.

The most frequent handshapes for SLN were also compared with those of other sign languages, such as ASL and British Sign Language. Apart from these frequency distributions, attention was paid to variation in handshapes within and between different regions.

Although the notation system and related analyses make use of clear subdivisions between categories, which is made possible by the adoption of a closed set of notation symbols, the descriptions do not yield much insight into the phonological structure of SLN other than the classification itself. The researchers remark that it is not clear whether all of the handshapes distinguished by the notation system are actually used distinctively in the language. Several handshapes are found to function as variants of each other (Schermer, Harder & Bos 1987, Harder 1989, Schermer 1990). The most commonly varying feature distinguishing these handshapes is the position of the thumb. Thus, the B, B0, and B1 handshapes (illustrated in Figure 3.3 below and in Appendix C), which differ only in thumb position, are in some cases used for the same sign in different realizations, and in some cases the variants are used depending on the other phonological or phonetic characteristics of the sign. The same holds for the A, AS, and money handshapes, which also differ only in thumb position (see below). Other features distinguishing related handshapes are spreading of the extended fingers (5 vs. B/B0), and the state of the middle, index and pinkie fingers in some handshapes (T0 vs. baby-O, T vs. baby-closed-beak). These variation data led the Dutch researchers to adopt further features to describe the handshapes, following Boyes Braem (1981).

One factor is presented that accounts for variation in thumb position between different signs: contact of a specific side of the hand with the body or the other hand (Schermer 1990, Schermer et al. 1991).47 _{Although this is not put very explicitly, the} claim seems to be that when the thumb would block contact with the body or the weak hand, it is folded over the fingers or into the palm (S, B), whereas otherwise it is not (AS, B1, B0). ‘Block’ should be interpreted as a position in between the rest of the hand and the contact surface. It is not clear if the above generalization accounts for all signs with one of the handshapes that are claimed to vary, and whether all those signs are invariant between the different regions. Given the emphasis on inter-region variation, it is implied that the thumb position differences grammatical or discourse purposes, and moreover, the choice of the strong hand (left or right) is not lexically distinctive.

47 _{These claims seem to be adopted from analyses of ASL (e.g. Wilbur 1979), although the authors do} not refer to any work on ASL in this discussion.

still differ between regions for some signs. Schermer (1990) mentions that there is also idiosyncratic variation in this respect: one signer from the Amsterdam region often used the B0 handshape where the other signers from this region used the B1 handshape, significantly skewing the frequency distribution. More importantly, the generalization still leaves the contrast between the two posited handshapes B0 and B1 unexplained.

It is implicit in the KOMVA publications that the related handshapes are in free variation within the regions: apart from the factor ‘contact’, there is no hint as to further factors determining the variation among the different forms. It is explicitly claimed, however, that the related handshapes that are mentioned above cannot freely vary in all signs that feature them: it has to be determined for each sign whether they can or not. The signs with varying handshapes are all claimed to be iconic, and this is advanced as the explanation for why the variation is possible (Harder 1989). No further discussion of this proposed explanation is presented.

Given this variation in handshapes, it is not clear how accurate the frequency distributions are: most of the handshapes that are included in the set of seven most frequently occurring ones are involved in one or more alternation pairs with each other: B0, B, 5, money, and S (but not T and 1). If all the variation data were taken into account, would the set of most frequent handshapes be different or not, and would the order within the set be different or not? These questions have not been answered in the literature that resulted from the KOMVA project.

A more fundamental problem also exists. It may be referred to as ‘the database paradox’ which is inevitably connected to phonetic transcription systems. To find out which distinctions in handshape are relevant, one has to be able to transcribe them in some way: it is impossible in practice to record and describe all the minute differences that occur between handshapes in different signs, in the same sign uttered by different signers, or by one signer on different occasions. But in order to make a transcription system, one already has to know which distinctions are potentially relevant and which are not. Thus, a phonological analysis ideally precedes the creation of a notation system, and not vice versa.

Although the KOMVA transcription system was built on the experience of researchers describing ASL, little analysis of phonological and phonetic variation had been carried out for this language, with the effect that researchers in the early 80s were essentially still using the categories proposed by Stokoe (1960), and adding subdivisions within those categories. Many of the distinctions made in the KOMVA transcription system are not explicitly motivated. This holds to a stronger extent for the many distinctions that are not made; that is, why are subtle differences such as in thumb position as they can be observed in surface forms not assigned a separate category and symbol? The choices now appear to be arbitrary in many cases. For example, consider the difference in thumb position mentioned above. Four handshapes are distinguished in the KOMVA system that all have the four fingers fully extended and adducted (non-spread); these are illustrated in Figure 3.3.

i. B1 ii. B0 iii. B iv. open-beak Figure 3.3

Four handshapes distinguished by the KOMVA notation system, varying in thumb position only

In fact, the last handshape is not included in the same group as the first three, even though thumb position is the only differing phonetic aspect. The open-beak handshape is included in the group of handshapes in which the thumb is opposed to the fingers. The interesting question now becomes, why are only these four thumb positions distinguished? Why not also distinguish one or two or sixteen others that are physiologically equally possible and also occur in the surface forms of signs? Take for example the following, all found in the SignPhon video corpus.

i. thumb 45deg abducted. (between 4a and 4b)

ii. thumb IP joint flexed, thumb tip against proximal phalanx of index

iii. thumb a little opposed (between B0 4b and open- beak 4d)

iv. thumb hyperflexed, IP joint extended (not flexed as in 4c)

Figure 3.4

Four ‘allophonic’ handshapes not distinguished by the KOMVA notation system

The handshapes in Figure 3.3 each form a category, but these categories and their boundaries are not defined and the reason for choosing the handshapes in Figure 3.3 as prototypes of their category are not made explicit. Of course, once these categories are assumed, one will be able to categorize almost any thumb position found in real life, but that need not be interpreted as evidence for the correctness of the categories. I surmise that this unclear definition of why certain categories are included and others are not is part of the reason why no further claims could be made about the occurrence of the different variants, or in other words, why they seemed to be in free variation. Neglecting small variations in articulation makes it hard to determine the context that determines the occurrence of a given variant. The same holds for variation in location which was found and the variation that was not found (i.e. distinctions that were not made).

I propose that the handshapes in Figure 3.4 are equally plausible realizations of certain articulator specifications as the handshapes in Figure 3.3. The model outlined at the end of Chapter 2 aims to generate both allophonic variation and phonetic

variation. However, the two do not occur at two sequential stages (respectively the phonological surface representation and the phonetic representation).

Aside from the research on the handshape parameter, most attention has been paid to the non-manual part of signs and especially ‘mouthing’ (or ‘word pictures’), in the work of Schermer (1990). It is concluded there that mouthings can serve to distinguish minimal pairs, such as in the SLN signs BROTHER vs. SISTER, which only differ in the mouthing that accompanies the manual part. These mouthings have been the topic of recent debate in the literature (e.g. Ebbinghaus & Heßmann 1990, Boyes Braem & Sutton-Spence to appear), but will not be dealt with in this book.