1. Problemáticas de la Enseñanza de la Ciencia del Suelo y la Difusión del Conocimiento
3.3. Funciones y servicios esenciales de la edafodiversidad, para el funcionamiento sostenible
3.3.1. Funciones
The features that I propose include many ‘exceptional’ values, which only occur in a few signs. These values often have a clear etymological background, being remnants of iconic motivation or fingerspelling. In this section I motivate why they are assigned the same status as all other features, that is, why they are included in the set of phonological feature values.
One of the core distinctions that a phonological model should express is that between signs that occur or might occur and signs that could not occur: only signs that are well-formed should be representable, whether they exist or not. In other words, the model should not overgenerate. Within the set of signs that we find, however, there are more and less frequent forms. Some features occur frequently and in combination with many different other features, while others are limited to one or a few signs. Many of the features or feature combinations that do not occur frequently in sign language can be argued to be iconic in nature (van der Kooij in prep.). In this section, I first discuss exceptions due to the different ways in which spoken language is represented in sign language, and second I discuss iconic exceptions.41
Users of SLN have two main means of representing words from spoken Dutch. Firstly, they can articulate a spoken word as hearing Dutch speakers do (typically without sound); this is perceived by lip reading, and can often serve to disambiguate SLN signs. Schermer (1990) shows that many of these ‘mouthings’ have become part of the lexicon, being the distinguishing phonological element between two variants of a sign such as SIBLING (meaning ‘brother’ when accompanied by [bRu:R], and ‘sister’ when accompanied by the mouthing [zYs]). All of these mouthings occur only once in the lexicon, and each is therefore ‘exceptional’. A second way to represent (written) Dutch is by means of the hand alphabet, with which Dutch words can be fingerspelled. Some of the 24 handshapes that are used to spell the 26 letters of the Dutch alphabet are regular SLN handshapes found in many lexical items as well. Some of the handshapes, however, occur only in the lexicon in so-called ‘initialized’ signs. In such signs, a letter from a Dutch gloss of the sign is represented by the handshape of the sign. For example, the F handshape only occurs in signs such as FAMILY and POOR-JOKE, in which the first letter of a Dutch translation (respectively familie and flauw) is represented. Such handshapes are exceptional, then. In most if not all cases, the regular set of features that is needed to describe other handshapes has to be expanded by ‘exceptional’ values to describe those fingerspelling handshapes.
A second main group of exceptions is related to the frequent iconic nature of signs, which has received much attention in the sign language literature (Frishberg
41 A third small group of exceptions is found in the counting system of SLN and lexical items incorporating counting handshapes. In particular, the selected fingers value [3] is needed to specify the form of the numbers THREE and EIGHT (and many derivatives thereof, such as EIGHTEEN, THIRTY, etc.). In the Groningen dialect of SLN, which is not further studied here, other exceptional handshapes occur in the counting system, featuring contact between the tip of the thumb and the tip of the pinkie, ring, and middle finger.
1975, Mandel 1976, 1977, DeMatteo 1977, Armstrong 1983, Taub 1997). In many signs, there seems to be a non-arbitrary relation between some part of the referent or concepts that are associated with the referent, and a part of the form of the signs; this has been referred to as ‘motivation’ or ‘iconicity’. The topic has received so much attention because iconicity in sign languages is so pervasive in comparison with spoken languages.
Several linguists have recognized the presence of iconicity, but at the same time they argued that since it does not seem to play a (facilitative) role in sign language acquisition or processing, it is not a relevant topic for linguistic research other than in etymological studies (Klima & Bellugi 1979, Poizner, Bellugi & Tweeney 1981). Iconicity, they argued, was a relic of the process of creating new lexical items, rather than an integral and accessible part of the lexical knowledge that signers possess (cf. Frishberg 1975 and Radutzky 1989, who established the decrease of iconic information in signs over time).
More recently, it has been argued that in studying word formation processes and the phonological structure of signs, iconicity needs to be an integral part of the investigation (e.g. Brennan 1990). It is misleading to speak of ‘iconic signs’, however. Signs typically are not iconic as a whole; rather a part of the sign resembles a part of the referent. It is never the case that all aspects of the referent are encoded, so to speak, in the linguistic form; our conceptual knowledge is much too rich for that. Some motivated form-meaning pairings of formal elements are not unique to a single lexical item but occur over and over again in the lexicon (e.g. Taub 1997, for ASL). An example is the temple as the location associated to signs expressing mental states or activities in both SLN and ASL. In this case the formal element [location: temple] could be seen as a bound morpheme meaning ‘related to mental activity’. Brennan (1990) argues that word formation processes in British Sign Language (BSL) show that much of the iconic information present in frozen lexical items is accessible to signers. Mandel (1977) and Taub (1997) discuss the precise way in which the selection of iconically represented information takes place.
As an example of an iconic sign, consider the sign BICYCLE, illustrated in Figure 3.2. In this sign the alternating movements of the hands resemble the alternating movement of the pedals or the feet. Many aspects of the concept ‘bicycle’ are not visible in the sign, such as the fact that they typically have two wheels, a saddle, and handlebars. Conversely, some aspects of the form of the sign do not mimic any aspect of the concept of bike: the size of the movement is not as large as the movement of the pedals, and the location in front of the upper body is not the location of the pedals in real life.
Figure 3.2
BICYCLE
One could consider such iconic signs to be ‘loan signs’, resembling loans from other languages: they are loaned from (our conception of) the visible world, just as words may be loaned from another language. In discussing phonology it is more appropriate to talk about exceptional features rather than treating the whole sign as exceptional. The iconic features in the sign BICYCLE are the movement features [alternating] and [circular] (mimicking the movement of the pedals or feet), and possibly the features representing the shape and orientation of the hand. None of these features are exceptional, though. By contrast, in a sign like KIDNEY, the location feature [side of the lower trunk] (which is close to the actual location of the kidney) is exceptional, while its other features, representing the contacting movement with an straight index finger, are not exceptional. In this way, these ‘iconic loans’ resemble loans from, say, spoken French in spoken Dutch. Most phonemes in Dutch lexical items borrowed from French are a regular part of the Dutch phoneme inventory, and do not lead to exceptions. Some phonemes though, such as the postalveolar fricative that appears in words like gêne, beige and genre, are exceptional phonemes in Dutch.
Many aspects of ‘iconic’ signs thus resemble regular phonological patterns and can most likely be represented by a model that aims to represent only the most frequent signs. Sometimes, however, some property is not representable by the standard set of features, and has to be considered an exception. The question then is how such exceptions should be accounted for, and whether a phonological model should be able to represent them or not.
Since iconicity is such a pervasive phenomenon, it is not surprising that some exceptions that seem to be iconic in nature occur in more than one sign. However, I would like to suggest that it is even the case that some features which are considered to be standard phonological features and which occur regularly can be argued to be iconically motivated in all signs in which they occur. This appears to be the case with the feature [alternating], which describes the out-of-phase movement of the two hands in otherwise symmetrical signs. This feature surfaces in most phonological
models and to my knowledge it is needed for every sign language described to date.42
The evidence for the claim that a feature such as [alternating] is indeed iconic in all signs is a rather delicate matter. The claim rests on the assumption that iconicity not only applies to the actual etymological origin of the sign, but also to the possibility for sign language users of pervasively seeing iconic aspects of the sign without knowing its actual etymology. This comes close to folk etymology, although this term is used mostly for spoken languages, where in my experience this phenomenon occurs much less frequently. Many hearing adults learning sign language report the eagerness of Deaf people to explain why a sign looks the way it does, whereas this is not a common phenomenon for adults learning a spoken language. I would argue that this fact may justify claims about iconicity by (hearing) researchers who do not have etymological evidence for the iconic status of signs: the presence of iconicity is not only a construction of the linguist studying the language, but is also experienced by the users of the language themselves. Etymological data are hard to find since in general historical data for sign languages are very scarce, because of the paucity of documentation of the early stages of most sign languages.43
Iconicity in sign language phonology is sometimes used to explain exceptional patterns (e.g. van der Kooij in prep.). The prediction that is explicit or implicit in resorting to iconicity in this manner is that the exceptional aspects of a sign’s form will gradually wear out, so that the sign conforms more and more to the phonological patterns of the whole lexicon.44 It may turn out that this process is relatively slow in sign language compared to changes in loanwords in spoken languages, because of the relatively great awareness of sign language users of the iconic status of the exceptional features. This remains to be studied; see Frishberg 1975 for a discussion of iconicity and historical change in ASL. Users of spoken languages are sometimes aware of the foreign origin of many words as well; it is not a phenomenon that is limited to sign language. What is different in sign language is that users tend to interpret signs as iconic loan signs even if they are already phonologically fully regular. Thus, it is possible that after language change has affected exceptional forms so that they come to conform to the rest of the lexicon, users will still invent iconic origins for the sign, or may have access to the original iconic motivation for the form.
42 Miller (1997) derives alternating movements from other prosodic properties of signs; see also footnote 65 and the discussion on page 86 below.
43 The present high speed expansion of the SLN lexicon, which is being stimulated by the development of new teaching programs and the planned recognition of SLN as a minority language, offers a wonderful opportunity for phonologists to study the use of iconicity in the creation of new words and in subsequent phonological change. The creation of sign language poetry and its appreciation among Deaf signers can also be a good source of data on the use of iconic aspects of signs by sign language users. To my knowledge, no study has specifically addressed the use of iconic aspects of signs in poetry.
44 Although this general expectation may be valid, Brentari & Padden (2001) argue that some ASL signs that incorporate exceptional handshapes from the fingerspelling alphabet (thus being able to indirectly ‘borrow’ words from written English) are quite stable, and should be considered to form a separate, but stable, subpart of the ASL lexicon. It is possible that (a subset of) exceptional iconic forms behave alike.
The actual analysis of iconic exceptional features can still be debated. In the case of [alternating], one can distinguish several semantic concepts, which stem from a superficial inspection of the 153 alternating signs in the SignPhon corpus (5% out of a total of 3076): alternating movement of the human body (e.g. SIGN, BICYCLE, WASH-HANDS, TYPE), exchange or repeated movement of physical objects (e.g. RESTAURANT, GROCERIES, ROLL), or the exchange or repeated movement of concepts (e.g. MATCH, TELL).
The feature [alternating] is needed in the phonological model because there is no other phonological property on the basis of which it is possible to predict the alternating movement. Nor is it the case that the form of the sign can be predicted on the basis of the semantic specification of the sign (which is the common lay conception of iconicity and even of sign language in general): not all signs which can be directly or indirectly associated with a concept related to alternating movement (such as actual alternating movement and exchange of information or objects) actually do have alternating movement. For example, in SLN the sign CAR can be argued to be iconic, in that the alternating movement of the hands mimics the stereotypical alternating movements of the hands on the steering wheel while driving. However, the phonologically and semantically related sign DRIVE-A-CAR does not have this alternating movement. Many other signs which could be fruitfully associated with the alternating movement of the hands do not have alternating movement either. One example is INFORMATION, which could be linked with the exchange of different pieces of information, conceptualized as the hands which move to-and-fro with respect to the body in an alternating pattern.
In SLN, then, there are some iconic phonetic features of signs that occur in only one sign, whereas others occur widely throughout the lexicon. I submit that this enforces the conclusion that the constant iconic ‘load’ of a feature itself is not an argument to deny the feature a phonological status in the lexicon, contrary to the claim made in van der Kooij (in prep.) and van der Hulst (2000). However, in the case of an exceptional phonetic form that cannot be accurately distinguished from other signs on the basis of any combination of phonological features, the iconic nature of the exception can be used to explain how the exception came to exist. This is an etymological issue and not a phonological one.
Other evidence is needed, then, to determine whether a feature is really exceptional or not. In many cases I do not have firm evidence from (either type or token) frequency indicating that a form really is exceptional, because the SignPhon corpus only contains a little over 3,000 signs. In most cases, the exceptional status of the sign only concerns part of the sign, for example an uncommon place of articulation. From the point of view of a traditional phonological model, it is not desirable to adopt a feature to account for every exception. Rather, van der Hulst (2000) and van der Kooij (in prep.) suggest that it is preferable to assume that these exceptional feature values are stored in the lexicon as phonetic surface representations. Van der Hulst refers to this option as “lexically specified phonetic implementation” (2000: 37), as the phonetic shape of such feature values is already present in the lexicon, while other phonological features are not assigned a concrete
phonetic form until the phonetic implementation stage of the derivation. It would seem, then, that such exceptional signs are partly made up of an abstract phonological specification, and partly of concrete phonetic information.
Van der Hulst remarks that he has “opened the door to simply phonetically pre- specify[ing] all signs in the lexicon, leaving their phonological representation uncompositional, and thus effectively absent. In some sense, perhaps, the phonology of sign truly balances on the edge between a compositional and a non-compositional phonology” (2000: 38; emphasis in original). The lexical specification in the present proposal does not have internal structure, such as feature geometries proposed. However, this is not to say that this specification is non-compositional. Each sign is still composed of a value for each perceptual (phonological) feature, in agreement with Stokoe’s (1960) initial claim that the form of signs is not a holistic entity but composed of multiple subparts.
From the point of view of a production model such as the speaker side of the Functional Phonology model, the many exceptions in sign language lead to a much less dramatic conclusion than the one van der Hulst suggests: there is not much of a difference between sign languages and spoken languages in this respect. This may be true for most Optimality Theory models available today. Every sign has to have a representation (the input in OT), and the distinction between ‘normal’ and ‘exceptional’ features is not a binary one. I assume that all relevant aspects of the sign are stored in the perceptual specification, including the exceptional ones, and that we need no distinction between regular and exceptional features for the production model. The task of the phonological model below is to express which perceptual features should be considered to be relevant aspects of signs. Knowledge on the part of the user about common vs. uncommon forms is not expressed directly in the representation of the sign, but has to be characterized in another way. Markedness constraints in OT are obvious candidates, but using these seems to imply that there is a constraint against every exceptional feature (cf. Boersma in press). Future research will have to determine to what extent this is problematic.
Regardless of how exactly knowledge of frequency of occurrence of phonological forms is modeled, an important aspect of not using standard phonological features for such modeling is that more subtle intuitions about frequency and prototypicality can be modeled than just a binary distinction between ‘regular’ and ‘exceptional’ (or ‘loan sign’). It has been argued for spoken language that users have fine-grained intuitions about frequency of phonological features (Frisch 1996, Frisch, Broe & Pierrehumbert 1997), and there is no reason to expect signers to behave differently.
In the phonological model I present in §3.4, I will indicate which features seem to occur only in exceptional cases. In most cases an iconic motivation for the exceptional status can be proposed. A far more thorough examination of the relation between phonological exceptions and iconicity in SLN can be found in van der Kooij (in prep.).