Does ASL Really Have Just Two Grammatical Persons

17240-SLS9.2 1/16/09 9:56 AM Page 166 P E Y TO N TO D D Does ASL Really Have Just Two Grammatical Persons? This article presents new data bearing on the analysis of the ASL pronominal system. One reasonable view would be that the language has no pronouns at all; it does not need them since all pronominal functions can be accomplished simply by pointing, except as supplemented in particular cases by conventions, such as the understanding that signers can express the equivalent of English we by pointing to two locations on their own chest instead of by pointing first to themselves and then at other persons, or vice versa (although pointing to each person in turn is employed as well). However, although several competing analyses of the ASL pronominal system have been proposed, most sign linguists seem to have accepted the analysis proposed by Meier (1990), which holds that the language distinguishes only two persons: first and other. This view has in fact been incorporated into the notation systems of two influential books on the grammar of ASL: Neidle et al. (2000) and Liddell (2003). In essence, it claims that ASL has first-person pronouns—morphemes like those in spoken languages, which are linked to their meaning by arbitrary convention— but it relies on pointing to express second- and third-person meanings. The new data reported here support only a weaker version of that view, in which greater weight is given to iconicity, and status as a morpheme is seen as a gradient phenomenon driven by automatization. This evidence is to be found in the speech and signs of Vincent, a hearing child of deaf parents, who was fluent in ASL by the time he first heard speech at about three years of age. Vincent learned the correct usage of first-person pronouns in English rather early, after a brief Peyton Todd is a retired computer programmer living in Atlanta, Georgia. 166 Sign Language Studies Vol. 9 No. 2 Winter 2009 17240-SLS9.2 1/16/09 9:56 AM Page 167 Does ASL Really Have Just Two Grammatical Persons? | 167 period of pronoun-reversal errors (saying me to mean “you”). But it took him several years to work out all the other distinctions signaled by English pronouns (e.g., second vs. third person, gender, singular vs. plural, near vs. far from the speaker). This is easily understood if he came to the data of English with a two-person pronominal system that distinguished only first and other grammatical persons as Meier proposes. Studies of second-language learning by children of approximately the same age Vincent was while he was learning English usually find that, instead of transferring patterns from their first language, children tend to repeat the typical stages of first-language acquisition (Ellis 2003; Bialystok 2001, 67–69; Wong Fillmore 1976); but importation of patterns from the child’s first language into the second does occur (Odlin 2003), and when the pattern imported is as distinct as this one is—to my knowledge it is unattested in any spoken language—it is a fairly safe bet that its origin lies in the child’s first language, even though this view of ASL as having just a two-person pronominal system fails to capture its essentials, as this article will try to show. I begin with a brief summary of previous analyses of ASL pronouns. Then in section 2, I present some particulars of Vincent’s background, showing his early (and continuing) fluency in ASL. The next four sections show how strongly Vincent’s ASL background influenced his use of second- and third-person English pronouns, including his failure to distinguish clearly among them (section 4), his tendency to view them as encoding a request to look (or imagine looking) toward their referent (section 5), and his tendency to regard them as variants of a single word (section 3, but especially section 6). In section 7, I evaluate Meier’s arguments, and conclude that they do not succeed in establishing what they purport to establish. Section 8 nods briefly to Berenz’s (2002) challenge to Meier’s arguments, and section 9 takes issue with some extensions of Meier’s views by Liddell (2003), relying once again upon data provided by Vincent, both his childhood speech and his childhood signs. Sections 10 through 13 expand on these points by arguing that the concept of a morpheme provides no more than a crude approximation to the reality of ASL pronouns, which inextricably combine arbitrary and iconic components and are better viewed in terms of automatization, and in terms of exemplar theory (Nosofsky 17240-SLS9.2 1/16/09 9:56 AM Page 168 168 | Sign Language Studie s 1986), which would see them as a collection of stored exemplars rather than a constant underlying form that the exemplars merely approximate. 1. Previous Analyses of ASL “Pronouns” The difficulty with equating pointing in ASL with pronouns in speech is that, in speech, the form of each pronoun does not show us what properties to look for in identifying its referent, so these must be linked to it by convention. Since our minds are finite, and a few properties are generally sufficient, the pronoun systems in speech usually make use of only two or three discrete values on each of a small number of dimensions, such as “the speaker” vs. “the addressee” vs. “neither”; or “the male one”; or “the one you know to be a group of multiple individuals,” and so on. It is a system of this kind, in which the shape of the form itself provides no information as to its meaning, that most people have in mind when speaking of a morpheme. In contrast, the form of “pronouns” expressed by pointing in sign languages does provide information as to what property to look for in identifying the referent—it is always the “property” of being located at the place the signer points to after naming or describing the object (or of being imagined as located there). Unlike the situation in speech, the relation between the form and its meaning is an analog one that varies continuously with changes in the form. The checkered history of attempts to analyze ASL pronouns stands as eloquent testimony to how difficult it is to equate them with pronouns in spoken languages. Although the positions can be complex, and later interpreters were not unanimous in their views of what had been said, it seems fair to say that Lacy (2003 [1974]) and Friedman (1975) (pace Berenz [2002, 204]) equated the imagined “loci” of referents in sign space with pronouns, which signers reference by pointing at them, while Fischer’s (1975) position appears to have been a hybrid one in which the fully executed pointings are indeed pronouns, but phonological features of these pronouns (namely, the loci) are “cliticized” onto the verb when the sign for the verb is moved from one locus to another. Kegl (2003 [1976]) equated the signer’s 17240-SLS9.2 1/16/09 9:56 AM Page 169 Does ASL Really Have Just Two Grammatical Persons? | 169 body, classifiers,1 and null anaphora with pronouns, while Edge and Hermann (1977, 142) (again pace Berenz [2002, 204]) equated the signer’s body, classifiers (“markers”), and the “loci” with pronouns. Ahlgren (1990) recognized no personal pronouns at all but essentially only pointing in Swedish Sign Language, and Lillo-Martin and Klima (1990) held a similar position in regard to ASL, namely that its pointings—all of them—correspond not only to a single grammatical category that they named PRONOUN but also to a sign (presumably they meant the extended index finger), so that pronominal reference is accomplished by “directing the PRONOUN sign toward” the loci (in short, by an act that looks just like pointing), thus leading them to the conclusion that “there are no contrasts for person in ASL” (ibid., 198). In the same volume with the Lillo-Martin and Klima article appeared the article by Meier (1990) mentioned earlier, which argued that ASL does indeed have a grammatical category of person but distinguishes only two: first person and non-first person. Although Meier did not say so explicitly,2 he seems to have considered firstperson pronouns in ASL to be individual morphemes (unanalyzable units of meaning) since he stressed that first-person signs have “the center of the chest as their place of articulation,” in common with “a large number of nondeictic signs—for example, LIKE, FEEL, and WHITE” (ibid., 188); and also since he said that movements toward the signer’s chest with the index finger, B hand, and thumb (respectively) “are better glossed as ME, MY, and MYSELF” (ibid., 185); and further, that “the first-person plural pronouns WE and OUR appear to 1. Lillo-Martin (2003, 273) seems to regard Kegl’s (1987) view that classifier handshapes used in verbs are clitics as a modification of her 1976 view that they are pronouns. If so, then this is yet another indication of how hard it is to match this or that feature of ASL with the pronouns of spoken languages. I suppose it depends on whether you think that him is still a pronoun in its usual cliticized pronunciation in hit’im. The fact that its vowel often changes from /I/ to /i/ (at least in my dialect)— evidently to more clearly distinguish hit’im (which would have been /hIdIm/) from hit’em (/hId0m/)—suggests that the ’im in hit’im is already moving past the clitic stage on its way toward that of a bound suffix in a paradigm. (Here and elsewhere, I use the more familiar /d/ in lieu of the /R/, or ‘flap’, sound.) 2. By my count, the actual word morpheme occurs just three times in his article, on pages 177, 180, and 189. 17240-SLS9.2 1/16/09 9:56 AM Page 170 170 | Sign Language Studie s be lexicalized” (ibid., 189). According to Meier, then, in none of these cases is the articulation of the sign on the signer’s chest to be equated with pointing. I am less certain whether he regarded “the non-first-person deictic pronoun” (ibid.) as a morpheme, but the phrase just quoted suggests that he may have, in which case he may have had in mind the extended index finger, in common with LilloMartin and Klima (1990). As noted earlier, Meier’s argument has been accepted and elaborated on by Liddell (2003). Since I am not persuaded by their arguments, I review them later in this article and provide an alternate account of ASL pronouns after illustrating and discussing Vincent’s pronouns in sections 3, 4, 5, and 6; but I begin with a brief account of Vincent’s background. 2. Vincent’s Background: Precocious and Continuing Fluency in ASL Vincent’s parents are both excellent signers who never speak while they sign—a trait regarded by many ASL purists as a shibboleth distinguishing signers who are fully oriented toward ASL from those whose signs have been corrupted by English. In fact, in the many years I have known them I have never heard them speak at all, either to me or to Vincent, except on rare occasions when they called him by an approximation of his name. In the metropolitan area where they lived, all of their relatives were deaf; Vincent had no older siblings and no friends near his age that I was aware of (or heard him mention) until well over a year after my work with him began, about three weeks after his third birthday. His first exposure to speech that we know of occurred a short time before that, in weekly visits by a hearing graduate student and via his entrance into a nursery school. By that time, he was at least as fluent in ASL as the typical threeyear-old child of hearing parents is in speech; in fact, as early as two years, three months, his use of ASL “classifiers” was arguably more advanced than any so far reported in the signs of deaf children at that age, if we may judge from recordings made by another researcher, himself a hearing child of deaf parents, who completed a study of Vincent’s signs without speaking to him. The details may be found at 17240-SLS9.2 1/16/09 9:56 AM Page 171 Does ASL Really Have Just Two Grammatical Persons? | 171 http://www.vincentasl.info; in brief, Vincent was controlling the hands separately to indicate figure and ground a full six months before the most advanced child reported by Slobin et al. (2003). Since I will occasionally refer to his adult signs and his intuitions about them, it is important to emphasize that his fluency in ASL has not diminished. Although his English is flawless today, and no one meeting him for the first time would have the slightest suspicion of his unique background, he has had frequent exposure to ASL ever since his childhood and has worked full time as a highly successful ASL interpreter for all of his adult life. Two deaf signers to whom I showed some video clips of his signs—both of them graduates of Gallaudet who have taught college-level courses in which ASL was the language of instruction—recognized him either from the video clips or by name and responded immediately with the ASL equivalent of “Oh yes, I know him. He’s an excellent signer.” Although I continued recording his development until he reached the age of 11;6, for a total of 264 recording sessions, most of the examples of his pronouns reported here belong to the seven sessions that I have analyzed in detail, spaced at approximately equidistant intervals from age 3;5 to 5;9 (sessions 22, 40, 57, 67, 80, 124, and 177). However, his indiscriminate use of non-first-person pronouns was obvious to me throughout the entire time that I studied his speech. 3. Vincent’s Variable Pronunciation of Pronouns Vincent’s pronunciation of English pronouns was variable, and his various pronominal forms shaded into each other. This is important as an indication that he regarded them as variants of a single word since, if he regarded them as different words, we would expect the pronunciations of each one to be fairly consistent and to differ rather clearly from his pronunciation of other pronouns. I pursue this question more fully later on. Here our concern with this fact is methodological since it relates to how I transcribed these items: I simply glossed each one as the English form it most closely resembled—including contracted forms like isa and thatsa, which I can state with confidence were amalgams not analyzed until at least recording session 177 (age 5;9) if even then. Exceptions to this rule were a few forms whose spelling I in- 17240-SLS9.2 1/16/09 9:56 AM Page 172 172 | Sign Language Studie s vented (e.g., yu, du, and uh). Yu was how I spelled /ju/ whenever it might have had a third-person meaning. Du = /d0/, and uh = /0/ may have derived from English the and a respectively, but Vincent used them as pronouns instead of as articles. The forms si, lu, and yi were rarer, partly because I granted them status as independent pronouns rather late in the transcription process; si and lu may have derived from see and look; as I discuss in section 5, they seem to reflect the link between ASL pronouns (or pointing) and an appeal to the addressee to look (or imagine looking) at a referent. Yi= /jI/ or /ji/ appeared to be a compromise between you and he (the latter being pronounced /i/ 60 percent of the time, /hi/ 40 percent), and was the most frequent (though not the most spectacular) of a number of such blends discussed in section 6. While my decisions as to which English words to equate these forms to were mainly determined by choosing the English word that sounded the closest, in borderline cases I also relied on what the forms appeared to mean in context. I did so particularly in distinguishing isa and heresa since everyday English does not always pronounce the /h/ in here, and I did so in distinguishing ita from thata since English sometimes leaves off the initial consonant of that (as in atta boy!). For example, if Vincent suddenly found something he had been looking for and announced it by saying /'IS0/ + (noun) with heavy stress on /IS0/ and none of the following noun, I glossed the pronoun as heresa despite the absence of an initial /h/. Letting considerations of meaning creep into my glosses in these ways might appear to compromise my claims later on about what these forms meant, but in fact it has the opposite effect: Since my point in this article concerns the ways in which Vincent’s use of these forms differed from their use in English, my occasionally letting his meaning influence the glosses biases the case against me. 4. Failure to Observe Distinctions Made in English As noted earlier, although Vincent learned the distinction between first person and other persons rather quickly, it took him years to properly distinguish between second and third person, singular vs. plural, near vs. far (e.g., this vs. that), object vs. place (e.g., this vs. here), and 17240-SLS9.2 1/16/09 9:56 AM Page 173 Does ASL Really Have Just Two Grammatical Persons? | 173 others illustrated later. Hearing children in more typical circumstances can take time to learn some of these distinctions, too—at least the near vs. far distinction, which comprehension tests by Clark and Sengul (1978) found to be acquired on average at the age of 3;10 for here vs. there and 4;9 for this vs. that. But Vincent’s failure to distinguish among pronouns was much more far reaching and lasted much longer. Moreover, the test used by Clark and Sengul must have been rather demanding since it revealed an imperfection in the knowledge of more typical children that is not obvious in our day-to-day experience with them. But Vincent’s indiscriminate use of pronouns was glaringly obvious in his everyday usage. I therefore take it as resulting from the influence of his native ASL, in which most distinctions are made by the direction of pointing, thus on transparent, analog principles; apparently it simply did not occur to him that discrete distinctions signaled by opaque morphemes could play a role in indicating these meanings. The ideal way to demonstrate Vincent’s indiscriminate use of English pronouns would be to identify all of his forms, then determine for each one, independently of its pronunciation, what grammatical person it refers to and whether it refers to an object or a place, near or far, and so on. But this would be difficult to accomplish using audiotape recordings, which were supplemented by videotape only about one-fourth of the time in the recording sessions examined here (and less so in other sessions). Even though the video cameraman and I transcribed them as soon as we could after they occurred, in too many cases we cannot be sure whether the object referred to was second or third person, near to Vincent or far from him, and so on. I therefore settle for listing some of the clearer examples, and even then only a small number so that we have more space for discussion. All the following examples and several others are presented in more detail at www.vincentasl.info, along with audio clips that let you hear how Vincent pronounced them, and sometimes video clips of their context or of Vincent’s translations of them into ASL as an adult. I selected most of these examples from relatively late in his development in order to illustrate how long the phenomenon lasted; for that reason many of them sound a lot less ASL-like than his speech did when he was younger. Most of the examples are taken from the sample I referred to earlier, which I call “the core sample”: sessions 22, 35, 40, 57, 17240-SLS9.2 1/16/09 9:56 AM Page 174 174 | Sign Language Studie s 67, 80, 124, and 177, which spanned the ages from 3;5 through 5;9. Their reference numbers are to facilitate finding their context later (e.g., 22.506 is the 506th utterance in recording session 22). First person. In the earliest session of the core sample, Session 22 (age 3;5), there were four instances of “pronoun reversal errors”, in which Vincent says me when he means “you” (e.g., No, me = “No, you do it” [22.506]), and two more occurred in Session 27 (age 3;6). After that, as far as I am aware, every single first-person pronoun was used only to make a first-person reference. Pronoun-reversal errors are a fascinating topic but are omitted from this article because of the space limitation; interested readers can find a discussion of their causes in Vincent and in other children, including the ASL examples reported by Petitto (1987), at www.vincentasl.info. Second vs. third person. Vincent was still not making this distinction at age 5;9, approaching three years after his exposure to English began (e.g., How come yu wanna stop? “Why does she want to stop?” 177.349). Gender. He can’t go to school because he got baby? “Can she not go to school because she’s had a baby?” 177.1127 (age 5;9). He was still making this error as late as about age 10.3 Number. Long time, or yu make it? “Has it been there forever, or did they make it artificially?” 177.710 (age 5;9), referring to an island in Oakland’s Lake Merritt. Near vs. far. This is . . . this is a island. “That’s an island.” 177.708 (age 5;9), pointing to the island just mentioned, which was quite some distance away from us. Objects vs. places. Here Howard? “Is that Howard?” 80.2526 (age 4;6, with regard to a picture in a book); No, there gimme. “No, give me that.” 124.1306 (age 5;0). Objects vs. directions. Peyton: Now I’m going to shoot [this rubber band] over you. Vincent: Thatone? (= “That way?”) 124.380 (age 5;0) (pointing back over his own head). 3. Somewhere between ages 9 and 11, on at least one still untranscribed tape and probably on others as well. 17240-SLS9.2 1/16/09 9:56 AM Page 175 Does ASL Really Have Just Two Grammatical Persons? | 175 Objects vs. people. Peyton: Houses. Vincent: He not, school. “That’s not, it’s a school.” 35.1080 (age 3;9). There were no clear examples of this error with he that I know of after age 3;9, and although yu could still refer to inanimate objects as late as 5;0, it, itsa, du, uh, and si seemed never to refer to people. 5. Relation to the Concept of Seeing Two of Vincent’s pronouns provide evidence that ASL non-first-person pronouns should indeed be analyzed as pointing in the minds of signers (at least in part). I noted earlier that his pronouns si and lu might have arisen as a grammaticalization of the words see and look, and in this section I report some evidence which suggests that this occurred. I finally accorded these forms official status as pronouns when I kept encountering syllables whose pronunciation was so weak that they seemed to lack enough phonetic substance to qualify as see and look. Since lu is a latecomer to official status (with only six examples currently recognized), I focus here only on si, and on see, its hypothesized progenitor. If the pronouns in Vincent’s childhood speech were equivalent to pronouns in ASL, and if pronouns in ASL are similar to pointing, then a grammaticalization path from see to si is a reasonable one since to point at something is to invite another person to look in that direction and see what is there (or to imagine doing so). Note first that the word see was very common in Vincent’s speech. Lumping see together with the amalgam seeit, it was more frequent than any other verb except go and more frequent overall than any of his pronouns except I and you. It was also more frequent than the word see in the speech of other hearing children. The following figures show the number of occurrences of see (just see, not see + seeit) per thousand words in Vincent’s speech compared to its rate for the more typical children studied by other investigators:4 Adam, Eve, and Sarah 4. My criteria for selecting them were that they be naturalistic and well known, and occur in an age range suggesting rough comparability to Vincent; this latter criterion had to be highly intuitive since the relative roles of mean length of utterance (MLU) and cognitive maturity in determining occurrences of the word see are unknown. In fact, there appeared to be no variation by age within any of the children except Vincent himself, so I averaged the figures for each child across sessions (not a global average but an average of the per-session frequencies). 17240-SLS9.2 1/16/09 9:56 AM Page 176 176 | Sign Language Studie s (Brown 1973); Naomi (Sachs 1983); Peter (Bloom 1970); Ross (MacWhinney 2000); and Shem (Clark 1978). The final column of the table focuses on Vincent’s sessions 57–80, where its frequency was greatest. That number is over five times the average of the other children’s frequencies and over three times that of their most seeprone member (Sarah); age ranges are shown just below each child’s name. Adam 2;3–3;11 7.1 Eve Sarah 1;6–2;3 3;3–5;1 8.6 9.2 Naomi 1;2–5;9 Peter 1;9–3;1 Ross 2;6–5;4 Shem 2;2–3;2 4.1 6.8 2.9 3.7 Vincent Vincent core sessions sample 57–80 21.3 32.1 Only 31 percent of Vincent’s 896 uses of see/seeit in the core sample are about whether something is seen (e.g., he can’t see snake [67.2389] or I wanna go see giraffe [177.687]). This contrasts with 69 percent (620 utterances) in which we find “pragmatic,” or conversationregulating, uses like those illustrated in example 1, where see has the approximate meaning of “do you see?” Example 1. See, too big. 57.1902 “See, the toy car is too big to fit in its garage.” ... See? 57.1914A Too big, see? 57.1914B “It’s too big, see?” Why were these pragmatic uses of see so frequent? Givón (1979, 89) points out that children learning their first language are under communicative stress since they do not yet know the language. This would be just as true of a child learning a second language, and a child as old as Vincent might be sufficiently aware of this predicament that he would frequently check for comprehension. Nonetheless, it seems likely that part of the explanation lies in the fact that sign language is visual. The addressee is expected to see what the signer means in a literal sense. It was surely Vincent’s habit of ensuring the addressee’s visual attention before signing that led him sometimes to 17240-SLS9.2 1/16/09 9:56 AM Page 177 Does ASL Really Have Just Two Grammatical Persons? | 177 say see see see . . . (in rapid succession) to get my attention before speaking to me. His “pragmatic” uses of see in example 1 (shown just above) are slightly different in that they appeal to the visual context of the utterance, a stratagem that is probably more important in sign than in speech (indubitably so if we include the imagined visual context set up by earlier signs). In 45 of these 620 “pragmatic” uses of see it appears, on intonational grounds, to belong in the same clause with other material. Its meaning in these cases usually seems to be “Do you see that p?” (where p is some proposition), as in example 2, where he indicates the bulge in his pocket to prove that he has something in it. (As Vincent used it, the construction too + adjective meant very + adjective, possibly reflecting the fact that his native ASL distinguishes these two meanings nonmanually, for example, in the case of “too big” by the sign BIG plus a pejorative facial expression and a negative, i.e., pejorative, shaking of the head.) Example 2. Peyton: Oh, is there something in it? Vincent: Too big. See too big? 57.223B “It bulges out. See how it bulges out?” In many of these uses, however, this intonational union was so tight—and the pronunciation of /si/ so brief—that it seemed surely to be a pronoun. Example 3 shows in microcosm the grammaticalization that I am suggesting might have occurred. (Refer to www.vincentasl .info for audio of this and other examples of both see and si.) Example 3. See, big truck. “See, there’s a big truck.” 67.248 (age 4;2) Si big truck. “It’s a big truck.” 67.249 I cannot say for certain that si grew out of see. My reasons for believing that it did are, first, that it lacked the phonetic substance a fullfledged verb ought to have. Second, examples of see shade along a continuous cline into examples that seem to compel us to recognize si. Third, we have a serendipitous source of evidence from adult Vincent’s renditions of his childhood speech into ASL. On a number of occasions, when I asked him to sign one of his childhood sentences 17240-SLS9.2 1/16/09 9:56 AM Page 178 178 | Sign Language Studie s containing see, he left out the sign SEE. When I asked him what happened to SEE, he objected that he had already signed it when he pointed. Exactly the same thing occurred in an ASL translation that his younger brother Toby (also fluent in ASL) did for me of a sentence containing the word look. I suggest that this link between pointing and inducing the addressee to see something was the basis of Vincent’s grammaticalization of see into the pronoun si, if that is what occurred. If so, then it suggests how similar ASL pronouns are to true pointing in the minds of signers. 6. Were Vincent’s Pronouns Forms of a Single Word? We have seen that apart from first person, Vincent used most English pronouns as though they were equivalent to each other. His pronunciation of them provides evidence that in fact they were not just synonyms for him but also forms of a single word, or at least in the process of becoming so. As noted earlier, it was sometimes hard to decide whether to gloss a given pronoun heresa vs. isa or thata vs. ita. Another example was thisone vs. thisway, which he also used interchangeably. Choosing between them was easy when he pronounced them as /dIsw0n/ or /dIswe/ but sometimes he said /dIswẼ/, a form that he never would have heard, so he must have produced it as a blend of the two forms that he did hear. And I have mentioned the highly frequent yi, which appeared to be a blend of yu with he. Other examples were less frequent than those just mentioned but more impressive. The form in example 4 may be a blend of here and the pronoun asai, which Vincent most likely derived from outside but used as a general deictic equivalent to here or there. The other fortyseven instances of asai in our core sample were all pronounced /asai/ (hence the spelling). Here, however, it appears with an initial /h/. Example 4. Puter water, hasai 40.1084 (age 3;10) /pV t0 w0: hasai/ standing beside a garden hose coiled up against the wall of a house near the water faucet that it is attached to; not just the physical context but also other utterances in the same situation make it clear that the second word here is water; puter is an all-purpose verb similar to English do. 17240-SLS9.2 1/16/09 9:56 AM Page 179 Does ASL Really Have Just Two Grammatical Persons? | 179 Additionally, in example 5, what I write as herethe was pronounced /hit0/, a sequence of sounds Vincent would not have heard but one which might be understood as a blend of he and the /it0/ variant of ita (which could be pronounced as /it0/, /id0/, or /Id0/), with perhaps some influence from his doubled pronoun it there. It is not to be identified with hita =/hId0/ (Vincent’s version of hit). The initial vowel of the latter was always /I/, not /i/, and he never used hita for striking a match. Examples 4 and 5 may be heard at www.vincentasl.info. Example 5. Yay! Herethe match! 57.1731 (age 4;0) cheering that he has persuaded me to play a matchstriking game Obviously this does not imply that Vincent was conscious of viewing these as forms of the same word. Nonetheless, a recent theory in phonology shows how storing them in his mind as though they were instances of the same word could unconsciously lead to the preceding pronunciations. It is rather complex, but expounding it here will be time well spent not only for explaining Vincent’s tendency to merge these diverse forms into a single word but also for its contribution to the view of ASL pronouns that I suggest in section 11 of this article. Exemplar theory, as it is called (Nosofsky 1986, 1992; Heit and Barsalou 1996), arose originally in psychology as a theory of concepts but found fertile ground in linguistics. Distinct not only from the classical view of concepts as necessary and sufficient conditions but also from Rosch’s (1973) view of concepts as represented by a single best example or ‘prototype’ and from Lakoff ’s (1987, 95) view of them as clusters of examples chained to a central one, exemplar theory holds that we simply remember a very large set of exemplars. We cannot keep a separate record of every examplar we encounter since memories fade, and we cannot detect differences of infinite subtlety. However, when a new exemplar is indistinguishable from one already stored, its occurrence causes the latter to gain in strength, making it in effect more prototypical, although it is still one of a cluster of a great many other more or less prototypical exemplars. In linguistics, this theory has been applied to phonemes by Miller (1994) and to larger units like the word by K. Johnson (1997) and by Pierrehumbert (2001). In Pierrehumbert’s version, our memory of a 17240-SLS9.2 1/16/09 9:56 AM Page 180 180 | Sign Language Studie s word is viewed as a “large cloud of remembered tokens” (ibid., 140) of its pronunciation, which occupies a region in the multidimensional parameter space defined by the dimensions along which the tokens vary (e.g., formant values). When a new token is encountered, its degree of similarity to each member of the set of stored tokens is computed, assigning decreasing weights to each one as distance from the encountered token increases. Stored tokens that are activated by the encountered token may belong to more than one word, and the word with the most stored tokens of the greatest strength wins, where strength is based on how many tokens indistinguishable from it were previously encountered. The important consequence for us here is that when this is calculated mathematically, the “best” exemplar of a category—here the best pronunciation of a word—is the one positioned so as to yield the highest weighted sum of the stored tokens of that word. And that may be a token which has never been previously experienced. The above examples from Vincent were of production, not comprehension. But Pierrehumbert extends her model to cover production by proposing that production selects a “window” as well (in neural terms, perhaps a region in the brain). Projecting this effect to the multidimensional set of Vincent’s pronouns would give us the result we need, for the effect occurs only with pronunciations regarded as belonging to a single word or to a collocation tantamount to a word (Bybee 2001). I conclude that Vincent’s tendency to blend his pronouns together in this way is good evidence that they really were a single word as far as he was concerned. Of course, it is a far cry from the blends described above to, say, collapsing herethe with thatway, but if the pronominal system that Vincent and his younger brother Toby used had been left to itself, in isolation from the English they heard spoken around them, such a thing could well have happened—unless, of course, the usefulness of such distinctions would have led them to forge distinctions of their own from some of their existing pronunciations. 7. Meier’s View of ASL Pronouns Now that we have described Vincent’s usage of pronouns, let us ask to what extent it supports Meier’s position. I argue here that the line between what is or is not a morpheme is being drawn too sharply. First- 17240-SLS9.2 1/16/09 9:56 AM Page 181 Does ASL Really Have Just Two Grammatical Persons? | 181 person pronouns in ASL may come closer to qualifying as morphemes than other pronouns do, and this may have helped Vincent catch on to the use of English first-person morphemes more quickly than he did non-first-person ones, but it is time to move past the either/or choice between morpheme status and its absence. I contend that ASL pronominal reference is accomplished by such a shifting mix of partially fixed physical configurations and partially novel pointing behaviors that the choice of some of them as morphemic to the exclusion of others is somewhat arbitrary. In reviewing the arguments of Meier here and of Liddell in section 9, I therefore point both to the possibility that some of the morphemes that they accept are more iconic (hence, less morphemic) than they suggest, and to the possible existence of more morphemes than they accept. If both of my arguments seem plausible even though neither seems ironclad (which is how I myself view them), I suggest that this is because status as a morpheme is a matter of degree. I begin with Meier’s arguments that ASL has a first-person morpheme, then turn to his contention that it does not make a grammatical distinction between second and third person. He presents three arguments for the first proposition and four for the second. 7.1. Does ASL Have a First-Person Morpheme? 7.1.1. First Argument Meier (1990) notes that the signs normally labeled WE, OUR, and OURSELVES are “only partially motivated” (ibid., 180) since the signer does not point to the real-world locations of those persons other than the signer to whom these signs refer. Instead, these signs contact two successive locations on the signer’s chest—with the index finger for WE, a B hand for OUR, and the thumb for OURSELVES. Moreover, he says, the form of OUR is idiosyncratic since, unlike THEIR and Y’ALL’s, it does not execute a sweeping movement but contacts one side of the chest with the thumb side of the B hand, then rotates before contacting the other side of the chest with the little-finger side of the hand. “In sum, the phonological form of WE and OUR . . . argues that ASL has a first-person category” (181). However, if these signs are “only partially motivated,” does that not already imply that we can regard them as partially nonmorphemic? It bears repeating that Meier (ibid.) did not refer outright to ASL first- 17240-SLS9.2 1/16/09 9:56 AM Page 182 182 | Sign Language Studie s person forms as morphemes. His concern was rather with how many grammatical persons ASL should be analyzed as having, and for this he evidently felt it to be sufficient that there be some specific physical configurations that at least partly require convention to connect them to their meanings, and that it be necessary to mention a particular grammatical person to describe their usage. But regardless of how he should be interpreted on that point, my purpose here is to argue against the overly simple approximation to the facts implied by any simple statement that ASL first-person forms are morphemes, whoever might subscribe to that position (Liddell 2003, 28, for example, calls them morphemes). To that end I argue below that first-person pronouns in ASL are less than fully morphemic because of the iconicity they still make use of, while at the same time suggesting that they are less than fully morphemic because they can be analyzed into a larger number of discrete “morphemes” at a lower level. As to the first point, we can certainly perceive some iconicity in the signs WE, OUR, and OURSELVES if we regard the signer as pointing, in a plural manner as it were, to himself as proxy for the group he presents himself as belonging to—in a manner adopted historically no doubt because it is easier to contact two locations on the chest than to point first to the self and then outward in some other direction. We could also see the twisting motion made by the B hand in OUR as an iconic representation of the boundaries of the group being referred to, like that which also occurs in CLASS, TEAM, and AGENCY. We do not know that this iconicity is employed in actual signing, but we also do not know that it is not. In cultures around the world, pointing is partially shaped by cultural convention without thereby ceasing to be pointing. In some cultures, people point only with the index finger, while in others they point with the lips (Labarre 1947). The Arrente of Australia distinguish three sorts of pointing with the hand: one-finger pointing for individuated objects, wide-hand pointing for areas or for objects spread out over an area, and flat-hand pointing (including a variant that resembles the ASL 3-hand vehicle classifier) to indicate paths of travel. Moreover, they are aware of these as culturally transmitted since they say that children must be taught to point (Wilkins 2003). Does this make the gestures of the Arrente morphemes? Although some might object that a 17240-SLS9.2 1/16/09 9:56 AM Page 183 Does ASL Really Have Just Two Grammatical Persons? | 183 given gesture is not a morpheme till it becomes part of a linguistic system, it is not obvious that incorporation into such a system changes the nature of a gesture itself. Meier, for example, considers it worth noting that pointing in ASL is “interspersed in the sign stream with other nondeictic, nonanalogic signs” (1990, 179). But pointing is interspersed in the speech stream as well, and is even subject to a grammatical rule specifying how far away we can point simultaneously with the words this vs. that. Are these gestures morphemes when used with speech but not at other times? Moreover, the first-person plural signs mentioned earlier do sometimes point to the locations of persons other than the signer. Liddell (2003, 29) observes that the sign typically glossed as WE can take on a different form depending on the location of the other members of the group being talked about. For example, if the other group members are to the signer’s right, the signer will lean toward the right while making the sign; also, the precise location of the points of contact on the chest will change. For a right-handed signer, instead of an initial contact on the right side of the chest followed by a contact on the left side, we will see a contact on the right side followed by one near the sternum. Thus the overall movement gestalt on the chest will move rightward—pointing, in effect, to the other members of the group. This intricate mixture of iconic and arbitrary principles is in fact quite characteristic of ASL in general.5 At the same time, we can also propose an analysis that regards these signs as consisting of five or six morphemes instead of the three rec5. I believe that my use of the word iconic to include pointing is appropriate. Not only is pointing more or less included when people ask to what degree sign language is “iconic,” but if we hark back to the father of these terms, C. S. Peirce, pointing also seems closer to his concept of an icon than to his concept of an index, which always stressed the real causal connection between the index and what is signifies (like a bullet hole and the bullet that made it). So, while pointing can, of course, be an index of a person’s intention to point or of an infant’s level of development, its relationship to the thing pointed at seems closer to being an icon since there is a relationship of analogy between the orientation of the pointing device (e.g., the index finger) and the direction in which we are to look (or imagine looking) to find the thing pointed at. Described more fully, in Peirce’s terms, since it functions in virtue of the habits and natural dispositions of the people using it (the defining feature of a symbol), we could consider pointing as an icon serving as a constituent of a symbol. 17240-SLS9.2 1/16/09 9:56 AM Page 184 184 | Sign Language Studie s ognized by Meier. For example, we could identify the three handshapes just mentioned (index finger, B, and extended thumb) as separate morphemes meaning “nominative/accusative,” “possessive,” and “self,” respectively. It does not defeat this analysis to point out that homonyms may exist, or that the same shapes are as meaningless elsewhere as the “cat” in cataract. We could also recognize a plural morpheme realized in the two contacts with the chest—the same morpheme as the “sweep” movement seen in non-first-person plural signs like THEY but subject to a different morphophonemic rule. That gives us four morphemes, and the first-person morpheme realized by contact with the chest would be another. All five of these retain their meaning when they combine with other possible morphemes like “singular” and “non-first-person,” just as proper morphemes should. Possibly the only reason Meier (1990, 180) did not adopt such an analysis is that it fails to predict the twist of the wrist that occurs in OUR. However, if we were to abandon for the moment our earlier explanation in terms of iconicity, we might see this motion as a sixth morpheme, identical to the “group” morpheme that also occurs in CLASS, TEAM, and AGENCY. Its occurrence in the multimorphemic sign OUR would thus be a contraction (on the model of is + not = isn’t) of the “plural” morpheme with the “group” morpheme (remember, we are viewing the latter morpheme for the moment as just a curved shape meaning “group” but with no iconic properties). If it seems reasonable to argue both that the several alleged firstperson morphemes are partially iconic and that more morphemes may exist than claimed, I suggest it is because these first-person forms in ASL are partially morphemic; at the very least we should not consider the issue to be resolved. 7.1.2. Second Argument Meier’s second argument for the existence of first-person pronouns in ASL is that in direct quotations the sign referring to the first person functions exactly as the first-person pronoun does in English, namely to refer to the person who uttered the quoted sentence, not to the person quoting it. I do not present the details of this second argument here since, in my view, this property of first-person reference would be just as true if it consisted entirely of pointing. As is well-known, 17240-SLS9.2 1/16/09 9:56 AM Page 185 Does ASL Really Have Just Two Grammatical Persons? | 185 in these situations the signer has adopted the role of the person being quoted, so it is to be expected that first person forms would refer to that person regardless of what principle determines them. Curious readers may evaluate this for themselves, if they wish, by reading the original (ibid., 181–85). 7.1.3.Third Argument Meier’s third argument for the first person in ASL is like the first but is presented separately (ibid., 188) and is not specific to the plural first person. It is the argument noted earlier that, unlike references to second or third person, where the pointing varies in continuous analog fashion with the referent’s real or imagined location, references to first person have the chest as a constant place of articulation, being in this respect not different from lexical signs like FEEL, LIKE, and WHITE. But this does not show that the signs making first-person reference must be morphemes. It merely removes a possible objection to their being morphemes. And intuitively it seems strange that a signer would experience the “singular first-person pronoun”—the one that looks just like pointing to himself—as being, in fact, something other than pointing to himself. Imagine that you are a signer pointing successively to two other persons and then to yourself. In this context, would you not feel that the third point—to yourself—is the same type of event as the two that preceded it? Is it plausible that a signer who directs his index finger toward himself at the start of a ME + Verb sentence is now doing something different? It may perhaps be slightly different. My point throughout this article is that these matters belong to a cline; it may be that we are farther along this cline when the handshape of the first-person pronoun assimilates to that of the upcoming verb than when it retains the extended index finger. A similar cline might be seen in the spoken words bow-wow and meow, depending on their use in a sentence: Meow seems more fully morphemic in The cat meowed than it does in The cat went “meow,” while bow-wow in The dog bow-wowed seems to lie between them, being more morphemic than it is in The dog went “bow-wow” but less so than meow is in The cat meowed; its conventionalized use as a verb was preempted by bark, but that merely explains—and does not alter— its location on the cline. Nor do the differences between these cases 17240-SLS9.2 1/16/09 9:56 AM Page 186 186 | Sign Language Studie s (pointing vs. quasi-quotation) detract from the general point that both manifest a cline in the degree of morphemehood. 7.2. Does ASL Have Second and Third Person Morphemes? Turning now to Meier’s arguments for the absence of a grammatical distinction between second and third person in ASL: as noted earlier, the problem in finding a second- or third-person morpheme is to find a form that distinguishes them but whose shape provides no information as to its meaning. As Meier (1990) points out, it does not suffice to describe the second-person pronoun as “pointing at the addressee” since the alleged description of the form includes a mention of its meaning. His first two arguments against a distinction between second and third person in ASL are aimed at refuting two possible attempts to escape this dilemma. 7.2.1. First Argument First, he says, suppose that we take our cue from the prototypical sign conversation in which the addressee is directly in front of the signer and any third person or object is off to the side. In this situation, we could identify the sign for second person with pointing straight ahead and third person with pointing to the side. Unfortunately for this attempt to solve the problem, in many conversations the addressee may not be directly in front of the signer. There could be two addressees, one on the left and one on the right. However, note that for each of these addressees, there is in fact an intrinsic shape—the pointing finger viewed end on—that tells him when he is being referred to. This is of course not the same as telling him that he is the addressee, but we will see how it can combine with other events in the sentence to accomplish this feat when we consider Meier’s second argument just below. And it will take its rightful place within the second-person schema I propose later on. 7.2.2. Second Argument Here Meier addresses the question whether we can solve the problem by including eye gaze at the addressee. Could the second-person pronoun in ASL consist of pointing to and gazing at a person at the same time? Unfortunately for that proposed solution, reference to one’s ad- 17240-SLS9.2 1/16/09 9:56 AM Page 187 Does ASL Really Have Just Two Grammatical Persons? | 187 dressee can occur without simultaneous eye gaze toward that person. Meier cites the following sentence from Baker and Padden (1978), who indicated the direction of eye gaze above the signs (where “at A” means at the addressee). In producing the first instance of YOU, the signer is looking downward, not at the addressee: Example 6. at A up, right down RAIN, YOU TOMORROW at A GO SWIMMING BEACH YOU In fact, however, there is no need to require that the pointing and the gazing be simultaneous. In the tradition of showing a particular feature of sign language to be really “linguistic” by finding an analogy in spoken languages, we can point to examples of discontinuous morphemes elsewhere. Negation in standard French (ne . . . pas) is an example, as are “separable verb” constructions (such as head ’em up and move ’em out) in English and German. This situation is made possible by the existence of rules (realized in performance as highly practiced procedures) for joining the two pieces together. No doubt the usual procedure for recognizing the second person in ASL is to look for pointing and eye gaze in the same direction at the same time, since most instances of YOU probably meet this criterion. For situations like that in example 6, one can look for a relation of identity between the direction of eye gaze occurring at the end of an utterance and a direction of pointing occurring earlier. From the addressee’s point of view, both of these are intrinsic physical shapes—the pointing finger viewed end on and the eyeball viewed end on. And there is a rule for putting them together. I conclude that we can speak of an unambiguous, specifically second-person pronoun in ASL, albeit a separable one like a separable verb (and one that we will later see as residing in a wider second-person schema, along with the variant where pointing and gazing line up, as well as other variants described in section 11.) 7.2.3.Third Argument Meier’s third argument is that certain “agreement” verbs (which move in space to identify their subjects and objects) lack first-person object forms, but apparently no verbs must be listed as having a specifically 17240-SLS9.2 1/16/09 9:56 AM Page 188 188 | Sign Language Studie s second-person object form but no third-person object form or vice versa. He suggests that only some of these gaps in agreement have “phonological explanations” (explanations in terms of the difficulty of making the requisite movements). Presumably his point is that the absence of such explanations means that they must depend on arbitrary convention and are therefore truly linguistic. On the contrary, I suggest that all of these gaps can be explained by factors other than arbitrary convention. The verbs he lists are FINGERSPELL-TO, SEE, PAY-ATTENTION-TO, PRAISE, RUIN, and CONVINCE. But note that the first four can be viewed as compromises between “agreement” and “body-anchored” verbs: Unlike classic body-anchored verbs, they include movement or orientation of the hands outward from the body, thus enabling them to move or orient toward real or imagined objects, while the other end of the movement remains anchored to the body. This anchoring to the body occurs because they contain a heavy iconic component: In FINGERSPELL-TO the hand is held beside the head in the location normally used in fingerspelling; in SEE, the V hand moves outward from the vicinity of the eyes to show the direction of gaze, in a metaphor as old as the ancient Greeks (Slobin 2006); in PAY-ATTENTION-TO, the B hands move out from the eyes as dictated by that same metaphor, while also portraying blinders restricting its focus; and the role-play in the second part of the two-part sign PRAISE seems obvious (GOOD + applauding hands according to Sternberg 1998; TRUE + applauding hands according to Costello 1998, but in truth applauding hands would usually be sufficient). While it would be physically possible to move or orient the hands toward the self in these cases, to do so would destroy the iconic pictures that they portray. If FINGERSPELL-TO or SEE were reversed, they would no longer have the same relation to the signer’s head; reversing the direction of PAY-ATTENTION-TO would contradict the existing metaphor (as it would also do for SEE); holding the hands near vs. far from the self is not something people usually do when they applaud. This is not to say these obstacles are insurmountable. Liddell (2003, 25n) points out that a different sign for fingerspelling (S opens to 5, not wiggling five fingers as in the sign discussed here) can be directed toward the self; and two Atlanta signers have corrected my perform- 17240-SLS9.2 1/16/09 9:56 AM Page 189 Does ASL Really Have Just Two Grammatical Persons? | 189 ance of PAY-ATTENTION-TO, insisting that the hands must move toward the self—not that they have turned it into an agreement verb with movement toward the self meaning “pay attention to me”: for them this is simply the constant form of the sign. Probably either they or the signers they copied it from have simply adopted a different metaphor based on taking in information. (Other Atlanta signers I have consulted use only the standard version as described earlier, although one did say that either version is acceptable.) In any case, while these examples show that the iconic pictures can be overridden, they do not show that they do not act as a brake on such innovations. That leaves the signs RUIN and CONVINCE. Meier cites Fischer and Gough (1978) for the claim that RUIN cannot move toward the self. But their account of that verb is puzzling. They describe it as being “made by a fist dez [designator] grazing the back of the tab [location] wrist and hand pronated as it sweeps outward” (ibid., 32). However, nine deaf acquaintances for whom I performed this sign were unanimous in stating that they had never seen it; when I told them it was alleged to mean “ruin,” two of them surmised that it might mean “ruin” in some other dialect of ASL. The sign usually glossed as RUIN is made with an X hand (not a fist) moving across the thumb side of another X hand (not the back of a wrist). In any case, Fischer and Gough contrasted this “fist dez” sign with TEASE, which can reverse “even though it is almost identical to RUIN except that the movement or action is less forceful and is repeated.” The sign usually glossed as TEASE does indeed relate to the sign usually glossed as RUIN in precisely the way they describe. But if that RUIN sign cannot be directed toward the self, this may be the result of a selection restriction similar to that which we find in English, where only things, not people, are generally spoken of as ruined (except for financial ruin, but that sense now seems to have passed into disuse). When the same sign is translated as “persecute,” it can sometimes be directed toward the self. I have asked two deaf Christian ministers whether, in signing the words of Jesus to the apostle Paul (“Why do you persecute me?” Acts 9:4), one could direct the sign toward oneself (while I demonstrated what this would look like). One answered with the equivalent of “Sure, no problem”; the other said that you could do it, but directing it away from the self would be more common. The ob- 17240-SLS9.2 1/16/09 9:56 AM Page 190 190 | Sign Language Studie s vious reason is that it is awkward to make such a forceful, hence large, movement toward the self. As for CONVINCE, there are two variants. One is performed by “placing an upright index finger toward the relevant person, then contacting the index finger with the ulnar surface of an active B hand making a diagonal downward movement” (Liddell 2003, 25–26); the gap here is that the sign supposedly cannot be made toward the self (but see below). The other variant is made by moving two hands diagonally downward and toward each other in the space in front of the signer. That is its secondand third-person form; it is described in an ASL internet dictionary (www.commtechlab.msu.edu/sites/aslweb/browser.htm) as chopping “at an imaginary neck as if to show that there is no room to move,” which would make this sign iconic for signers who are aware of that image as they use it (the standard first-person object form does chop with both hands at the signer’s own neck). Here the gap lies in the fact that, according to Meier, there is a dialectal variant of the first-person form that contacts the neck with only one hand, but no one-handed version second- or third-person form. This is indeed a matter of dialect since adult Vincent’s dialect is exactly the opposite: He rejects the one-handed chop at the neck since it looks too much like the sign for BROKE (= poor) but accepts a one-handed chop in neutral space oriented toward second or third person. And he accepts making the upright index finger variant toward the self as long as it is performed not by placing that finger near the self but rather by making the diagonal, B hand, upright-finger-striking movement toward the self. This can be somewhat physically uncomfortable, but that is easily solved by turning the B hand over so that its radial surface (index finger side) strikes the upright index finger of the passive hand—a version that he says he has seen some deaf people perform. On the other hand, he does not accept the one-handed chop in neutral space made toward the self. This is indeed a gap since he does accept it made toward (or rather oriented toward) others. However, it is a gap that is explained by its lack of distinctiveness, just as we saw with the one-handed chop at the neck. He tested this in a real conversation with a deaf person, setting up the conversational context so that “convince me” would have made sense, 17240-SLS9.2 1/16/09 9:56 AM Page 191 Does ASL Really Have Just Two Grammatical Persons? | 191 but the deaf person responded with the equivalent of “What? Are you inviting me somewhere?” Without the upright index finger of the other hand in the picture, it looks too much like INVITE. A similar explanation might apply to the signers who reject all variants of the one-handed chop in space: Either they or those they learned it from may see it as insufficiently distinctive without a neck or an upright index finger to chop at. Nonetheless, despite all these points, which help to clarify the ultimate reasons for these gaps, we should not conclude that these gaps are not a matter of convention in the present. On the contrary, the mechanism by which these signs operate in the present is surely the same as that employed for any conventional linguistic unit, namely habit. Even assuming the truth of the explanations I have offered, it does not seem likely that the routine for expressing first-person object meanings with these verbs begins with an impulse to move or orient them toward the self, which is then found to be either too awkward to perform, incompatible with their iconic components, or insufficiently distinctive to be recognized. It is more likely to depend simply upon the desire to express a first-person meaning. Nor (if put into words) would it have a form like “If you’re expressing a first-person object meaning with this verb, don’t try to reverse it.” Rather, it consists in the absence of any routine for reversing the verb for expressing that meaning and in the presence of a routine for expressing it in some other way. Let us grant, then, that the presence of these gaps is a matter of convention (habit)—just not arbitrary convention. Are there really no gaps of this kind other than first-person gaps? In fact, some do appear to exist. It has been recognized for some time that subject-verb agreement is not absolutely obligatory in ASL (e.g., Neidle et al. 2000, 34, citing references that go back to Fischer and Gough 1978). According to Vincent, agreement with the object is not always obligatory, either; for either of these one-handed-chop variants of CONVINCE, agreement is not obligatory unless the object is second person. The situation is diagrammed in example 7. (To make these examples easier to read, I write HIM and YOU for the object pronouns realized by pointing.) 17240-SLS9.2 1/16/09 9:56 AM Page 192 192 | Sign Language Studie s Example 7a. ME CONVINCE-rt HIM-rt “I convinced him” 7b. ME CONVINCE-cntr YOU-cntr “I convinced you” 7c. ME CONVINCE-cntr HIM-rt “I convinced him” 7d. * ME CONVINCE-rt YOU-cntr “I convinced you” Sentences 7a and 7b show full agreement, where in fact the object pronouns (and the first-person pronoun as well) are likely to be omitted when they do not express a contrast; 7d is the unacceptable case and could even be given extra stars since it looks downright weird to sign CONVINCE off to the right and then point directly ahead to the addressee. However, it is quite acceptable to sign either of these versions of CONVINCE toward the center and then point off to the right as in 7c. Signing them toward the center and then pointing to the self to say “He convinced me” is also acceptable. So agreement is required in this case only with second person, and we seem to have a rule of grammar that cannot be stated without mentioning second person explicitly, at least if this sort of agreement is part of the grammar, as it is usually taken to be despite its analog implementation. Another way to see this might be to regard the center-directed version of these signs as their citation form. Viewing it from that perspective, we can say that agreement is available as an alternative to the citation form only in the case of third person, and then we have a rule of grammar that cannot be stated without mentioning third person. And it is not just the signs for CONVINCE that work as diagrammed in example 7: According to Vincent, even GIVE works like that for its second- and third-person object meanings. I conclude that the rules for verb agreement in ASL require us to recognize a distinction between the second and third grammatical persons. 7.2.4. Fourth Argument Coming finally to Meier’s fourth argument for the absence of secondand third-person categories in ASL, he pointed out that “the linguistic category of third person is not sufficient to distinguish among the 17240-SLS9.2 1/16/09 9:56 AM Page 193 Does ASL Really Have Just Two Grammatical Persons? | 193 available pronominal forms” (1990, 188) since there are indefinitely many directions the signer might point toward in making third-person reference. Liddell (2003, 26) has pointed out that the same is true of second-person reference. Of course, it is not really the multiplicity of possible directions in itself that matters but whether that leads to indefinitely many possible physical configurations as candidates for a second- or third-person pronoun. While it may do so for third person, we have already seen in section 7.2.2 that it does not for second person since a second-person pronoun consisting of two specific physical configurations joined by a grammatical rule does indeed exist. This forces us to acknowledge a distinction between second and third person. 8. Berenz’s Arguments Against Meier’s View For completeness, I briefly review here the recent article by Berenz (2002) in which she attacks some of the same arguments that I have. While I agree with her in spirit, I confess that, as far as I can see, her counterarguments do not refute Meier’s position. She points out (210, bottom) that signers express a distinction between second and third person, but Meier did not deny this. She stresses the importance of lining up eye gaze and the angle of pointing with the midline of the body to express the second-person meaning (211), but Meier acknowledged this to be the typical case (1990, 185). Nowhere did he deny the point that, as she puts it (2002, 212), “gaze and midline are crucial determiners.” It is just that exceptions exist, at least insofar as their simultaneity is concerned, such as example 6 above. Nonetheless, we will be able to resuscitate Berenz’s stress on the midline of the body in section 11 after more discussion of what a morpheme is. 9. Liddell’s View of ASL Pronouns Liddell’s (2003) view of ASL corresponds well to my own in that it sees the language as a hybrid of arbitrary and iconic components. He is also aware of the possibility that some signs may be “partially morphemic” (ibid., 274). However, in discussing pronouns, he writes as though this were an either/or matter, and I insist that it is not. 17240-SLS9.2 1/16/09 9:56 AM Page 194 194 | Sign Language Studie s 9.1. Personal Pronouns Liddell is quite explicit in claiming that ASL pronouns are morphemes (ibid., 28). And he is clear in adopting the view, also expressed by Lillo-Martin and Klima (1990), that these pronouns are directed toward their referents (2003, 66, 96), while at the same time equating this with pointing (ibid., 68, 96). He also extends a suggestion by Meier (1990, 189) that the pronominal forms that incorporate numeral handshapes (TWO-OF-US, etc.) are compositional, and he provides a table (2003, 231) that contains twenty-six distinct pronouns. I present a subset of that table below and argue that his proposal accords morphemic status to forms that are in fact analyzable, while also obscuring their iconicity: Singular First person Non-first person PRO-1 PRO POSS-1 (MY) POSS Dual PRO-DUAL-1 PRO-DUAL Multiple {PRO-MULT-1}{THREE} {{PRO-MULT-1}{FOUR} {PRO-MULT-1}{FIVE} {{PRO-MULT}{THREE} {{PRO-MULT}{FOUR} {PRO-MULT}{FIVE} Plural PRO-PL-1 (WE) PRO-PL POSS-PL-1 (OUR) POSS-PL It is important to be clear about the claims that this table embodies: The curly brackets represent bound morphemes. For example, {PROMULT}{THREE} (= “a group of three individuals other than the signer”) combines the {PRO-MULT} morpheme (a rotating hand held palm upward some distance out from the signer’s body but with no specified handshape) with the {THREE} morpheme (three extended fingers with no specified movement or orientation). They are regarded as bound morphemes because each can occur only in combination with the other or with some other morpheme. And each retains its meaning in these combinations, as a proper morpheme should: {THREE} still means “three” when twisted in the air in {THREE}{DOLLAR(S)} or slid outward along the upturned palm of the other hand in {THREE}{WEEK(S)}. 17240-SLS9.2 1/16/09 9:56 AM Page 195 Does ASL Really Have Just Two Grammatical Persons? | 195 However, calling these sign components morphemes ignores whatever iconicity they may have since a morpheme by definition is a shape with only an arbitrary connection to its meaning. To represent the signs for numbers as words inside of brackets—e.g., as {THREE}, {FOUR}, and {FIVE}—is to treat them as being just as opaque as the English words three, four, and five. Calling the contents of these curly brackets morphemes also means, again by definition, that we cannot analyze them further into meaningful elements. The hyphenated expressions inside these brackets, therefore, cannot represent combinations of anything but are there only as mnemonics to help us researchers identify which unanalyzable morpheme is being referred to. Placing them inside curly brackets also denies the possible iconicity of MULT and POSS: for example, that the circular motion in MULT might define a group in the same manner as that in CLASS, TEAM, and AGENCY, or that the flat hand in POSS might define a barrier that the possessed object cannot cross to leave its possessor. Their possible status as separate morphemes is also denied: As mere mnemonic labels they cannot even be said to mean the same thing in their repeated occurrences since they do not mean anything at all in the notation system itself. Every one of the signs whose mnemonic label contains MULT includes a circular motion made by an upturned palm, and every one of the signs whose mnemonic label contains POSS includes a B hand, but according to this notation these commonalities are no more meaningful in themselves than the /tæk/ sound that is common to the English words tactics, contact, and syntax or, for that matter, no more meaningful than the X handshape common to the signs TAX, DRY, ONION, and APPLE. In fact, it is clear from the examples that the notation is not intended to represent the fact that many of these signs contain the same or similar handshapes or movements. Unlike the cases of MULT and POSS just described, the signs in the left-hand column of the table— every one of which contains the numeral “1” in its mnemonic name—do not all contain a common handshape or even movement or location. PRO-1 and PRO-PL-1 are made with a single extended index finger, but none of the others have that handshape. All those in the third row of the column have a circular movement, but none of 17240-SLS9.2 1/16/09 9:56 AM Page 196 196 | Sign Language Studie s those in rows one, two, and four have that movement. All the signs in the first and fourth rows of that column are made on the chest, but those in rows two and three of the same column are made in neutral space. 9.1.1. {PRO-MULT}{THREE} It is true that our ability to discern possible iconicity or meaningful subparts of these signs does not prove that signers make use of them in ongoing signing, but there are arguments and evidence which suggest that they do. Consider once again the alleged morphemes {PROMULT-1} and {PRO-MULT}. They differ in that {PRO-MULT-1} consists of a circular movement made near the self, while in {PRO-MULT} the circular movement is held outward from the self. Despite this seeming iconicity, the notation entails that this difference in place of articulation is no more meaningful in itself than the difference that distinguishes /k/ and /t/. But consider the following thought experiment. Imagine that I am standing very close to a large plate-glass window, addressing two other individuals in the room, each one standing six feet from me and six feet from the other person. I want to describe a contrast of some sort between the three of us in the room and another group of three individuals on the other side of the window some distance away from us (much farther than six feet). Given how close I am standing to the window, and the distance between me and my addressees, I will execute the sign {PRO-MULT}{THREE} closer to my body than {PRO-MULT-1}{THREE}. At the same time, since I am prevented by the window from executing {PRO-MULT}{THREE} closer to the other group, I might supplement this sign by leaning my body away from the window to show that group’s distance from me in that way and/or by craning my neck or squinting as I look toward them. The point is that not even the iconicity of positioning the sign near the self or far from it is crucial. An addressee’s understanding of our pointing—to two different groups in this case—is best seen as the comprehension of an intentional act that we perform by whatever means we may have available, and not as wedded to a particular physical configuration. We will choose the simpler and more conventionalized means of indicating a group’s greater distance by positioning the 17240-SLS9.2 1/16/09 9:56 AM Page 197 Does ASL Really Have Just Two Grammatical Persons? | 197 sign outward from us if we can, but if we cannot, we will adopt other means to the same end.6 9.1.2.“Yu two me two” Another example of underrated iconicity can be seen in PRO-DUAL-1 and PRO-DUAL. Representing their names as hyphenated expressions implies that they do not incorporate number two in the way that {PROMULT} {THREE} incorporates {THREE}. And this is clearly what Liddell intends since he says (2003, 27) that what looks like the number two in this sign is in fact made with the K handshape (two fingers extended with the thumb contacting the middle finger), while TWO is made with the V handshape (two fingers extended but no contact with the thumb). This leads him to suggest that PRO-DUAL-1 and PRO-DUAL may have incorporated {TWO} historically but no longer contain it as an analyzable component. However, in the case of PRO-DUAL-1, we have empirical evidence from Vincent’s childhood speech that {TWO} is still a living part of this sign in the minds of signers. In the example from his speech in example 8 he breaks this sign into its meaningful components: Example 8. Yu two me two. 67.2357 (age 4;2) Asking that his younger brother Toby join him where he is; I have glossed the first word as yu (= “he”) on the assumption that he is addressing me since I would have to lift Toby to put him up there, but we cannot be sure that he was not addressing Toby. It might be objected that what I gloss as two might actually be the word too since “you too” and “me too” are common expressions in 6. This recalls an experiment conducted long ago by D. A. MacFarlane (1930), who was a student of psychologist Edward Tolman, to counter the belief of fellow psychologist Clark Hull that learning consists of connections between stimuli and physical responses. MacFarlane had his rats learn their way through a maze to find food by swimming (rats are natural swimmers). Then, when a raised floor in subsequent trials forced them to wade instead—entirely different physical movements— the rats had no trouble finding their way to the food right away. I am arguing that signers’ understanding of ASL pronouns is no less insightful. 17240-SLS9.2 1/16/09 9:56 AM Page 198 198 | Sign Language Studie s English. But example 9 seems immune to this objection. If we do not recognize an intonation break at the parenthesized comma, it combines two (or too) with Toby to form a unit quite distinct from any he would have heard in English. My own inclination is to perceive a break between the two words, though I suspect most others who heard it would not; I also perceive a break between the two signs in adult Vincent’s ASL rendition of the sentence (viewable at www.vincentasl.info). On this interpretation, the word two would stand alone. Nonetheless, while the word two could be used by itself with this meaning in English, the word too could not. In addition, when I asked adult Vincent to render the sentence in ASL (without mentioning the too vs. two issue), he unhesitatingly translated it using the PRO-DUAL–1 sign Example 9. Two(,) Toby, okay? 80.466B (age 4;6) Having asked me to buy him a certain toy, he is now asking that I buy one for his brother, too. I conclude that the sound /tu/ in example 8 does mean two, hence that PRO-DUAL-1 contains {TWO} as an analyzable component. The wording of example 8, together with its rather rapid pronunciation (audible at www.vincentasl.info), suggests that the iconic component of pointing is part of the sign as well—not just that the axis along which it moves points to the other person in the pair (which Liddell does regard as pointing) but also the fact that the handshape rocks back and forth in such a way that it points to the second- or third-person individual being referred to as it rocks outward from the signer and points to the signer himself as it rocks back inward. In short, the sign might be better represented in Liddell’s notation system as {YOU} {TWO} {ME} and {3P} {TWO} {ME}. Its iconicity may go even further if the back-and-forth motion defines the boundaries of the group in the same way as the circular motion does for TEAM and AGENCY, and for the sign component MULT. In any case, according to adult Vincent, PRO-DUAL–1 is not always performed with a K; sometimes a 2 hand is used instead. And there is a simple reason why the 2 handshape tends to become a K in this sign: 17240-SLS9.2 1/16/09 9:56 AM Page 199 Does ASL Really Have Just Two Grammatical Persons? | 199 It is easier to rock the hand back in forth if the thumb extends along with the index and middle fingers. Try it both ways, and if your arm and hand are built like mine, you should feel the difference in how your muscles are stretched. 9.2. Deictics for Distance In addition to the pronouns just described, Liddell (2003, 176–78) presents an analysis of THERE, BEHIND-THERE, FAR-THERE, and WAYOVER-THERE as four separate and distinct morphemes. To my eye these signs reek with iconicity. To show this, in the following three sub-sections I present evidence from Vincent’s childhood signs that some of the components that Liddell presents as parts of these individual morphemes—in the same way that “c,” “a,” and “t” are parts of the word “cat”—can be substituted for similar ones, or added to, or suppressed entirely without resulting in an entirely different sign. It is as though “cad” could be used to refer to a type of cat (BEHINDTHERE), “cate” could refer to one as well (FAR-THERE), and so could “ca” (WAY-OVER-THERE). 9.2.1. BEHIND-THERE In Liddell’s example of BEHIND-THERE, the signer is pointing to a location behind a plastic tub just in front of her and “moves her hand outward then flexes her wrist so that at the completion of the sign her index finger points through the plastic tub toward a location behind it” (ibid., 177). But instead of seeing this sign as pointing through the tub at the completion of the sign, why not see it as depicting a curved trajectory for arriving at that location? In Vincent’s childhood performance of what I identify as essentially the same sign (viewable at www.vincentasl.info), he indicates the location of a tire around the corner of the house and shows the curved trajectory by a sweeping movement of his arm instead of his wrist (analogous to substituting “d” for “t” in “cat”) since the location behind the corner of the house is farther away than something behind a much smaller tub. This strongly suggests the presence of meaningful iconicity that can be flexibly varied to accord with particular conditions—as with our {PRO- 17240-SLS9.2 1/16/09 9:56 AM Page 200 200 | Sign Language Studie s MULT-1} vs. {PRO-MULT} thought experiment and MacFarlane’s rats described in note 6. Furthermore, if we were to interpret the direction of his finger at the end of this movement as pointing, it would take us through the house to the front of it, not through the house across its corner to where the tire is. 9.2.2. FAR-THERE In Liddell’s description of FAR-THERE, the signer’s pointing hand begins with “a sharp backward movement, opposite to the direction the finger is pointing” (ibid., 177), while turning his head away from the direction he is about to point in, with “his eyes nearly closed.” I suggest that the backward movement represents “cocking” the hand before its launch, as one pulls one’s arm backward before throwing something, and that the signer’s squinting is iconic either of the difficulty in seeing something far away or of the effort it would take to go that far. Liddell notes that at the completion of the sign, the finger is pointing upward at an angle, not directly at the referent: “However, by translating the slightly elevated vector of the fingertip into a horizontal vector” (ibid.), that vector would lead to what he is pointing at. I suggest that this upward angle should be interpreted ballistically, as the higher initial angle that a trajectory needs in order to attain greater distance. This seems confirmed by the fact that, in Vincent’s performance of a sign that I equate to FAR-THERE (viewable at www.vincentasl.info), he has the same initial backward movement and squinting eyes, but he completes the arc-shaped trajectory, and does so several times (analogous to adding “e” to “cat”) to show that the location is very far away. 9.2.3. WAY-OVER-THERE Finally, in Liddell’s description of WAY-OVER-THERE (pictured at ibid., 178), the pointing finger moves at an upward angle, accompanied by an “oscillating twisting of the forearm” (ibid., 177) and a “repeated lateral tongue flap” (ibid.,178). It ends with the arm still angled upward but with “the index finger nearly horizontal or perhaps pointing slightly down from horizontal” (ibid., 177). Here, too, 17240-SLS9.2 1/16/09 9:56 AM Page 201 Does ASL Really Have Just Two Grammatical Persons? | 201 Liddell notes that “the directionality of the sign translates into a horizontal vector leading toward the faraway place even though the finger does not actually point at the place itself ” (ibid., 178). And here, too, I claim that the upward angle should be interpreted ballistically, while the slightly downward tilt of the finger represents the downward continuation of the trajectory. In Vincent’s variant of this sign (www.vincentasl.info), we have the same upward angle of the arm and downward tilt of the finger described by Liddell but neither the oscillating twisting of the forearm nor the lateral flapping of the tongue (analogous to removing the “t” from the end of “cat”). 9.2.4. Summary of Section 9.2 It seems likely that we are dealing here with a number of different distance-connoting devices that signers can mix and match. If so, then it is of course a little artificial to line up Vincent’s childhood signs side by side with those described by Liddell. I equated those which shared a backward movement, and those which both included an arm held stiffly outward with downward-tilting index finger. I also argued that there is an iconic equivalence between describing a trajectory by movement vs. by the angle and/or curved position in which arm and hand are held. We could go a little further and see the repeated movements of the arm and the widely open mouth in Vincent’s FAR-THERE sign as performing the same intensifying function in a broadly similar way as the oscillating forearm and lateral tongue flapping in Liddell’s WAY-OVER-THERE sign. If these various distance-connoting devices are free to apply themselves to whatever pointing gesture they choose, no doubt one could say something is really, really, really far away by applying them all at once. The parallels between Vincent’s signs and those described by Liddell may be a little clearer in the following table, where the gestures that are essentially identical or (I claim) iconically equivalent are listed side by side, except for those inside the rectangles—the semantically equivalent (and broadly iconically equivalent) gestural patterns that young Vincent and Liddell’s consultants did not happen to group with the same sets of other distance-connoting devices. 17240-SLS9.2 1/16/09 9:56 AM Page 202 202 | Sign Language Studie s BEHIND-THERE FAR-THERE WAY-OVER-THERE Liddell Vincent • • • • • • • • • • • curved arm (via flexed wrist) initial backward movement squinting eyes elevated vector (= beginning of a ballistic trajectory) • upward angle of arm • with finger tilted downward curved trajectory initial backward movement squinting eyes completed ballistic trajectory repeated arm movement + wide open mouth • upward angle of arm with • finger tilted downward oscillating forearm + lateral tongue flap I take it as further evidence of the iconic and gradient nature of these signs that Vincent did not find separate words to express their equivalents in his childhood speech. The degree to which some of the details of his versions of them match the distinctive components that Liddell describes shows that he had been exposed to the same cultural conventions as Liddell’s signers. However, if these signs were in effect four separate words for Vincent, we would expect him to have tried to connect them to four separate words in the speech he heard. But he did not. He not only failed to distinguish here from there, and this from that, but also failed to distinguish any of these from over-here and over-there, while treating the latter two words as equivalent to each other. 10. Morphemes and Automatization None of what I have said should be taken as implying that the “pronouns” discussed here are purely iconic to the exclusion of their also being partly apprehended as physical shapes with no intrinsic relation to their meaning. On the contrary, I am suggesting that these two ways of expressing meaning are copresent in varying degrees, often closely intertwined. Describing exactly how this operates is perhaps our greatest challenge in sign language research. One approach to this problem has been to acknowledge the arbitrariness of linguistic forms while 17240-SLS9.2 1/16/09 9:56 AM Page 203 Does ASL Really Have Just Two Grammatical Persons? | 203 holding out for their also being “motivated,” but that just names the problem and may even avoid it if it encourages us to relegate a form’s motivation to dead etymology. The forced choice implied in deciding whether a given form is a morpheme is even worse since, when a form is declared to be a morpheme, this entails that its internal complexity is not meaningful. A better way would be to enlist the concept of automatization, while reserving the term morpheme for a very late stage in that process. All our actions become smoother and freer of conscious attention when we practice them. This is certainly true of pointing, and probably truer of pointing toward the self than of pointing to others. When we point to ourselves there is less need to calculate just where to point; meanwhile, the addressee does not have to pay as much attention to imagining the trajectory of the pointing finger. Pointing to the self can therefore come closer to being a morpheme than pointing in other directions can. But even pointing at others can be more or less automated, and the same can be true of forms like {PROMULT}{THREE}. If the morpheme {THREE} itself can be automated— as surely it can if a signer has only to intend “three” and his fingers automatically go where they should—there is no reason why the same principle could not apply to {PRO-MULT}{THREE}, which we would then write as PRO-MULT-THREE. How do we know whether {PRO-MULT}{THREE} or PRO-MULTTHREE is the correct form? I suggest that the shades of gray we face in answering this question are essentially the same as those faced by dictionary makers, who perennially must decide which English collocations or compounds ought to be written with spaces, which with hyphens, and which with neither. It might appear that the situations are different since the ASL case relates to iconicity while the English case does not (except for the “diagrammatic” iconicity described by Peirce). However, I suggest that the general problem in describing how the two cases operate is the same. Nor is the problem of combining the iconic and morphemic aspects of ASL seamlessly in a common theoretical account necessarily more difficult; if it seems so, that is only because the situation in spoken language is more familiar. And at least there is another familiar situation that may not be so different in the relevant respect from the one in ASL: Many of the objects in the room around 17240-SLS9.2 1/16/09 9:56 AM Page 204 204 | Sign Language Studie s you are probably occluded in part by other objects in front of them. In some cases only a little sliver may protrude, but if you know (have automated your concept of ) the room and the objects in it, that will be enough to enable you to recognize an object’s full shape and its identity at the same time. The shape and the use (the meaning?) of each object inhabit our picture of the room at the same time, with either one of them predominating depending on our purposes at a given instant. It is true that automatization can reach a stage at which nothing is left but a physical shape (including, of course, a sound “shape”) that is merely associated with its meaning. Even those who know that lord derives from (the earlier pronunciation of ) loaf-warden (keeper of the bread) or that stirrup derives from stair-rope make no use of that knowledge in their day-to-day use of those words. But there is every position on the continuum below the extreme end at which the true morpheme resides. Farther down the continuum, everyone understands that pencil sharpener is composed of two distinct words, but the phrase is no doubt sufficiently automated that one might carry out a request to find one by seeking an object that looks like the image the phrase calls to mind without actually resolving it into two words. But when one does not see an object that matches that image close by, an immediate analysis of the phrase could lead to a search for a different sort of pencil sharpener. Likewise, performing the {PRO-MULT} motion at a distance from oneself may be as automated as extending three fingers when intending to express “three,” while other ways of expressing distance from oneself might come to the fore only in a situation like my plate-glass window example. But they will come to the fore so rapidly that they must have been there all along in some sense, in no less a state of readiness than the two parts of pencil sharpener. 11. Second- vs.Third-Person Pronouns Revisited I hope that the examples of the preceding section have reminded us that our knowledge of a complex word or sign can be automated only in part, and more automated in some situations than in others. This means that it can exist as multiple variants in our minds at the same time. In fact, according to exemplar theory as discussed in section 6, multiple variants can exist in a single person’s mind even for a com- 17240-SLS9.2 1/16/09 9:56 AM Page 205 Does ASL Really Have Just Two Grammatical Persons? | 205 pletely unanalyzable spoken morpheme used in a recurring situation. When we allow for multiple situations, such variations are well known, particularly in the case of the pronunciation of morphemes as conditioned by their phonetic environments. It would be a mistake to think that this always takes place through “rules” such as phonetic assimilation operating upon an underlying form in real time. We know that once phonological processes are operating, their output becomes part of the stored representation they had operated upon; for an English example, compare wife and wives, where the voicing in /v/ arose by assimilation to the following vowel, which is still written but no longer pronounced. From other examples cited by Bybee (2001, and particularly Bybee 1999, 219), we also know that changes to the stored representation can occur before the conditioning environment is lost. With reason, this is typically referred to as “morphologization.” Similar considerations should apply to an alternation like that between the “midline of the body” variant of the ASL second-person form emphasized by Berenz (2002) and what I called the “separable” variant seen in Meier’s example cited in example 6. Either of these can be automated in varying degrees, and the more prototypical “midline” variant seems particularly likely to have approached the status of a morpheme. In one of Vincent’s performances of it (viewable at www.vincentasl.info), it is so rapid that it is hard to see how there could be time even for a highly automated routine in the mind of the addressee to imagine following the trajectory of the pointing finger to identify the referent. Often it appears to be no more than an instantaneous “burst” occurring in his hand somewhere near the body’s midline, though of course this may be just my own shortcoming as a nonnative signer viewing video at 29.97 frames-per-second. A very short pause at the appropriate location appears to be another variant. The second-person examples at www.vincentasl.info even include one where the eye gaze focuses on the addressee but the pointing finger is off the midline and does not point directly to the addressee. That any one of these is only a variant of the second-person form is no more detrimental to its status as a morpheme than the occurrence of a synonym is detrimental to that status for any word you choose. More accurately, perhaps, rather than as distinct synonymous morphemes, each of them should be seen as occupying one among 17240-SLS9.2 1/16/09 9:56 AM Page 206 206 | Sign Language Studie s many positions in an overarching second-person pronoun schema that includes both specific physical configurations and pointing from various angles. Similar comments would apply to third-person forms. 12. Sign in Three-and-a-Half Persons? Berenz (2002, 205) points out that if Meier’s view is correct, it would mean that “sign languages differ in a crucial way” from spoken languages, which seem to have forms for all three persons. However, if the account in the preceding section is correct, it would mean that ASL (and other sign languages) do indeed have forms for all three persons at least when signer and addressee are not situated at an angle from each other. If we add my notion of the “separable” second-person form as discussed earlier in regard to sentence 6, we seem to have second-person morphemes for all situations. Nonetheless, let us take another look at this “separable” form. Its decoupling of pointing and eye gaze from each other highlights a situation that is already present in spoken languages, namely, that determining the addressee and the referent are different functions. We determine who is being addressed based on who the speaker is looking at, who is on the other end of the phone line, and even who the information in the sentence would be appropriate for. It is only given that knowledge that we can use a form like you to determine who is being referred to. However, the index finger viewed end on seems to provide sign languages with a form that does not depend on this indirect means of telling a conversational participant that he is being referred to. Seeing it through the eyes of conversational participants other than the signer, let us temporarily call this the ego form (= “he’s referring to me”). Shall we conclude that sign languages differ in a crucial way from spoken languages after all since the latter have only speaker, addressee, and other, while sign languages have signer, addressee, ego, and other? On the other hand, since the ego and addressee forms are often identical, and this characterization of the system applies only for participants other than the signer, should we compromise and call it a three-and-a-half-person system instead of one with four persons? That expresses it as a joke, of course, but the point is that even a grammat- 17240-SLS9.2 1/16/09 9:56 AM Page 207 Does ASL Really Have Just Two Grammatical Persons? | 207 ical category so seemingly basic as person may not be the same in languages as different from each other as English and ASL. I believe this illustrates a general truth about language, which I continue to stress in the following section. 13. Analogy vs. Identity I regard Meier’s account of ASL pronouns as an improvement over treating all ASL pronouns as on a par with each other. I see Liddell’s account as a further improvement in that it provides more detail and allows for the presence of both morphemic and iconic properties in the same pronominal sign at the same time: in a sign’s handshape and type of motion for the one and in the direction of its movement along a vector for the other. I would like to think that my own account represents yet a further improvement since I believe that the most accurate description of any language requires a lot of detail—detail that brings to light the differences that distinguish its constructions from analogous ones in other languages, while not losing sight of their similarities. For the relation between constructions or categories across languages is always one of analogy, never of identity, in my opinion. ASL-style pointing in particular should not be equated with any category in another language (such as pronouns), or indeed with any other behavior (such as pointing as performed by speakers of the ambient spoken language). I believe that the data reported here support that opinion since they show not only that ASL-style pointing is sufficiently analogous to pronouns in English to have led young Vincent to try to equate them, but also that it has enough in common with an invitation to look or see to have led Vincent and his brother to use pointing to translate the English words look and see into ASL as adults, and to have led young Vincent to grammaticalize the word see into a pronoun, assuming that is what occurred. References Ahlgren, I. 1990. Deictic Pronouns in Swedish and Swedish Sign Language. In Theoretical Issues in Sign Language Research. Vol. 1, Linguistics, ed. S. Fischer and P. Siple, 167–74. Chicago: University of Chicago Press. 17240-SLS9.2 1/16/09 9:56 AM Page 208 208 | Sign Language Studie s Baker, C., and C. Padden. 1978. Focusing on the Nonmanual Components of American Sign Language. In Understanding Language through Sign Language Research, ed. P. Siple, 24–57. New York: Academic Press. Berenz, N. 2002. Insights into Person Deixis. Sign Language and Linguistics 5: 201–27. Bialystok, E. 2001. Bilingualism in Development: Language, Literacy, and Cognition. New York: Cambridge University Press. Bloom, L. 1970. Language Development: Form and Function in Emerging Grammars. Cambridge, Mass.: MIT Press. Brown, R. 1973. A First Language. Cambridge, Mass.: Harvard University Press. Bybee, J. 1999. Usage-based Phonology. In Functionalism and Formalism in Linguistics. Vol. 1, General Papers, ed. M. Darnell, E. Moarvcsik, F. Newmeyer, M. Noonan, and K. Wheatley, 211–42. Philadelphia: John Benjamins. ———. 2001. Phonology and Language Use. New York: Cambridge University Press. Clark, E. 1978. Awareness of Language: Some Evidence from What Children Say and Do. In The Child’s Conception of Language, ed. R. J. A. Sinclair and W. Levelt, 17–43. Berlin: Springer. ———, and C. J. Sengul. 1978. Strategies in the Acquisition of Deixis. Journal of Child Language 5: 457–75. Costello, E. 1998. American Sign Language Dictionary. New York: Random House. Edge, V., and L. Hermann. 1977. Verbs and the Determination of Subject in American Sign Language. In On the Other Hand: New Perspectives on American Sign Language, ed. L. Friedman, 137–79. New York: Academic Press. Ellis, N. 2003. Constructions, Chunking, and Connectionism. In The Handbook of Second Language Acquisition, ed. C. Doughty and M. Long, 63–103. Malden, Mass.: Blackwell. Fischer, S. 1975. Influences on Word Order Change in American Sign Language. In Word Order and Word Order Change, ed. C. N. Li, 1–25. Austin: University of Texas Press. ———, and B. Gough. 1978. Verbs in American Sign Language. Sign Language Studies 18: 17–48. Friedman, L. 1975. Space, Time, and Person Reference in American Sign Language. Language 51: 940–61. Givón, T. 1979. On Understanding Grammar. New York: Academic Press. Heit, E., and L. Barsalou. 1996. The Instantiation Principle in Natural Categories. Memory 4: 413–51. Johnson, K. 1997. Speech Perception without Speaker Normalization. In Talker Variability in Speech Processing, ed. K. Johnson and J. W. Mullennix, 145–166. San Diego: Academic Press. 17240-SLS9.2 1/16/09 9:56 AM Page 209 Does ASL Really Have Just Two Grammatical Persons? | 209 Kegl, J. 1976. Clitics in American Sign Language. In Syntax and Semantics. Vol. 19, The Syntax of Pronominal Clitics, ed. H. Borer, 285–309. New York: Academic Press. ———. 1987. Coreference Relations in American Sign Language. In Studies in the Acquisition of Anaphora, ed. B. Lust, 135–70. Boston: Reidel. ———. 2003 [1976]. Pronominalization in American Sign Language. Sign Language and Linguistics 6: 245–65. Labarre, W. 1947. The Cultural Basis of Emotions and Gestures. Journal of Personality 16: 49–68. Lacy, R. 2003 [1974]. Putting Some of the Syntax Back into Semantics. Sign Language and Linguistics 6: 211–43. Lakoff, G. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press. Liddell, S. 2003. Grammar, Gesture, and Meaning in American Sign Language. New York: Cambridge University Press. Lillo-Martin, D. 2003. Pronouns, Indices, and Agreement. Sign Language and Linguistics 6: 267–76. ———, and E. Klima. 1990. Pointing Out Differences: ASL Pronouns in Syntactic Theory. In Theoretical Issues in Sign Language Research. Vol. 1, Linguistics, ed. S. Fischer and P. Siple, 191–210. Chicago: University of Chicago Press. MacFarlane, D. A. 1930. The Role of Kinesthesis in Maze Learning. University of California Publications in Psychology 4: 277–305. MacWhinney, B. 2000. The CHILDES Project:Tools for Analyzing Talk, 3d ed. Mahwah, N.J.: Erlbaum. Meier, R. 1990. Person Deixis in American Sign Language. In Theoretical Issues in Sign Language Research. Vol. 1, Linguistics, ed. S. Fischer and P. Siple, 175–90. Chicago: University of Chicago Press. Miller, J. 1994. On the Internal Structure of Phonetic Categories: A Progress Report. Cognition 50: 271–85. Neidle, C., J. Kegl, D. MacLaughlin, B. Bahan, and R. G. Lee. 2000. The Syntax of American Sign Language. Cambridge, Mass.: MIT Press. Nosofsky, R. M. 1986. Attention, Similarity, and the Identificationcategorization Relationship. Journal of Experimental Psychology: General 115: 39–57. Nosofsky,R. M. 1992. Exemplars, Prototypes, and Similarity Rules. In Essays in honor of William K. Estes, ed. A. F. Healy and S. M. Kosslyn, 149–67. Hillsdale, NJ: Lawrence Erlbaum Associates, Odlin, T. 2003. Cross-linguistic Influence. In The Handbook of Second Language Acquisition, ed. C. Doughty and M. Long, 436–86. Malden, Mass.: Blackwell. Petitto, L. 1987. On the Autonomy of Language and Gesture: Evidence from the Acquisition of Personal Pronouns in American Sign Language. Cognition 27: 1–52. 17240-SLS9.2 1/16/09 9:56 AM Page 210 210 | Sign Language Studie s Pierrehumbert, J. 2001. Exemplar Dynamics: Word Frequency, Lenition, and Contrast. In Frequency and the Emergence of Linguistic Structure, ed. J. Bybee and P. Hopper, 137–57. Philadelphia: John Benjamins. Rosch, E. 1973. Natural Categories. Cognitive Psychology 4: 328–50. Sachs, J. 1983. Talking about There and Then: The Emergence of Displaced Reference in Parent-child Discourse. In Children’s Language, vol. 4, ed. K. E. Nelson, 1–28. Hillsdale, N.J.: Erlbaum. Slobin, D., N. Hoiting, M. Kuntze, R. Lindert, A. Weinberg, J. Pyers, M. Anthony, Y. Biederman, and H. Thumann. 2003. A Cognitive/Functional Perspective on the Acquisition of “Classifiers”. In Perspectives on Classifiers in Signed Languages, ed. K. Emmorey, 271–96. Mahwah, NJ: Lawrence Erlbaum. Slobin, D. In press. Relations between Paths of Motion and Paths of Vision: A Crosslinguistic and Developmental Exploration. In Routes to Language: Studies in Honor of Melissa Bowerman, ed. V. Gathercole. Mahwah, N.J.: Erlbaum. Sternberg, M. 1998. American Sign Language Dictionary, 3d ed. New York: HarperCollins. Wilkins, D. 2003. Why Pointing with the Index Finger Is Not a Universal in Sociocultural and Semiotic Terms. In Pointing:Where Language, Culture, and Cognition Meet, ed. S. Kita, 171–209. Mahwah, N.J.: Erlbaum. Wong Fillmore, L. 1976. The Second Time Around: Cognitive and Social Strategies in Second Language Acquisition. PhD diss., Stanford University.