Laryngeal Traces without Laryngeals in Vedic Metre?

Laryngeal Traces without Laryngeals in Vedic Metre? MARTIN JOACHIM KÜMMEL 1. THE THEORY OF LARYNGEALS ‘MAKING POSITION’ Already in 1927, ‘the young Kuryłowicz’ made the observation that in some words the presence of the newly found IndoEuropean ‘laryngeal’ consonants still seemed to have a consonantal effect causing metrical irregularities in Vedic (see also Kuryłowicz 1928). But his idea remained largely unnoticed until the late Jochem Schindler revived it (in lectures), and, lately, the problem was mentioned in the metrical edition of Van Nooten and Holland (1994: vi) and made the object of two papers by Jost Gippert (1997; 1999), and of several by Gary Holland (most recently, at our conference). What is this impact of Pre-Vedic ‘laryngeal’ consonants on Vedic metre? It is clear that a Pre-Vedic, Early Proto-IndoIranian structure *VC.HV = – × lost its laryngeal without a trace, leading to a Common Indo-Iranian and Vedic structure *V.CV = ◡ ×. But the earlier scansion – × could have survived as a metrical licence (lexically or ‘phrasally’ conditioned, not general): a synchronically light syllable may appear in the place of a heavy syllable (brevis in longo scansion for which I will use the abbreviation BIL and the symbol ᴖ ). 26 MARTIN JOACHIM KÜMMEL NB: This allegedly includes cases of the Indo-Iranian ‘Vocalization’ *CHC > *CHьC (> *C(h)ьC > Vedic CiC) – a fact that has been used as an argument to show that a svarabhakti vowel developed following the consonantal laryngeal. Gippert (1997: 78) claims that ‘at least’ 100 out of 215 cases of irregular triṣṭubh cadences can be explained by laryngeal effects. However, in this counting he explicitly included other laryngeal effects as well as doubtful cases where the laryngeal could only be reconstructed circularly because of the metrical data (e.g., for áhan- ‘day’). If we subtract these, the number is reduced to 76 out of 215, i.e., about 35%. Now it is crucial to compare this to the general frequency of such syllables containing a laryngeal following a short vowel and a single consonant – which Gippert did not do (though he was aware of the problem, 1999: 125). To get an idea, I simply counted the verbal roots with a synchronic structure °aC- in the RV, and thus I arrived at approximately 80 seṭ roots vs. 120 aniṭ roots, i.e., roughly 40%. Since this is the same percentage as for ‘brevis in longo’, we may conclude that there is no significantly heightened frequency of seṭ cases showing brevis in longo when compared to their general frequency. But in fact, the locus classicus of the problem has never been the triṣṭubh cadence alone. 1.1. Suggestive data present themselves when we have a look at the words and verses collected by Oldenberg (1888: 11–12, 64) and Arnold (1905, §190 ii & §220 ii) for purely metrical reasons: A laryngealist’s eyes will soon recognize many ‘laryngeal’ cases, like savit́ at the end of the gāyatrī or jagatī cadence, standing in the place of – ◡ ×, as a preform *sau̯Hь-t́ would have counted, e.g., RV 5.82.9c prá ca suv́ti savit́ ◡ ◡ ◡ – | ◡ ᴖ × (opposed to more frequent regular scansion ◡ ◡ – of this form after the caesura); and similarly but much more rarely in the triṣṭubh cadence, e.g., jánāḥ for – ×, e.g., RV 6.51.11b pūṣ́ bhágo áditiḥ 27 Laryngeal Traces without Laryngeals in Vedic Metre páñca jánāḥ = – – ◡ ◡ ' ◡ ◡ – | – ◡ ᴖ ×, as if we still had the old form *ánH-ās. Significantly these irregular scansions are most common in Arnold’s ‘Archaic’ period of Vedic metre, continuously decreasing in later periods (Arnold 1905, §221), so it might well be called an archaism. Of course, Arnold’s chronology is not proven (and partly based on exactly these scansions), but it would fit into our theory, and we might accept it for the time being. Now before we proceed, we must bear in mind that what we are speaking about are metrical rules, not linguistic reality, as shown by the very special distribution. Most often, only one of a lexeme’s possible prosodic structures displays the phenomenon. E.g., it is attested in disyllabic forms of śáru- and the plural of jána- but never in trisyllabic forms of these words nor even the singular of jána-. On the contrary, it is never found in disyllabic forms of hári-, ávas-, kaví- or the simple noun háva- while trisyllabic forms of these word or roots may show BIL. And while trisyllabic forms of duhitár- do show it (as against quadrisyllabic ones), structurally identical words of similar frequency do not, e.g., janitár-. It is far less frequent that we find a word or root that shows BIL in more than one position. The data for some important words may be summarized in this table: Scansion of first syllable Disyllabic Trisyllabic Quadrisyllabic partly heavy śáru-, jána- pl. śávas, rath́-, sákhi- savitár-, duhitár-, hári-, ávas-, háviya-, kavíśávas-, rath́-, sakhiyá- -- only light hári-, ávas, háva-, kaví-, jána- sg. śáru-, harít-, háritajanitár-, avitár- savitár-, duhitár- 28 MARTIN JOACHIM KÜMMEL We can conclude that normally BIL scansion is not a quality of the word or root itself but of some forms in metrically defined places. This behaviour is different from, e.g., genitive plural *-aām in all metrical positions (Arnold 1905, §143). Already Oldenberg had noticed the main area of the phenomenon: The first syllables (second compound members included) of a. trisyllabic forms of words beginning with two light syllables in the cadence of 8/12 syllable verses (Arnold 1905, §190 ii & §220 ii), b. disyllabic forms of words with a light first syllable in the triṣṭubh cadence (Arnold 1905, §220 ii). NB: It is important to note that from a metrical perspective we cannot assume brevis in longo except when the brevis follows a regular light syllable in the cadence, i.e., when the text gives ◡ ᴖ × or ◡ ᴖ ◡ ×. When an irregularly long syllables precedes the brevis, i.e., in the case of – ◡ ◡ × for ◡ – ◡ × or – ◡ × for ◡ – × (see Arnold 1905, §190v & §220v), the phenomenon could rather be understood as a kind of anaclasis ( ◡ – > – ◡ ) and must be kept apart (pace Gippert). Now, there are some other problems in this account of ‘laryngealistic’ metrics. 1. It is all about frequency or probability, not absolute rules: Is there really a difference in frequency between laryngeal cases and non-laryngeal cases? If there is, how to explain the non-laryngeal cases? Are they caused by secondary metrical analogy? Or could there be a third factor causing higher frequency of laryngeal cases? 2. The problem of metre itself: The rules would have to be inherited from Proto-Indo-Iranian, but was syllable quantity relevant in the metre then? As is well known, Old Avestan Laryngeal Traces without Laryngeals in Vedic Metre 29 metrics is syllable counting irrespective of quantity (some slight tendencies left aside). If this should be old, IndoIranian metrics could hardly have been different, and we would have to assume a special Indo-Aryan development before the loss of laryngeals – in spite of the fact that the loss of these in postconsonantal position seems to be a common or even Proto-Indo-Iranian process. Unfortunately, it is less than clear whether the nearly total neglect of quantity in Avestan is an archaism or an innovation. 3. Individual problems: Why do some ‘laryngealic’ words show irregular scansion, and others do not? Why is there not a single instance of a heavy first syllable of words as frequent as prathamá-, janitár- or the singular of jána-? 4. Relative chronology: The aspiration of preceding stops may presuppose (at least to my understanding of aspiration processes, pace Schindler as well as Gippert 1997: 63–64) the loss of segmental *h2: *Th2 > *Th > *Th, at least in Proto-IndoIranian. So the metrical rule would have to be older than that. 5. Last but not least: What is our heuristic profit in concrete cases? Can Vedic metre prove or disprove original postconsonantal laryngeals in the absence of other evidence? To answer these questions let us now have a more precise look at the data. I started from Arnold’s collection of irregular cadences and collected the respective words, concentrating on those that occur rather frequently (more than once in the material). Details about these and some other cases may be found in the appendix at the end. 1.2. First we shall consider lexical distribution: Which words do show BIL scansion? We may divide them into four classes: a. Words with a – fairly – certain postconsonantal laryngeal in PII, e.g., ávas-, śávas-, savitár-, duhitár-, sakhiyá-/sákhi-, 30 MARTIN JOACHIM KÜMMEL átithi-, hári-, háviya-, rátha-/rathiya-, jána-, kaví- (rare) – in sandhi or compounds, the preservation of original laryngeals is less probable, as these combinations might be younger (e.g., áditi-, úd V, duritá-). b. An equally large group of words that quite certainly did not exhibit a postconsonantal laryngeal, e.g., ajára-, udára-, mánas-, apás-, br̥hat-, abhítas, -vacas-, uṣás-, r̥tú-, amŕ̥ta-, bhŕ̥gu-, namasíyà-, ap-, nár-. c. Words where a postconsonantal laryngeal in PII or its postconsonantal position is possible but uncertain, often without an assured etymology, e.g., váruṇa-, ŕ̥ṣi-, marút-, aruṣá-, yaśas-, sanutár, mánuṣ-, áhan-, suffix -(t)ama-, -mahas-. d. Some cases where a heavy scansion might be explained by original vowel or consonantal length, thus not providing evidence at all, e.g., gr̥hápati- (for *gr̥h́-pati- < *°ó-poti- with a residue of lengthening by Brugmann’s law?), ácidhuvam (< *áciddhuvam, Korn 1998: 178, n. 275), dádr̥śe (< *d́dr̥ś-?), vavr̥mahe (for*vu-vūr-m°, see Gippert 1997: 75; Kümmel 2000: 459), ruruhur (<*Hru-Hrudh-), áraruṣ- (*á-rā-r° or *á-ra-rH-?), no ava (< *nō (H)áu̯a < *nas Háu̯a). Of course, the groups a and b are the most relevant, but on the whole, the picture looks rather like what Oldenberg said (1888: 12): Es ist kaum zu viel gesagt, dass es in häufigeren Worten keine Kürze giebt [– sofern natürlich ihre Umgebung zu den umgebenden Stellen des metrischen Schemas passt –], die nicht gelegentlich an drittletzter Stelle des achtsylbigen Pâda statt der Länge gebraucht würde, i.e., every common word form with two successive short syllables might be used at the end of the cadence of 8 syllable verses (cf. Arnold 1905: 159). Indeed, in ‘dimeter verse’ BIL scansion is more than thrice as frequent as in trimeter verse; compare the counting of Gippert (1999: 107): 7.15% (8 syllables) against Laryngeal Traces without Laryngeals in Vedic Metre 31 1.64% (11 syllables) and 2.21% (12 syllables) in the trimeter verse. Why this difference? 2. THE METRICAL POSSIBILITIES Here the constraints of the metre are crucial. They are quite different depending on the verse type. 2.1. Word forms beginning with two light syllables followed by a (potentially) heavy third syllable (i.e., with the structure ◡ ◡ – ) are rather difficult to place, as can be inferred from the metrical schemata (unpreferred syllable weights are written above the preferred): Dimeter verse (8 syllables) Regular anuṣṭubh, gāyatrī :  –̆ ◡ ̅ –̆ | ◡ – ◡ || Trimeter verse (11 syllables, triṣṭubh type) Early caesura:  –̆ ◡ ̅ –̆ ' ◡ ◡ –̆ | – ◡ – || Late caesura:  –̆ ◡ ̅ –̆  ' ◡ ̅ ◡ | – ◡ – || Trimeter verse (12 syllables, jagatī type) Early caesura:  –̆ ◡ ̅ –̆ ' ◡ ◡ –̆ | – ◡ – ◡ || Late caesura:  –̆ ◡ ̅ –̆  ' ◡ ̅ ◡ | – ◡ – ◡ || 2.2. The relevant metrical constraints can be displayed in an Optimality Theory type table (cf. McCarthy & Prince 2001: 5ff.) which lists the constraints and their violations triggered by the ‘candidates’ (the possible placement of two light syllables followed by a heavy one). The constraints are ordered from left to right according to their decreasing rank (i.e. the frequency of their being violated increases from left to right). The relevant constraints are indicated as follows: w = word boundary before the caesura; constraints as to syllable weight are indicated by the respective number of the syllable within the 32 MARTIN JOACHIM KÜMMEL verse followed by l = light or h = heavy. As elsewhere, * means violation of a constraint. Trimeter verse with {early caesura} (late caesura) w 5–7 11l 9l 8h 10h (7l) (*) 6l {5l} 2or3h {7h} 4or5h 2h (*) * 1–3 {*} 6–8 * 2–4 * 10–12 * 8–10 * * 9–11 7–9 4–6 3–5 4h * * * * * {*} * {*} * Normal dimeter verse 7l 5l 2or3h 6h 2h * 1-3 * 6-8 * 2-4 * 3-5 5-7 4h * 4-6 * * * If we rank the constraints according to frequency of their violation in Arnold’s collection, we see an important difference. While for all variants of trimeter verse the respective forms can easily be placed after the caesura, in dimeter verse there is no such place, so violation of a constraint is necessary. But word forms like that are too frequent in the language to avoid them completely, and after the middle and the beginning of the line, the third best candidate for their placement would be the cadence. Thus, the metrical need for irregular scansion would Laryngeal Traces without Laryngeals in Vedic Metre 33 be very much stronger in dimeter verse, and the respective scansions might be considered as artificial as ‘metrical lengthening’ in Homeric Greek. 2.3. As to the relative frequency of irregular scansion, different approaches are possible. 2.3.1. First, we may compare with a given word forms of identical prosodic structure. Here we see that words with or without laryngeals may show the same range from very high to very low percentages, e.g., from Groups b and c: udára- 66.7%; ajára30.8%; namasi yà- 30%; áruṣī- 21.4%; abhítas 12% – very similar to Group a: yávasa- 40%; údita-/úditi- 28.6%; hávi ya- 20.5%; ávas- 19.4%; átithi- 18.6%; and we find some possible candidates for a without BIL: avitár-, prathamá-, harít- (or hári-, kaví- when disyllabic). 2.3.2. Second, we might compare the frequency of individual words relative to other words of the respective derivation type. Here we will immediately see that some types only comprise seṭ roots. This is because of what we might call the ‘laryngeal bias’: Derivation from seṭ roots (i.e., roots ending in postconsonantal laryngeal) produces more light syllables than from aniṭ roots, because of two special effects: 1. If a full grade root is combined with a consonantal suffix, seṭ roots become disyllabic, producing two successive short syllables, e.g., *°aCH-C° > °aCi-C° ◡ ◡ ×: savitár-, jánitár-, átithi-, while aniṭ root derivatives will have one heavy syllable instead: *°aC-C° > °aC-C° – × : sotári, bhartár-. This increases the frequency of seṭ cases that might cause metrical problems. 2. The common type of simple thematic nouns or any other type with o grade of the root will have a short root syllable only with seṭ roots. This is because their laryngeals blocked Brugmann’s Law in Proto-Indo-Iranian: whereas *°oC.H-o- led to 34 MARTIN JOACHIM KÜMMEL *ăC(H)-a- ◡ ×, in aniṭ roots *°o.C-o- the o stood in an open syllable and was lengthened, leading to *°āC-a- – ×, see, e.g., seṭ svará-; gáya-, jána-, ráṇa-, vára-, háva- vs. aniṭ ghāsá-, tyāgá-, vāká-, vāsá-; ḿna-, ś́ka-. 3. Compounds with prefixes like úd or duṣ- will only appear in the material when the root begins with a vowel where, normally, a laryngeal has been lost. These facts additionally increase the statistical preponderance of seṭ roots in our material, as these cases were quite frequent in Indo-Iranian word formation. 2.3.3. So we must use the ‘unbiased’ formation types to compare seṭ and aniṭ roots. These would be vocalic suffixes after a zero or an e-grade root, and looking at these we arrive at the following relation between seṭ and aniṭ cases (percentage of BIL in brackets): 1. For s-stem nouns and adjectives, the seṭ cases ávas- (19.4) and śávas- (19.5) together with derived yávasa- (41.2) are clearly on top, but otherwise there is no difference in frequency: aniṭ apás- (8), vácas- (7.4) ≈ seṭ tavás- (6.4). 2. Non-o-grade thematic nouns (including derived ī-stems) and adjectives are not a very common type, so we should not be surprised that there is no relevant evidence. 3. As to thematic verb forms, we can see that seṭ and aniṭ roots show nearly the same figures in trisyllabic forms, and disyllabic forms are attested in only three single aniṭ cases! However, the thematic suffix (at least in the subjunctive) has been reconstructed as *-h1o/e- (Tichy 2002: 202–204; 2004: 101–102). 4. i/u-stems, trisyllabic: the scarce evidence for trisyllabic forms shows no difference, see bhŕ̥gu- (aniṭ, 13) vs. hári- (seṭ, 11.9); as to disyllabic śáru- (seṭ?, 25) and sákhi- (seṭ, 4), both might have o-grade and thus are not really ‘unbiased’. Laryngeal Traces without Laryngeals in Vedic Metre 35 5. We find only two root nouns, both aniṭ ap- (trisyllabic 0.6, disyllabic 6.2) and nar- (2.5). 6. When we come to other vocalic suffixes, the most frequent is -iya-, where we find an apparent difference in háviya- (seṭ, 20.5) vs. gúhiya- (aniṭ, 6.3), but -así yà- (aniṭ, 25) is as frequent as háviya-. Other suffixes are rarely found, we have -u(-)ṣá- in an unclear case (aruṣá-/áruṣī) and a probable aniṭ case in udára- (66.7); with -aná-/-anā̆- we meet śaraṇá- (seṭ, 12.5%) vs. four single aniṭ cases; otherwise there is no relevant evidence. What remains are some cases that begin with a prefix rather than a root: 7. The negative prefix *ń̥- > *á- is frequently measured long like that in ajára- (aniṭ, 30.8), amŕ̥ta- (3.9) and 5 other single cases (aniṭ); as a seṭ case we can only adduce áditi- (5.5). 8. The homophonous verbal augment á- seems to show long scansion in 5 single forms, the roots of which never begin with *CH° (or, *HC°). 2.4. If we summarize the ‘unbiased’ types, we find positive examples with a clearly high frequency of brevis in longo in four words containing v and two compounds of i, yávasa-, ávas-, śávas-; háviya-, údit°, duritá-; on the other hand we also find examples of aniṭ roots with a similarly high frequency: udára-, ajára-, yáśas- (?), bhŕ̥gu-, and -asíya-; and we have to note that some seṭ words like prathamá-, harít-/hárita-, páyas- are negative examples. When we look at the ‘biased’ types, we find a similar quantity of good examples, e.g., átithi-, sánitar- and (disyllabic) jánain the plural, rátha-/rath́-, sákhi- – but as many counterexamples without heavy scansions: avitár-, jánitár-, kaví-, háva- or jána- in the singular. 36 MARTIN JOACHIM KÜMMEL 3. CONCLUSION There is no clear correlation of frequency for *°aCHV° as against *°aCV°, when we consider the overall frequency of seṭ root derivatives and their specific syllable structure. Clear cases of frequent irregularity are attested in both cases, and when counted by word, there is no difference at all. When counted by instance, the frequency of two words, ávas- and śávas-, leads to a slight preponderance of seṭ instances. This slight dominance of the type ávas- might be a residue of a more original distribution, but the figures do not prove that. The evidence for the case of originally interconsonantal laryngeals *°aCHC° > *°aCHьC° is even slimmer – in spite of their overrepresentation within all words with two successive short syllables. Even the most frequent case of this type, namely átithi-, is clearly less frequent than aniṭ cases like ajára-, namasíya-. As for most other cases, for instance, the often cited scansion duhit́ = ᴖ ◡ – is too rare (and possibly rather late) to have any significance for the reconstruction of this kind of structure. What is most clear, however, is that there is no heuristic benefit for the judgement of unclear cases: if a word shows a high percentage of Vedic brevis in longo scansions, it may contain an old laryngeal, or it may not. If it does not show the phenomenon, we still have both possibilities. It cannot be excluded that the whole phenomenon of this licence in Vedic metre could be explained as a residue older than the loss of laryngeals. But it is only the licence itself, not its distribution in the RV, that can have such an explanation. REFERENCES ARNOLD, E. Vernon 1905. Vedic Metre in its Historical Development. Cambridge: Cambridge University Press. GIPPERT, Jost 1997. Laryngeals and Vedic metre. In: Alexander Lubotsky (ed.), Sound Law and Analogy. Papers in Honor of Laryngeal Traces without Laryngeals in Vedic Metre 37 Robert S. P. Beekes on the Occasion of His 60th Birthday: 63– 79. Amsterdam: Rodopi. ——— 1999. Neue Wege zur sprachwissenschaftlichen Analyse der vedischen Metrik. In: Heiner Eichner & Hans Christian Luschützky (eds.), Compositiones indogermanicae in memoriam Jochem Schindler: 97–125. Prague: Enigma corporation. KORN, Agnes 1998. Metrik und metrische Techniken im Ṛgveda. Streckformen in Trimeter-Versen. Graz: Leykam. KURYŁOWICZ, Jerzy. 1927. Les effets du ə en indoiranien. Prace Filologiczne 11: 201–243. ——— 1928. Quelques problèmes métriques du Rigvéda. Rocznik Orientalistyczny 4 (1926): 196-218. KÜMMEL, Martin Joachim 2000. Das Perfekt im Indoiranischen. Eine Untersuchung der Form und Funktion einer ererbten Kategorie des Verbums und ihrer Weiterentwicklung in den altindoiranischen Sprachen. Wiesbaden: Dr. Ludwig Reichert. MCCARTHY, John. J. & Alan PRINCE 2001. Prosodic Morphology. Constraint Interaction and Satisfaction. Rutgers Optimality Archive: http://roa.rutgers.edu/view.php3?roa=482 Rutgers University, New Jersey. (Originally circulated in April 1993 and available as Technical Report no. 3 of the Rutgers University Center for Cognitive Science). LUBOTSKY, Alexander M. 1997. A Ṛgvedic Word Concordance, I–II. New Haven, Connecticut: American Oriental Society. NOOTEN, Barend A. van & Gary B. Holland (eds.) 1994. Rig Veda – metrically restored text with an introduction and notes. Cambridge, Mass.: Harvard University Press. OLDENBERG, Hermann 1888. Die Hymnen des Ṛigveda, I: Metrische und textgeschichtliche Prolegomena. Berlin: Verlag von Wilhelm Hertz. TICHY, Eva 2002. Zur Funktion und Vorgeschichte der indogermanischen Modi. In: Heinrich Hettrich (ed.), Indogermanische Syntax – Fragen und Perspektiven: 189–206. Wiesbaden: Reichert. ——— 2004. Indogermanistisches Grundwissen für Studierende sprachwissenschaftlicher Disziplinen. 2., überarbeitete Auflage. Bremen: Hempen. 38 MARTIN JOACHIM KÜMMEL APPENDIX: Material Metrical behaviour of words with potential brevis in longo (‘BIL’) scansion in the cadence Symbols used in the first column: + postconsonantal laryngeal assured – postconsonantal laryngeal excluded ! postconsonantal vocalized laryngeal () an alternative explanation for ‘BIL’ may be available *VCH – Stem(s) Total ‘BIL’ Periods Frequency A S N Maṇḍalas C P 1 2 3 4 5 6 7 8 9 10 udára- 6 4 66.7% 1 3 - - - 2 - - - - - - 2 - - +!? (-)*nayi-tár- 6 4 66.7% 4 - - - - - - - - 3 - - 1 - - (+)? á-ra-r-uṣ- 8 4 50.0% - 2 2 - - 2 - - - - - 1 - 1 - +? yáv-asa- 17 6 41.2% 5 1 - - - 1 - - - 3 1 - - - 1 – a-jára- 65 20(8) 36.9% 16 1 - 3 - 7 - - - 2 5 - 4 - 2 +? táru-tár- 6 2 33.3% 1 1 - 1 - - - - - - 1 - - –? -yaśas- 12 4 33.3% 4 - - - - 1 - - - 1 - - 1 1 - + úd-i-ta- + úd-i-ti- 21 6 28.6% 2 4 - - - - - - - - - 3 3 - - – -as-íyà- 20 5(2) 25.0% 2 - 3 - - - 2 1 - 1 - - 1 - - + háv-iya- 32 8(3) 25.0% 7 1 - - - - - - - 1 - 1 6 - - + áv-as- 165 33(8) 20.0% 21 7 5 - - 7 - 1 - 5 3 - 15 2 - + (-)śáv-as- 113 22(1) 19.5% 17 3 1 - - 4 - 2 - 5 2 1 - 2 (–) 6 21 4 19.0% 1 2 - 1 - 1 - - - 1 1 - - - 1 ap-ám 17 3(1) 17.6% 2 1 - - - 1 - - - - - - 1 1 - + sáni-tar-/sáni-ti- 24 4 16.7% 2 1 1 - - 2 - - - 1 - - 1 - - + dur-itá- 37 6 16.2% 4 1 1 - - - - - - 1 - 1 - - 4 39 gr̥há-pati- – Laryngeal Traces without Laryngeals in Vedic Metre Light antepaenultima before light paenultima in the RV (in brackets: trimeter verse) 40 +! rath-íy- 33 5 15.2% 3 1 1 - - - - - - - - 1 3 1 - +! áti-thi- 43 6 14.0% 8 - - - - - - - - 1 2 - 3 - - –? bhr̥g ́ -u- 23 3(2) 13.0% 3 - - - - 2 - - - - - - - 1 - +? śar-aṇá- 16 2 12.5% 1 - 1 - - - - - - - 1 - 1 - - – abhí-tas 25 3 12.0% 1 - 1 1 - 1 - - - - - - - 1 1 + vidh-at- 25 3 12.0% 3 - - - - - - - - 1 - - 2 - - +! távi-ṣī- 25 3(2) 12.0% 3 - - - - 2 - - - - - - 1 - - + hár-i- 59 7(1) 11.9% 1 3 3 - - 1 - 1 1 - - - 2 2 - aruṣá- + áruṣī 57 5 8.8% 2 - 3 - - 2 - 1 - - - - 2 - - sakh-iyá- 57 5(1) 8.8% 5 - - - - - - - - 1 1 - 2 - 1 – ap-ás- 25 2(1) 8.0% 1 - - 1 - 1 - 1 - - - - - - - +! savi-t́r- 144 11(1) 7.6% 5 3 1 - 1 2 - - - 4 - - - 2 3 – (-)vác-as- 27 2(1) 7.4% 1 1 - - - 1 - - - - - - 1 - - + (-)táv-as-, tav-ás- 49 3(1) 6.1% 1 - 1 1 - 1 - - - - - 1 - 1 - + á-di-ti- 165 9(2) 5.5% 7 2 - - - - - - - - - 1 7 - 1 – mán-as- 147 7 4.8% 6 - - - 1 - - - - - - - 7 - - ? hav-íṣ- 72 3 4.2% - - - - 3 - - - - - - - - - 3 + hár-i-vant- 49 2 4.1% 1 - 1 - - 1 - - - - - - 1 - - +! duhi-tár- 54 2 3.7% - 1 1 - - - - - 1 - - - - 1 - MARTIN JOACHIM KÜMMEL ? +! váruṇa- 287 13 3.5% 8 2 2 1 - 1 1 - - 3 - 2 6 - - nám-as- 88 3(2) 3.4% 2 1 - - - - - - - - - 1 2 - - – a-mŕ̥ta- 102 3 2.9% 2 - 1 - - - - - - - - 1 2 - - ? marút- 324 7(3) 2.2% 3 3 1 - - 1 - - - 2 - 1 1 2 - – uṣ-ás- 205 3(1) 1.5% - 3 - - - 3 - - - - - - - - - – br̥h-at- 134 2 1.5% 1 - 1 - - 1 - - - - - - 1 - - +! pr̥thi-v́- 241 2 0.8% 1 1 – - - 1 - - - 1 - - - - - (–) gúh-iya- 16 1 6.3% - - 1 - - - - - - - - - 1 - - (–) pŕ̥tanā- 18 1 5.6% - - 1 - - - - 1 - - - - - - - (+)! path-í- 29 1 3.4% 1 - - - - - - - - - - - 1 - - (+)! kav-í- 53 1 1.9% - - - - 1 - - - - - - - - - 1 +! avi-tár- 32 0 -- - - - - - - - - - - - - - - - + prath-amá- 185 0 -- - - - - - - - - - - - - - - - + harít- 27 0 -- - - - - - - - - - - - - - - - +! jánitār-/janit́r- 34 0 -- - - - - - - - - - - - - - - - Verbs: – áy-a-te 4 3(2) 75.0% 2 1 - - - 3 - - - - - - - - - – kr̥-ṇáv-a-te 6 2 33.3% 1 - - - 1 - - - - - 1 - 1 - - – vás-a-ti 3 1 33.3% 1 - - - - - - - - - - - 1 - - Laryngeal Traces without Laryngeals in Vedic Metre ? – 41 42 – rám-a-ti 3 1 33.3% 1 - - - - - - - - - - - 1 - - +? dhám-a-ti 4 1 1 = 25.0% 1 - - - - - - - - 1 - - - - - 11 2 18.2% - 2 - - - 1 - - - - - - 1 - - 14 2(1) 14.3% 1 - 1 - - - - - - - - - 1 - 1 29 4(1) 13.8% 3 1 - - - 2 - - - 1 - 1 - - - –? pát-a- ti – sr̥j-á- + náy-a-ti ti – yám-a- 8 1 12.5% 1 - - - - - - - - - 1 - - - - + tár-a-ti 10 1 10.0% 1 - - - - - - - - - 1 - - - - + áv-a-ti 64 6 9.4% 4 1 1 - - - - - 1 2 1 - 2 - - ti sád-a- 13 1 7.7% 1 - - - - - - - - 1 - - - - - – d-áy-a-te 18 1 5.6% 1 - - - - 1 - - - - - - - - - - - 1 - - - - - 1 - - - - + 19 1 5.3% + át-a- ti 1 1 100.0% - 1 - - - 1 - - - - - - - - - – va-náv-a-te 1 1 100.0% 1 - - - - - - - - - 1 - - - - dá(-)d-a- ti MARTIN JOACHIM KÜMMEL – t(i) *VCH +? Stem(s) Total ‘BIL’ Periods Maṇḍalas Frequency A S N C P 1 2 3 4 5 6 7 8 9 10 - - - 1 - - - - - - - 1 2 25.0% 2 - 99 21 20.2% 14 2 1 3 - 4 1 - 1 1 10 - - 1 2 – iṣ(-)i(-)rá- 28 5 17.9% - - 3 1 1 - 1 - - 1 - - - 1 2 – riṣ- 14 2 14.3% 2 - - - - - - - - - 2 - - - - ? -(t)ama- (sum) 23 2 1 3 - 4 1 - 1 1 10 - - 1 2 ján-a- (plural) 13 11 - 2 - - 2 1 1 1 1 7 - - - - + (-)śáv-as- 3 11.7% 8.9 (24.1)% 7.7% 14 +! 197 147 (54) 39 3 - - - - 1 - - - - 1 - - - 1 – ap- 129 8 6.2% 4 - 3 1 - 1 3 1 - - 2 1 - - - +! (-)sákh-i- 100 4 4.0% - 1 3 - - - - - 2 - - 1 - 1 - +! (-)ráth-a- + (-)rath-́- 392 12 3.1% 7 1 2 2 - 4 1 0 1 2 2 0 0 0 2 – nár̆ °VC-amá- 161 4 2.5% 3 1 - - - 1 1 - - 1 - 1 - - - ? 98 2 2.0% 2 - - - - 1 - - - - 1 - - - - – br̥h-at- 134 2 1.5% - - 1 1 - - - - - - - - - 1 1 57 1 1.8% - 1 - - - - - - - - - 1 - - - 116 1 0.9% 1 - - - - 1 - - - - - - - - - (–) uṣás- (–) (-)mán-as- - 43 śár-u°V̆-tama- 8 ? Laryngeal Traces without Laryngeals in Vedic Metre Light paenultima syllables in the RV 44 (–) br̥hán 161 1 0.6% 1 - - - - - - - - - 1 - - - - + hár-i- 82 0 -- - - - - - - - - - - - - - - - +! kav-í- 133 0 -- - - - - - - - - - - - - - - - Verbs: – yu-ná-j-a-t(i) 3 1 33.3% 1 - - - - - - - - - 1 - - - - – t(i) 6 1 16.7% - - - - 1 - - - - - - - - - 1 t(i) 17 1 5.9% 1 - - - - - - - - 1 - - - - - – áy-aás-a- Relevant nominal forms (sum of the above, without single instance cases) ‘BIL’ = A S N C P 1 2 3 4 5 6 7 8 9 10 + 2074 186 9.0% 126 32 23 3 2 27 1 5 5 28 18 10 56 10 15 – 1362 85 6.2% 51 11 13 8 2 22 7 3 - 6 9 4 22 6 6 Thematic verb forms in general Total ‘BIL’ = A S N C P 1 2 3 4 5 6 7 8 9 10 + 127 14(1) 11.0% 9 3 2 - - 3 - - 1 4 3 1 2 - - – 107 18(3) 16.8% 12 3 1 - 2 5 - - - 2 4 - 5 - 2 MARTIN JOACHIM KÜMMEL Total