Evidential-perceptual transfer by a blind speaker? Or: what do the Ladakhi markers for “visual” and “non-visual” perceptual experience, ḥdug and rag, actually encode?

1 Introduction

Evidentiality or the linguistic coding of information source (for this definition see Aikhenvald 2004: xi, 1, and passim, 2015: 239, 2018: 1) is a linguistic category that does not play an important role in most South Asian languages. It is, however, a prominent feature of the Tibetic languages, and among them of the various Ladakhi dialects, spoken in the Union Territory of Ladakh, India.

Ladakhi (locally known as Ladakse skat) is used here as a cover term for the Tibetic dialects spoken in Ladakh (with the exception of the varieties of the Exile Tibetan community and the Baltipa dialects, spoken at the western border to Pakistan). Until recently, Ladakh was one of the three provinces of the state Jammu & Kashmir, but on October 31 2019, it became a Union Territory of India. The Ladakhi dialects fall into three groups: the Balti–Purikpa group with the dialects of western Purik; the Shamskat group with the dialects of Eastern Purik, Sham (or Lower Ladakh), and Ldumra (a.k.a. Nubra); and the Kenhat group, with the dialects of Central Ladakh and Leh, Zanskar, Upper Indus, Lalok, and the border region to Tibet. The most obvious difference between the groups is that the dialects of the first two groups formally differentiate between an agent and a possessor, while the Kenhat dialects do not. The Balti–Purikpa group, however, shows a less developed evidential system.

The modern Tibetic languages are known for having a very particular complex verbal system of marking the knowledge base of a statement. At the risk of over-simplifying, the different knowledge bases can be described as

1. personal, namely

   1. intimate and/or authoritative knowledge of one’s personal or cultural sphere or through personal involvement or through acquaintance over a certain time

   2. knowledge through numerically limited sense perceptions

   3. knowledge based on inferences, reasoning, guessing etc.

2. non-personal, namely

   1. second-hand knowledge

   2. knowledge shared or to be shared with the addressee or the larger community, particularly about generic facts

These modes of knowledge are linked to the perspective of the speaker in statements and, by a natural perspective shift, the addressee in information-seeking questions. These two speech act-related perspectives will be covered by the term “main speech act participant” (msap). Persons, items, or more generally situations not under the control or responsibility of the msap will be referred to as other. These two terms do not refer to some weird syntactic person category, but to the epistemic authority and the stance of commitment the speaker is willing, or allowed, to take.

In most Modern Tibetic languages, speakers do not differentiate between different perceptual channels, but use the linking verb or auxiliary ḥdug (or the corresponding dialectal counterpart) for all sense perceptions (see among others, Garrett 2001; Sun 1993; Tournadre 1996; and more recently the edited volume by Gawne and Hill 2017). In the Shamskat and Kenhat Ladakhi dialects, however, speakers differentiate between visual perception, using the auxiliary ḥdug or, in the Ldumra dialects, snaṅ, and sense perception through other channels, using the auxiliary rag. This is at least the standard view (cf. Koshal 1979; Bielmeier 2000, followed until recently by Zeisler, e.g., Zeisler 2012a).

The markers expressing intimate and/or authoritative knowledge of the msap, often covered by the term “egophoric”, will be subsumed under the neutral term Set 1; those indicating knowledge through immediate sense perceptions will be termed Set 2. Both sets consist mainly of linking verbs. Set 1 contains the attributive and identifying copula yin and the existential linking verb yod. Set 2 contains the existential linking verbs ḥdug, and rag, and in a few Kenhat dialects also the bleached verb soṅ lit. ‘went’, plus in even fewer Kenhat dialects the bleached verb byuṅ ‘appeared’.

There are further various ‘evaluative’ markers (em) for inferences, various types of presumptions, mental distance, as well as for shared or shareable knowledge, varying from region to region, plus a marker for quotation or hearsay, which scopes over all other markers. That is, the quotation retains the (perceived or remembered) original marker of the reported speech, deixis, and illocutionary force as in direct speech, while pronouns are shifted as in indirect speech.

A first detailed description of the Leh dialect including the evidential and epistemic markers, albeit not without substantial errors, is found in Koshal (1979). Bielmeier (2000) gives a first overview of the evidential usage in identificatory, attributive and existential constructions. More details are given in an extended handout available online (Zeisler 2012a) and in an exemplary questionnaire available online (Zeisler 2016). More specialised questions are treated in Zeisler (2017, 2018a, 2018b, to appear). For the Western Purikpa dialect of Kargil(o) and its less fine-grained evidential system see Zemp (2017, 2018).

The neutral distribution of these markers as linking verbs and auxiliaries is shown in Table 1.

What is presented in Table 1 is not at all a strict paradigm, but the prototypical background against which pragmatic variation can occur. With few exceptions, the forms listed under other can be used for the msap, when the speaker rejects his or her active involvement and responsibility for the situation or may distance him-/herself from the situation for other reasons. The forms listed under msap can be used for the other, when the speaker claims particular epistemic rights through acquaintance or through active involvement and/or responsibility. Such usages, however, are highly marked. See particularly Häsler (2001) for the notion of stronger or weaker empathy.

For Ladakhi Shamskat and Kenhat speakers, the visual channel normally overrides all other senses, that is, a previously visually observed activity triggers ḥdug or snaṅ, even if the situation is momentarily out of view, but still accessible, as in (1), while yod is used under the condition of previous involvement when the observed situation is no longer accessible. rag is used for a, typically immediate and individual, non-visual sense perception, only when vision is excluded or insufficient. Accordingly, when hearing a cat purring while seeing it, ḥdug or snaṅ is used. Only when the cat is out of view, behind one’s back or outside the room, and thus invisible, rag is used, see (2) further below.

When reanalysing (1a), which was formulated for another research question, I was quite surprised, because when the informant formulated the sentence, the workers were outside my room, out of view, and we had heard the noise caused by their work. The reason for the use of ḥdug is, to various informants, that the speaker had seen the workers already in action when she came into the house, and so the comparatively fresh visual input dominates over the auditory channel, cf. also (1b). If the workers had started to make the noise after the informant had come into my room, she would probably have used the auxiliary for auditory perception rag. However, if she had been talking about the work in her own house, she would have used the non-experiential auxiliary yod, indicating that she had left an ongoing situation of her sphere without observing the end.

2 A blind speaker’s choice

How would a blind person make the distinction? Would s/he never use the auxiliary ḥdug? Or would s/he perhaps “feel” some perceptions like visual ones, just because they are also processed in the visual cortex, due to brain plasticity (see also Section 4)? Would s/he perhaps simply copy the linguistic behaviour of others, transferring non-visual perceptions into the domain of ḥdug? Would s/he reinterpret the auxiliaries in some way or another and adapt them to her perceptions? Or could his or her usage point to something that is hidden behind the opposition of ḥdug and rag? As I have been able to observe in the case of the speaker portrayed below, most of her choices correspond to that of non-handicapped speakers, that is, in most contexts where all other speakers use the “visual” marker, she would likewise use ḥdug, and in most contexts where all other speakers use the “non-visual” marker, she would likewise use rag, see Section 2.2. However, there are also significant deviations, as shown in Section 2.3. From a superficial perspective, her choices appear to be idiosyncratic. At a closer look, her choices turn out to be meaningful, and they hint at a more subtle distinction underlying the opposition between the two markers. This distinction shows up infrequently also among individual sighted speakers in a few contexts, see Section 3.1. It may, at least partly, also underlie the general preference for the “visual” marker when describing habitual experiences of a non-visual nature, see Section 3.2.

3 Comparative data from other speakers

The cases where the blind speaker differs from non-handicapped speakers indicate that she does not simply copy the speech behaviour of others without understanding its implications. Her choices make perfect sense: she uses rag for endopathic perceptions like other speakers (6b, e), but she uses rag also for the perception of a mountain or creeks, which she experienced more indirectly, where non-handicapped speakers would use ḥdug (or snaṅ), cf. (6a, d) with (7). On the other hand, she uses ḥdug for more immediate perceptions through a feeling of closeness, often through some kind of active or passive involvement, that is, through touching or because of a movement towards her (5), (6c). When I explained these usages to informants who had judged the Sharapa speaker’s linguistic behaviour negatively, they agreed that her use of ḥdug is well motivated and not a case of pretension.

What is less clear, is: a) are her choices triggered by the intensity of the activation of the visual cortex, that is, how intensely “seeing”-like her perceptions are, particularly her perceptions of space and of her movements in space, b) does she merely reinterpret the opposition signalled by the two auxiliaries according to the intensity of her perceptions, or c) do her choices perhaps reflect a more abstract and supramodal spatial conceptualisation, in terms of closeness or immediateness or intensity, lying behind the opposition of ḥdug and rag?

A sensual impression of “seeing” can likely be ruled out. The Sharapa speaker never claimed that she perceived the situation visually. In one case, she justified her use of ḥdug with the comment that she likes the person.

4 A brief note on neural plasticity

When I started thinking of writing about this topic, I had the vague impression that it was common knowledge that blind persons might talk about their environment as if they were seeing it. A google-based search revealed a lot of fascinating data, but nothing on that particular question. If the effect exists at all, it might be restricted to persons who grew up seeing and became blind later.

What is documented, however, is that due to brain plasticity, parts of the visual cortex are recruited for the processing of non-visual stimuli (Cattaneo et al. 2008). Cross-modality processing in the visual cortex is to a certain extent also observed in un-impaired persons, but it is unmasked or enhanced in the case of sensory deprivation (Merabet et al. 2004; Merabet and Pascual-Leone 2010). More crucially, parts of the visual cortex can then be involved in language processing, particularly semantics and sentence structure (Röder et al. 2002; Bedny et al. 2012).

Congenitally blind persons develop the same internal category-selective mapping of face-, body-, scene-, and object-related categories as seeing persons (Van den Hurk et al. 2017) and their reasoning about mental states based on vision is similar to that of seeing persons (Bedny et al. 2009; Koster-Hale et al. 2014). To a certain extent, this is due to the fact that “[b]lind children adapt to the language spoken in their surroundings” (Vinter et al. 2012: 6b). Language, therefore, “plays an important role in the construction of knowledge and can partially compensate for the absence of visual experience in blind children during conceptual development” (Vinter et al. 2012: 6b with further reference).

Under optimal conditions, a blind Ladakhi child would thus learn the correct use of the auxiliaries ḥdug and rag. If these two auxiliaries referred strictly to different perceptual channels, like the verbs of seeing and feeling, s/he would learn that ḥdug is to be used whenever vision is involved. The child would thus also learn that it would not be appropriate to use ḥdug for his or her own non-visual perceptions, in the same way as, say, a blind English child would learn that he or she cannot say I see the cat etc. If the caretakers, that is, the family members around the child, do not object to his or her use of ḥdug, this means that either they have adapted their own speech to the impairment of the child or that the two auxiliaries do not refer exclusively to different perceptual channels, but instead or also to different intensities of perceptions, so that the blind child’s use of ḥdug makes perfectly sense to the seeing person.

5 Conclusion

The connotations of greater or lesser objectivity associated with ḥdug and rag when used for habits, as shown in Section 3.2, fit well with the observation that in other contexts, ḥdug may indicate a somewhat more immediate, more intense, and thus also more certain knowledge than rag, as in the case of groping and looking for coins, see Section 3.1.

One may certainly object that the notions of greater certainty follow automatically from visual perceptions as compared to other perceptions. The experience of greater certainty would then also automatically lead to the connotation of immediateness (or perhaps also the other way round). However, neither ḥdug nor rag refer explicitly to sense perceptions, not to speak about different sense perceptions. ḥdug as a linking verb and auxiliary originated from a lexical meaning ‘stay, dwell, sit’. Elsewhere, I have argued that ḥdug and snaṅ started their evidential career as markers of uncertain knowledge, especially for predictions and inferences. I have further suggested that the form rag may have developed out of ḥdug as a marker of uncertain knowledge and may thus have inherited its notion of greater uncertainty or less immediate knowledge from an earlier stage of ḥdug (Zeisler 2017, 2018a).^[9]

The use of ḥdug by at least four non-handicapped speakers for clearly non-visual, but nevertheless relatively immediate and/or certain sense perceptions, as in the case of groping inside a pocket, indicates that the notion of immediateness associated with ḥdug is primary, and that ḥdug does not necessarily encode a particular perceptual channel. When showing these examples to speakers who may not use ḥdug in this context and when further explaining to them how the Sharapa speaker interpreted her own use of ḥdug, most of them agreed that such usages are meaningful, and that the use of ḥdug and rag may have something to do with the immediateness of the perception.

While the Sharapa speaker never claimed that she was seeing things or persons, her use of ḥdug is also not an implicit pretence that she is seeing. It is further most likely that her particular use of the two auxiliaries is not a question of transfer or adaptation. Nor is she taking creative license. She is an ordinary pragmatic language user, not a creative poet who reflects her own language and strives for novelty. Furthermore, unlike in the idealised American individualism (see Johnstone 1996: 125–127, 182), there is no expectation for, even less so pressure on, Ladakhi speakers to express themselves as individuals.^[10] Quite to the contrary, the Sharapa speaker’s use of the “visual” marker points to her implicit, perhaps even unconscious, wish to be like all others around her, despite being criticised for exactly this attitude.

Therefore, I think that she could not have intentionally transgressed the framework of la langue, and that also means she exploits available (albeit usually ignored) connotations of the two markers in question, which are visible also in some idiosyncratic and some fairly common usages by other speakers. I do not want to suggest that every blind Ladakhi speaker would necessarily make the same choices as the Sharapa speaker. Her choices are individual, inasmuch as she was brought up with the license to use ḥdug in the way she does. As a result, she uses both auxiliaries, ḥdug and rag, “literally” as markers for immediate and less immediate sense perceptions.

The Sharapa speaker’s individual usage, as well as the comparative data from other, non-handicapped speakers, allows a reanalysis of ḥdug (or snaṅ) and rag as markers for immediate and less immediate sense perceptions rather than as markers for different perceptual channels. The association of ḥdug (or snaṅ) with the visual channel follows only secondarily from its association with the most immediate or most certain perception. The Sharapa speaker’s impairment turns out to be like a magnifying glass, unmasking a fine-grained distinction otherwise hidden beyond conventionalised preconceptions.

To conclude, I should like to add a remarkable comment that I received when I presented the above data at the workshop Hearing and Linguistics, Oldenburg 13–14 May 2022: Why insist on a difference between visual and auditory or more generally non-visual perception of space and movement? Which is to say that perception of space and movement is always multimodal (albeit for normal seeing people masked by conscious visual experiences). It is thus not so surprising that the Sharapa speaker, relying on what she hears or feels, and upon how she moves, discriminates between dynamic events that are moving towards her (stones falling) and static situations she literally happens to come across (rivulets, steep slopes), and again between this more indirect coming across and the more direct touching of pebbles, not to speak of actively handling the showcase items.

Abbreviations and conventions

“-”

segmentable

“=”

functional equivalence or morphological “sum”

ā

high tone (first syllables only)

a

neutral tone

a̱

low tone (first syllables only)

aes

aesthetive (experiencer subject marking)

all

allative

ass

authoritative-assertive

cnt

continuative morpheme

com

comitative

cond

conditional

cop

copula

ctr

controlled action

df

definiteness marker

em

evaluative marker

emp

emphatic

erg

ergative

excl

exclusive

exp

experiential (visual and non-visual)

foc

focus marker

gen

genitive

grd

gerundive

hab

habitual

hon

honorific

hum

humilific

intj

interjection

ints

intensifier

lb

lhag.bcas morpheme for clause chaining

lq

limiting quantifier (‘a’, ‘some’)

msap

main speech act participant

nctr

non-controlled event or state

ng

negation marker

nml

nominaliser

nvis

non-visual

pl

plural

prf

perfect

prs

present

pst

past

ppos

postposition

rm

remoteness marker (shifts situations further into the past and marks assertive past)

qm

(sentence) question marker

vis

visual