Contrast instead of comparison: Evidence from West Tibetan differentiating property ascriptions

 

Bettina Zeisler

Universität Tübingen

 

I think that one reason we fail to notice, when we do field research, the fundamental differences between languages is because linguistic theory over the last 50 years—maybe even longer—has been primarily directed towards understanding how languages are alike, as opposed to how they are different. Dan Everett[1]

 

Non-equative comparisons are typically interpreted in terms of degree semantics. That is, the comparee is thought to have the same property as the standard, but to a different degree. In this paper I should like to introduce a different way of conceptualising differences, namely categorical contrasting, where one focuses more on the contrast than on the gradualness of the difference. Two items are described as being essentially different with respect to a certain property, and this can imply that the standard against which an item is contrasted lacks the property in question. In order to show that this approach is more suitable for the Tibetic languages, especially the West Tibetan varieties spoken in Ladakh, I will not only discuss the standard ways of expressing differences, but also some more marginal constructions at the limit of acceptability.

 

1. General remarks[2]

 

Non-equative comparisons are typically interpreted in terms of scalar or degree semantics. That is, the comparee is thought to have the same property as the standard, but to a higher (or lesser) degree. In the Standard European languages, one would usually say something like Peter is rich, but Mary is richer, rather than Peter is poor, but/and Mary is richer. Dixon (2008: 787, 2012: 341) speaks of “the prototypical scheme in which two participants are compared in terms of the degree of some gradable property associated with them” (emphasis added). A similar position is held by Stassen (2013): “In semantic or cognitive terms, comparison can be defined as a mental act by which two objects are assigned a position on a predicative scale. If the positions on the scale are different, then we speak of the comparison of inequality, which finds its linguistic encoding in comparative constructions”.

 

This approach does not take into account that ascribing a property to an item already implies some kind of comparison or contrast with an implicit standard, namely of what is not worth mentioning because it is average or expected (Andersen 1983: 100, Beck 2006; see also example (1) for Ladakhi, as well as Hahn 1996 or any other edition, Lektion 12.3f for Classical Tibetan). A notion of degree is thus already involved in most neutral property ascriptions, except perhaps in those languages where property ascriptions are “norm-related” (for this notion see Bochnak & Bogal-Allbritten 2015: 118-123).

 

On the other hand, scalar semantics are not applicable to all usages of comparative constructions in the Standard European languages, as when one says in German Paß nächstes Mal besser auf! ‘Take better care next time’. In this case, the person in question is typically not thought of having taken bad care or good care to a lesser degree, but of not having taken care at all (cf. example (66) and the subsequent discussion). If the property is absent in the standard, it cannot be shared by the comparee, and there is also no degree that could be lessened or heightened. Or to say it differently, ordinary speakers are not mathematicians operating with zero and negative values. Thus, when saying in French il est très mauvais, mais sa sœur est plus gentille ‘he is very bad, but his sister is nicer’ (something that may not be possible in all languages), it seems to be farfetched to suppose that the speakers conceive of the two properties ‘bad’ and ‘nice’ as being different degrees on a scale, with minus and plus values, rather than being categorically opposite.

 

Scalar semantics, furthermore, do not seem to be universally applicable to all languages. Many languages use what has been described as a conjoined strategy, juxtaposing opposite values, see also examples (22) and (23) below for Modern Hybrid Literary Tibetan and Ladakhi. Why should such contrasting necessarily imply a scale and a notion of degrees? Just because we cannot help to translate such expressions with a comparative construction in English or any other Standard European language, and just because philosophers of language and formal linguists have decided on the base of Standard European languages that there is always a scale implied? Do human beings in all cultures really always use a mental scale when confronted with two items that are quite apparently different in size, beauty, or quality? Do languages without explicitly encoded scalarity really “lack” something, and thus have to resort to some kind of “comparative strategies” (as suggested by Dixon 2008: 790, 802, 2012: 342, 359), or do they perhaps reflect a different kind of conceptualisation? One possible alternative way of conceiving of differences is what I should like to term here “categorical contrast”.

 

As a cover term for both, scalar comparing and categorical contrasting, one might speak neutrally of “differentiating property ascriptions”. I shall argue that both strategies are not fundamentally opposite, but share common features and an area of overlap, that is, a form implicating scalar comparison may well be used to express a categorical contrast (as in the case of take better care next time) and a construction implicating categorical contrasting can be applied to situations where the difference is measured (see also section 7). Both, scalar comparing and categorical contrasting involve a relation between a standard (S), against which a difference is measured or a contrast is established, and a comparee or contrastee (C) for which a property, also called parameter (P), is predicated. The relation itself may be signalled with a relational marker (M), which may or may not be specific for the comparison or contrast.

 

When contrasting two items with respect to a particular property, e.g. when saying A is beautiful but B is not, A is beautiful rather than B, or A is beautiful in contrast to B, it is positively stated that the contrastee has the property in question, but nothing is said about the standard. It is simply left open whether the standard shares the property, but to a lesser degree, whether it does not have that property at all, or whether it has an opposite property (e.g. being ugly). One focuses more on the difference itself, rather than quantifying it, much in the way as different colours or shapes are perceived. One would not normally say that the green is bluer than the red, even if the difference in wavelength is scalar (and even though one might say that a particular green has more of a blueish shade than another green). One might also say that contrastive constructions aim at differences only between individuals.

 

Contrasting two items as being different, however, does not necessarily imply that the property of the contrastee is absolute, and so it does also not necessarily preclude that the property of the contrastee is only a relative one, and that it might be given only in relation or contrast to the standard. This is at least true for the Tibetic languages, where the properties remain relative properties, see also the discussion of the conjoined construction in section 5.1.

 

Categorical contrasting may be understood as the true opposite of equative comparison. While the latter describes two items as equal with respect to a certain property, the latter describes them as unequal. Non-equative comparison, of course, shares the notion of difference with categorical contrasting, but it also shares with equative comparison the notion that the items compared share the same property, although to a different degree. One might thus say that equative comparison and categorical contrasting are the extreme ends of a continuum with non-equative contrasting somewhere in between. But I would rather think that categorical contrasting and non-equative comparison are different ways of perspectivising differences. Fig. 1 is an attempt to visualise the relation between simple property ascriptions and differentiating property ascriptions, on the one hand, and the relation between contrasting and comparing on the other.

 

Figure 1. Contrasting and comparing

 

Although they do not use juxtaposition as their main “comparative strategy”, the large family of Tibetic languages or at least some of its members challenge the general Eurocentristic concept of grade semantics. If my understanding of what happens in the Tibetic languages is correct, it may turn out that speakers of other languages with no “dedicated” comparative constructions may similarly conceptualise differences not so much in terms of degrees, but in terms of a categorical contrast or simply as an indefinable relation of difference.

 

2. Background information

 

The Tibetic languages[3] are counted among the Tibeto-Burman or Sino-Tibetan languages (perhaps a convergent rather than a genetically related divergent group). The written language is attested since the mid 7^th century (Old Tibetan until the end of the 10^th century, Middle or Classical Tibetan since the early 11^th century).  

 

The Western Tibetan languages are spoken from Baltistan (in Pakistan) along the Himalayan range up to Western Tibet (in China). Ladakh is part of the Indian state Jammu & Kashmir. The Ladakhi dialects fall into two main groups, the Shamskat (or “Lower Language”) dialects, spoken in the north-western or lower part of Ladakh (Sham, Ldumra, a.k.a. Nubra, and Purik) and in Baltistan and in Balti enclaves in Ladakh, and the Kenhat (or “Upper Language”) dialects, spoken in the upper or south-eastern part (Leh, Upper Indus, Zanskar, Lalok, and the Changthang dialects of the Nyoma Block). Shamskat is represented in this paper by the dialects of Sham: Domkhar, Khalatse, Skindiang, and Teya, by the Ciktan dialect of Purik and the Turtuk dialect of Balti. Kenhat is represented by Gya-Mīru from the Upper Indus area and Shachukul from Lalok. See Fig. 2.

 

Figure 2. Map of Ladakh and her dialects

(map designed by Adella Edwards, approximate location of places by author)

 

The Kenhat group is closely related to the West Tibetan varieties spoken in Himachal Pradesh and Uttarakand (of India) and in parts of Western Tibet. The dialects spoken by the nomads in the Nyoma Block close to the Chinese border have only recently been established as belonging to the Kenhat group, but cannot be considered here. The two groups do not only differ with respect to their phonology, but also with respect to their grammar. The most notable difference between the two groups is that the Shamskat dialects differentiate between an actor and a possessor, while the Kenhat dialects do not. There are also minor differences with respect to their morphology (see Zeisler 2011), as in the case of the relational marker.

 

The Ladakhi dialects are under pressure from two sides. On the one hand, the state language is Urdu, while the medium of instruction is English. Furthermore, English is (still) the dominant lingua franca in all Indian media. The impact of these languages is not only reflected in a host of loanwords for all modern items, but to some extent also in syntactic borrowings and changes. On the other hand, Buddhist scholars insist that the Tibetan script was invented for the holy books, and the orthography, therefore, cannot be modified to write the local language. Ladakhi is thus barely written and appears to be threatened in the long run.

 

2.1. The data

 

The Ladakhi data presented here is based on more than 50 months or nearly two decades of field work.[4] Many of the examples are taken from recorded non-elicited speech, narrations, personal narratives, and monologues on various issues (more than 20h of transcribed recordings). Other examples, especially those in section 9, have been elicited in 2007 on behalf of the partner project SFB 441, Sigrid Beck, Comparative Constructions.[5] In this connection, I collected about 250 examples from various dialect speakers for simple property ascriptions, equations, and differentiating property ascriptions. Some of the examples have also been elicited undesignedly in the context of my work for a Valency Dictionary of Ladakhi Verbs. The latter contains about 180 contrastive constructions among the more than 25,000 example sentences.[6] None of the elicited examples has been recorded. The elicitation language is usually English, but I also often formulate or reformulate examples on my own in Ladakhi and let them be judged by the informants (see also Zeisler 2016). Except for the occasional drawing, I do not use any particular stimuli. The elicited examples will be marked here by the abbreviation FD for field data and the year of elicitation, the recorded examples are provided with a title or a content description and the year of recording.

 

3. Differentiating property ascriptions in Tibetic languages - the formal side

 

3.1. Adjectives, adjectivals, and the alleged “degree” marker

 

Descriptions of Tibetic varieties often talk about the positive, comparative, and superlative “degree” of adjectives (e.g. Denwood 1999: 179, 181 for Classical Tibetan; Tournadre & Sangda Dorje 1998: 201, 233 for Standard Spoken Tibetan; Häsler 1999: 118 for Dege (also known as Derge or Sde.dge); Haller 2000: 55 for Shigatse; Huber 2005: 78 for Kyirong). However, the word class of (nominal) adjectives is typically derived from adjectivals, and it does not regularly take part in non-equative comparison. The original verbal character of adjectival roots is evident from several facts. In Old Tibetan, they could take two stem forms like other inchoative-resulting state verbs, e.g. che, ches ‘be, get big’ or maŋ, maŋs ‘be, become much, many’.[7] In Old Tibetan and in Classical Tibetan, they appear in the verbal slot (the last position in a clause) and may take several non-finite markers, cf. (3), as well as the verbal proclitic negation markers mi and ma. In some modern Tibetic languages, adjectival stems can still take the proclitic negation markers mi and ma (Hu Tan 1989: 406f). In Ladakhi, full verbal usage is attested, particularly in contexts that imply a development or a difference between two items, but it seems to be in the process of becoming obsolete.

 

Nominal attributive adjectives are derived from monosyllabic verbal roots in several ways, most often by the non-productive nominalisers {‑po}, ‑mo ,[8] and ‑ma, frequently also by the productive nominaliser {‑pa}, yielding verbal nouns, e.g. gjokspa ‘fast, quick’ in example (27). Shamskat rgyalba, other dialects gjal(l)a ‘good’. They can be derived also by other means, such as reduplication or by adding the derivational suffix ‑can ‘having’ or the negated verb med ‘not exist, not have’. In a few cases, an archaic derivational morpheme -d/-n is inserted between verb stem and nominaliser, e.g. Old Tibetan che, ches ‘be, get big’ vs. che-d-po or che-n-mo ‘big’.

 

What is usually counted as the “comparative degree” or “comparative form” of the adjective is a nominalised form of the adjectival, e.g. Written Tibetan che-ba ‘big-ing, being, getting big’ or ‘the big-ing one’.[9] Combined with the allative marker ‑la, this form appears in exclamatives, such as Written Tibetan che-ba-la or Ladakhi ʧhe-a-la ‘how big!’ or rather ‘[Look] at that big one!’. Such exclamatives can also appear with nouns[10] and in Ladakhi, also with verbs.[11]

 

The exclamative usage shows that the nominaliser {-pa}, which appears with the adjectival stem in differentiating property ascriptions, has no inherent degree semantics comparable to the English and German degree marker -er.

 

In Ladakhi, complex derived adjectives, such as ɲalbaʧan ‘poor’, appear unmodified in the contrastive construction, and one can also observe a tendency to use the simple adjectives, such as ʧhenmo ‘big’, unmodified, cf. examples (4), (5), (47), (52), (53), and (62). Most probably this is due to the influence of the neighbouring Indoaryan languages, especially Hindi and Urdu, where the adjectives remain unmodified.

 

If a (nominal) adjective is used for simple property ascriptions, it is followed by an auxiliary. The predication with the verbal noun (or a derived adjective) in differentiating property ascriptions follows the same rules. That is, the verbal noun (or the derived adjective) is followed by an (evidential) auxiliary, cf. (1) [12] for a simple property ascription and (4) and (5) for differentiating property ascriptions.[13]

 

(1) Shamskat: Domkhar (FD 2007)

C	P	aux
ʦheriŋ	riŋbo	duk.
[name]	long	vis.be
‘Tshering is tall (lit. long; visual evidence).’ (As the informant explained: The person is tall not in relation to a specific person, but taller than the average.)

 

3.2. Marking the relation with respect to the standard

 

The modern Tibetic languages have developed different ways to indicate a contrastive relation. In many Tibetic varieties, the standard is followed either by an ablative marker corresponding to Old and Classical Tibetan ‑nas or ‑las, or a related morpheme. The Old and Classical Tibetan ablative markers are derived from two locational markers ‑na and ‑la through a reduced form of an originally syllabic morpheme *‑se or *‑so, the same that derived the instrumental (and ergative) marker {‑kyis} from the genitive {‑kyi}.[14] In Old and Classical Tibetan, the standard is either marked with the ablative marker ‑las or, more commonly, with the morpheme {‑pas}, which seems to be analysable into a nominaliser {‑pa} and the same reduced ablative-instrumental -s element. This morpheme is also used to indicate causal relations between events. With respect to using a (kind of) ablative as relational marker, the Tibetic languages obviously follow a common “separative” strategy (cf. Stassen 2013).

 

The most prominent exceptions to the ablative strategy are Amdo with either a genitive marker (Hu Tan 1989: 404 for Zeku alias Tsekhog, Rtse.khog; Haller 2004: 54 for Themchen)[15] or with the verbal expression ɸtina ~ htina (bltas.na) ‘if one looked’ (Hu Tan 1989: 404 for Guide alias Thrika, Khri.ka; Haller 2004: 54 for Themchen; Sandman & Simon 2016: 112 more generally), Sherpa with sina, and the Kham dialect Chayu = Zayü (Dzayül, Rdza.yul) with jī’na (Hu Tan 1989: 404). The Sherpa form could be from zer.na ‘if one says’, the Chayu form perhaps from yin.na ‘if it is’. Cf. example (29) and (30) for a similar construction with yet two other verbs in Ladakhi. The Kham dialect Batang (Ḥbaḥ.thaŋ) seems to use the comitative marker daŋ (Hu Tan 1989: 404); other Kham dialects and Rutog (Ru.thog) in Ngari use the allative marker la[16] (Hu Tan 1989: 403; Causemann 1989: 69-70 for Nangchenpa; Häsler 1999: 118-119 for Dege (Sde.dge)).

 

Balti uses the morphemes ‑pa, (‑pa)-ʦe or ‑baʦek (Read 1934: 22, Grierson 1909: 27, 35) or, as in Turtuk, (‑a)-paʦa (own data).[17] In Purik, the morpheme is attested as ‑baʦik in the dialect of Kargil (Rangan 1979: 146f., Zemp 2013: 319), and as ‑p/batsek in Ciktan (own data). The latter form appears infrequently also in the Western Sham dialects, where the element ‑ʦek can be used in equative property ascriptions besides ‑ʦoks ‘like’ (cf. examples (2), (50), (56), and (57)). ‑ʦek is also found in the second part of relative clause constructions in compounds such as de-ʦek ‘that much’ or dena-ʦek ‘that very much’. ‑ʦa, ‑ʦe, and ‑ʦek seem to be contractions of ʦam ‘as/how much’ plus the limiting quantifier {‑ʧik} ‘a, some’. Cf. also Sprigg’s (2002: 126) statement that when following verbs, ‑paʦe means ‘as far as, as much as’. While the Western Sham informants stated that this form is used when focusing on a measurement or amount, it often appears when contrasting two actions, cf. (38) and (39).

 

Many West Tibetan varieties use a morpheme ‑saŋ (also ‑sa:n or ‑su:m). This may follow the standard directly as in the Ari/Ali = Ngari (Mŋaḥ.ris) dialects Gar (Sgar), Tsamda (Rtsa-mdaḥ), Gergye (Dge.rgyas), Purang (Spu.hreŋ), cf. Hu Tan (1989: 404), as well as in the Himachal Pradesh varieties Spiti (Grierson 1090: 27) and Nako (Saxena in preparation), infrequently also in Ladakhi. In the Kenhat dialects, it typically follows the genitive (‑e), while in the Shamskat dialects it typically follows the morpheme {‑pa}. Arguably, the element ‑saŋ contains the same element ‑s < *‑so or *‑se that was used in deriving ablative and instrumental case markers from the locational and genitive cases. In at least one dialect of Lahul, namely in Koksar, ‑saŋ is found both as a contrastive and as an ablative marker (Roerich 1933: 108). Several Ladakhi dialects use a clause-chaining marker {‑pasaŋ} for a (mostly) causal relation, which apparently contains the same elements. The {‑pasaŋ} construction seems to retain an earlier form of the Old and Classical Tibetan morpheme {‑pas}. Table 1 gives an overview over the relational markers in the Tibetic languages. The numbers in the rightmost column refer to the sources specified in note 17.

 

The use of verbal constructions for the standard is not accounted for by the common classifications, such as Stassen (1984, 2013), and the use of a comitative or genitive case marker is also not very prominently discussed. The attested variability further contradicts some of the typological predictions: it is apparently not always the case that “[I]f a language has an allative comparative, then it is VSO” (Stassen 1984: 159, no 18b, 173) and it is also not always the case that “[l]anguages with an allative comparative are languages with absolute posterior consecutive deranking and total identity deletion” (Stassen 1984: 172, no. 1B, 173; no Tibetic language subordinates posterior events to anterior events in sequential chains and identical verbs are hardly ever deleted).

 

Table 1. The relational marker across Tibetic languages[18]

 

3.3. Word order

 

If one did not know how to express oneself in a foreign language, one would possibly first point to a standard, then to the comparee, and then make a gesture signalling big or bigger or small or smaller (at least I would try to do so). Some sign languages follow exactly this principle, see Özsoy & Kaşıkara, this volume, for Turkish Sign Language. This order corresponds to the common structure of topic and comment. Jacques (2016: 21) accordingly observes that “in comparative constructions, the comparee is more often the focus than the standard”. I should like to call this the iconic order. The Tibetic languages, by and large, follow the iconic order: the neutral order for differentiating property ascriptions is S-M C P, cf. (3) and (4). However, in Ladakhi, the aniconic word order C S-M P is strongly preferred for asymmetric equative property ascriptions, cf. (2). Both word orders contradict the prediction that the order between noun (N) and adjective (A) is inverted with respect to the adjective (parameter) and the standard: NA > AS, AN > SA (see here Andersen 1983: 103 with further references).

 

(2) Shamskat: Domkhar (FD 2007)

C	S-M	P	/	S-M’s	P	aux
l̥ʧaŋma˖o	naŋ-po-ʦek	thonbo		naŋ-po-ʦeg-i	thonbo	duk.
tree˖df	house-df-as.much	high		house-df-as.much-gen	high	vis.be
‘The tree is high as much as the house / is [of] as much the high[ness] of the house (visual evidence).’ ~ The tree is as high as the house. / is of the same height as the house.

 

(3) Old Tibetan: ITJ 0730 Mother Sumpa’s sayings (l. 14f.)

S-M	C	P-verb
pha-bas	bu	ḥdzaŋs-na-ni ¦	spaŋ-la	mye	thar-ba	bžin-la ¦¦
father-rel	son	be.clever-cd-top	meadow-all	fire	pass-nls	like-all
S-M	C	P-verb
pha-bas	bu	ŋan-na-ni	mʦhal.chus		ded-pa-daŋ		ḥdraḥo ¦¦
father-rel	son	be.bad-cd-top	vermilion.water.erg		chase-nls-com		be.like.sf
‘If, in contrast to the father, the son is clever, it is like fire spreading on the meadow. If, in contrast to the father, the son is bad, it is like being chased by vermilion water.’ ~ If the son is/wants to be more intelligent than the father, this is like a meadow catching fire. If the son is worse than the father, this is like facing the muddy waters of a flood.[19]

 

(4) Shamskat: Skindiang (FD 2007)

S-M		C	P-verb	/	P &		aux
zgo-e	riŋbo-basaŋ	ʦheriŋ	riŋ-ok.		riŋ-[b]a ~	riŋmo	duk.
door-gen	length-rel	[name]	be.long-inf		be.long-nls	long	vis.be
‘In contrast to the length of the door, Tshering tall-s (generic) / is tall-ing ~ is tall (visual evidence).’ ~ Tshering is taller than the door.

 

(5) Shamskat: Turtuk (FD 2015)

S-M		C		P &	aux
gji	kore-(a)paʦa	gji	kore	phraŋo	naŋ.
this	cup-rel	this	cup	small	vis.be
‘In contrast to this cup, this cup is small (visual evidence).’ ~ This cup is smaller than that one.

 

When the contrastee is already given, it can be shifted to the topic position: C S-M P. In (6), the speaker had been banished to a foreign country, where she was received with great honours, but her heart was with the people she had to leave behind. These people naturally occupy the topic position:

 

(6) Shamskat: Khalatse, Ñilza Aŋmo (recorded 1996)

C					S-M		P-verb
ŋi	sem	de-la	jot-pa-ri[g-i]	mi-ŋun	rinʧan-i	ser-basaŋ	r̥kon.
I.gen	mind	that-all	exist-nls-lq-gen	people-pl	costly-gen	gold-rel	be.scarce
‘The people in my mind are scarce-ing [i.e., precious] in contrast to the costly gold.’~ The people [who I bear] in my memory are dearer to me than [all] the costly gold.

 

In modifying or embedded differentiating property ascriptions, the predication precedes the contrastee, cf. (9).

 

Independent of word order, there is generally no problem to stack two properties, if there is a feasible context.

 

(7) Kenhat: Gya-Mīru (FD 2007)

ka̱lte	ʧārfa	go-saŋ		ʒaŋ		ʧhe-a	ja̱ŋ	riŋ-a	ɦot-na,
if	bed	door-rel		width		be.big-nls	again	be.long-nls	ass.be-cd
te̱ne	ɦoγa(ː)		ka̱χfo		ʧh˖en.
then	we.aes		difficult		go˖ass.be=fut
‘If the bed, in relation/contrast to the door, is big-ing with respect to [its] width and also longing, then we will get difficulties (assertive).’ ~ If the bed is (not only) wider, (but) also longer than the door [is high], then we will get difficulties.

 

4. The inherent meaning of the relational marker

 

The relational markers ‑esaŋ /‑basaŋ, ‑(a)patsa, and ‑p/batsek establish an unspecific relation between two items, places, or points in time. Most often it is a relation of difference or contrast with no scale implied, as when stating that something differs from something or is other than something, as in (8) and (9). The particles from and than in English are likewise unspecific, and do not involve a scalar notion. Items that differ from each other, may do so particularly with respect to non-scalar and non-shared properties, one item might be round, the other square, one might have a sonar system, the other not, etc.

 

(8) Shamskat: Teya (FD 2010)

 

(9) Shamskat: Teya (FD 2010)

 

The marker may also be used to express non-scalar relations of time and space in competition with other constructions. The relation ‘before’ is expressed with the postposition (genitive plus clitic) sŋonla ~ sŋanla (ŋōna ~ ŋāna) ‘earlier’ when referring to a short interval, such as in Teya daŋ-i-sŋonla, in Gya-Mīru daŋ-e-ŋana ‘just before yesterday’, but with the relational marker and the adverb when referring to a longer interval: Teya daŋ-asaŋ sŋonla, Gya-Mīru daŋ-esaŋ ŋāna ‘some time before yesterday’. Cf. also Purik saq-batsik snan-la ‘earlier than all’ (Zemp 2013: 406, ex. 116). Similarly, a locational postposition (genitive plus clitic) is preferred for a direct relation, such as Gya-Mīru ʧōktse-(ː)-ɦoga ‘below, under the table’, whereas the use of the relational marker indicates a less direct relation, such as Gya-Mīru ʧōktse-(ː)saŋ ɦoga ‘somewhere near the space below the table’ if the item is not exactly under the table, but somewhat on the side on the floor. In both, the temporal and the spatial usage, there is no gradable property early or late or down or up implied. Instead there are fixed anchor points against which the relation is established. In (10), the fixed anchor time or standard is ‘now’, which is neither late nor early, and in (11), the fixed anchor location or standard is the village Mīru. The first alternative with the shortened form ‑saŋ is used when the speaker is in Mīru, that is, on the same level, and in that case the village position is neither high nor low in any meaningful sense. The second alternative with the full form ‑esaŋ is used when the speaker is at some other place, and in that case, the anchor location might be even higher up than the speaker, but the position relative to the speaker plays no role for the relation between the anchor location and the place that is referred to, cf. also (12).

 

(10) Kenhat: Gya-Mīru (FD 2008)

ta̱ksa	ʃam˖e	ʧūli	ʦhaŋma	ʦhe˖re-duk,	kūʃu	ʧūli˖(ː)saŋ	tīŋne	ʦhe-ʒen.
now	[name]˖gen	apricot	all	ripe˖cc-vis.be=prf	apple	apricot˖rel	after	ripe-fut
‘Now the apricots of Sham (Lower Ladakh) have become ripe (visual evidence). The apples will get ripe after the apricots.’

 

(11) Kenhat: Gya-Mīru (FD 2007)

mīru-saŋ	thur(r)a	/	mīr˖isaŋ	thur(r)a
[name].(abs)-rel	downward		[name]˖(gen‑)rel	downward
‘in relation to Mīru downwards’ ~ below Mīru  (With abs: the speaker is in Mīru; with gen: the speaker is somewhere else, either in Gya, the next village further down, or in Leh, much further down than the place referred to.)

 

(12) Shamskat: Turtuk (FD 2015)

tjakʃi-paʦa	thurla	pakistan	in.	go-ʧuk-pa-met.
[name].(abs)-rel	downward	[name]	be	go-cs-nls-ng.ass.be
‘In relation to Tyakshi downwards is Pakistan. [The army] won’t let [you] go (assertive).’ ~ Below [downriver] Tyakshi comes Pakistan …

 

The marker may further indicate an unspecific, typically non-scalar relation of ‘beyond, in addition’. It is quite commonly used to express relations between generations, such as in Teya api-(b)asaŋ ama ‘mother in relation to grandmother’ ~ grand-grandmother or in Turtuk apo-patsa apo ‘grandfather in relation to grandfather’ ~ grandfather’s grandfather, cf. also (13).

 

(13) Shamskat: Khalatse, Langdarma (recorded 2006)

memeˈgjap˖e,	d˖o-basaŋ	memeˈgjap˖e	[ʧhagˈraps]
ancestor.king˖gen	that˖df-rel	ancestor.king˖gen	hon.genealogy
‘[the genealogy/history] of the ancestor king(s) (and) in relation to that/those, of the ancestor king(s)’ ~ the history of the ancestor king(s) and again of the ancestor(s) of that/those king(s).

 

In other cases, the relation marker should be translated as ‘not only x, but (even) y’, as in examples (14) and (15). With numerals, the meaning can also be ‘more than x’ (a numeral is a fixed anchor point, it does not have a scalable property), cf. (16).

 

(14) Shamskat: Teya, proverb (FD 2010)

hapo-(ba)saŋ	hupo-aŋ	ʧhat-soŋ!
morsel-rel	sip-fm	get.finished-happen.pa
‘In relation to the morsel also the sip happened to finish!’ ~ Not only the morsel but also the sip has finished![20]

 

 

(15) Shamskat: Domkhar (FD 2003)

kho	ʃi-a-basaŋ-nik	ʦhat-po-aŋ	jal-e-mi-nak.
s/he	die-nls-rel-top	heat-df-fm	disappear-cc-ng1-nvis.exist=prf.compl
‘In relation to his/her dying, also the heat has completely disappeared (non-visual experience).’ ~ Not only has s/he died, but also the [body] heat has completely vanished. ~ Not to talk about his/her dying, even the heat has completely left [his/her body].

 

(16) Shamskat: Domkhar (FD 2004)

ladaksˈpa-ŋun-la,	kargil-i	r̥mak-ʦana,	sipa	ɲiʃu-basaŋ	maŋbo	ʃi.
[name].people-pl-aes	[name]-gen	war-when	soldier	20-rel	many	die.pa
‘During the Kargil war, the people of Ladakh had to suffer that in relation to 20 soldiers many died.’ ~ The people of Ladakh suffered the death of (much) more than 20 soldiers during the Kargil war.

 

Furthermore, the marker is very frequently used to express a fundamental contrast, instead of or rather than (cf. also Rangan 1979: 147 and Zemp 2013: 406-407, exx. 118, 119, 721, ex. 115):

 

(17) Shamskat: Turtuk (FD 2015)

tibi-paʦa	gonʧas-ʧi	khjoŋ!
hat-rel	dress-lq	bring.imp
‘In contrast to a hat, bring a dress!’ ~ Bring a dress instead of a hat!

 

(18) Kenhat: Gya-Mīru (FD 2012)

daŋ	ŋ˖e̱	khimʦep˖e	ʦōgdan	ʒak	ʧū-sesaŋ	ʒak	dun-a	go	lā˖fen.
yesterday	I˖erg	neighbour˖gen	pile.carpet	day	10-rel	day	7-all	position	raise.pa˖rm
‘Yesterday I finished off the neighbour’s pile carpet in contrast to 10 days in 7 days.’ ~ Yesterday, I finished off the neighbour’s pile carpet in 7 instead of 10 days.

 

(19) Kenhat: Gya-Mīru (FD 2007)

 

(20) Shamskat: Domkhar (FD 2013)

las	rgjas-en˖uk,	ʦhar-ba-basaŋ.
work	increase-cnt˖vis.be=prs	finish-nls-rel
‘The work increases (visual evidence), in contrast to finishing.’ ~ The work increases, rather than getting finished.

 

(21) Shamskat: Khalatse Pakistan war (recorded 2006)

« di-aŋdu

hinduˈstan-is

bam

tã-ok.

odi-aŋdu

baγo-ek

duk.

this-ppos

[name]-erg

bomb

give-inf

this.very-ppos

cave-lq

vis.exist

de-aŋ

ʧha-[r]gos-ok. »

zer-e,

di-aŋ

khjoŋs.

that-ppos

go-need-inf

say-cc

this-ppos

bring.pa

deana

braŋsa

di-aŋ

duks-pasaŋ-na,

then

lodging

this-ppos

stay.pa-cc/rel-abl

ne

d˖o-basaŋ

di-aŋdu

bam

joŋs-pa,

thoγ-eka

bap-sok.

then

that˖df-rel

this-ppos

bomb

come.pa-nls

roof-ppos

come.down-inf

‘[The soothsayer] having said: « The Indians will bomb this place (inferential). Over there is a cave (visual evidence). [You] should go there (inferential) », took [us] here (observed). Then, after settling in this “lodging”, then in contrast to that [predicted place], the bomb came in here, it fell [on the rocks] above (inferential).’ Shamskat: Khalatse Pakistan war (recorded 2006) ~ … Instead of [falling on] that [predicted place], the bomb came in here, …

 

None of the relations just presented implies a difference in terms of shared properties and degrees, but a categorical positioning of one item in time or space or in a more abstract sense in relation to another standard or anchor point. Even the relation ‘earlier’ or ‘before’ does not imply any kind of graduality, but simply a positioning on the time arrow ‘left’ of the anchor point (one of two possible different positions, cf. also the use of the Indo-European contrastive marker *­tero­ for the meanings ‘left’ and ‘right’, as discussed in section 11).

 

5. Alternative strategies

 

5.1. Juxtaposition

 

In order to express a difference or a contrast, speakers may also juxtapose one property or situation with an opposite or fundamentally different one in two clauses. This strategy can be used when one has to decide which one of two items has the property one is looking for. In other contexts, however, this strategy seems to emphasise the contrast. Example (22) from a modern textbook has clearly an overtone of surprise and disapproval, as its content is against the modern Tibetan values of peacefulness. An emphatic overtone can also be observed in (23), which was given as an exemplification of the verb rgjas ‘increase’, before reformulating it into an ordinary construction of differentiating property ascription, as in (20) above.

 

(22) Modern Hybrid Literary Tibetan (Bod.gžuŋ Šes.rig Las.khuŋs 1994: 20.15-16)

raŋ.bžin-gyis	ši-bar	ŋan-par			brʦis ¦
natural-ins	die-nls.loc	be.evil-nls.loc			count.pa
g.yul-du	bsad-pa-la		bzaŋ.por	brʦis		žes	gsal ¦
battle.field-loc	kill.pa-nls-all		good.loc	count.pa		qom	be.clear
‘It becomes clear [from the documents] that [in olden times] to die from natural [causes] was considered as evil, [while] to be killed in the battlefield was considered as (morally) good.’ ~ … it was considered to be (morally) better to be killed in the battlefield than to die from natural [causes].

 

(23) Shamskat: Domkhar (FD 2013)

las	ʦam-ʃik	ʧo-na,	ʦhar-ba-mi-nuk,	rgjas-en˖uk.
work	how.much-lq	do-cd	finish-nls-ng1-vis.be=prs	increase-cnt˖vis.be=prs
‘However much [one] works, [the work] does not finish (visual evidence), [it] increases (visual evidence).’

 

When one decides which entity has a certain property and which not (following an alternative question), the most common interpretation is that the more positive element (e.g., the big one, the high one) constitutes the contrastee, while the opposite element (e.g., the small one, the short one) constitutes the standard, independent of whether the question focuses on the positive property (Is X big or Y?) or on its counterpart (Is X small or Y?; Leh, Shachukul FD 2016), see example (24) for the positive variant. Unlike in the case of Washo (see Bochnak & Bogal-Allbritten 2015: 119), this construction does not imply a norm-related contrast, it may also be used when both items are relatively small, e.g., when deciding about two tree saplings that are less than 1m high (Leh, Shachukul FD 2016).

 

In individual cases, speakers may prefer the opposite interpretation, that is, the positive property is related to the standard and its opposite to the contrastee, see example (25). Such individual variation may have something to do with what kind of mental image, related to their own experiences, speakers have in their mind. This kind of hidden context is usually not accessible to the researcher (cf. also Zeisler 2016).

 

(24) Shamskat: Domkhar (FD 2007)

naŋ-po	thon-bok.	l̥ʧaŋma	ʦhuŋ-bok.
house-df	be.high-inf	tree	be.small-inf
[Is the house high(er) or is the tree high(er)?] - ‘The house high-es (generic), the tree small-es (generic).’ ~ The house is higher than the tree. Not: *The tree is smaller than the house.

 

(25) Kenhat: Gya-Mīru (FD 2007)

khamba	tho-ɦak.	ʧāŋma	thuŋ-gak.
house	be.high-inf	tree	be.short-inf
[Is the house high(er) or is the tree high(er)?] ‘The house high-es (generic), the tree short-es (generic).’ The interpretation in terms of: The tree is shorter than the house was preferred to: The house is higher than the tree by this speaker, at this occasion.

 

5.2. Relative clause constructions

 

Relative clauses of the type tsam - detsam ‘as/how much - that much’ for equative property ascriptions, as in (26), or tsam - do-rel ‘as/how much - in relation to that’ for differentiating property ascriptions, as in (27), are a common alternative strategy for more complex relations between two items or situations.

 

(26) Kenhat: Gya-Mīru (FD 2007)

aba	ʦām-ʃik	thonbo	ɦot,	te̱zam-ʃik	ʈūu-aŋ	tho-ɦanak.
father	as/how.much-lq	high	ass.be	that.much-lq	child-fm	be.high-inf
‘As much as the father is tall (assertive), that much also the child will get tall (inferential).’ ~ The child will probably get as tall as his/her father.

 

(27) Shamskat: Domkhar (FD 2007)

aŋmo-a	ʦam-ʃik	gjokspa		sikel			ʃrul-ba-ɲan-et,
[name]-aes	as.much-lq	quick		cycle			ride-nls-be.able-ass.be=prs
d˖o-basaŋ	gjokspa-(rik)		ʦheriŋ		/	ʦheriŋ-a		baŋ	t eaŋ-ba-ɲan-en˖uk.
that˖df-rel	quick-(lq)		[name]			[name]-aes		run	give-nls-be.able-cnt˖vis.be=prs
‘As much as Angmo can ride fast on the bicycle (assertive), in relation/contrast to that, Tshering can run fast (visual evidence).’ ~ Tshering can run faster than Angmo can ride on the bicycle.

 

5.3. Explicit expression of comparison

 

Infrequently, the notion of comparing is mentioned explicitly, as in (28), or in an elliptical construction, as in (29) and (30). This might have been inspired by the English usage of compared to. However, the Ladakhi construction does not express the idea that the comparee has only a relatively low degree of the property in question. To express this latter notion, one might use a construction where the property is negated for the standard, cf. (58) below.

 

(28) Kenhat: Gya-Mīru (FD 2015)

kh˖e	khimʦep˖e	aʧi	no̱mo-a	dur-de,	no̱mo	de-ɦak	lo̱.
s/he˖erg	neighbours˖gen	elder.sister	younger.sister-all	compare-cc	younger.sister	be.beautiful-inf	qom
‘S/he compared the neighbours’ elder sister with the younger sister and said the younger sister is beautiful (inferential).’ ~ S/hei compared the neighbour’s elder daughterj with herj younger sister and said that the younger one was more beautiful.

 

(29) Kenhat: Gya-Mīru (FD 2015)

khimʦep˖e	aʧi-a	dur-na,	no̱mo	de-ɦak	lo.
neighbours˖gen	elder.sister-all	compare-cd	younger.sister	be.beautiful-inf	qom
‘If one compares [her] with the neighbour’s elder sister, the younger sister is beautiful (inferential) [s/he] said.’ ~ Compared to the neighbours’ elder daughter, the younger one is more beautiful, [s/he] said. ~ The neighbour’s younger daughter is more beautiful than her elder sister, [s/he] said.

 

(30) Shamskat: Domkhar (FD 2012)

le-a	spes-e,	domkhar	goŋma˖(ː)	ɲildab-is	silmo	ɖak.
[name]-all	compare.pa-cc	[name]	[name]˖all	twice-ins	cool	nvis.be
‘Compared to Leh, Domkhar Gongma is cool by a double (non-visual evidence).’  ~ Compared to Leh, it feels twice as cold in Upper Domkhar.

 

5.4. Verbs expressing difference or excess

 

Ladakhi, like other Tibetic languages, has a few verbs that indicate some kind of difference. The most common of them express the idea that something happens in excess to what is normal, expected, or sanctioned. The standard may thus remain unexpressed. If expressed, it commonly receives the relational marker, but an ablative postposition is also frequently found. In such cases, the ablative postposition positively indicates that the situation is singular or exceptional, (34) and (35), while the relational marker is used neutrally both for individual and general situations. Such verbs are usually quite restricted in their application and are not generally used for differentiating property ascriptions. Example (31) illustrates the notion of excess with respect to an implicit moral standard, examples (32) to (34) the meaning extension for numerical values and actions. The Ladakhi (and Tibetan) exceed construction contradicts Stassen’s (1984: 157) claim that “the standard NP is invariably constructed as the direct object of a special transitive verb” (emphasis added).

 

(31) Shamskat: Domkhar (FD 2003)

kho	thal-duk.
s/he	overshoot-vis.be=prs
‘S/he always exaggerates/goes over the top/crosses the limit (visual evidence).’

 

(32) Shamskat: Khalatse, Changing Ladakh (recorded 2006)

deʦana-si	kirmo-ŋun-la	daksa	rgja-basaŋ	thal-e-in-ʦog_	_le.
that.time-gen	rupee-pl-all	now	100-rel	exceed-cc-be=prf-inf	hon
‘For the rupees of that time [what one would get] now is more than/exceeds one hundred [rupees] (inferential).’ ~ The value of one rupee of those times would be more than 100 rupees now.

 

(33) Kenhat: Gya-Mīru (FD 2009)

ma̱ŋʧhea		aŋm˖esaŋ		riŋzin	(ma̱ŋ-a)			pheʃn	ʧē˖ruk.
mostly		[name]˖rel		[name]	(be.much-nls)			fashion	do˖vis.be=prs
ɦinaŋ	te̱riŋ		aŋmo	riŋzin-ehane		~	riŋzin-esaŋ		thal.
but	today		[name]	[name]-ppos:abl			[name]-rel		exceed.pa
‘Mostly, Ringzin does (much) fashion in relation/in contrast to Angmo (visual evidence). But today Angmo (exceptionally) surpassed Ringzin (observed).’ ~ Usually Ringzin is more fashionable than Angmo, but today Angmo has (exceptionally) surpassed her.

 

(34) Kenhat: Gya-Mīru (FD 2009)

kh˖esaŋ	/	khe˖hane	aŋmo	ʈi̱-ʒe˖(ː)naŋˈa	thal.
s/he˖rel		s/he˖ppos:abl	[name]	write-nls˖ppos:loc	exceed.pa
‘In relation/contrast to him/her / From him/her, Angmo exceeded in writing (observed).’ ~ (As an exception,) Angmo was better than him/her in writing [i.e., wrote faster, nicer, or with less mistakes].

 

To some extent, such verbs can also be used adverbially to express the meaning ‘do something in excess’. Again it is not necessary to explicitly mention a standard.

 

(35) Shamskat: Khalatse, Discourse on religion (recorded 2007)

koa˖(ː)	sku-ʧaz-la			zdeps-e	kher-ʧe˖n				lo,	ɲeraŋ-a,
leather˖all	rub-nls-all			barter-cc	take.away-grd˖ass.be=fut				qom	hon.you-all
lʤakma-la		mar	ɖanɖa-basaŋ			thos	taŋs-e,	potpa-s
grease-all		butter	equal-rel			be.high	give-cc	[name]-erg
‘[The monks would collect the excess grease from the butter tee and the Tibetans] would barter [it] and would take [it] for rubbing [it] into leather (assertive), [it] is said, you know. And having given for the grease butter in excess in relation/contrast to [what] equals, [that is] the Tibetans …’ ~ … And since the Tibetans gave more butter for the grease than what would be the equivalent, …

 

(36) Shamskat: Khalatse, Discourse on religion (recorded 2007)

kho-e	lʤakpo-la	mar	thos	taŋs-e	kher-ʧe˖n	zer-ed_	_are,...
s/he-gen	grease-all	butter	be.high	give-cc	take.away-grd˖ass.be=fut	say-ass.be=prs	intj
‘For their grease, [the Tibetans] would give butter in excess, and take [it] along (assertive), they say (assertive), hey, …’

 

6. Differentiating property ascriptions in complex situations: contrasting situations, participants of situations, and different properties

 

When contrasting different situations or options, the relational marker may follow a nominalised verb, cf. (37) to (40), or any constituent of a clause. In the latter case, conjunction reduction may lead to the omission of case markers, and this may yield ambiguous interpretations, as in (41).

 

(37) Kenhat: Gya-Mīru, proverb (FD 2013)

semŋan-ʒik	ma̱ne	tōn-a-saŋ	semzaŋ-ʒik	lū	tāŋ-na	gjal.
mind.bad-lq	prayer	utter-nls-rel	mind.good-lq	song	give-cd	be.good
‘In relation/contrast to uttering prayers evil minded, if a song is given noble minded, [it] is good.’         ~ It is better to sing a song with a noble mind, than to utter prayers with an evil mind.

 

(38) Shamskat: Domkhar, Tale of Khimbo Skambo (recorded 2007)

khje(t)-ʦokspa˖(ː)	bagma˖(ː)	joŋ-ba-ʦek	ŋa	ʧhu	ma-khur-ba	ʧh˖et
you-like˖all	bride˖all	come.prs-nls-rel	I	water	ng2-carry.prs-nls	go˖ass.be=prs
‘As much as to coming as a bride for someone like you, I go [back home] without carrying water (assertive).’ ~ I’d better go/I prefer to go [home] without the water, rather than becoming the wife of someone like you.

 

(39) Shamskat: Domkhar (FD 2014)

dziŋzmo		t eaŋ-in-duk-pa-ʦek			~	teaŋ-in-duk-pa-basaŋ
quarrel		give-cnt-stay-nls-rel			~	give-cnt-stay-nls-rel
ta	ɲentaŋ		ɲiska	bes-aŋ!
now	fam.you.incl		both	separate[intr].imp-dm
‘As much as/in contrast to continuously quarrelling, now you two separate!’ ~ Instead of always quarrelling, you’d better separate!

 

(40) Shamskat: Turtuk (FD 2015)

de	ri-a	thul-ba-paʦa	ʧok	duk-na	gjal.
that	mountain-all	climb-nls-rel	onom	stay-cd	be.good
‘In relation/In contrast to climbing that mountain, if [you] stay completely, [it] will be good.’ ~ [You]’d better stay were you are, instead of going up that mountain.

 

(41) Shamskat: Domkhar (FD 2007)

 

If the two situations are of a similar type, such as ‘eating’ and ‘drinking’, only one verb needs to be mentioned, as in (42). On the other hand, relative clauses may be preferred, as in (43), when the relation between the two situations is less intuitive.

 

(42) Shamskat: Domkhar (FD 2007)

ʧhaŋ-po-basaŋ		kharʤi	maŋ-ba-rik		zo!
beer-df-rel		food	be.much-nls-lq		eat-imp
kharʤi-basaŋ	ʧhaŋ-po		ɲuŋ-ba-rik	thuŋ!
food-rel	beer-df		be.few-nls-lq	drink-imp
‘In relation/contrast to the beer, eat somewhat much food! In relation/contrast to the food drink somewhat little beer!’ ~ Eat some more food than [you drink] beer! Drink somewhat less beer than [you eat] food!

 

(43) Kenhat: Leh (FD 2007)

ʦam-ʃik		zuk	ʧhenmo			ʧha˖t,
as.much-lq		body	big			go/become˖ass.be=prs
te˖(ː)saŋ	gonlag-e			rin	maŋ-a		ʧha˖t.
that˖rel	cloth-gen			price	be.much-nls		go/become˖ass.be=prs
‘As much as the body becomes big (assertive), in relation/contrast to that the price of the clothes is going much-ing (assertive).’ ~ The prices for clothes rise faster than the body grows.

 

Ladakhi speakers have a clear preference for talking about properties of the same category over contrasting unrelated properties. One informant put it bluntly: “Why do you want to compare unequal things?” (FD, Leh 2007). Artificial sentences with no support from a realistic background, such as The bed is longer than the door is wide, were rejected even by well-educated informants: “Why don’t you just turn the bed round?” (FD, Leh 2007).[21] All informants felt somehow relieved when I offered them a more verisimilar context, e.g. a Western marriage-bed, not fitting through a Ladakhi door or exaggerated statements about oversized tourists.

 

Although the relational marker can be combined with verbs, it cannot follow the auxiliaries. This is perhaps not so much the “fault” of the auxiliaries, but an outcome of the fact that Ladakhi speakers do not compare across scales or that the language, like Japanese, “does not allow degree abstraction in the syntactic standard constituent” (cf. Kennedy 2009: 153) and that it likewise does not allow binding of degree variables (cf. Kennedy 2009: 148).

 

The property of the standard, however, can be expressed by an abstract measure noun. The difference is then expressed with an adjectival denoting a quantity. Symmetry effects lead to the use of a measure noun also for the property of the contrastee. But some speakers prefer the more economic construction with only one measure noun for the standard. Relative clause constructions may also be used.

 

(44) Kenhat: Leh (FD 2007)

i

ʈebel-e

ʒaŋ-naŋ

rinbo

go-e

ʒaŋ-saŋ

maŋ-a

duk.

this

table-gen

width-com

length

door-gen

width-rel

be.much-nls

vis.be

tefia

golok

ʧos-te

naŋkug-a

toŋ!

/

kher-in.

therefore

canted

do.pa-cc

inside-all

send.imp

carry-ass.be=fut

‘The width and the length of the table, in relation/contrast to the width of the door, is much-ing (visual evidence). Therefore put [it] / let [us] carry [it] inside (assertive) by canting it.’ ~ The table is broader and longer than the door is wide. …

 

7. Measuring the difference

 

If an integral factor, such as twice or thrice is combined with a comparative construction, ambivalences might arise. I, for my part, know that, when people say in German zweimal größer als ‘two times bigger than’, they actually mean ‘twice as big as’, but it sounds wrong, and I immediately start to wonder whether they did not mean ‘thrice as big’. I would definitively prefer the equative construction zweimal so groß wie ‘twice as big as’. Although Ladakhi speakers use the relational marker, they likewise intend an equation in terms of x-times as not a multiplication of the difference.

 

(45) Shamskat Khalatse, Religious traditions (recorded 2006)

samjas	mana	deana-niŋ
[name]	totally	then-top
l̥a-naŋ	l̥u-naŋ		norʧin-is			ona	ɲima	ʦam-ʦek		r̥ʦig-na,
deity-com	spirit-com		[name]-erg			well	day	as.much-much		build-cc
ʦhan-la	d˖o-basaŋ			ɲildap	rhʦiks-e-jot-khainʦog_				_le
night-all	that˖df-rel			two.time	build-cc-ass.be=prf-dst				hon
‘As for Samyas, at that very time, sky-born deities and earth spirits (nāga-s) and the Wealth-Bestower (Norcin, i.e., Kubera), well, how much [the people] had built up at day time, in relation to that [they, the benevolent spirits] had built up two times in the night time (distanced story mode).’ ~ As for Samyas [the first Tibetan monastery], how much [the people] had built up during day time, twice as much than that was built by the sky-born deities, nāga-s, and Kubera in the night time.

 

Non-integral differences in measurement and amount (or a lack in size or amount) can be expressed by the instrumental case, cf. (46). The construction is not very common, and some speakers avoid it.

 

(46) Shamskat: Domkhar (FD 2011)

aŋmo-a	aʒaŋ-i	bulon-po	stoŋ	ʒi-s	ma-ŋgok.
[name]-aes	uncle-gen	loan-df	1000	4-ins	ng2-be.able.to.pay.back.pa
‘Angmo could not pay back [her] uncle’s loan, with 4.000 [rupees still missing].’

 

Similarly, quantitative differences are hardly ever mentioned in the context of differentiating property ascriptions, and I have not yet come across an example from natural, non-elicited speech. If expressed, the measurement of the difference is often in the instrumental case. For some speakers it may alternatively remain unmarked, as in (47), while other speakers would not use the unmarked form, cf. (48). If both the instrumental and the unmarked form can be used, the unmarked form is used for a neutral statement, while the instrumental emphasises the smallness or greatness of the difference and may thus convey a connotation of surprise as in (47), second alternative.

 

(47) Kenhat: Leh (FD 2007)

ʦheriŋ	aŋm˖esaŋ	inʧ	ɲiʃu /	ɲiʃu-i	riŋmo /	riŋ-a	duk.
[name]	[name]˖rel	inch	20	20-ins	long	be.long-nls	vis.be
‘Tshering, in relation to Angmo, is tall / is tall-ing 20 (neutral) / by 20 (surprise) inches (visual evidence).’ ~ Tshering is 20 inches / as much as 20 inches! taller than Angmo.

 

(48) Kenhat: Gya-Mīru (FD 2007)

ʧāŋma	khamb˖esaŋ	miʈar	ʧīg-e /	*ʧīk	tho-ɦak.
tree	house˖rel	meter	1-ins	*1	be.high-inf
‘The tree, in relation to the house, high-es by one meter (generic).’ ~ The tree is 1m higher than the house.

 

It should be noted that the explicit mentioning of a difference in measurement does not entail a gradable property or predicate. This has been shown in example (46), but is also true for an English sentence like Peter missed the target by 2cm (cf. Pearson 2010: 366 with further references).

 

8. Having no match

 

Absolute property ascriptions (superlative or elative) can be expressed by negating the possibility or existence of a match of a contrastee. Expressions such as graŋs.med ‘numberless’ dpag.med or gžal.med ‘measureless’, etc. are very frequent in the written language. In Ladakhi, mindra ‘not being like, incomparable, different’ is often used. One speaker also suggested sammiɲanʧese ‘of not being thinkable, inconceivable’. The following constructions convey the same idea.

 

(49) Classical Tibetan (Hahn 1996, Lektion 12.3 f.)

 

(50) Shamskat: Domkhar (FD 2007)

su-aŋ	ʧhorol-ʦoks-i	rdemo /	ʧhorol-ʦek	rdemo	mi-nuk.
who/someone-fm	[name]-like-gen	beautiful	[name]-as.much	beautiful	ng1-vis.be
‘Whosoever is not beautiful of Chorol-likeness / as much as Chorol (visual evidence).’ ~ Nobody is beautiful the way Chorol is. ~ Chorol is the most beautiful (girl).

 

Another option is to indefinitely quantify the standard and mark it either with the relational morpheme or an ablative postposition. The ablative postposition ‑i-aŋ ~ ‑e-naŋa ‘from among’ implies that the contrastee is in some way part of the standard group. If that is not the case, the relational marker must be used.

 

In the Kenhat dialects as in many other Tibetic varieties, the absolute property can be expressed by a compound form of the adjectival, such as Classical Tibetan che ‘be big’ + šos ‘the other one’ > ‘unsurpassed big’ or Kenhat ʧhe + ʃok. Like a superlative in Standard European languages, the compound with ʃok can only be used if the contrastee is part of the standard group, thus the form cannot be used to express that the stranger is the tallest compared to all my friends. The compound can be used like a derived adjective, cf. (51). In the Shamskat dialects, the compound form is not used, the derived adjective or a verbal form is used instead, (52). A non-specified (or not contextually given) standard tshaŋma ‘all’ plus relational marker alone implies that the standard is a human or at least a living being. In the case of non-animate items, tshaŋma plus relational marker cannot be used alone without further specification. More commonly, however, one would use the formula P-M P ‘X in relation to X’, as in (53) or one would use the world as the absolute standard location, as in (51).

 

(51) Kenhat: Gya-Mīru (FD 2015)

ŋe̱	mi̱ŋbo		ri-a		dza(k)-kan		ɦin-pen.
I-gen	brother		mountain-all		climb-nls		be-rm
na̱niŋ		kho		ʤikten-enaŋne		ri		tho-ʃog-a	dzak˖ʃe-a	thuk.
last.year		s/he		world-ppos:abl		mountain		high-most-all	climb˖nls-all	meet.pa
‘My brother was a mountaineer. Last year, he met with the climbing of the high-most mountain from among the world.’ ~ … He was about to climb/almost climbed the highest mountain in the world.

 

(52) Shamskat: Turtuk (FD 2015)

puʦa	ʦhaŋma-paʦa	kho	riŋmo	naŋ.
boy	all-rel	he	tall	vis.be
‘In contrast/relation to all[22] boys, he is tall (visual evidence).’ ~ He is the tallest of/among the boys.

 

(53) Hybrid Ladakhi (All India Radio Leh, 31.08.2015)

khoŋ-is		« pakistan-i		sanʦhams-ika		r̥poŋgol		ʧo-khan-la
hon.s/he-erg		[name]-gen		border-ppos:loc		attack		do-nls-all
lan	ʈakp˖esaŋ		ʈakpo		taŋ-ʧes-in »		mol-tok.
answer	strong˖rel		strong		give-grd-ass.be=fut		hon.say.pa-inf
‘He [the army spokesperson] said (non-witnessed): « [We] will give an answer that is strong in relation to being strong to those who attack on the border of Pakistan (assertive) ».’ ~ We shall retaliate in the strongest possible manner to those who attack across the Pakistan border.

 

9. At the limit of acceptability: contrasting with nothing

 

If we only look at the most common way in which Ladahki speakers translate English comparative constructions or translate their constructions of differentiating property ascriptions into English, there does not seem to be an obvious difference. They just use a different construction. Or perhaps they only use an exotic “strategy” to express what we do with our comparative construction. However, most informants do not really know what implications the English construction has and whether it really represents what they mean. We researchers, on the other hand, usually have no understanding of what the informants mean when they use their specific construction, and this independently of whether we let them describe a picture or whether we let them translate an English model sentence. This rough matching is, of course, usually sufficient when speakers of different languages simply interact. The infrequent instances of mismatch might even go unnoticed, or if not, one might not understand why one does not understand, and start a fight or simply move on.

 

However, if we, as linguists, want to know what speakers of a structurally different language really mean or how they conceptualise situations of difference (or anything else), it is not enough to compare the most common constructions or those that apparently easily translate into what we think is the corresponding construction in English. It is rather necessary to test border cases, that is, constructions that are not so common or only marginally acceptable or perhaps not acceptable at all. We may then find out that our preconceptions derived from our own usages or the linguistic mainstream are not fully suitable or that they even hinder us to analyse the uncommon constructions or to understand why these are used.

 

I made this initially very frustrating experience when I collected data for the partner project on comparative constructions. I was told to look out for constructions with a negated standard. This posed no problem for the equative constructions, cf. examples (50) and (54) to (57). However, when trying to elicit constructions that might roughly correspond to the sentence X is bigger than nobody, the first problem I faced, was that I had no idea what this could possibly mean. Hence I could not explain to the informants what kind of meaning I was looking for. Of course, they could not make sense of the English sentence either.

 

The second, and perhaps even greater, problem was that the Tibetic languages do not have constituent negation, but only sentence negation. The negation of a single constituent involves using an indefinite pronoun plus the focus marker ‑yaŋ, Ladakhi ‑aŋ ‘ever’ or the limiting quantifier {‑cig}, Ladakhi {‑ʧik} ‘a, some’ and a negated verb, e.g. su‑s‑aŋ las ma‑ʧos ‘anybody-erg-ever work not-done’ > ‘nobody worked’.

 

In order to establish such a “nobody” as a standard, one would have to nominalise the negated clause. In Ladakhi, one could use the nominaliser ‑k(h)an. But the result does not have the same logical implications as the use of ‘nobody’ in English. I tried all possible permutations of the negation. Not all worked. Some were classified as ungrammatical or meaningless, others were declared to be “too crooked” or as grammatically possible, but not used.[23] Below I present only those constructions that have been accepted by at least one informant. Most of these constructions look quite bewildering, to the extent that one might ask with the reviewers: does anybody really use them? However, they are used, if only infrequently.

 

In the context of differentiating property ascriptions, the informants interpreted a phrase like su-aŋ met-k(h)an-e/basaŋ never in the sense of ‘in relation to somebody who does not exist’ = ‘nobody’, but always in the sense of ‘in relation to anybody who does not have [the property in question]’, yielding a rather modest degree of the property, as in (58), or, when changing this to su-e/basaŋ met-k(h)an, always in the sense of ‘in relation to anybody in a way that nobody has [the property in question]’, yielding an exaggerated property, as in (59) and (60). What is more important, whenever the informants did not reject the construction as “too crooked”, they always interpreted it as an equative construction, despite the use of the relational marker ‑e/basaŋ. Compare the equative constructions in (54) to (57) with the use of the relational marker ‑e/basaŋ in (59) and (60).

 

(54) Shamskat: Skindiang (FD 2007)

su-aŋ	riŋmo	met-khan	ʦheriŋ	(riŋmo)	duk.
who-fm	long	ng2.ass.exist/be-nls	[name]	long	vis.be
‘Whosoever not being tall, Tshering is (tall) (visual evidence).’ ~ Tshering is (tall) in a way that nobody is tall. ~ Tshering is as tall as nobody else. ~ No one is as tall as Tshering. ~ There is no one as tall as Tshering.

 

(55) Shamskat: Skindiang (FD 2007)

su-aŋ		met-khan	ʦheriŋ	riŋmo	duk.
who-fm	ø	ng2.ass.exist/be-nls	[name]	long	vis.be
‘Whosoever not being ø = [tall], Tshering is tall (visual evidence).’[24] ~ Tshering is tall in a way that nobody is. ~ Tshering is as tall as nobody else. ~ No one is as tall as Tshering. ~ There is no one as tall as Tshering.

 

(56) Shamskat: Domkhar (FD 2007)

ʧhorol	su-aŋ	met-khan-ʦek	rdemo ~	rde-a	duk /	rde-ok.
[name]	who-fm	ng2.ass.exist/be-nls-as	beautiful	be.beautiful-nls	vis.be	be.beautiful-inf
‘Chorol is as beautiful as whosoever non-existing (visual evidence / inference).’[25] ~ Chorol is beautiful like nobody else. ~ Chorol is the most beautiful (girl).

 

(57) Shamskat: Domkhar (FD 2007)

ʧhorol	su-aŋ	met-khan-ʦoks-e	rdemo	duk.
[name]	who-fm	ng2.ass.exist/be.-nls-like-gen	beautiful	vis.be
‘Chorol is beautiful of like whosoever non-existing (visual evidence).’ ~ Chorol is beautiful like nobody else. ~ Chorol is the most beautiful (girl).

 

(58) Shamskat: Skindiang (FD 2007)

su-aŋ		met-khan-basaŋ	ʦheriŋ	riŋmo	duk.
who-fm	ø	ng2.ass.exist/be-nls-rel	[name]	long	vis.be
‘In relation/contrast to whosoever not being ø = [tall], Tshering is tall (visual evidence).’[26] ~ Tshering is tall only in relation to those who are not. ~ Tshering is the tallest of the short-grown people. (Looking for a tall child among the school children, but most of the children are too small, only Tshering is reasonably tall.)

 

(59) Kenhat: Gya-Mīru (FD 2007)

s˖isaŋ /	su-ʒig-esaŋ		riŋmo	me̱t-kan	ʦhiriŋ	(riŋmo /	riŋ-a)	duk.
who˖rel	who-lq-rel	ø	long	ng2.ass.exist/be-nls	[name]	long	be.long-nls	vis.be
‘In relation/contrast to anyone, Tshiring is (tall) in a way that ø = [anyone] tall does not exist / that ø = [anyone] is not tall (visual evidence).’ ~ Tshering is taller than anybody else in a way no one is tall.

 

(60) Kenhat: Gya-Mīru (FD 2007)

s˖isaŋ /	su-ʒig-esaŋ			met-kan	ʦhiriŋ	riŋmo /	riŋ-a	duk.
who˖rel	who-lq-rel	ø	ø	ng2.ass.exist/be-nls	[name]	long	be.long-nls	vis.be.
‘In relation/contrast to anyone, Tshiring is tall in a way that ø = [anyone] ø = [tall] does not exist / that ø = [anyone] is not ø = [tall], (visual evidence).’ ~ Tshering is taller than anybody else in a way no one is. ~ Tshering is extraordinarily tall.[27]

 

Several informants stated that example (60) is, in principle, not different from the as tall as nobody constructions in (54) and (55) or the as beautiful as nobody constructions in (56) and (57) above (FD 2007).

 

The last two examples appear over-complex and are not easily analysable. But apart from the fact that the Gya-Mīru informant is one of the most reliable informants I worked with, and one who readily objects to constructions that are not suitable, both examples have been confirmed by speakers from other dialects. They may however disagree which one is the more suitable one. Several informants described example (59) as sounding like a slogan (FD 2007), which means that the construction would rarely be used. However, when re-discussing it with an informant from yet another dialect and presenting it as being perhaps a bit problematic, the informant spontaneously stated “we use it” (Kenhat: Shachukul (FD 2016). She offered a few more common alternative constructions. The one coming closest to the intended meaning of (59) would be example (61).

 

(61) Kenhat: Shachukul (FD 2016)

kho	ʤikten-e	met-kan-e	riŋmo	duk.
s/he	world-gen	ng2.ass.exist-nls-gen	long	vis.be
‘S/he is tall/has the height of [somebody] not existing in (lit. of) the world (visual evidence).’ ~ S/he is tall like nobody else in the world.

 

Despite this exaggerating expression, the informant noted that the construction in (59) is more expressive and the only construction that gives the idea that the person has a supernatural height of, say 2.50m or even 3m, whereas example (61) would be suitable if the height of the person is still in the range of human beings, say, 2.10m.

 

The informants’ descriptions point to the fact that neither the form of the property ascription (adjective vs. verbal noun) nor the relational marker have an inherent scalar semantics, and further that the Ladakhi speakers do not automatically conceptualise the observed differences in terms of degrees. This can also be demonstrated with a non-elicited example.

 

In the immediately preceding context of (62), the narrator describes a representative house that, although possessing attributes of wealth and modernity from outside, is not very beautiful in his eyes, because it is “empty”. He then contrasts it with a traditional house with an old-fashioned balcony where barley is heaped up in the corners (as if this could make the house more homelike) and continues with (62). His statement cannot be understood in the sense that the old house was only relatively beautiful. And since the old house could not have been far beyond the limits expected for a house, the most likely interpretation is again one in terms of a categorical contrast:

 

(62) Shamskat: Khalatse, Village history (recorded 2006)

den	d˖o_	_rdemo	dug_	_jaŋ,	ʧaŋ	met-khan-i	naŋ-ʧig-basaŋ.
then	that˖df	beautiful	vis.be	fm	what.fm	ng2.ass.exist/be-nls-gen	house-lq-rel
‘Then that was beautiful (visual evidence), again, in relation/contrast to a house that does not have anything.’ ~ Now, THAT one is (really) beautiful, NOT any other house that hasn’t anything [special].

 

I could not make sense of this passage, as long as I tried to analyse it along the scalar semantics of European comparative constructions. It was only when I noticed that the relational marker ‑e/basaŋ has other functions in other contexts, (see section 4, examples (17) to (21)) that I was able to get an idea of what it possibly meant. But if no scalar notions are involved in this example, why should we suppose that there are scalar notions involved in those Ladakhi expressions that we can, or have to, translate into English with comparative constructions?

 

10. Negative islands

 

In English, sentences like John bought a more expensive book than anybody else [did buy (an expensive book)] are fine, while the opposite: John bought a more expensive book than nobody [did buy (an expensive book)] does not work. This is called the “negative island effect”. It is thought that this effect arises in English “because the comparative clause […] should return a maximal degree, but the degree description fails to provide one” (Kennedy 2009: 146 with further references). Languages like Japanese do not show this effect, because they compare (or perhaps rather contrast) individuals rather than degrees and/or they compare or contrast unequal things via “relative clauses” or embedded nominalisation. In such constructions, arguably no maximality operator interferes, as can also be demonstrated with the corresponding relative clause in English: John bought a book that is more expensive than the book that nobody bought (Kennedy 2009: 146). Although using the framework of formal semantics, Kennedy seems to point at the same difference in focus that I am arguing for. 

 

The situation is quite similar in Ladakhi. There are again several ways to formulate the situation, none of which is very common. The constructions were acceptable when the set of books not bought was clearly limited, either because it contained a very limited number of books or because it was the set a particular person did not buy, examples (63) and (64). Unlimited standard sets would yield a connotation of boasting or exaggeration, but such sentences were rather questionable. Accordingly, example (65), where no definite set is available, was judged to be a madman’s speech.

 

(63) Shamskat: Domkhar (FD 2007)

 

(64) Shamskat: Domkhar / Kenhat: Leh (FD 2007)

 

(65) Shamskat: Domkhar (FD 2007)

 

11. Conclusion

 

West Tibetan (and more generally: Tibetic) differentiating property ascriptions might be best understood as categorical relations of difference with respect to individuals, rather than comparisons implying a scale. The standard either lacks the property totally (e.g., a very small person to which a tall person is contrasted) or remains unspecified with respect to the particular property (e.g., the standard could be of average height, hardly worth mentioning in a neutral context).

 

The Tibetic languages are certainly not the only languages to do so. Differentiating property ascriptions with similar properties have been described for other languages, such as Japanese or Chinese (cf. Kennedy: 2009). Because these languages defocus from the inherent graduality of properties, unequal properties or situations with different scales cannot be treated in single elliptical clauses as in English subdeletions, but need more complex constructions (embedded nominalisations, relative clauses, or explicit compared-to constructions). The contrastive constructions resemble thus comparisons of similarity and difference, which likewise do not allow subdeletions (cf. Alrenga 2010: 172). On the other hand, just because these language defocus from the inherent graduality of properties, negative island effects do not appear. This is probably also one of the reasons why the Ladakhi sentences with standards that explicitly do not have the property in question, discussed in section 9, are possible. There are thus also some “benefits” for not viewing differences solely through a scalar filter.

 

Both, categorical contrasting and non-equative comparison, conceptualise differences, but they do so from different perspectives: the former focuses more on the difference or contrast as such (defocussing from, or even denying, a shared property), the latter focuses more on the basic similarity, the shared property. The difference between these two types of conceptualisations is not necessarily a fundamental one, and one can observe extensions from both sides. That is, categorical contrasting can be used, and, in fact, is commonly used, for the representation of quite minor differences - which would be judged as being gradual from our European perspective. It is compatible with explicit measurements, although such constructions might be rare in natural speech. Non-equative comparison, on the other hand may also be used in cases of fundamental differences, as in the context of (66).

 

(66) Shamskat: Khalatse, Pakistan war (recorded 2006)

deʒak-ʧik		ze˖ːn˖ak,			mana,
these.days-lq		say˖cnt˖nvis.be=prs			ever
«siaʧen-i	kaŋri-ŋun-la,			mana,		elmet		ma-ʧo!
[name]-gen	glacier-pl-all			ever		carelessness		ng2-do.prs=proh
sŋon-i-basaŋ			intizam	ʧos-e			ʃruŋs-e-duk!»
early-gen-rel			preparation	do.pa-cc			guard.pa-cc-stay.imp
‘These days (I) heard [the Prime Minister] saying, truly: « Do not, ever, be careless at the Siachen glacier etc.! In relation/contrast to earlier, guard [it] by being prepared! ».’ ~ … guard it, better prepared than/in contrast to last time.

 

In 1999, the Indians had retreated from the Siachen, as they had done every winter. When the Pakistan army invaded it, they were completely taken by surprise. One could not say that they had guarded the Siachen with a not-so-good preparation. There was no guarding and no preparation. Nevertheless, in German (or other Standard European languages), one would typically say be better prepared next time or guard it better next time in such situations, using the comparative construction glacier. While there might still be a gradual interpretation possible from the point of view of a logician, an ordinary speaker is not a logician (nor a linguist, for that matter), and s/he does not conceive of such situations in terms of degrees of better or worse. There are also other usages of besser in German that are not meant to imply a scalar comparison, but a contrast: Das hättest Du besser nicht getan! ‘You (really) shouldn’t have done that.’[31]

 

The Indo-European languages further supply some evidence that degree semantics can develop out of contrastive expressions. There is also evidence that speakers might (repeatedly) chose a contrastive expression over a scalar comparative expression or reinforce a non-scalar notion by using contrastive expressions with comparatives. This is another indication that differences are not solely conceptualised in terms of degrees.

 

In Ancient Greek and in the Old Indo-Iranian languages, the comparative degree marker developed from a contrastive formation with the IE suffix *‑tero-. The suffix was originally used besides some apparently more comparative suffixes (­yes-/‑yos‑ and -isto-) mainly to indicate a difference between, or a separation of, two elements of a pair (see here Szemeréni 1990: 210f.). Compare Sanskrit i-tara ‘other’, ka-tara ‘who of the two (question)’, ya-tara ‘who of the two (co-relative)’ with the comparative priya-tara ‘more liked/loved’, Greek he-teros ‘other’, po-teros ‘who of the two’, protos ‘first’ vs. deu-teros ‘second’, heme-teros ‘our’ vs. hyme-teros ‘yours pl.’, dexi-teros ‘right’ vs. aris-teros ‘left’, with the comparative makro-teros ‘big(g)-er’, or also Latin u-ter ‘which of the two’, ne-u-ter ‘none of the two’, nos-ter ‘our’ vs. ves-ter ‘yours pl.’, dex-ter ‘right’ vs. sinis-ter ‘left’ (here, no comparative usage developed).

 

This original binary contrastive meaning of *­tero was also underlying the earlier use of German we-der and English whe-ther as a question pronoun ‘who of the two’ and of German weder in the sense of ‘none of the two’, similar to English nor < nother < nâhwäðer ‘none of the two’. In some older German varieties as well in some non-standard English varieties it was possible to use weder (and nor) with comparatives, hence to say I am greater nor he in the sense of ‘I am great(er) and not he’, and it was similarly possible in German to use weder in connection with an-ders ‘o-ther’ or ‘different’, thus es kan vor abends wol anders werden, weder es am morgen war (Luther) ‘until evening it may well be other than/different from how it had been in the morning’. For this Germanic data, cf. Grimm (1854-1961, Bd. 27, Sp. 2834-2848).[32] Stassen (1984: 178) points to a similar “underlying negative element” in English that may appear even overtly in Gaelic and Latvian comparatives. Cf. also Andersen (1983: 127, 128 with further references) for Indic and Slavic. Negation may also show up in the French subcomparative: La table est plus longue qu’elle n’est large ‘The table is longer than it is not wide’. The semantic analysis of comparatives can thus be broken down to a reformulation in the sense that A is X to an extent that B is not (Stassen 1984: 179). However, given this negative element, it should be rather logical that speakers might focus more on the negation, that is, on the contrast, than on the shared property and its degrees. In that case, the semantic analysis breaks down to A is X and B is not, and it has been argued that such conjoined construction can be reconstructed for Proto-Indo-European (Cuzzolin & Lehmann: 2004: 1214).

 

If the Standard European languages are almost unique among the world languages for having a special affix for the comparative degree of adjectives (see Stassen 2013), the Indo-European origin of this affix from a marker of binary contrast and the reintroduction of contrastive suffixes or particles in the context of differentiation, as well as the above observations in Ladakhi (and also other languages), indicate that the notion of contrast or otherness is at least as fundamental to human thinking as the conceptualisation of differences in degrees. Apparently, the categorical contrast between good and bad, small and big, few and many, can be broken down into smaller units, until one reaches a scale of infinite degrees. However, it does not seem to be necessary to do so for interacting successfully with nature and other human beings, and hence speakers of other languages do not necessarily need “comparative strategies”. We may rather have to ask, why speakers of European languages developed notions of degrees for talking about differences.

 

The differences in focus between categorical contrasting and non-equative comparison may be subtle, but that does not mean that they can be neglected. More generally, before subsuming a language-specific construction under a category established for Standard European languages (or any other language, for that matter), it might be useful not only to look for the most common, every-day applications of this construction, those that translate easily into English, but also to test the fringes of acceptability. It will be exactly at the limits of what can be said, that different conceptualisations of situations or relations may get revealed. We linguists should appreciate such diversity more, rather than levelling it out under claims of universality.

 

Appendix: Some characteristics of Tibetic languages

Tibetic languages are generally treated as monosyllabic languages, although words (or intonation units) are often polysyllabic. However, when forming compounds, derivational syllables are deleted, so that the compound ideally consist of only two syllables. All major word classes (noun, pronoun, adjective, adverb, verb) are found; (nominal) qualitative and quantitative adjectives, however, are only secondarily derived from (verbal) adjectivals (see also section 3.1 above). Modifying adverbs are typically derived from adjectives through the addition of an oblique case marker, but in Ladakhi, the plain adjectives are used as adverbs. The Tibetic languages lack articles, but use pronouns or other morphemes to (non-obligatorily) mark definiteness or indefiniteness. They do not have gender distinctions, and they do not use classifiers, except in a few cases.

 

The unmarked word order is verb-final and subject-initial, but the order of the nominal constituents is flexible and reflects the theme-rheme relations of the discourse. That is, new information comes closest to the verb, while given information is either found sentence initially or, more commonly, is simply left implicit. Within the noun phrase, the word order is as follows: (modifying syntagm) > noun > adjective[33] > numeral/ plural marker[34]/ indefiniteness marker or demonstrative pronoun > case marker or postposition. In Ladakhi, however, the demonstrative pronoun appears at the beginning of the noun phrase, while a special definiteness marker may appear in the slot of the classical demonstrative pronoun.

 

The Tibetic languages thus show group inflection, that is, only the last element of the nominal phrase bears the relevant case marker or postposition. The modifying syntagm may consist of an adjective or an embedded nominalised clause, both followed by a genitive case marker. This construction corresponds to a restrictive relative clause in English.

 

Tibetic languages originally showed a somewhat atypical ergative alignment,[35] but many modern languages have reduced agent case marking to a minimum or show a split related to temporal reference and pragmatic features. The Ladakhi dialects vary along this cline. There is no voice distinction and no syntactic pivot.

 

Apart from main and dependent clauses, the Tibetic languages also show an intertwined clause chaining construction of co-subordination (cf. Haspelmath 1995: 9, 20-27 with further references), where the non-finite first verb triggers the choice of the “subject” case marker (the “subject” is “deleted” in the following clauses), while the finite last verb bears the mode and tense morphemes. Clause chaining and subordination is indicated by morphemes added to the verb stem or to a complex verb syntagm.

 

In the written language, nominalised embedded clauses may contain further embedded clauses. While these clauses usually appear where English speakers use relative clauses, relative clauses in the strict sense, involving an indefinite pronoun in the first clause and possibly a demonstrative pronoun in the second clause did exist as a marginal construction in the oldest attested stages. Such constructions are also commonly used for the more complex relations of difference in Ladakhi. Under the influence of English and Urdu, an inverted construction with an indefinite pronoun plus definiteness marker (ka-bo ‘that which’) in the second clause is spreading in Ladakhi.

 

The older stages of the language (Old and Classical Tibetan) show partial verb stem inflection for relative tense and mode (up to four verb stems), but otherwise, the languages are agglutinating. They show no traces of person marking. The modern languages, however, developed a special kind of evidential-cum-attitudinal marking. The opposition between self-related, intimate, or authoritative knowledge and (mere) sense perception is mainly displayed by auxiliary verbs. The whole finite verbal syntagm consists of the lexical verb stem or a complex verbal expression, mostly followed by an auxiliary verb. Between verb stem and auxiliary a nominaliser or some other linking morphology may appear. The stem or the auxiliary may be further compounded with elements for inferences and epistemic evaluations.[36] Polar questions are marked at the end of the syntagm.[37] All this may be followed by a quote marker.

 

Tibetic languages have no constituent negation, only sentence negation. The two negation markers (mi and ma) precede either the lexical verb stem, the modal verb in a more complex construction, or the last auxiliary.[38]

 

References

 

Alrenga. Peter. 2010. Comparison of similarity and difference. In Patricia Cabredo Hofherr and Ora Matushansky (eds.), Adjectives: Formal analyses in syntax and semantics, Amsterdam, Philadelphia, PA: Benjamins, pp. 155-186.

Andersen, Paul Kent. 1983. Word order typology and comparative constructions. Amsterdam, Philadelphia, PA: Benjamins.

Beck, Sigrid. 2006. Positively Comparative. Manuscript. Tübingen.

Bielmeier, Roland. 1985. Das Märchen vom Prinzen Čobzaṅ. Eine tibetische Er­zählung aus Baltistan. Text, Übersetzung, Grammatik und westtibetisch ver­gleichendes Glossar. St. Augustin: VGH Wissenschaftsverlag.

Bochnak, M. Ryan and Elizabeth Bogal-Allbritten. 2015. Investigating gradable predicates, comparison, and degree constructions in underrepresented languages. In M. Ryan Bochnak and Lisa Matthewson (eds.), Methodologies in semantic fieldwork. New York: Oxford University Press, pp. 110-134

Bod.gžuŋ Šes.rig Las.khuŋs/ Department of Education, Central Tibetan Administration of His Holiness the Dalai Lama. 1994. ḥdzin.grwa drug.paḥi slob.deb ¦ rgyal.rabs.daŋ chos.­ḥbyuŋ ¦ History and religious history reader Part II. S.l. Šes.rig par.khaŋ [Educational Press].

Causemann, Margret. 1989. Dialekt und Erzählungen der Nangchenpas. Bonn: VGH Wissenschaftsverlag.

Cuzzolin, Pierluigi and Christian Lehmann. 2004. Comparison and gradation. In Geert Booij, Christian Lehmann, Joachim Mugdan, and Stavros Skopeteas (eds.), Morphologie. Halbband 2. Berlin, N.Y.: Mouton de Gruyter, pp. 1212-1220.

DeLancey, Scott. 1982. Lhasa Tibetan: a case study in ergative typology. Journal of Linguistic Research 2.1: 21-31.

DeLancey, Scott. 1984. Etymological notes on Tibeto-Burman case particles. Linguistics of the Tibeto-Burman Area 8.l: 59-77.

Denwood, P. 1999. Tibetan. Amsterdam, Philadelphia, PA: Benjamins.

Dixon, R.M.W. 2008. Comparative constructions. A cross-linguistic typology. Studies in Language 32.4: 787-817.

Dixon, R.M.W. 2012. Basic linguistic theory: further grammatical topics. Vol. III. Oxford: Oxford University Press.

Francke, August Hermann. 1901. Sketch of Ladakhi grammar. Journal of the Asiatic Society of Bengal 70: 1-63.

Grierson George Abraham (ed.). 1909. Linguistic survey of India. Vol. III: Ti­beto-Burman family, Part I: General introduction, specimens of the Tibetan dialects, and the North Assam group. Calcutta: Superintendent of Government Printing. Reprint 1967, Delhi etc.: Motilal Banarsidass.

Grimm, Jakob und Wilhelm Grimm. 1854-1961. Deutsches Wörterbuch. Leipzig, Stuttgart: Hirzel. Online-Version vom 11.04.2015. http://dwb.uni-trier.de/de/die-digitale-version/online-version/

Hahn, Michael. 1996. Lehrbuch der klassischen tibetischen Schriftsprache. 7. Aufl. Bonn: Indica et Tibetica.

Haller, Felix. 2000. Dialekt und Erzählungen von Shigatse. Bonn: VGH Wissenschaftsverlag.

Haller, Felix. 2004. Dialekt und Erzählungen von Themchen. Bonn: VGH Wissenschaftsverlag.

Häsler, Katrin Louise. 1999. A grammar of the Tibetan Dege (Sde dge) dialect. Zürich: Selbstverlag.

Haspelmath, Martin. 1995. The converb as a cross-linguistically valid category. In: Martin Haspelmath and Ekkehard König (eds.), Converbs in cross-linguistic perspective. Berlin, N.Y.: Mouton de Gruyter, pp. 1-55.

Herrmann, Silke. 1989. Erzählungen und Dialekt von Diṅri. Bonn: VGH-Wissenschaftsverlag.

Hu Tan. 1989. Comparative sentences in Tibetan. Acta Orientalia Academiae Scientiarum Hungaricae 43.2-3: 399-408.

Huber, Brigitte. 2005. The Tibetan dialect of Lende (Kyirong). Bonn: VGH Wissenschaftsverlag.

Jacques, Guillaume. 2016. From ergative to comparee marker: multiple reanalyses and polyfunctionality. Diachronica 33.1: 1-30.

Jäschke, Heinrich August. 1881. A Tibetan-English dictionary. With special ref­erence to prevailing dialects. London. 4^th reprint 1992. Delhi: Motilal Ba­narsidass.

Kennedy, Christopher. 2009. Modes of comparison. In: M. Elliott, J. Kirby, O. Sawada, E. Staraki, and S. Yoon (eds.), The main session: proceedings from the main session of the forty-third annual meeting of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society, pp. 141-165.

Koshal, Sanyukta. 1979. Ladakhi grammar. Delhi etc.: Motilal Banarsidass.

Özsoy, A. Sumru & Hüner Kaşıkara this volume. Comparatives in Turkish Sign Language (TİD).

Pearson, Hazel. 2010. How to do comparison in a language without degrees: a semantics for the comparative in Fijian. In: M. Prinzhorn, V. Schmitt, and S. Zobel (eds.), Proceedings of Sinn und Bedeutung 14, Vienna: Universität Wien, Institut für Sprachwissenschaft, pp. 356-372.

Rangan, Krishnasamy. 1979. Purki grammar. Mysore: Central Institute of Indian Languages.

Read, Alfred Frank Charles. 1934. Balti grammar. London: The Royal Asiatic Society.

Roerich, Georges de. 1933. The Tibetan dialect of Lahul. Journal of the Urusvati Himalayan Research Institute 3: 83-189.

Sandman, Erika and Camille Simon. 2016. Tibetan as a “model language” in the Amdo Sprachbund: evidence from Salar and Wutun. Journal of South Asian Languages and Linguistics 3.1: 85-122.

Saxena, Anju. In preparation. A linguistic sketch of Nako Kinnauri (book chapter).

Simon, Walter. 1940. Certain Tibetan suffixes and their combinations. Harvard Journal of Asiatic Studies 5.3/4: 372-391.

Sprigg, Richard Keith. 2002. Balti-English English-Balti dictionary. London, NY: Rout­ledge­Curzon.

Szemeréni, Oswald. 1990. Einführung in die vergleichende Sprachwissenschaft. 4. Aufl. Darmstadt: Wissenschaftliche Buchgesellschaft.

Stassen, Leon. 1984. The comparative compared. Journal of Semantics 3: 143-182.

Stassen, Leon. 2013. Comparative Constructions. In: Matthew S. Dryer and Martin Haspelmath (eds.), The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. Available online at: http://wals.info/chapter/121.

The American Heritage. 1996. The American Heritage Book of English Usage: A Practical and Authoritative Guide to Contemporary English. Boston: Houghton Mifflin.

The American Heritage. 2005. The American Heritage Guide to Contemporary Usage and Style. Boston: Houghton Mifflin.

Tournadre, Nicolas. 1995. Tibetan ergativity and the trajectory model. In: Yoshio Nishi, James A. Matisoff and Yasuhiko Nagano (eds.), New Horizons in Tibeto-Burman Morphosyntax. Senri Ethnological Studies 41: 261-275. [Originally in: Hajime Kitamura, Tatsuo Nishida and Yasuhiko Nagano, eds., Current Issues in Sino-Tibetan Linguistics. Osaka 1994: The Organising Committee, pp. 637-648.]

Tournadre, Nicolas. 2014. The Tibetic languages and their classification. In: T. Owen-Smith and N.W. Hill (eds.), Trans-Himalayan linguistics. Historical and descriptive linguistics of the Himalayan area. Berlin, Boston: De Gruyter, pp. 105-129.

Tournadre, Nicolas and Sangda Dorje. 1998. Manuel de tibétain standard, langue et civilisation. Bod.kyi spyi.skad slob.deb [Textbook of colloquial Tibetan]. Paris: Langues & Mondes/L’Asiathèque.

Zeisler, Bettina. 2006. The Tibetan understanding of karman: Some problems of Tibetan case marking. In: Christopher I. Beckwith (ed.), Medieval Tibeto-Burman languages II, PIATS 2003: Tibetan studies: Proceedings of the Tenth Seminar of the International Association for Tibetan Studies, Oxford 2003. Leiden etc.: Brill, pp. 57-101.

Zeisler, Bettina. 2007. Case patterns and pattern variation in Ladakhi: a field report. In: R. Bielmeier and F. Haller (eds.), Linguistics of the Himalayas and beyond. Berlin etc.: Mouton de Gruyter, pp. 399-425.

Zeisler, Bettina. 2009. Mainstream linguistics for minor(ity) languages? Or: What is it like to speak Ladakhi? In: Anju Saxena and ÅkeViberg (eds.), Multilingualism. Proceedings of the 23rd Scandinavian Conference of Linguistics. Uppsala: Uppsala Universitet, pp. 305-319.

Zeisler, Bettina. 2011. Kenhat, the dialects of Upper Ladakh and Zanskar. In: Mark Turin and Bettina Zeisler (eds.), Himalayan languages and linguistics: stud­ies in phonology, semantics, morphology and syntax. Leiden etc.: Brill, pp. 235-301.

Zeisler, Bettina. 2012. Practical issues of pragmatic case marking variations in the Kenhat varieties of Ladakh. Linguistics of the Tibeto-Burman Area 35.1: 75-106.

Zeisler, Bettina. 2016. Context! Or how to read thoughts in a foreign language. Journal of South Asian Languages and Linguistics 3.2: 197-221.

Zemp, Marius. 2013. A historical grammar of the Tibetan dialect spoken in Kargil (Purik). PhD-thesis, University of Berne. [Now enlarged as A grammar of Purik Tibetan. Leiden, Bostom 2018: Brill. DOI: 10.1163/9789004366312.