Printer Friendly

tt in Old Khotanese.


Khotanese orthography and double tt, gg, ss, ss

In the orthographic system of the majority of Old Khotanese texts there are four cases where a double consonant graph is used for a single phoneme, tt=/tl, gg=/gl, ss=/s/, ss-/s/. This doubling avoided ambiguity as it allowed the singly written consonant graphs to have dilferent values, t=/d/, g=/y/, s-lzl, s-lzl. This may be called the classical Old Khotanese orthography. Some documents survive in an older orthographic system first noted by E. Leumann in his 1920 edition of the Eurangamasamadhisutra, where he remarked that, "die Laute g s s nur selten doppelt geschrieben werden" (Neb: 92). This oldest attested spelling system may be called archaic. In it the three graphs mentioned by Leumann are ambiguous, with g=/g, y/, s'=/s, z/, s=/s, z/. But tt for /t/ is already in the oldest Khotanese documents.

The archaic system is amply attested. It is featured in manuscripts 1 to 5 of the Sanghatasutra. In his edition of that text G. Canevascini provides a useful discussion of the relationship between the archaic and classical systems (not his terms). Importantly, he shows that the paleographic age of the script of a particular document parallels the age of its orthographic system. This confirms that the archaic precedes the classical (Sgh: xv-xvii). A unique folio from The Book of Zambasta (Z) found in Sorcuq near Qarasahr, T III S 16, also features old script and archaic orthography. Using L. Sander's paleographic criteria, M. Maggi recognized that the folio is "the oldest witness of Z and altogether one of the earliest Khotanese manuscripts" (Maggi 2004: 184b). (1) No edition of the folio has been published,

Documents in archaic orthography also show the most conservative grammar. Leumann noted that in the Eurangamasamadhisutra the nominative singular masculine (NSm) of a-stems was always -a, and the genitive-dative singular (GDS) was always -i (Neb: 92). In later texts these are confused. Similarly, he noted that the third singular present indicative active ending -ata is kept distinct from middle -ate. R. E. Emmerick later confirmed and expanded these findings in his edition of the text (Sgs: xix-xxi). T III S 16 also shows this conservative morphology, consistently using NSm -a for the a-declension, e.g., vairocana (= vairocani 13.10) and sravaka-yana (= sravaka-yani 13.16), as well as NSm -ei twice for the aa-declension, an archaic feature noted by Emmerick, buddhavalamtsei (= buddhavalamtsai 13.13) and barei (= barai 13.13). (2)

While the oldest attested Khotanese orthography, the archaic, does not use double gg, ss', or ss, it does use double tt. The device was introduced in an earlier, unattested period of orthographic development. The three other double consonant graph spellings entered the system after the archaic period and are distinctive features of the classical system. They were likely designed on the model of tt, which makes determining the origin of tt more important.

Writing Khotanese dental obstruents

Ernst Leumann seems to have been the first to correctly note the functions of single t and double tt in Khotanese orthography. Single t is mostly /d/ except in some clusters like st where it is It/. He thought this was due to sound change within Khotanese wherein intervocalic voiceless consonants became voiced:
   so hat t zwischen Vokalen den Wert von d bekommen, ist aber doch,
   weil bekanntlich die Orthographie leicht hinter der Lautentwicklung
   zuruckbleibt, weiterhin als t geschrieben worden, worauf man denn,
   um von dieser Pseudo-Tenuis die wirkliche zu unterscheiden, fur
   letztere die Verdoppelung des Zeichens einfuhrte. (1912: 38)

That is, he proposed that after t had become /d/ intervocalically, tt was introduced to distinguish the voiceless phoneme /t/ from the "pseudo" or orthographic voiceless one.

The use of tt for [t] is easily seen in Khotanese in loan words. The Indian word ttathaggatta-(Z 5.103) from Skt tathagata shows tt for single [f] initially and intervocalically. Leumann also recognized that the signs for Sanskrit voiced stops, b, d, g, could stand for voiced fricatives /[??], [??], [??]/ (his symbols b, d, g; 1912: 42).

These values have essentially been followed by all scholars. In his last statement on the value of the dental graphs, Sten Konow in "A Primer of Khotanese Saka" wrote,
   Of the dentals t is a voiceless stop only after s and s; ...
   Intervocalic t was voiced, i. e. d. Elsewhere tt is written for the
   voiceless dental stop.... D was the voiced dental stop after
   consonants and apparently before r.... Elsewhere d was a spirant,
   English th in "thou." (1949: 15)

Konow deviated slightly from the Leumann values with his view that dr was /dr/ rather than /[??]r/. This is not a large difference, as there is no contrast between /d/ and /[??]/ before /r/. It may not be possible to choose between those values for Old Khotanese but it is also not critical to do so.

Gercenberg (1965: 47-51) had tt=[t] and d=[[??]] but t=[[??]], probably because he thought [d] was written dd. For the latter value he gives two examples, both Indian loans but each having a different explanation. The first, ssadda- 'faith', reflects Skt sraddha- and simply had double /dd/ as shown from its metrical use in the The Book of Zambasta (Z) where the first syllable is always heavy. In the examples below the word occurs as H(eavy)L(ight) in cadence 1 HLLHL, first as HLLHL and then as HLLHL:

13.160ab padamjsi I hettu bodhi-citta [] u vayslnamjsiya ssadda . (A:5+12+7) (3)

23.370ab ttana ma pratalbimbai vlri ml kye ma udisa I ssadde jsa yande (A:5+7+5+7)

The stem is also spelled ssada- with single -d- six times in Z, a practice I call "hidden double d." The practice is clear from the Khotanese spelling of what is Suddhodana in Sanskrit, which occurs eight times as ssadutana-, where the meter proves that singly written -d- stands for [dd] in every instance (Hitch 2014: 34-35).

The other loan, reflecting Skt mudita- 'joy', appears once in Z with dd as NSf mudditta 10.14 but also once with d as ASf muditto 15.124. In both cases the metrical count is LLL, precluding an internal consonant cluster. (4) Further, the related form pramuddtto occurs twice with single d (10.12, 16.60). Emmerick was certainly correct to observe, "No doubt the doubling of d in mudditta was an attempt to represent [d] unambiguously at a time when d had become [8] between vowels, but it was a sporadic effort" (1981: 187). Gercenberg's first example has dd=/dd/ while the second is an effort to write [d] between vowels in a loan word where the Skt spelling with d was likely known. Neither example proves Gercenberg's value for Khotanese.

Although Emmerick 1981 (185-88) essentially follows the values of Leumann, he only refers to Konow and Gercenberg. His description for the motivation for the introduction of tt may be seen as a paraphrase of that given above from Leuman: "tt was invented to represent [t] because t had developed in intervocalic position to [d]" (1981: 186). Also like Leumann, but not like Konow, he regarded dr as [Sr] (1981: 188).

Emmerick 1989 (213-14) continues these values. His chart of initial consonants that were retained appears to contain a confusing error. In the line "/t/ /d/ It'/ spelled tt d th" ( the /d/ should likely be replaced by /8/ as shown by his phonemic transcription "drainu /Srainu/ 'of three'" two paragraphs later ( He does not specifically discuss values in clusters or between vowels but he uses /8/ for intervocalic orthographic d in the transcription "padajs- /pa[??]adz/ 'to burn'" (

E. Leumann in 1912, Gercenberg, Konow, and Emmerick agree on tt=[t] and, between vowels, t=[d]. All but Gercenberg agree on r/=[S] in initial and intervocalic position (except for Konow's dr-=[dr-]). The consensus view on these segments implies that there was a shift in the values of the Brahmi symbols relative to Sanskrit: Skt i=[t] > Kh t=[d], and Skt d-[d] > Kh d=[8], This shift made it necessary to adopt or develop a new device for Kh [t].

Earlier ideas on tt

Until now perhaps the most accepted explanation for the introduction of the tt-device has been from Oskar von Hiniiber. In "Die Paisaci und die Entstehung der sakischen Orthographie" (1981) he outlined the evolution through Middle Indie of the intervocalic stops originally written with t, d, and k, g. Although this is not his formulation, the sense is that the segments weaken over time along the hierarchies [t > d > [??] > y > [??]] and [k > g > [??] > y > [??]]. As additional proof he pointed out that g, d, and t in Khotanese can represent lyl, 181, and /d/. He concluded, likely correctly, that this use "kann die Lautwerte des indischen Alphabets in Indien selbst zur Zeit seiner Ubemahme widerspiegeln" (p. 124). About the "-device in Khotanese he suggested the cases of double tt writing in Prakrit could have served as a model (p. 125).

Emmerick 1981 (186-87) proposed a contrasting idea:
   tt representing the result of the secondary contact of two OIr.
   dentals, e.g. butte 'he knows' < *budatai, pitta 'he falls' <
   *patati, may not be identical with tt from OIr. *t since tt in such
   cases makes position: butte can occur in spondaic but not in iambic
   endings. The tt in butte etc. may have been a long consonant [t:]
   rather than a full double consonant [tt] and thus not greatly
   different from initial [t] phonetically. Since this combination
   became common in Khotanese it is likely that the idea of writing tt
   for [t] after intervocalic [t] had become [d] was suggested by the
   fact that tt [t:] resulting from secondary contact was phonetically
   similar to initial [t].

Emmerick's proposal was based on two kinds of evidence, historical and metrical. As far as I can tell, he did not elaborate on this idea for the origin of the tt device. He may later have deferred to von Hiniiber. In his 1989 description of the phonology Emmerick employed phonemic transcriptions rather than the phonetic transcriptions of 1981 and now transcribed /tt/ in those words: pitta /pi:tte/ 'he falls', butte /butte/ 'he knows', hamuttu /hamette/ 'he changes (intr)'. He made no comment about the origin of the tt-device, and it is possible that he had abandoned his earlier phonetic theory, but he still regarded intervocalic << as ambiguous. He gave historical derivations to suggest the ambiguity but did not mention the metrical evidence (1989: 216 [section]

The view proposed in the concluding pages of this paper is partly a blend of the ideas of von Hiniiber and Emmerick, which are complementary rather than mutually exclusive. The Middle Indie model provided VttV=fVttV/, which was initially applied by the Khotanese to their /VttV/. From that use it spread to writing single III.

My interest in this problem arose from studying the synchronic morphophonology of Khotanese present stems. In the early 1990s I came to the conclusion that stems traditionally listed as ending in *-d- or *-h- actually ended in a vowel in the synchronic grammar. This development was due to a morphophonological reanalysis or metanalysis that also led to the development of suffixes beginning with /tt/. Briefly, a word like 3Sp.m butte 'he knows', historically *bud- + -te, was reanalyzed to bu- + -tte. It was essential for this theory to find other evidence that butte contains double /tt/. I noticed that the meter in the Book of Zambas ta contained that evidence, but had overlooked or forgotten Emmerick's prior observation about butte occurring in spondaic endings. I had further overlooked, as apparently had Emmerick, the fact that a more elaborate version of that observation was first proposed already at least eighty years ago. From etymology and the study of Khotanese meter, Ernst and Manu Leumann maintained that " was sometimes /tt/.

The Leumanns on tt

Ernst Leumann's lifework on Khotanese was directed towards interpretation of the long Khotanese Buddhist poem that he called "E" but today is known as The Book of Zambasta (Emmerick 1968a). He did not live long enough to finish an edition. From his father's Nachlass, Manu published essentially his father's edition adding an introduction and a glossary (E. and M. Leumann 1933-36). Where it is not possible to tell whether father or son is the source of a particular idea in that edition, they are jointly credited here.

In the introduction to the Leumanns' edition of the poem we find the theory about tt and meter. It is expressed in few words and perhaps partly for this reason it seems to have been passed over in silence. After discussing the use of two consonant letters to write one phoneme, including tt for /t/, they noted (p. xxxii),
   Abweichend hiervon stehen tt s's gelegentlich als
   Doppelkonsonanten, sie bilden dann also nach kurzen Vokalen
   Position. Im Druck der Ausgabe (nicht im Glossar) werden diese
   doppelwertigen tt ss durch ein kleines Spatium zwischen den beiden
   t oder s gekennzeichnet (utta-, bis' s'o); nach kurzem Vokal deuten
   sie also auf Langmessung der Silbe. Solche tt stehen einerseits in
   sanskritisch-prakritischen Lehnwortem mit indischem tt wie citta
   Uttaravata utta-ruva; andrerseits handelt es sich um etymologischen
   durch Synkope enstandenen Doppeldental, so besonders in der 3. sg.
   der Dentalverben: but te - w aus budatai, dit te, nach Vokallange
   but te hot te samkhilt te, oder aktive pit ta rrvit ta sait ta pv'
   ait ta; und im ppp. der Labialverben: aksut te, nach Lange
   suhautta- dukhotta-; auch in Komparativen wie battaru ysittara-.

It is worthwhile unpacking this compact statement and rigorously testing its various claims. The Leumanns identified four categories of vocabulary where VttV may show /VttV/ (the role of ss will not be examined here):

1. Sanskrit and Prakrit loan words containing orthographic tt where it stands for double [tt] in the original language, as in citta (Skt citta), Uttaravata (Skt Uttarapatha), utta-ruva (Skt Uttarakuru (5));

2. What they describe as an etymologically arising double dental especially in the 3S of "dental verbs," as in butte, ditte, butte, hotte, samkhiltte, pitta, rrvitta, saitta, pv'aittu;

3. In the past stem, ppp, of "labial verbs," as in aksutte, suhautta-, dukhotta-',

4. In tara-comparatives like battaru, ysittara-.

Within these four categories they saw metrical evidence that the first syllable in orthographic -VttV "made position" (was metrically long) and thus could be syllabified /-Vt-tV/. Their expressions "dental verbs" and "labial verbs" refer to the consonants that end the present stem in Old Iranian, or, in the case of the loan words suhev-:suhautta- and dukhev:dukhotta-, apparently to the originally foreign consonant ending the present stem in Khotanese (for dukhev- cf. BS duhkhapayati, SGS: 46).

Except for the one sentence in Emmerick 1981 involving category 2, the Leumanns' alltoo-brief observation appears to have subsequently been neither confirmed nor challenged. There may be several reasons why. The silence may be partly because the topic of Old Khotanese meter has been contentious (see Hitch 2014: 2-3). Scholars dubious of the Leumanns' metrical analyses might think that they invoked double /tt/ in some words just to make their metrical theory work better. Another reason may be because the majority of the examples in the quotation above are not testable. The words butte, hotte, pitta, rvitta, saitta, pv'aitta, suhautta-, dukhotta-, ysittara- contain a long vowel before tt, which would make metrical position regardless of whether tt stands for one or two consonants. The word samkhiltte has a consonant cluster, either /lt/ or /ltt/, both of which would make position with the short Other than the three loanword examples, only butte and ditte in category 2, aksutte in 3, and battaru in 4 can be tested metrically. The Leumanns' inclusion of the untestable examples alongside the testable ones is distracting rather than helpful.

A further reason why the Leumanns' theory has not been investigated may be because Emmerick's edition of the Book ofZamabasta has superseded theirs. For most purposes it is not necessary to refer to the earlier edition, although the Glossar is still essential. The Leumanns' brief comment can be easily overlooked. I only noticed it relatively recently.


It is possible to test the Leumanns' four categories where they thought tt stood for double /tt/ through the moraic structure of Z. In doing this, it is useful to expand the categories to their limits. The aim is to find every possible word that may contain an orthographic short vowel followed by tt and test the sequence VttV against the meter. If the syllable with the short vowel is heavy (H) this suggests tt=/tt/; if the syllable is light (L) this suggests tt=/t/. The analytical metrical apparatus is that developed in Hitch 2014.

Category 1: Sanskrit and Prakrit loanwords

The Leumanns' first category of forms may be the easiest to examine and may contain the clearest evidence for tt-/tt/. An orthographic -tt- reflecting a Sanskrit or Prakrit double -tt- does appear from metrical position in Z to show /tt/. Here are included also any cases of orthographic -tt- that correspond to a Sanskrit consonant cluster, which in practice appears to involve only -tt-, -pt-, and -kt-. The words I have noticed in Z that fit this description, with Sanskrit correspondants, are devadatta- < Devadatta, anuvatatta- < Anavatapta, andvatta- < anapatti, anutpattia- < anutpattika-(dharma-ksanti), bodhi-citta- < bodhicitta, upaggutta- < Upagupta, uttara-pamcala- < Uttarapaacala, patavutta- < praty-upta, uttaruvu < Uttarakuru, uttaravata- < Uttarapatha, nimittana < nimittajna, nimitta- < nimitta, citta-matra- < cittamatra, maitra-catta- < maitracitta, vyatta-6 < vyakta, citta- < citta. There are fifty-seven examples. These are examined below by meter type, A, B, or C, and then by further criteria. Where the example occurs in a clearly definable cadence, the cadence type, following Hitch 2014: 5-10, is given and the position in the cadence is noted. For example, the bold face in HLLHL shows that -VttV occurs at the end of a cadence 1, while HLLHL shows that it comes at the beginning of a cadence 1. Where the example occurs outside a clearly definable cadence, it is examined by mora counting. This is most convincing where the examples occur in five-mora segments, but is also useful where segment boundaries are not observed and hemistich totals can instead be counted. A maximum of three examples of each is given here, but all cases listed in the footnotes have been examined.

Meter A

Cadence 1, HLLHL (23 cases (7)):

4.8cd anamdu | u vajrrapanu [??] datai upalggutta vikalpo (5+7+5+7)

4.119cd ttana bodhi-lcitta utara [??] I ggambhira I masta aggamjsa (5+7+5+7)

23.111ab muho varata | mastu aretu [??] | yade devaldatta jadina (5+7+5+7)

Cadence 3, LHLHL 1x

2.12a kye kamala paltavutta tslndi [??] I k[y]t va jala I bulysa kye mulysga (5+7+5+7)

Five-mora segment (5 (8) cases):

2.51ab ka vyattu (9) | sarvani iya [??] | ttu butte I sa ulatana (5+7+5+7)

13.68ab devadatltl vata sana ? I sei sravaka-lyano tta hvinde (5+7+O+7)

13.108cd ssei uttalruvi hva'nda |0 ha[??]rstaya | achai nista . (5+7+5+7)

Segment boundary not observed and mora count with -Vt-tV makes 24 moras (11 cases (10)):

1. 82ab tt[a]na sanana | uttara-pamca[??]la ttraste I sarvani balysa (5+12+7)

4.120cd ttana bodhi-|citta upevani [??] maste | mulsde jsa arsta (5+12+7)

11.41ab bassa satva | viri maitra-ca|tta ksamolttati masta. | cu [??] (5+12+7)

Segment boundary not observed and mora count with -Vt-tV is below 24 (2 cases):

11.57ab ku bodhi-citta ne ttusse jsate [] kho hve' (11) | bumata ssura . (16+7)

11.58ab ku bodhi-citta pata'jsi ka[??]rmyau hve' | pitta avaya (15+7)

Irregular meter and mora count with -Vt-tV is above 24 (2 cases):

13.50ab anutpattiye | ksande a[??]stanna (12) | nau ro ssai | handare ksandi (7+7+5+7)

13.75ac ci svarna-|paksa-raya ham[??]jsasde anuvatattu | nagu u[??]sthamji (5+18+7)

Meter B

Cadence [B.sub.1], HL LLL, HL HL (2 cases)

12.64ab tti ana|vatta [??] ku hve' | masta ce heda . (5+6+7)

12.83cd samu ana|vatta [??] kui (13) jsa | ttartha khijinda (5+6+7)

Cadence [B.sub.1], HL HL, HL LH (5 cases (14))

16.38ab tcuratasa | diva [??] utta|ruva tcoho[ra] (5+6+7)

24.226ab kvi hajva | hva'nda [??] namat|tana (15) ditanda . (5+6+7)

24.269ab bassa ssu va | ttiya [??] utta|ravata trande . (5+6+7)

Irregular meter and mora count is short even with reading Vt-tV (1 case):

12.73cd nimitta | hvanate ggulna [??] cira yinda (4+6+7)

Meter C

Cadenee [C.sub.1], HLLL (2 cases)

7.40ab cu ro va brata [??] | sei' bodhi-|catta raysa (7+5+5)

12.109cd ttara [??] buljsa-jsera a[??]na|vatta ttana (12+5)

Cadence 3, LHLHL (1 case)

7.36a abhava catta [??] | nairatma | dharma ttana (7+5+5)

Cadence 3, LHLHL (1 case)

24.327c sa uttaruva [??] | sarbanda | seitta hama (7+5+5)

Segment boundary not observed but mora count requires -Vt-tV to make 17 moras (1 case):

8.34cd ttatvatu hara nasta [??] citta-[??]matra samu (12+5)

Summary of Indian loan word evidence

The evidence is solidly in favour of tt in these words standing for double /tt/. Of the fifty-seven examples fifty-two clearly show Vt-tV. Of these, thirty-six occur in a cadence needing HL, and a further sixteen require a heavy syllable to make up a regular mora count. There are three examples where a segment boundary is not observed but even with counting HL the mora count for the hemistich is low. The only place one might find counterexamples would be with the two examples from type A meter where the count is high. 13.50ab anutpattiye is seven moras where one expects a five-mora segment. This is probably an example of a metrically awkward foreign word (see Hitch 2014: 14-15). There are six moras too many in 13.75ac. According to Mauro Maggi, "the hemistich is probably to be restored as ci svarna | hamjsasde anuvatattu | nagu u[??]sthamji (5+12+7), with -paksa-raya- being an interpolation due to the common combination Skt. suparni-paksi-raja- in Divyavadana etc." (p.c.). Maggi's restoration of 13.75ac to a normal 5+12+7 count is convincing. There are thus no counterexamples. The Leumanns' first category of forms is valid.

Category 2: 3S of "dental verbs" like butte, ditte

This category has the largest number of forms that can be tested in Z. The Leumanns' definition of the category can be improved and expanded.

In his description of verb morphology in Saka Grammatical Studies, Emmerick identified four types of verbs by their endings in the 3Sp.a: "A. -iti; B. -ti', C. -aitta; D. -aiya" (p. 192). Type B arose as the unaccented thematic vowel was lost and "the -t- of the 3 sg. came into secondary contact with the stem final" (ibid.). Emmerick prepared a chart showing the historical derivations of the resulting consonant clusters. Of interest to us are those Old Khotanese forms of this origin that show orthographic -tt-. These correspond to the "dental verbs" of Leumann. The historical categories and the OKh representatives Emmerick listed are
*-t-t-    pitta
*-t-t-    hotte
*-d-t-    rrvitta
*-d-t-    butte
*-rt-t-   galtte
*-rd-t-  *spaltte
*-nd-t-  *khimtta
*-nd-t-   nvamtte
*-[??]-t- hamutta (SGS:193 and n. 1).

The stems historically ending in clusters *-rt- (galtte), *-rd- (*spaltte), *-nd- (*khimttu, nvamtte) show reflexes with clusters that would make position in any case, so they cannot be examined metrically for evidence of the value of -tt-. The stems historically ending in *-Vt-(pitta, hotte), *-Vd- (rrvitta, butte), and *-V[??]- (hamatta) show Old Khotanese reflexes with -tt- between vowels and so may offer forms to examine against the meter of Z. In traditional grammatical descriptions these groups are given with synchronic stems ending in -Vt- (pat-, hot-), -Vd- (rrud-, bud-), and -Vh- (hamah-). The -Vt- rendering is synchronically accurate; the -Vd- and -Vh- renderings are not. Skjaervo also recognized this, stating, "no stems end in h or d" (SuvII: 229). In my view, as mentioned above, all of these stems end synchronically in a vowel. Where I list vowel stems below in "Vowel stems with tt-initial suffixes," for reference I also give the more traditional consonant-final shapes from Emmerick (SGS) or Bailey (DKS) and, where available, stem shapes suggested by Skjaervo (Suv).

The aim here is to compile a list of forms from these verbs that can be examined in the meter of Z for /tt/. Leumann identified the third singular suffixes (3S) and listed present indicative active (p.a) and middle (p.m) examples. But there are also forms showing -VttV- in the second plural (2P), and in the imperative (iv) and injunctive (ij) moods. The chart below is intended to contain all attested forms (with just one spelling for each). For convenience, I retain Emmerick's type C as a distinct type. (16) The suffixes are all <<-initial. The rows for 2Pp.m and 2Piv.m are included for completeness but there are no <<-initial suffixes attested for these categories.
Conjugated present stems showing -VttV-.

Class    Ending       Type B                     Type C

3Sp.a    -i tta       datta < da-                khaitta < kha-
                      natta < na-                ttahvaitta < ttahva-
                      pajatta < paja-            * [pa]hai[tta] < paha-
                      pajutta < paju-            pvai' tta < pusa-
                      pitta < pat-               baitta < ba-
                      bitte < bi-                bvai' tta < busa-
                      rrvitta < rru              saitta < sa-
                      hambruitta < hambru-       hvaitta < hva-
3Sp.m    -tte         nuvatte < nuva-            --
                      butte < bu-
                      butte < bu-
                      hamatte < hama-
                      hamatta < hama-
                      hotte < hot-
2Pp.a    -tta         patta < pat-               --
2Pp.m    (-iru)       --                         --
3Siv.a   -tto, -ttu   (jindo)                    pva' ttu < pusa-
3Siv.m   -tto         hautto < hot-              --
2Piv.a   -tta         natta < na-                pahatta < paha-
                      hambitta < hambi-          puva' tta < pusa-
2Piv.m   --           --                         --
3Sij.a   -tta         (jinda)                    --
3Sij.m   -tta         butta < bu-hautta < hot-   --

Stems in -Vt- /-Vd-/ with tt-initial suffixes

There are three stems in OKh ending in -Vt-: pat-, hot-, and vavat- 'fall down'. The last is not attested with a tt-initial suffix. The short list below is intended to contain all known cases of -Vt- + -tt-. Synchronic phonological derivations in // follow the bullet *.

pat- 'fall'

3Sp.a pitta Z 2.29 + 9x, Suv 1.24, Sgs 3.12r3, SI P65.2v3, pi[tta] Sgh 99[7] * /bitte < bad- + -itte/

2Pp.a patta Z 23.95 * /batta < bad- + -tta/

hot- 'be able'

3Sp.m hotte Z 12.19, 13.35, 14.78, 23.38, 126, h[otte] Suv 16.21, [h]au[tt]e Suv 17.51 * /hotte < hod- + -tte/

3Siv.m hautto Neb 141.8 [=IOL Khot 153/4rl hauttio; read 'he can' by Bailey DKS: 498a] * /hotto < hod- + -tto/

3Sij.m hautta Z 24.437, Suv 18.99, Sgh 92.3 * /hotta < hod- + -tta/

Only one form from this group, 2Pp.a patta, has the shape -VttV- and can be examined in the meter of Z. The others feature a long vowel before -tt- which makes the syllable in question heavy.

Vowel stems with tt-initial suffixes

The list below is intended to present all known instances of OKh vowel stems attested with a tt-initial suffix. Included for reference are stem shapes and definitions from Emmerick (SGS), Bailey (DKS), and Skjaervo (Suv), and one definition from Canevascini (Sgh). kha-: SGS khad- 'wound, hurt' C, DKS s.v. khasta- 'beaten, pained'.

3Sp.a khaitta IOL Khot 28/5 * /khaitte < kha- + -itte/ (17)

ttahva-: 'cross; penetrate', SGS ttdhvah- 'cross' C, DKS ttahvah- 'tread over, cross'.

3Sp.a ttdhvaitta Z 13.21, 22 * /tehwaitte < tehwa- + -itte/

da-: SGS did- 'appear' B, Suv diy- 'appear', DKS seems to not distinguish this from da-'see' and has forms from both under dai-: di- 'see'.

3Sp.m datte Sgs 2.3v3, 3.12r4, Z 1.38 +9x, ditte Z 1.37 +5x, dattu Z 9.27, ditta Z 6.6 * /[??]ette < [??]e- + -tte/

na-: SGS nad- 'sit down' B, DKS nasa'd- 'sit', Suv nay- 'sit'.

3Sp.a natta Z 13.28, 22.215, Suv 10.9, 40, 12.10 * /nette < ne- + -itte/ 2Piv.a natta Rk 154.12 * /netta < ne- + -tta/

nuva-: SGS nuvad- 'lie down' B, DKS nuvad- 'lie down'. (18) 3Sp.m nuvatte is regular.

3Sp.m nuvatte Z 4.72 * /nuwatte < nuwa- + -tte/

paja-: SGS pajad- 'ask for' B, DKS pajad- 'beg, ask for, demand'.

3Sp.a pajatta Z 12.42 * /bajette < baje- + -itte/

paju-: SGS pajud- 'conceal' B, DKS pajud- 'cover'.

3Sp.m pajutta Z 2.28 * /bajutte < baju- + -itte/

paha-: SGS pahad- 'to strike' (C assumed), DKS s.v. khasta- 'beaten, pained', Sgh pihad-'to strike, fell (tree)'.

3Sp.a *[pa]hai[tta] Sgh 31.6 * /behaitte < beha- + -itte/ 2Piv.a pahatta Z 24.414 * /behatta < beha- + -tta/

pusa-: SGS puva'd- 'fear' C, Suv puvay'- (s.v. puva'na-, pvay-), DKS puvgd-, pvgd-. That IVzV > VV/ is a synchronic process and that the Brahmi subscript hook (transliterated by apostrophe) shows that a word contains a sequence of two vowels have been briefly described in Hitch 2014: 23-27. In Z a sequence /CuV/ may sometimes be read as if /CwV/. For example, pvai'tta 2.101 is HL while 4.113 is LHL (as if *puvai'tta). (19)

3Sp.a pvai'tta Z 2.101, 4.113, 115, 7.29, 9.19, 12.130, 24.428; puvai'tta Sgh 99.5, 207.4(2x) * /(bwaitte <) buaitte < buzaitte < buza- + -itte/

3Siv.apva'ttu Z 2.101 * /(bwattu <) buattu < buzattu < buza- + -ttu/

2Piv.apuva'tta Z 24.474, IOL Khot 17/13, *puva'tta' Sgh 199[1], *puva'tta Sgh 199[5] (both Sgh forms restored or read with single -t- by Bailey KT3 77, by Canevascini Sgh[7], and by Skjaervo IOL Khot 170/4, but double -tt- is expected; has incorrect image) * /(bwatta <) buatta < buzatta < buza- + -tta/

ba-: SGS bad- 'be bound' C, DKS bad- s.v. ban-'bind', baitta 'binds'.

3Sp.a baitta Sgh[16, 24) 79.9, Z 5.55, 57, 12.94, Vim 8.9c, 8.1 Id (Skjaervo transcribes bainta in Vim 8.1 Id), baitta IOL Kh 8/7 r3 = baitti Or. 6402B/1.2 r3 (as noted by Skjaervo), [baitti IOL Kh 148/2 r1, r4, r7 probably belongs here but cf. DKS: 303a with two definitions, 'longs' and binds'.] * /[??]aitte < [??}a- + -itte/

bi-: SGS bid- 'pierce' B, DKS bid- 'throw, shoot, pierce'. See also hambi- 'pierce'. 3Sp.a bitte Z 2.138 * /[??]itte < [??]i- + -itte/

bu-: SGS bud- 'perceive, know' B, Suv buv-, bv- 'know, perceive, realize (bodhi)', DKS buv-, bv- 'be aware, know, awake, smell [sic]'.

3Sp.m butte Z 2.3 +56x, Sgs 3.8r3, 13r3, Sgh 39.2, 202.2, 243[35], 253[87], SI P4.11r4, 12v1, 14r4, 20r4 * /[??]utte < [??]u- + -tte/

3Sij.m butta Z 2.25, 117, 8.36, Sgh 18.3, 137.2 * /[??]utta < [??]u- + -tta/

busa-: SGS buva'd- 'mount, ride' C, Suv buva'y- (also bva'y-, buva'd-) 'mount, enter', DKS buvad- 'mount'.

3Sp.a bvai'tta Z 14.92, bvaitta Z 22.150 * /([??]wa'tte <) [??]uaitte < [??]uzaitte < [??]uza- + -itte/

bu-: SGS bud- 'be fragrant' B, DKS butte 'give scent, smell'.

3Sp.m butte Z 3.57, 85, 20.3, Or. 12637/57.2a3 * /[??]utte < [??]c- + -tte/

rru-: SGS rrud- 'grow (intr)' B, DKS rruv-, rrv- 'grow'. Short -u- in the stem is shown by the diphthongization in the 3Sp.a u + i > vi /ui/.

3Sp.a rrvitta Z 15.4, 22.117, 125 2x, 126 * /ruitte < ru + itte/

sa-: SGS 'appear, seem' C, DKS sad- 'appear, seem', Suv say- 'appear'.

3Sp.a saitta Z 1.35 + 50x, Sgh 91.1, Suv 18.166, IOL Khot 155/3 v2, 149/2 v4 (sai[tta]), 17/1 r4, seitta Z 4.83, 5.38, 15.6, 22.240, 24.327. IOL Khot 173/8 b2 * /saitte < sa- + -itte/

hambi-: SGS hambid- 'pierce' B, DKS hambitta 'pierce'. Here given as active on the basis of the relationship with bi- 'pierce'.

2Piv.a hambitta Z 24.414 * /hambitta < hambi- + -tta/

hambru-: SGS hambrud- 'heal (intr)' B, DKS hambruitta 'grows together, joins up', Sgh hambrud- 'heal'.

3Sp.a hambruitta Z 22.128, Sgh 39.1, 2 * /hambruitte < hambru- + -itte/

hama-: 'sober up' not in SGS, DKS 'be freed from intoxication'. Related to ma-, SGS mad-'be intoxicated'.

3Sp.m hamatte Z 2.167 * /hamatte < hama- + tte/

hama-: SGS hamah- 'change (intr.)' B, Suv hamiy- 'change', DKS hamih- 'change' and hamy- 'change' "in plural."

3Sp.m hamutta Z 15.93, 22.131, hamatti SI P65.1r2 * /hamette < hame- + tte/

hva-: SGS hvah- 'strike' C, Suv *hvay- 'beat', DKS hvah- s.v. hvaitta 'he beats'.

3Sp.a hvaitta Z 2.16, IOL Khot 155/5 * /hwaitte < hwa- + -itte/

From the list above there is no point in examining the metrical distribution of forms with a long vowel before the tt like ttahvaitta, butte, or rrvitta, since the long vowel already makes for a heavy syllable whether or not it is closed by a consonant. We need to examine the forms in the list with penultimate short vowel -VttV that occur in Z. These are 3Sp.m datte 1.38 + 9x, ditte 1.37 +5x, datta 9.27, ditta 6.6, 3Sp.a natta 13.28, 22.215, nuvatte 4.72, 3Sp.a pajatta 12.42, 3Sp.a pajutta 2.28, pahatta 24.414, 3Siv.apva'ttu 2.101, 2Piv.apuva'tta 24.474, 3Sp.a bitte 2.138, 3Sp.m butte 2.3 +56x, 3Sij.m butta 2.25, 117, 8.36, 2Piv.a hambitta 24.414, 3Sp.m hamatte 2.167, 3Sp.m hamatta 15.93, 22.131. Add to these the t-stem form 2Pp.a patta 23.95 mentioned above and we have ninety-one instances of -VttV to examine with the meter of Z.

All hemistichs containing one or more of these instances have been analyzed. The resulting patterns are organized in the same fashion as the loanword hemistichs above. At least one example of each conjugated form is given, which means the first group has four examples, but otherwise a maximum of three examples is given for each group.

Meter A

Seven-mora segments. All structures 5+7+5+7 unless noted.

Cadence 1, HLLHL. (36x (20))

2.101cd ttai hvana | badr ma pva'ttu [] | avulatu | ggamdyo bada

2.167ab samu kho hulsandi baysenda [] | o (21) masta | mau n<a> hamatte

4.72ab samu kho hve' | ciya nuvatte [] | vanautai | ttamdr na samna

15.93ab mulsde jsa na | huna hamatta [] | svida alksuta hamana

Cadence 1, HLLHL (7x (22))

2.220cd cera hamata | butte dukhe[]va | uysano | satva-hataya .

23.9ab sa ju na ssaru | butte na diru [] | kalyana-lmattro ni butte

23.95cd haysa va | avitsara ttita [] | ciya pusso | patta avaya

Cadence 2, HXHL (5x (23))

5.67cd padamjsi | karma vivata [] | sati prata|bimbai ditte

6.54cd data paslkalna hamamgga . [] | biss'u avai|vartti butte

13.28ab kho ju ysalyslno (24) dalysu [] I hve' banite | bendi natta . (4+7+5+7)

Cadence 3, LHLHL (3x)

2.28cd ce tta hvate | ggamtsu ye kamggindi [] u ysalysanai pajutta (4+12+7)

2.138ab aysu hasta | manama jau[]ysa | kyeri halci | purnyau bitte

2.232ab tvi balysi | aysmya butte [] | khano va | dyanite ttiya (khano LH, ASm khanaa- 'smile, laugh')

Cadence 4, LLLLHL (2x)

1.37ab satvanu | kadana asta badi [] ku ssai | ttusuvo' (25) ditte (5+12+7)

2.25cd cu va ne ko | sate bissu bu[]tta | ttye klaisa | bissu jita aro

Five-mora segments requiring three moras (7x (26)):

2.3cd bissu butte | sarvani ban[]lysa | bisye ysamalssandai ttrani

6.50cd samu nama-|matrai hvinde [] | ne ju butte | busta ne bvaka

11.24ab cu ttu butte | sei mama tcera . [] | sei' ma harlstaya ne tceru .

Segment boundary not observed but mora count requires -Vt-tV to make 24 moras (3x):

6.56ab aggunau | butte samahanu [] ttlya | rrasta tcohora (5+12+7)

22.148ab ttamdvi yi | butte se ma ju [] ssando (27) skaulyate payau jsa (5+12+7)

22.215cd sata ggampha | vasta parsa natta [] patana | mattrau balysu (5+12+7)

Irregular meter and mora count with -Vt-tV is below 24 (1x):

4.70cd sate gyadi | cu ditte . []| o saitta lo cu ye pyusde (5+4+5+7; three moras missing)

Irregular meter and mora count with -Vt-tV is above 24 (1x):

10.33ab ssuramggamu | butte samahanu [] u vaj|rropamu rrasu (6+12+7; note three long words of foreign origin)

Meter B

The structure is noted where it deviates from the ideal 5+6+7 mora segments. Note that 24.414cd has two target forms so is listed twice.

Seven-mora segments:

Cadence 1, HLLHL (2x):

12.65cd ysurre jsa o | hisani [] dildete ku butte

24.414cd halahala | hoda na [] ham|bitta pahatta (4+6+7)

Cadence 1, HLLHL (6x (28)):

12.42cd kvi asta | haysda [] cai pa|jatta ne heda

14.29ab tramu hama | balysi [] cu ttera (29) | datte vicattre

24.414cd halahala | hoda na [] hamlbitta pahatta (4+6+7)

Cadence [B.sub.1] HL LLL (lx):

14.71cd samu tramu | datte [] kho purra | hambada sslve

Five-mora segments, LLHL (2x):

14.68ab [ne ne] butte | balysa [] havyo | hotu gyadina

24.474cd ma ne puva'tta (30) | nyau ju [] ye yudu | ylnda dukhauttu

Irregular meter but mora count with -Vt-tV in butte is below 18 (2x):

12.82ab ku bodhisatva ttu [] butte | ttye tta saitta . (11+6)

Meter C

The ideal or most common count is 7+5+5 with cadence [C.sub.1] as the last segment. 7+5+6 with cadence [C.sub.3] also occurs (there is no case with [C.sub.2]). The seventeen-mora count with [C.sub.1] is exceeded only once (17.32ab), but even in that exception the position of the target may ensure HL.

Five- or six-mora segments at the end of the hemistich:

Cadence C, HL LL (7x (31))

7.15cd eu kraya sa trama [] | kho ju ca'ya | ditte samu . (7+5+5)

8.36cd varju na byode | ce va ju [] ttate | butta hara (7+5+5)

15.123ab crrama vasutana [] | atina | datte teara . (7+5+5)

Cadence [C.sub.3] HL LLL (1x):

8.50ab cude sa ttye arru [] kyai ne | butte hamata . (12+6)

Seven-mora segments:

Cadence 1 HLLHL (1x):

8.42ab huna ne butte [] | sata ttussai | huna samu . (7+5+5)

Cadence 3 LHLHL (2x):

2.105cd kho rahu butte . [] | o blmalcatri aysuri (7+5+6; ends in [C.sub.3])

18.18ab ssanomu butte [] | kalyana-lmatr kade (7+5+5)

Cadence 4 LLLLHL (2x):

3.143ab tcamana ye butte [] | thatau ballysustu bisso (7+5+5)

24.322c kho purra ne datte [] | ne hade ttana | hama jata (7+5+5)

Mora count with -Vt-tV is above 17 (1x):

17.32ab bitanda ne butte [] | kho ye dukhyau | parsta pusso (8+5+5; possible counterexample but butte is at end of segment in HL position)

Summary of Category 2, "dental verbs" with -VttV.

Of ninety-one cases, seventy-five have the sequence -VttV counting as HL in a cadence. A further nine cases occur in a five-mora segment where the sequence counts as three moras, not two. Another two instances require a three-mora reading to make the correct hemistich mora count. Three instances have a short hemistich mora count even with a three-mora reading and cannot serve as counter evidence. There are only two cases where a three-mora count pushes the hemistich count too high. 10.33ab has three long words of foreign origin so metrical awkwardness is expected (see Hitch 2014: 14-15). 17.32ab is the only possible counterexample but butte occurs at the end of a seven-mora segment where we find HL in all cadence types. The evidence is essentially one hundred percent in favour of the -VttV sequence in these words showing HL. This should prove beyond doubt that the spelling -tt- in these forms denotes double /tt/. The Leumanns (and Emmerick) were correct about this category.

Category 3: Past stems ending in -Vtta-

The third grammatical category mentioned by the Leumanns involves the past stems (ppp) of what they call "labial verbs": "und im ppp. der Labialverben: aksutte, nach Lange suhautta- dukhotta-" (p. xxxii). As mentioned above, their phrase "labial verbs" refers to the historical final consonant of the present stem. This is a loose description even from a historical point of view as some "labial verbs" do not have a ppp in Khotanese with tt, such as ttav-:ttauda- 'be hot' < *tap- (SGS: 38) or byeh-:byauda 'obtain (tr)' < *abi-afya- < *abi-ap- (SGS: 106). From a synchronic point of view, this category may be described as past stems ending in -Vtta-.

In SGS Emmerick listed twenty past stems ending orthographically in -Vtta- (p. 174). Of these twenty, only aksutta- has a short vowel before tt and so can be used to test for metrical position. It occurs six times in Z in aksutte 'he began' (2.171 aksuttai = aksutte+-i 'his') and once as aksuttanda 'they began', aksutte counts as four moras and aksuttanda as six, which itself strongly suggests that the syllable with -u- is L. Further, the five times aksutte occurs in a cadence, it has the value HLL:

Meter A

Cadence 1, aksutte HLLHL

2.159cd vaisramanu | aksutte pulsu []| kama ma na | ttatvatu balysa . (5+7+5+7)

2.169cd pacadanau | aksutte bussa []| ssarvi spatu | bis'ye bilsangi (5+7+6+7)

5.26c van va | aksutte brema []| kho ju ye vib|randau daiya (5+7+5+7)

23.107b edu jsi|danda sta rrunda []| rasayyo jsa | aksutte kusa (5+7+5+7)

Five-mora segment, aksuttai HLH

2.171cd kade saggo|ravana aysmu[]na | aksuttai | buljse (32) hvani (5+7+5+7)

Meter B

Cadence 1, aksutte HLLHL

24.234cd namasatai | ttere jsa [] po' (33) | aksutte brema . (5+6+7)

Five-mora segment, aksuttanda HHL|HL

24.220ab tti aksulttanda [] pajsama | kadana ysanaja (5+6+7)

Summary of Category 3, the ppp of "labial verbs"

The one ppp of the "labial verb" type suggested by the Leumanns that may be tested, aksutta-, has clear metrical patterns, aksutte occurs five times as HLL in cadence 1 HLLHL. The two five-mora segments, aksuttai and tti aksu, show that the u-syllable is light. The conclusion seems unavoidable that the past stems with -Vtta- had just single /t/ and that the Leumanns were incorrect to assume double /tt/ here. (34)

Category 4: turd-comparatives, tama-superlatives, and tati-nouns

It is puzzling why the Leumanns suggested that /tt/ occurs "auch in Komparativen wie bat tara ysit tara-" (p. xxxii), but did not suggest it occurs in the related superlatives like battamu ASm 'least'. Both comparatives and superlatives in OKh exhibit suffixes that are reflexes of OIr suffixes beginning with *-t-: the comparative *-tara- and the superlative *-tama-. There is in addition a third common OKh suffix seen in words like bumattati- 'strength', ttadati- 'darkness', and rastatati- 'rightness, truth' that reflects another OIr *-t-initial suffix, the de-adjectival noun suffix *-tdti- (KhSuf: [section]45.2.1). These three suffixes may exhibit similar morphophonological behavior in OKh. It is useful to examine them together when describing the variety of attested shapes and looking for evidence of double /tt/.

Historically the three suffixes beginning *-tV- attached directly to the stem. But synchronically we also have the same suffixes now showing the orthographic forms -atV-, -atV-, -ttV-, -attV-, and -attV- after the stem. A summary explanation for this variety is given below (see "Synchrony, diachrony, /t/ and /tt/"). The focus in this section is on those forms that orthographically show a short vowel before -ttV- (-VttV-) and occur in Z. Their metrical behavior may be examined to determine if they contain double /tt/.

In Z, in tara-comparatives, tama-superlatives, and tati-nouns, seven stems show orthographic -VttV-: battama- 'least' 1x, battara- 'less' 3x, bumattati- 'strength' 1x, ggamattara-'faster' 1x, tcarsvattati- 'splendor' 1x, vasutattati- 'purity' 3x, haj(u)vattati- 'wisdom' 21x. All of these forms are examined below in their hemistich with the exception that only four representatives of haj(u)vattati- are illustrated.

Meter A

Seven-mora segments

Cadence 1, HLLHL (1x)

23.40cd cvi ttisa | tcarsvattata (35) [] | tto ye ce yudu | tindi padandu (5+7+5+7)

Five-mora segments requiring target syllable to be L (12x (36))

9.5cd ttramu ava|ssarsta ya[]na | samu vasuta|ttete vassesa (A:5+7+5+7)

9.6bd tra[]ma avalssarsta yana [] | samu vasuta|ttete vissesa (A:5+7+5+7)

11,24cd tta tta pare|haui paraha []| sa hajva|ttata dyana (A:5+7+5+7)

23.101cd bissa pata'ni | rruyata rra[]su | biss'o buma|ttetu dahinu (A:5+7+5+7)

Segment boundary not observed but mora count requires single /t/ to make 24 moras (5x37):

1.45ab [da]tu hvate | budaru mahauyanu battaru | sravaka-yanu (A:5+12+7)

11.42bd si [] jana | masta hajvattata [] cu sam|tsari puva'sta (A:5+12+7)

Meter B

Six-mora segments

Cadence [B.sub.1], HL LLL (1x)

14.73cd kye va gratu | pyusde [] hajva|ttete (38) padamgyo (B:5+6+7)

Cadence [B.sub.1], HL LLL (2x)

14.87cd ssai rro brah|mana ba[]ttaru | hota kho balysa (B:5+6+7)

19.82cd raksaysyo | mulysda. [] battaru | aysmya striye (B:5+6+7)

Five-mora segment (1x)

24.228ab kyai battamu | bvate [] bistu | hastamu datu (B:5+6+7)

Meter C

Five-mora segment (1x)

18.45 anacca hista [] | ggamattaru | skaugye vate (7+5+5)

Segment boundary not observed but mora count requires -VtV to make 17 moras (6x (39)):

3.93ab datena (40) vasutatte[]ta busse jsa | skutana bissa (C:12+5)

8.44ab ttramu hajuvattata [] kho dai | ciya hamate . (C:12+5)

Summary of Category 4: tarn-comparatives, tama-superlatives, and tati-nouns

The evidence of the twenty-nine examples is convincingly against there being /tt/ in these words. The syllable before the -tt- is always light. The target occurs four times in a cadence. The seven-mora segment tcarsvattata constitutes a cadence 1 HLLHL. There are two examples of battaru and one of haj[u]va- at the end of cadence [B.sub.1] HL LLL. Fourteen examples occur in five-mora segments probably because almost all cases contain the sequence LLL and that sequence does not fit into most cadences. Only tcarsvattata and bumattetu do not contain LLL. The eleven cases where the segment boundary is not observed require the syllable in question to be L to make the correct count. There are no hemistichs where a value H would be preferred or even acceptable. There is no evidence suggesting any forms have double /tt/. The Leumanns were incorrect about this category.

Evaluating the Leumanns on tt

The Leumanns' brief and overlooked statement from 1933 that orthographic tt sometimes stood for /tt/ partly holds up under close scrutiny. Two of their four categories of forms with tt, those with the most examples, clearly show that orthographic -VttV sometimes "made position" (was metrically long) and thus was in those cases syllabified /-Vt*tV/.

The Leumanns' category 1 involves the Sanskrit and Prakrit loan words where tt denotes a cluster in the original language. Of the fifty-seven examples, fifty-two clearly show Vt*tV: thirty-six occur in a cadence needing HL, and a further sixteen require a heavy syllable to make up a regular mora count. The three cases with short mora counts cannot be counterexamples. The two places with long counts could be counterexamples, but other explanations are apparent. The evidence is essentially completely in favor of tt here standing for /tt/.

Category 2 the Leumanns called the 3S of "dental verbs." This category was extended here to its maximum range by including all present stems ending in -Vt- /-Vd-/ (the Leumanns' hotte and pitta), all type B present stems ending in a vowel (the Leumann's butte, ditte, etc.), and all suffixes beginning with -tt-. Of the ninety-one examples with orthographic -VttV in Z, seventy-five occur in metrical positions where the -VttV is clearly scanned HL confirming that the string is syllabified /-Vt*tV/ and that orthographic tt in these instances stands for double /tt/. Eleven examples require a three-mora reading to make up the count. Three short hemistichs cannot be counterexamples. The two long hemistichs have clear explanations. Again the evidence is essentially one hundred percent in favor of tt=/tt/.

The Leumanns' category 3, "ppp. der Labialverben," may be described synchronically as past stems ending in -Vtta-. Although at least twenty past stems fit this description, only aksutta- has a short vowel before tt and so can be used to test for metrical weight. Of the seven cases in Z, aksutte occurs five times as HLLHL in cadence 1. The other two cases are in a five-mora segment where the -u- syllable is counted L. It is clear that -Vtta- in past stems contained single /t/. The Leumann's were incorrect about this category.

The Leumanns' last category, 4, originally concerned the tara-comparatives but was here expanded to include the tama-comparatives and the tati-denominal nouns, because all three derivational suffixes should display similar morphophonological behaviour. All twenty-nine examples show that the target syllable is L and thus in these words -VttV- always contains single /t/. The Leumanns were also incorrect about this category of forms.

Although the Leumanns mistakenly supposed that their categories 3 and 4 of forms with orthographic tt contained double /tt/, they were correct about categories 1 and 2. The metrical evidence from the Book of Zambasta provides solid support for the fact that Old Khotanese orthographic tt was ambiguous, not only standing for single /t/ but also sometimes for double /tt/.


An explanation, diachronic or synchronic, is not obvious for the disparate phonological developments in the forms above. A consonant in contact with *t produced double /tt/ in category 2 (butte, patta) but single /t/ in categories 3 {aksutta-) and 4 {battara-). An in-depth discussion might be out of place here but a couple of lines of inquiry may be suggested. One possibility is that the contact occurred at different times in the evolution of the language. Category 2 contact could be more recent than the others. This idea is likely correct for the past participle aksutta- of category 3, which is an inherited simplex like most past stems. In most cases the relationship between present and past stems is historical and relatively old. Compare daj-:dista- 'ripen (intr)' puls-':brasta- 'ask', bud-:busta- 'know', panam-:panata- 'rise', all showing a reflex of the historical *-t- in the past stem but none showing a synchronic phonological relationship between the past and present stems. There clearly exists a synchronic suffix deriving a past stem from a present stem, but this is -ata-, not *-ta-. The suffix -ata- is used with borrowed words like car-:carata- 'practice', namas-:namasata- 'worship', and with synchronically derived causatives like handajan-:handajanata- 'ripen (tr)' (< handaj'- be ripened'), dyan-:dyanata- 'make appear' (< da- 'appear').

This first idea is however perhaps not correct with category 4. Three of the words containing reflexes of *-tara-, *-tama-, and *-tati- examined in Z show -tt- < *-t- + *-t-, plausibly synchronic /-t- < -d- + -t-/:

battama- 'smallest' < bata- 'small' + -ttama- * /[??]atam- < [??]ad- + -tam-/

battara- 'smaller' < bata- 'small' + -ttara- * /[??]atar- < [??]ad- + -tar-/

bumattati- 'strength' < bumata- 'strong' + -ttati- * /[??]umatad- < [??]umad- + -tad-/

The second line of inquiry involves the stem + suffix metanalysis that saw the development of a set of present stem suffixes that synchronically begin with /tt/. Pre-Khotanese 3Sp.m *bud-te became Old Khotanese bu-tte, *pahad-ta became paha-tta (2Piv.a pahatta Z 24.414), etc. This suffixal double /tt/ may have been preserved because it was morphologically important. In contrast, there was less need for distinct consonantism in the category 4 suffixes. Those suffixes evolved a variety of forms with regard to their historically initial *-t-. Following a consonant cluster, we find an epenthetic vowel and voicing of the suffix initial, -CC- + *-tV- > -CCVtV-. All forms below are from Degener, KhSuf.

tsastatati- 'peace' < tsasta- 'peaceful' * /tsastadad- < tsast- + -adad-/

balondatara- 'stronger' < balonda- 'strong' * /[??]alondedai- < [??]alond- + -edai/

The epenthetic vowel was reanalyzed as part of the suffix so that in OKh -atV- and -atV- stood beside the -tV- forms as alternates:

kalyanatara- 'more beautiful' < Skt kalyana- * /kalyanadai- < kalyan- + -adai-/

bvamatiyatara- 'more knowledgeable' < bvamatiya- 'knowledgeable' * /[??]wamadiyedai- > [??]wamadiy- + -edai-/

ssaratati- 'goodness' < ssara- 'good' * /se.iadad- < sei- + -adad-/

Another reanalyzed type arose containing both epenthetic vowel and -tt-, -attV-, and -attV-, blending the /t/ from cases like tsattara- 'richer' < tsata- 'rich' with the vowel either introduced by epenthesis or through reanalysis as above.

garkhattati- 'respect' < ggarkha- 'heavy' * /gaikhetad- < gaikh- + -etad-/

ggamattara- 'faster' < ggama- 'swift' + -attara- * /gametar- < gam- + -etai-/

tcarsvattati- 'splendor' < tcarsua- 'shining' + -attati- * /tsaisuwatad- < tsaisu- + -atad-/

vasutattati- 'purity' < vasuta- 'pure' + -attati- * /wasudatad- < wasud- + -atad-/

hajuvattati- 'wisdom' < hajua- 'wise' + -attati- * /hajuwatad- < haju- + -atad-/

A final reanalyzed type, -ati- (<*tati-), abandons the original suffix-initial dental altogether:

avajsamati- 'dishonoring' < avajsama- 'dishonor' * /awadzamad- < awadzam- + -ad-/

asadati- 'wickedness' < asada- 'evil' * /ase[??]ad- < ase[??]- + -ad-/

skalsati-~skalsatati- 'arrogance' < skalsa- 'pride' * /skalzad- < skalz- + -ad-/

From an original *-tV-~suffix initial, OKh attests -tV-, -VtV-, -VttV-, -V-. At least one stem can show more than one type, vasuta- 'clean' shows -tV- in vasuttara- 'cleaner' but -VttV- in vasutattati- 'purity'. This variety demonstrates that the original suffix-initial consonantism was not essential in maintaining the morphological distinctiveness of the suffixes. Although a form like battama- plausibly has two dental segments early in the synchronic derivation, /bad- + -tam-/, this is resolved to one segment, /batam-/, on the surface without loss of semantic distinctiveness. In contrast, the double /tt/ in the various type B suffixes was apparently essential to the morphology and was therefore maintained. (41)


As mentioned above, in 1981 Oskar von Hinuber and R. E. Emmerick offered distinct ideas about the origin of the tt-device in Old Khotanese: von Hinuber proposed that the source was the isolated cases of double tt writing in Prakrit, while Emmerick suggested it was an internal development. Here, supported by the metrical evidence, the roles of both the Middle Indic model and internal Khotanese development are recognized.

Emmerick 1981 thought that words like butte and pitta, that is, the Leumanns' category 2, contained "a long consonant [t:] rather than a full double consonant [tt]" and because "tt [t:] resulting from secondary contact was phonetically similar to initial [t]" the tt was transferred from those cases to initial position (pp. 186-87). That idea can be improved in several ways.

An initial improvement is that the spelling tt was probably inherited directly from the Indian model for writing intervocalic /tt/. That is, by using principles in the parent orthography, butte was written with tt. This differs slightly from von Hinuber's suggestion that Middle Indic tt for double /tt/ led directly to Khotanese tt for single /t/.

Another improvement is to use phonemic analysis. We can simply say that tt was first adapted to writing voiceless double /tt/ and was later adapted to writing voiceless single /t/. There is no need to distinguish a long consonant from a double consonant. Emmerick later did write phonemic /tt/ in these words but does not mention whether he still regarded his earlier phonetic theory as valid (1989: 216 [section]

Two refinements result from the inclusion of evidence from the Indian loanwords listed above in the Leumann's category 1, where a tt that corresponds to a cluster in Sanskrit or Prakrit counts metrically as two consonants. On the one hand this confirms that the Middle Indie model had tt=/tt/ as suggested by von Hinuber. On the other hand, this increases the frequency of the use of tt for /tt / in the early stages of the adaptation of the writing system. Writers would have encountered many more examples of tt-/tt/.

A final difference has to do with the way the device developed within Khotanese. Emmerick thought it was first applied from /VttV/ directly to initial position. In contrast, here it is proposed that tt from /VttV/ was first applied to cases of intervocalic single /t/, then to tit in initial position, and finally to III in consonant clusters. Each step has a reasoned explanation.

The chart below and the discussion that follows sketch the evolution of the use of the graphs t and d from Middle Indic to Old Khotanese.
Evolution of d and t from Middle Indie to Old Khotanese


1   phonological values         V[??]V   VdV   VtV     VttV

2   M. Indian orthography       VdV      VtV   --      VttV

3   earlier Kh adaptations      "        "     * VtV   "

4   adaptation of tt to /VtV/   "        "     VttV    "

5   adaptation of tt to /tV-/   "        "     "       "

6   oldest attested Khotanese   VdV      VtV   VttV    VttV

                                      in clusters

1   phonological values         Cd/dC   Ct/tC

2   M. Indian orthography       Cd/dC   Ct/tC

3   earlier Kh adaptations        "             "

4   adaptation of tt to /VtV/     "             "

5   adaptation of tt to /tV-/     "             "

6   oldest attested Khotanese   Cd/dC   Ct/tC and Ctt/ttC

                                 initial before

1   phonological values         [??]V   dV    tV

2   M. Indian orthography       --      dV    tV

3   earlier Kh adaptations      dV      --    * tV

4   adaptation of tt to /VtV/   "       --    "

5   adaptation of tt to /tV-/   "       --    ttV

6   oldest attested Khotanese   dV      --    ttV

Row 1 lists phonological values of segments for both languages. These are organized by position in the word. The remaining rows show how the spelling of these segments evolved from Middle Indic usage to the oldest-attested Khotanese orthography.

Row 2 shows the spellings of the segments in the Middle Indic source orthography. There the intervocalic weakening as outlined by von Hinuber (discussed above) resulted in a shifting of the values of some graphs in intervocalic position relative to their original values. With the weakening, the spelling VdV was phonetically (42) [V[delta]V], VtV was [VdV], and there was no *[VtV]. As demonstrated by the loanwords in the Leumanns' category 1, the phonetic sequence [VttV] was likely relatively common in the Middle Indic source language. In clusters and in initial position the graphs d and t probably had the values [d] and [t] familiar from Sanskrit.

Row 3 contains an inferred stage in the evolution. This would have been a short, experimental period of time when the script was first borrowed for Khotanese with few modifications. Where the languages shared phones, the same spellings were naturally used. This includes writing initial /t-/ with single *t, a spelling that is not attested in Khotanese but was possibly used for a short time. In contrast, Middle Indic did not have phonetic [VtV], as older III had become voiced in that position. Here it is assumed that the script borrowers took the easiest route to writing this sound and at first simply used *VtV, although again there are no Khotanese documents attested with this spelling. At this point in the evolution both /VdV/ and /VtV/ would have been written ambiguously as VtV.

To write /VttV/, the Khotanese borrowed the spelling used in Middle Indic for the same sequence. This use may have been reinforced by the presumably Middle Indic principle that single t was voiceless in clusters with a voiceless consonant. /tt/ is also a consonant cluster.

Also in the third row there is innovation regarding voiced segments in initial position. Middle Indic had no initial *[[delta]V-], while Khotanese had no initial */dV-/. The absence of initial */dV-/ in Khotanese is disguised in the normalized transcription where it appears that some words begin with tV-*/dV-/, but these are actually all enclitics. The Leumanns' Glossar to Z, for example, lists tan- enclitic form of yan- 'make, do', -tarana- compound form of karana- 'deed', tatu~tavu~gavu 'time, while', tamu~gamu '-wise, -ways', and ta GDSm enclitic pronoun 'him' (p. 430). As there was no initial *ldV-l, the Khotanese adapted the spelling dV- to initial /[delta]V-/ without creating ambiguity. It is possible the adaptation was unconscious. That is, the adapters just used the sign for the foreign initial voiced dental obstruent for the native one. Also, as intervocalic d already stood for /[delta]/, now d stood for /[delta]/ everywhere but in clusters.

Row 4 shows the proposed origin of the tt device. In the stage represented by the third row there was ambiguity in the spelling *VtV, which stood for both [VdV] and [VtV], This ambiguity, it is proposed here, was the motivation for the development of the device. The ambiguity was resolved by transferring the use of double tt to also write single /t/between vowels. The method for representing double voiceless /tt/ between vowels, VttV=/VttV/, in both native words and words of foreign origin was transferred to show the voicelessness of single t between vowels, VttV=/VtV/. In practice this meant that the spelling VttV used in the Leumanns' categories 1 and 2 for /VttV/ was transferred to their categories 3 and 4 for /VtV/. (43) This adaptation at the same time created new ambiguity. The orthographic sequence VttV could now stand for either /VtV/ or /VttV/. Presumably this new ambiguity created less difficulty for readers than the one it replaced. It was never resolved in Old Khotanese.

Row 5 shows the adaptation of the tt device also to initial position before a vowel. This was a natural extension of the device. It made the system simpler. There was no difficulty in reading since there was no initial /tt/. At this point, except in clusters, every III was written with tt.

Row 6 in the chart shows the oldest-attested Khotanese orthography, the archaic, which was described in the introduction. Up to this point in the evolution, Khotanese consonant cluster writing with regard to /t/ and /d/ simply followed the Indian practice. But now that single /t/ was written tt everywhere else, the use of tt for /t/ spread, incompletely, into clusters as well. This gave rise to three kinds of spelling variation, initial tr-~ttr-, medial -tr-~-ttr~, and medial -nt-~-mtt-. Examples from Z:

Initial tr-~ttr-: trana-~ttrana- (3x~2x) 'stronghold', tram-~ttram- (23x~10x) 'to cross over; enter', trama-~ttrama-(ca. 170x~90x) 'such', tray-~ttray- (11x~3x) 'to rescue; deliver', tri-adhva-~ttri-adhva- (5x~1x) 'three times', tr-adhva-~ttr-adhva- (5x~11x) 'three times'.

Medial -tr-~-ttr- (all loans): ksetra-~ksettra- (5x~3x) 'field', ksatra-~ksattra- (7x~3x) 'umbrella', buddha-ksetra-~buddha-ksettra- (4x~2x) 'Buddha field', vicitra-~vicittra-(28x~18x, vacitra- 1x) 'various', vacatra-vacattra- (10x~2x, vicuttra- 1x) 'various', ttavatrisa-~ttavattrisa- (6x~2x) 'trayastrimsa-god', pattro-patro (1x~1x, pattaru 1x) 'bowl', sutra-~suttra- (33x~9x) 'textbook', asupatra~asapattra-vani (1x~1x) 'asipattras-Asipattravana', ggutra-~gguttra- (6x~2x, gguttara-1 ()x~gguttr 1x) 'gotra' vajjiputtru 1x 'Vrjiputra', naksattra-~naksatra- (1x~1x) 'lunar mansion'.

Medial -nt-~-mtt-: ananta-~anamtta- (8x~2x) 'endless', bihante-bihamtte (1x~1x) 'he smiled', branta-bramtta (1x~1x) 'confusion'.

The variation t~tt in clusters may be understood as a conflict between two principles, the principle that clusters be written in the Middle Indic fashion and the principle that every /t/ be written tt. Curiously, there is no instance of *VrtV in Z but there are twenty-nine cases of VrttV. (44) That is, /[??]t/ is always written with double tt. Similarly, there is no *VltV in Z but there are ten cases of VlttV. (45) In contrast, /st/ and /st/ are always written in Z with single t as st and st, never as *stt or *stt.

The metrical system of the Book of Zambasta is an important source of information on Old Khotanese morphophonology. It reveals that E. and M. Leumann about eighty years ago were in some cases correct to regard the spelling VttV as showing /VttV/. This finding in turn makes it possible to propose an origin for the tt-device which includes the ideas of von Hinuber and Emmerick. With regard to this device, the Old Khotanese orthography attests both a Middle Indic model and internal development.
Technical abbreviations:

ab, ac,        a hemistich may span two (ab, cd) or three (ac, bd)
bd, cd         graphic columns in the four column arrangment of the
               main manuscript of Z.

A, B           1) meter types A and B; 2) SGS present stem types A
               and B

C              1) any consonant; 2) meter type C; 3) SGS present
               stem type C

D              SGS present stem type D

V              any vowel, short, long, or diphthong

V              any short vowel

V              any long vowel or diphthong

1S, 2S, 3S     first, second, third person singular

1P, 2P, 3P     first, second, third person plural

p, iv, op,     present indicative, imperative, optative, subjunctive,
sj, ij         injunctive

a, m           active, middle

PPP            past passive participle, past stem          participle of necessity

p.stem         present stem

NS, AS, GDS,   nominative, accusative, genitive-dative,
IAS, LS        instrumental-ablative, locative singular

NAP, GDP,      nominative-accusative, genitive-dative,
IAP, LP        instrumental-ablative, locative plural

m, f           masculine, feminine

H, L           metrically Heavy, Light syllable

X              two-mora syllable in cadences 2, [B.sub.3] and
               [C.sub.5], filled only by contracted vowels or the IAP
               suffix -yau

[]             space between graphic columns

I              metrical segment boundary

II             metrical hemistich boundary

Bibliographic abbreviations:

DKS         Bailey 1979

Glossar     Glossar, in E. and M. Leuman 1933-36, pp. 385-530

IOL Khot    transcription (and translation) in Skjaervo 2002

KT3         Khotanese Texts III in Bailey 1969

KhSuf       Degener 1989

Neb         E. Leumann 1920

Rk          Skjaervo 2003, Maggi in press

Sgh         Canevascini 1993

SGS         Emmerick 1968a

Sgs         Emmerick 1970

SI          transcription and translation in Emmerick and
            Vorob'eva-Desjatovskaja 1995

Suv         Skjaervo 2004a and 2004b

SuvII       Skjaervo 2004b

Vim         Skjaervo 1986

Z           Emmerick 1968b


Bailey, Harold Walter. 1969. Indo-Scythian Studies Being Khotanese Texts I-III. 2nd ed. Cambridge: Cambridge Univ. Press.

--. 1979. Dictionary of Khotan Saka. Cambridge: Cambridge Univ. Press. Canevascini, Giotto. 1993. The Khotanese Sanghatasutra, A Critical Edition. Beitrage zur Iranistik, vol. 14. Wiesbaden: Reichert.

Cheung, Johnny. 2007. Etymological Dictionary of the Iranian Verb. Leiden Indo-European Etymological Dictionary Series, vol. 2. Leiden: Brill.

Degener, Almuth. 1989. Khotanische Suffixe. Alt- und Neu-Indische Studien, herausgegeben vom Institut fur Kultur und Geschichte Indiens und Tibets an der Universitat Hamburg, vol. 39. Wiesbaden: Reichert.

Emmerick, Ronald Eric. 1968a. Saka Grammatical Studies. London Oriental Series, vol. 20. London: Oxford Univ. Press.

--. 1968b. The Book of Zambasta, a Khotanese Poem on Buddhism. London Oriental Series, vol. 21. London: Oxford Univ. Press.

--. 1970. The Khotanese Eurahgamasamadhisutra. London Oriental Series, vol. 23. London: Oxford Univ. Press.

--. 1981. The Consonant Phonemes of Khotanese. In Monumentum Georg Morgenstierne, vol. 1, ed. Gilbert Lazard. Pp. 185-209. Acta Iranica, vol. 21. Leiden: Brill.

--. 1989. Khotanese and Tumshuqese. In Compendium Linguarum Iranicarum, ed. Rudiger Schmitt. Pp. 204-29. Wiesbaden: Reichert.

Emmerick, Ronald Eric, and Mauro Maggi. 1991. Thoughts on Khotanese e and o. In Corolla Iranica: Papers in Honour of Prof. Dr. David Neil MacKenzie on the Occasion of His 65th Birthday on the Eighth of April 1991, ed, Ronald E. Emmerick and Dieter Weber. Pp. 67-73. Frankfurt am Main: Peter Lang.

Emmerick, Ronald Eric, and Margarita I. Vorob'eva-Desjatovskaja. 1995. Saka Documents, Text, vol. III: The St. Petersburg Collections. Corpus Inscriptionum Iranicarum, pt. II: Inscriptions of the Seleucid and Parthian Periods and of Eastern Iran and Central Asia, vol. V: Saka. London: School of Oriental and African Studies.

Gercenberg, L. G. [??]. 1965. Xotano-Sakskij Jazyk [Xomano-Cakckuu ??]. Moscow: Nauka.

von Hinuber, Oskar. 1981. Die Paisaci und die Entstehung der Sakischen Orthographie. In Studien zum Jainismus und Buddhismus: Gedenkschrift fur L. Alsdorf ed. Klaus Bruhn and Albrecht Wezler. Pp. 121-27. Wiesbaden: Steiner.

--. 2001. Das altere Mittelindisch im Uberblick. Osterreichische Akademie der Wissen schaften, philosophisch-historische Klasse, Sitzungsberichte, vol. 467 = Veroffentlichungen der Kommission fur Sprachen und Kulturen Sudasiens, vol. 20. Vienna: Verlag der Osterreichischen Akademie der Wissenschaften.

Hitch, Doug. 1990. Old Khotanese Synchronic Umlaut. Indo-Iranian Journal 33: 177-98.

--. 2014. Meter in the Old Khotanese Book of Zambasta. Ars Metrica 2014/11: 1-43. http://

Konow, Sten. 1941. Khotansakische Grammatik mit Bibliographie, Lesestucken und Worterverzeichnis. Mit einer Schrifttafel. Leipzig: Harrassowitz.

Konow, Sten. 1949. Primer of Khotanese Saka: Grammatical Sketch, Chrestomathy, Vocabulary, Bibliography. Ed. G. Morgenstieme. Norsk Tidsskrift for Sprogvidenskap 15: 1-136.

Leumann, Ernst. 1912. Zur nordarischen Sprache und Literatur: Vorbemerkungen und vier Aufsatze mit Glossar. Strassburg: Deutsche Morgenlandische Gesellschaft. (Rpt. Nendeln, Liechtenstein: Kraus, 1966.)

Leumann, Ernst. 1920. Buddhistische Literatur: Nordarisch und deutsch, pt. 1: Nebenstucke. Abhandlungen fur die Kunde des Morgenlandes, vol. 15.2. Leipzig: Brockhaus. (Rpt. Nendeln, Liechtenstein: Kraus, 1966.)

Leumann, Ernst, and Manu Leumann. 1933-36. Das nordarische (sakische) Lehrgedicht des Buddhismus: Text und Ubersetzung von Ernst Leumann. Aus dem Nachlass herausgegeben von Manu Leumann. Abhandlungen fur die Kunde des Morgenlandes, vol. 20.1-3. Leipzig: Deutsche Morgenlandische Gesellschaft. (Rpt. Nendeln, Liechtenstein: Kraus, 1966.)

Maggi, Mauro. 2004. The Manuscript T III S 16: Its Importance for the History of Khotanese Literature. In Turfan Revisited: The First Century of Research into the Arts and Cultures of the Silk Road, ed. D. Durkin-Meisteremst et al. Pp. 184-90 and plate 36 p. 457. Berlin: Dietrich Reimer.

--. in press. A Folio of the Ratnakuta (Kasyapaparivarta) in Khotanese. (Draft of 11 August 2013.)

Skjaervo, Prods Oktor. 1986. Khotanese Fragments of the Vimalakirtinirdesasutra. In Kalyanamitraraganam: Essays in Honour of Nils Simonsson, ed. Eivind Kahrs. Pp. 229-60. Oslo: Norwegian Univ. Press.

--. 2002. Khotanese Manuscripts from Chinese Turkestan in the British Library, a Complete Catalogue with Texts and Translations. With Contributions by Ursula Sims-Williams. Corpus Inscriptionum Iranicarum, pt. II, vol. V: Saka, Texts VI. London: The British Library.

--. 2003. Fragments of the Ratnakuta-Sutra (Kasyapaparivarta) in Khotanese. In Religious Themes and Texts of Pre-Islamic Iran and Central Asia: Studies in Honour of Professor Gherardo Gnoli on the Occasion of His 65th Birthday of 6th December 2002, ed. Carlo G. Cereti, Mauro Maggi, and Elio Provasi. Pp. 409-20. Beitrage zur Iranistik, vol. 24. Wiesbaden: Reichert.

--. 2004a. This Most Excellent Shine of Gold, King of Kings of Sutras, the Khotanese Suvarnabhasottamasutra, vol. I: The Khotanese Text with English Translation and the Complete Sanskrit Text. Sources of Oriental Languages and Literatures, vol. 60. Central Asian Sources, vol. V. Cambridge, Mass.: Department of Near Eastern Languages and Civilizations, Harvard University.

--. 2004b. This Most Excellent Shine of Gold, King of Kings of Sutras, the Khotanese Suvarnabhasottamasutra, vol. II: Manuscripts, Commentary, Glossary, Indexes. Sources of Oriental Languages and Literatures, vol. 61. Central Asian Sources, vol. VI. Cambridge, Mass.: The Department of Near Eastern Languages and Civilizations, Harvard University.


I have seen the proofs for my article "Contracted Diphthongs in Old Khotanese" (Indo-Iranian Journal 58.4 [2015]). It introduces some improvements to the metrical tools such as recognizing meter types A1 (5+9+3+7) and C1 (9+3+5), and a cadence 1AHLLHHL. It demonstrates that the spellings vi and ui reflect a diphthong /ui/ and explains why. It shows that hajvattati- 'wisdom' and tcarsvattati- 'brilliance' contain a contracted diphthong /ua/ written va which counts as two moras in meter. These advances would simplify the presentation here, but not affect the conclusions.

Doug Hitch

Whitehorse, Canada

These pages have benefitted greatly from review by Mauro Maggi, Dieter Maue, Elan Dresher, and two anonymous reviewers for the Journal. I also thank Hiroshi Kumamoto for making available electronic versions of Khotanese texts and a word list, without which the studies here would not be possible. Phonemic transcriptions are essentially those used in Hitch 1990. Translations, transcriptions (including emendations in italics), and chapter and line numbers from The Book of Zambasta are from the edition by Emmerick 1968b, unless otherwise noted. The metrical apparatus used here, including abbreviations, symbols, and definitions of cadences, was developed in Hitch 2014.

(1.) Z appears to have had a long history within the Budhist literary tradition of Khotan. Besides the main manuscript, a number of variant fragments exist. Emmerick transliterated seventeen in his edition (Z: 424-36), and Skjaervo transliterated a large number of certain and possible fragments from Z in his catalog (listed at 2002: 609). but from the images provided by Maggi it is clear that it uses exclusively archaic spellings. For instance, akasagarbha (s, g) corresponds to akassaggarbha (ss, gg) in the main document of Z (13.9), and sava (s) to ssava (ss; 13.14).

Some are older than the main manuscript, some younger, and some of roughly the same age but from a different copyist stream. An older example is T III S 16 (discussed below). A younger fragment is Variant 4 featuring, for example, hira- 'thing' twice (8.15, 16), which is always spelled hara- in the main ms, and the Late Khotanese-influenced spelling hamye lya for main hamata iya 3S perfective optative masculine 'would be' (8.22). Variant 14 is perhaps the same age as the main ms but each has its own errors in marking the hemistich boundaries (see Hitch 2014: 10-12). Variant 14 has incorrect division at 22.109 and 117 where the main ms has it correct, and it has correct division at 108 and 110 where the main has it incorrect. Both have incorrect division at 22.119, but in different places.

(2.) Besides old script, orthography, and language, T III S 16 also appears to have a layout that is a precursor to the four-column layout of the main manuscript of Z (Hitch 2014: 12 n. 14).

(3.) 13.160=chapter and line number; ab=graphic columns occupied by the hemistich (a line has four columns, a-d); I=metrical segment boundary; [] = graphic column break; A = meter type (there are also B, C types); 5+12+7=moras per segment in the hemistich. 4. Both occur in a five-mora segment. The first counts LLLH and the second LLLLL: 10.14ac prabhamkaro I bhuma matra u []| mudditta salrayana drraya (A:5+7+5+7) 15.124ab ttatana hamlggarggana ttari []| muditto vate | ana vimana (A:5+7+5+7)

(5.) M. Leumann had utta-ruva- 'ukta-rupa' (Glossar: 401b). M. Maggi points out (p.c.) that the corresponding Skt is Uttarakuru, as already correctly read by Emmerick Z 13.108, 16.38.

(6.) Z 2.51 vyattu 'wisely' <Pkt<Skt vyakta 'wise, learned, clever' is listed under v[i]yattu in the Glossar.

(7.) 2.231ed, 4.8cd, 4.94ab, 4.119cd, 5.110cd, 11.lab, 11.49ab, 11.50cd , 11.51cd, 11.56cd, 11.58cd, 11.59ab, 11.60ab, 11.60cd, 13.41ab, 13.41cd, 13.42ab, 13.69ab, 13.72ab, 13.74ab, 13.75cd, Ib.10ab, 23.11 lab.

(8.) 2.51ab, 4.68ab, 11.52cd, 13.68ab 13.108cd.

(9.) The Leumanns probably correctly suggested v[i]yattu. It counts as four moras, probably LHL, and involves what I call "post-consonantal glide resolution" (Hitch 2014: 17-23).

(10.) All eleven cases feature a count 5+12+7 with the second segment boundary not observed: 1.82ab, 11.41ab, 4.120cd, 11.48ac, 11.49bd, 11.53cd, 11.69ab, 13.17cd, 13.68cd, 13.154ab, 13.160ab. A better metrical definition of these hemistichs could likely be made.

(11.) Whether spelled hve' (32x) or huve' (3x), this word is always LL (Hitch 2014: 25).

(12.) astanna is subject to the uysnora-effect, which is defined and illustrated in Hitch 2014: 15-17. Briefly, at the end of seven-mora cadences, a three-syllable word that is nominally HHL is read as if LHL. usthamji in the example that follows also undergoes this effect.

(13.) kui is more commonly spelled kvl in Z (35x) and is always two moras. The spelling kui occurs nine times in Z (2.85, 5.105, 12.46, 66, 83, 93, 111, 115, 24.379). Examples from five-mora segments in the three meter types:

2.85cd kailaysu | ggaru vate jsani [??]| kui brahmalkaya nimalaa (A:5+7+5+7)

12.46cd kui patco | naste . [??] hamate | trami patco (B:5+6+7)

12.93cd kade anarra tti[??]yi | kui tta mata | jyare pussu (C:7+5+5)

(14.) 16.38ab, 24.226ab, 24.269ab, 24.395cd, 24.399cd.

(15.) namuttana (cf. Skt nimittajna) contains what I call a "hidden palatal geminate" (Hitch 2014: 35-40) and counts LHHL.

(16.) Emmerick's type C verbs, when reanalyzed as vowel stems, should be regarded as type B. They all have active morphology, so there are blanks in the chart for the Type C middle.

(17.) The symbol traditionally shows that the suffix causes stem umlaut. For instance Emmerick lists the 2Sp.a ending -ia (SGS:192), which gives yana Z 24.43 from yan- do' and hvira from hvar- 'eat'. An extensive treatment of Old Khotanese umlaut is given in Hitch 1990. But the element behind the is more complex. This is reflected when an umlaut-inducing suffix is attached to a vowel stem ending in /-a-/. The result of /-a- + -i-/ is /ai/, not /i/ as with consonant stems. Similarly, with stems ending in /-u-/, an -i- produces a high-to-high diphthong /ui/, e.g., 3Sp.a rrvitta < rru- 'grow (intr)' * /ruitte < ru- + itte/. That the spelling vi or ui often represents this diphthong will be defended elsewhere.

(18). There is a second stem nu- meaning 'lie down' attested as 3Sp.m nutte Z 22.129 and 3Pp.m nuyare Z 24.168.

(19.) 2.101ab badr kade I dratai pvai'tta [] balysa alnanda tta parste . (A:5+7+5+7)

4.113cd kho ju brata I kara ne pvai'tta []l eu dukha date I huna nariya (A:5+7+5+7)

Similarly, the stem hva'nd- 'man' may be read /huand-/ or /hwand-/ (Hitch 2014: 25-26).

(20.) 1.38ab, 2.20cd, 2.26cd, 2.59ab, 2.101cd, 2.133cd, 2.161ab, 2.167ab, 2.212cd, 2.230cd, 2.238ab, 4.15bd, 4.28cd, 4.41cd, 4.72ab, 4.100ab, 5.60ab, 5.61cd, 5.63ab, 6.50ab, 9.17cd, 9.27cd, 10.22cd, 10.27ab, 10.31ab, 11.26cd, 11.46cd, 13.23ab, 13.42cd, 15.93ab, 22.129ab, 22.131cd, 22.272ab, 22.328ab, 23.9ab, 23.371ab.

(21.) o 'or', as noted by the Leumanns is always H. They transcribe o (Hitch 2014: 14).

(22.) 2.187cd, 2.220cd, 10.30ab, 13.18ab, 13.158cd, 23.9ab, 23.95cd.

(23.) 2.54ab, 5.67cd, 6.6cd, 6.54cd, 13.28ab.

(24.) The ASm -au, -o of inaa-denominal adjectives, as a long vowel resulting from contraction, should always be two moras. It is clearly H in the six-mora segments 14.78, 19.91, 92, and X in cadence 2 HXHL 5.42, 50, 82, 89, 11.72, 13.28, 14.70, 22.218, 239.

(25.) The LPm word meaning 'among the Tusitas' is spelled variously in Z as ttusuvo' 1.37, ttusato' 13.92, ttussato' 14.18, ttusatu' 24.182, ttusato' 24.186, all reflecting /tuse?uo < tuseduzo < tused- + -uzo/ which counts LLLL. In the example below, the seven-mora segment ttusatu' asta is also in cadence 4 LLLLHL:

24.182ab ttlyi sad|dhartha [] ttu skyatu | ttusatu' asta (B:5+6+7)

(26.) 2.3cd, 2.158cd, 6.50ab, 6.50cd, 11.24ab, 2.51ab, 13.107ab.

(27.) ssando ASf or LSf from ssandaa- 'earth' has been a metrical puzzle (cf. Emmerick and Maggi 1991: 69). The distribution matches that of the LAP suffix -yau when it is not followed by jsa. In the X position (see Hitch 2014: 6, 29-33) both the -o in ssando and the morpheme -yau without jsa count as two moras. But elsewhere, this -o and -yau without jsa count as one mora. The puzzle is better defined, but still a puzzle.

(28.) 12.42cd, 12.66cd, 14.29ab, 19.79cd, 24.199cd, 24.414cd.

(29.) itera 'so' occurs 56x and tteri 13x in Z, apparently always LL.

(30.) As mentioned above under pusa- 'fear', the sequence /CuV/ may be read /CwV/ in meter. Also, in this phrase, the ne may have been added later as ma puva'tta is sufficient for "Do not fear!" This hemistich does not provide evidence for a light reading of the target syllable.

(31.) 2.116cd, 2.117cd, 2.121ab, 7.15cd, 8.33cd, 8.36cd, 15.123ab.

(32.) buljse NAPf < buljsaa- 'virtue' is in cadence 2 HXHL.

(33.) The LP po' (12x) 'feet' is also spelled pau' (1x), pvo' (4x), and pato' (3x) in Z. All count as two moras. The apostrophe shows a sequence of two vowels so these spellings probably all scan LL (Hitch 2014: 26-27 n. 39).

(34.) I also examined past stems ending in -Vda-, which historically derive from present stems ending in *-Vr- with the suffix *-ta- (KhSuf: 224), for the possibility that they contain double /dd/. Nine stems with orthographic -Vda- occur in Z: aphar-: aphada-, aphida- 'be disturbed' 7x; aspar-:aspuda- 'tread' 2x; 1. bar-:buda- active 'carry' 1x; 2. bar-:buda- middle 'ride' 4x; mar-:muda- 'die' (and related adjectival forms) 16x; yan-:yada-, yuda'- make, do' yada- 57x, yida- 24x, yuda- 21x; haphar-:haphada-, haphada- 'be distracted' 3x; hayar-:hayada- rejoice, be happy' 2x; hvar-:hvada- 'consume' 4x. The evidence of these 141 cases overwhelmingly shows that the words contain single /d/ not double.

(35.) The word is given as tcars[u]vattata by the Leumanns, which is a correct emendation as shown by the eight cases in Z of the adjective tcarsua- 'splendid', which all show the high rounded vowel, e.g., NAP tcarsuva 22.124.

(36.) 2.197ab, 6.2cd, 9.5cd, 9.6bd, 11.21ab, 11.24cd, 11.26cd, 11.29cd, 11.31ab, 11.31cd, 11.32cd, 23.101cd.

(37.) 1.45ab, 2.224ab, 4.115ab, 11.39ab, 11.42cd. All are 5+12+7.

(38.) The Leumanns transcribe haj[u]vattette (p. 29, III.73), which is likely correct.

(39.) 2.119ab, 3.93ab, 8.43cd, 8.44ab, 8.45cd, 20.26ab. All are 12+5 or 12+6.

(40.) IASm dutena 'appearance' (7x in Z) for regular datana (11x, ditana 1x) is always three moras, presumably LLL. Oddly, many cases of each occur where HL is expected as in the examples below in seven-mora segments counting HLLLLL where HLLHL is expected:

22.165cd brrahmanu | rrijite datena [] | laksanai | dvavare dirsa . (A:5+7+5+7)

22.155cd sumara | tcarsuva datana [] | bisaa padya | dasta aggamjsa (A:5+7+5+7)

At least two other words with internal -cita- show a similar metrical distribution, hutanaa- 'red' (3x with -ata-) occurs in the seven-mora segment hatanai uri 22.144 LLXHL where cadence 2 HXHL might be expected. It is once spelled hatenai 4.33 and once henei Z 23.44. NSf butuva 'lightning' occurs 3x, once in the seven-mora segment ucai batava 1.87 HXLLL where cadence 2 HXHL might be expected. It is also spelled bateva- twice (and batava- 1x), once in the seven-mora segment huni kho bateva 6.15 HLLLLL where cadence 1 HLLHL might be expected. There is a pattern of CataCV counting as if CVCV with -ate- or even -e- being found for -ata- in alternate spellings. It is probably the intervocalic -t- that is at the heart of the matter. We also find CutaCV where HL is expected in vasutuna 15.123, vasutana 22.243, and skutana 3.37, 53, 93. And there is CataCV where HX is expected in hvatanau 23.4. At the time of Z, intervocalic -t- was likely phonologically a glottal stop. It is lost in Late Khotanese. These forms appear to be somehow in anticipation of the evolution to Late Khotanese.

(41.) Two other features separate category 2 from category 4 forms. All category 2 suffixes are inflectional and one syllable in length, while all category 4 suffixes are derivational and more than one syllable in length (on the surface with the required inflectional ending). I do not see how these features can help explain the disparity in treatment at issue here.

(42.) It is possible but not certain that these were also phonemic values in Middle Indic. With Khotanese we can be certain that the values in the chart for Khotanese were phonemic.

(43.) There are few cases of VttV outside of these categories. I have noticed only asottana- Z 5.14 hap.leg Emmerick 'to be despised'; pattara- Z 2.78 < Skt patra 'bowl'; ssotte Z 22.187 Eramerick 'he will descend', Leumann 3Sm 'eilt, schwebt' (Glossar: 509b), Bailey vussotte 'he descended' (DKS: 390b); bumattona-~bumattauna- 'strength' Sgs 3.4r3, 4.10v3); h[a]mattauna- 'being, state of having become' < hamata- ppp to ham- 'be, become' (transcription and translation Emmerick apud KhSuff: 164a; Skjaervo gives 'state of having been changed' IOL Khot 190/2v3 Sgs 58.3); ttattika (Z 2.99, 23.133, 24.255) 'here, hither', ttatti (Z 13.106, 23.1313) 'id.' The ghost form [dagger]tattita 'then' (DKS: 122a), Sgh 213.1[mss 10,11] is now transliterated ttatita by Skjaervp ([10]=IOL Khot 184/1v2, [11]=IOL Khot 169/2v4).

(44.) All twenty-nine cases in Z are Indian loans containing (-)vartt-: pravartt- 1.187, 4.69, 77, 81, 103, 104, 14.10, 25, 22.192, 193, 216, cakkravartta- 2.6, 13.59, 66, 112, 14.82, 22.141, 23.167 i-akra-), 23.168 (-akra-), 24.198, samantamukha-parivartta- 4.39, amvartt- 4.108, 14.39, 81, 85, avaivarttia- 6.54, vartt- 12.14, 24.191,492.

(45.) All ten cases in Z are native words: hamgaltte 4.47, 5.84, pataltta 12.49, nyultte 17.14, 24.419, baltte 20.57, 24.504, samkhiltta 22.259, samkhiltte 24.227, ggaltte 24.450.
COPYRIGHT 2015 American Oriental Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2015 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Hitch, Doug
Publication:The Journal of the American Oriental Society
Article Type:Critical essay
Geographic Code:1USA
Date:Oct 1, 2015
Previous Article:The Calm before the Storm: Selected Writings of Itamar Singer on the End of the Late Bronze Age in Anatolia and the Levant.
Next Article:The cat demon, gender, and religious practice: towards reconstructing a medieval Chinese cultural pattern.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters