So far, we have looked at cases of DOI in the inflection of simple (i.e., underived) nominal bases (cf. 1.2.45 arthavad adhātur apratyayaḥ prātipadikam). Now let us look at some cases of DOI in the inflection of complex (i.e., derived) nominal bases such as kr̥t ‘primary derivative’, taddhita ‘secondary derivative’, and samāsa ‘compound’ (cf. 1.2.46 kr̥ttaddhitasamāsāś ca).
Generally speaking, as compared to the inflection of simple nominal bases, which we have seen in the previous chapter and in this chapter, and verbal inflection and primary derivatives, which we will see in the following chapters, we find a smaller number of examples of conflict in taddhita derivations, and even fewer examples in samāsa derivations. I will explain why this is the case towards the end of chapter 4.
We will see that the tradition manages to avoid dealing with conflict in the first four examples. However, it has to rely on certain external (post-Pāṇinian) metarules to correctly derive these four forms. I will show that my solution for DOI (my interpretation of 1.4.2) can help us perform these derivations without relying on such external metarules. In the following four examples, we do find cases of conflict. Here too, I use my solution for DOI (cf. my interpretation of 1.4.2) to get the correct answer and also mention the traditional solution where it is known.
pratīcas
(1) Consider the genitive singular form of prati-ac1 ‘turned towards, facing’: pratīcas. By 2.2.18 kugatiprādayaḥ, prati-ac is a tatpuruṣa compound made of prati, which takes the technical designation gati by 1.4.60 gatiś ca and ac, which is derived as follows: añcU + KvIN
91
(3.2.59 r̥tvigdadhr̥ksragdiguṣṇigañcuyujikruñcāṁ ca2) 🡪 ac + v (6.4.24 aniditāṁ hala upadhāyāḥ kṅiti3) 🡪 ac (6.1.67 ver apr̥ktasya4).
The Siddhāntakaumudī (SK) completes all the operations within the base before adding the genitive singular affix Ṅas5: prati-ac (2.4.71 supo dhātuprātipadikayoḥ) 🡪 pratyac (6.1.77 iko yaṇ aci). If the derivation is stopped at the addition of the genitive affix Ṅas to pratyac, that does not give the correct answer: pratyac + Ṅas 🡪 *pratyacaḥ.
The tradition has found a way to work around this. In pratyac + Ṅas, pratyac takes the designation bha by 1.4.18 because it is followed by a non-sarvanāmasthāna affix beginning with a vowel. Then, 6.4.138 acaḥ teaches that the a of verbal base ac (from añc), when designated as bha, is replaced with LOPA: pratyc + Ṅas. To get the correct form, it takes recourse to the metarule nimittābhāve naimittikasyāpy abhāvaḥ16, which teaches that when the cause of an operation is lost, the impact or effect of that operation too is lost. In other words, if X causes A to change to B, upon the deletion of X, B becomes A again. Thanks to this paribhāṣā, since the cause of the operation 6.1.77 iko yaṇ aci, namely a, has been deleted, the preceding y will go back to its original form i. Thus, we get pratic + Ṅas. At this step the tradition applies 6.3.138 cau, which teaches that, the final aṆ of the preceding pada in a compound is replaced with its dīrgha equivalent when c (from añc) follows. This gives the correct form: pratīcaḥ.
Another Pāṇinian paribhāṣā, which makes this very argument in terms of antaraṅga and bahiraṅga operations, is cited by the Siddhāntakaumudī6 when discussing this derivation: akr̥tavyūhāḥ pāṇinīyāḥ ‘The Pāṇinīyas do not insist that a rule should take effect if its causes disappear’. Nāgeśa (Pbh 56, Paribhāṣenduśekhara), while discussing this paribhāṣā in antaraṅga and bahiraṅga terms, says: bahiraṅgeṇāntaraṅgasya nimittavināśe paścāt
92
sambhāvite antaraṅgaṁ neti yāvat ‘An antaraṅga operation (here, 6.1.77 iko yaṇ aci) should not be undertaken if its cause would disappear later due to the bahiraṅga operation (here, 6.4.138 acaḥ)’.
These two paribhāṣās require one to go a step back into the derivation and undo a previous operation. This runs contrary to the idea that derivations should move in one direction, and that each operation should take us one step forward (rather than backward) into the derivation. Besides, if Pāṇini wanted us to use these metarules, he would have taught them explicitly in the Aṣṭādhyāyī. For these reasons, I do not accept these two paribhāṣās. Now I will derive this form using my method. Two rules are simultaneously applicable to prati - ac:
(6.1.77→) prati - ac (4.1.2→)
4.1.2 svaujasamauṭchaṣṭābhyāmbhisṅebhyāmbhyasṅasibhyāmbhyasṅasosāmṅyossup18 6.1.77 iko yaṇ aci: iK (i, u, r̥, l̥) is replaced with yaṆ (y, v, r, l) when aC (vowel) follows.
This is a case of DOI. By my interpretation of 1.4.2, we apply the RHS rule 4.1.2 and get prati - ac + Ṅas. Here two rules are applicable:
(6.1.77→) prati - (6.4.138→) ac + Ṅas
6.1.77 iko yaṇ aci: same as above.
6.4.138 acaḥ: the a of ac which has taken the technical designation bha is replaced with LOPA.
This is a case of DOI. By my interpretation of 1.4.2, the RHS rule 6.4.138 wins, and we get: praticaḥ 🡪 pratīcaḥ (6.3.138 cau), which is the correct form.
93
seduṣaḥ
(2) Let us derive the genitive singular of the perfect participle of sad ‘to sit’, namely sad + vas ‘one who had sat’. The Siddhāntakaumudī attaches the declensional affix Ṅas to the base only after the base is fully ready.19 The base is derived by replacing LIṬ with KvasU: sad + LIṬ 🡪 sad + KvasU (3.2.108 bhāṣāyāṁ sadavasaśruvaḥ20). Now, (i) by 6.1.8 liṭi dhātor anabhyāsasya, (which teaches that the un-reduplicated root undergoes reduplication when followed by LIṬ), (ii) by 6.1.1 ekāco dve prathamasya (which teaches that the first syllable of the root undergoes reduplication) and (iii) by 1.1.56 sthānivad ādeśo’nalvidhau (which teaches that the substitute should be treated like the substituendum except when an operation relative to the original sound is to be performed), we get sadsad + vas. By 7.4.60 halādiḥ śeṣaḥ, which teaches that all but the first consonant of the abhyāsa (first half of sadsad) are deleted, we get sasad + vas. Now, by 6.4.120 ata ekahalmadhye’nādeśāder liṭi21, we get sed + vas. At this point, 7.2.67 vasv ekājādghasām is applicable, which, according to the tradition22, teaches that the augment iṬ should be attached to vasU when it occurs after a root which, after doubling, consists of a single syllable, or a root ending in ā, or ghas ‘to eat’. By applying this rule, we get the base sedivas, but, if at the next step we add the genitive singular affix Ṅas, we get *sedivasaḥ, which is the incorrect answer.
Here, again, the tradition uses the two paribhāṣās discussed above to circumvent this problem. In sedivas + Ṅas, sedivas takes the designation bha because it is followed by a non sarvanāmasthāna affix beginning with a vowel (cf. 1.4.18 yaci bham). To this, the tradition applies 6.4.131 vasoḥ samprasāraṇam, which teaches that the semivowel of the affix vasU in an item termed bha is replaced with the corresponding vowel u. This gives sediuas, and the augment i in sedivas, which is attached to vas by 7.2.67 vasv ekājādghasām, is lost, because its cause v no longer exists (cf. akr̥tavyūhāḥ pāṇinīyāḥ and nimittāpāye naimittikasyāpy apāyaḥ). Then, the a of seduas is deleted by 6.1.108 samprasāraṇāc ca which teaches that both the
94
samprasāraṇa replacement and the vowel following it are together replaced with the former. This gives us sedus + Ṅas 🡪 seduṣaḥ, which is the correct form.
Again, like in the previous example, I reject the use of these two anitya paribhāṣās. I perform this derivation as follows. I add the affix LIṬ to sad by the following rule:
3.2.115 parokṣe liṭ: affix LIṬ occurs after a verbal root when an unwitnessed (parokṣa) action which is not current (anadyatana) is denoted in the past (bhūta).
Then, the following rules become applicable:
sad + LIṬ
6.1.8 3.2.108
6.1.8 liṭi dhātor anabhyāsasya: an un-reduplicated root undergoes reduplication when followed by LIṬ.23
3.2.108 bhāṣāyāṁ sadavasaśruvaḥ: the affix LIṬ is optionally replaced with KvasU in classical Sanskrit after the roots sadA ‘to sit’, vasA ‘to inhabit’ and śru ‘to listen’ when the action has taken place in the past.
This is a case of DOI. By my interpretation of 1.4.2, the RHS rule 3.2.108 wins and we get sad + vas. Multiple rules are applicable here:
sad + vas
6.1.8 7.2.67 4.1.2
6.1.8 liṭi dhātor anabhyāsasya: same as above.
7.2.67 vasv ekājādghasām: (my interpretation) augment iṬ is introduced to vasU when it occurs after a root which either consists of a single syllable, or ends in a, or else, is constituted by ghas ‘to eat’.24
95
4.1.2 svaujasamauṭchaṣṭābhyāmbhisṅebhyāmbhyasṅasibhyāmbhyasṅasosāmṅyossup25
This is a case of DOI. By my interpretation of 1.4.2, the right-most rule 4.1.2 wins and we get sad + vas + Ṅas. Multiple rules are applicable here:
sad + vas + Ṅas
6.1.8 6.4.131 7.2.67
6.1.8 liṭi dhātor anabhyāsasya: same as above.
7.2.67 vasv ekājādghasām: same as above.
6.4.131 vasoḥ samprasāraṇam: vasU of an item termed bha undergoes samprasāraṇa.26 There is an SOI between 6.4.131 and 7.2.67. Let us find out which rule is more specific. 6.4.131:
monosyllabic root / ākārānta root / ghas + vas (termed bha)
other bases + vas (termed bha)
7.2.67
monosyllabic root / ākārānta root / ghas + vas
The conditions highlighted in bold are relevant to this SOI. Since 6.4.131 has been taught specifically for a bha-saṁjñaka vas, it is more specific and thus wins.
Now let us consider the DOI relationship between 6.1.8 and 6.4.131. By my interpretation of 1.4.2, the RHS rule 6.4.131 wins and we get: sad + uas + Ṅas. Here, again, two rules are applicable:
cannot be inferred by anuvr̥tti either. The tradition takes the liberty to read this phrase into this rule purely on the basis of certain derivational considerations. I do not think we should make such assumptions and therefore I do not include ‘after doubling’ in my interpretation. 25 1.2.46 kr̥ttaddhitasamāsāś ca.
96
sad + uas + Ṅas
6.1.8 6.1.108
6.1.8 liṭi dhātor anabhyāsasya: same as above.
6.1.108 samprasāraṇāc ca: both the samprasāraṇa replacement and the vowel following it are together replaced with the former.
This is a case of DOI. By my interpretation of 1.4.2, the RHS rule 6.1.108 wins and we get sad + us + Ṅas. Thereafter, the derivation proceeds as follows: sadsad + us + Ṅas (6.1.8 liṭi dhātor anabhyāsasya) 🡪 seduṣaḥ (6.4.120 ata ekahalmadhye’nādeśāder liṭi), which is the correct form.
gargāḥ
(3) Let us derive the nominative plural of ‘descendant of garga’, first through the traditional method and then through mine.
The tradition27 adds the declensional affix only after the base is ready. As per the traditional method, we first add the affix yaÑ to garga + Ṅas by 4.1.105 gargādibhyo yañ28; then by 2.4.71 supo dhātuprātipadikayoḥ29, Ṅas is deleted, which gives us garga + yaÑ. At this juncture, 7.2.117 taddhiteṣv acām ādeḥ prescribes the vr̥ddhi substitution of the first vowel of garga given that the following affix is marked with Ñ. Thus, we get gārga + yaÑ. The a of gārga is deleted by 6.4.148 yasyeti ca, which teaches that the final i or a of a bha item is deleted when it is followed by ī or a taddhita affix. Thus, we get our base gārgya.
At this point, the tradition prescribes the addition of the affix Jas to the base gārgya: gārgya + Jas. This leads to the application of 2.4.64 yañañōś ca, which teaches that the gotra affixes yaÑ and aÑ are replaced with LUK when the following declensional affix denotes plural, except when the base is feminine. Stopping here gives us the incorrect form: *gārgaḥ.
97
On 2.4.64, the Bhaimī commentary on the Laghusiddhāntakaumudī suggests the metarule, nimittāpāye naimittikasyāpy apāyaḥ, which we have discussed in the previous two examples, to solve this problem: because yaÑ is deleted, the vr̥ddhi of the first vowel (cf. 7.2.117) and the deletion of the final a (6.4.148), which were caused by yaÑ, also must be undone, thereby giving us the correct form: garga + Jas 🡪 gargāḥ. However, I do not accept this metarule, as stated above. I perform this derivation as follows. Upon adding the affix yaÑ to garga, the following rules are applicable:
g a rg a + yaÑ
7.2.117 6.4.148 4.1.2
7.2.117 taddhiteṣv acām ādeḥ: the first vowel of the base undergoes vr̥ddhi when an affix marked with Ñ or Ṇ follows in taddhita derivations.
6.4.148 yasyeti ca: the final i or a of a bha item is deleted when it is followed by ī or a taddhita affix.
4.1.2 svaujasamauṭchaṣṭābhyāmbhisṅebhyāmbhyasṅasibhyāmbhyasṅasosāmṅyossup
This is a case of DOI. By my interpretation of 1.4.2, the right-most rule 4.1.2 applies and we get gargya + yaÑ + Jas. Here multiple rules are applicable:
[ g a rg a + yaÑ ] + Jas
7.2.117 6.4.148 2.4.64
7.2.117 taddhiteṣv acām ādeḥ: same as above.
6.4.148 yasyeti ca: same as above.
2.4.64 yañañoś ca: LUK replaces the gotra affixes yaÑ and aÑ introduced after a nominal stem when that nominal stem ending in these affixes itself denotes plurality and is not followed by a feminine affix.
This is a case of DOI. By my interpretation of 1.4.2, we apply the right most rule 2.4.64 and get: garga + Jas 🡪 gargāḥ (6.1.102 prathamayoḥ pūrvasavarṇaḥ30), which is the correct form.
98
Note that, at this point, 7.2.117 and 6.4.148 no longer have a chance to apply. So, unlike the traditional solution, mine does not require us to go backwards to undo the application of rules like 7.2.117 and 6.4.148. Therefore, my solution is more acceptable than the one provided by the tradition.
pañcālāḥ
(4) Now let us derive the nominative plural of ‘a kṣatriya descendent of the country of the pañcālas’ first through the traditional method, and then through mine. The tradition first derives the base and then adds the declensional affix at the end. Consider the following rule:
4.1.168 janapadaśabdāt kṣatriyād añ: the taddhita affix aÑ is added to a syntactically related base ending in the genitive which stands for both a janapada and its class of kṣatriyas, in order to denote the sense of apatya ‘descendent’.
The tradition31 starts by adding the affix aÑ to pañcāla + ām by 4.1.168: [pañcāla + ām] + aÑ. ām is deleted by 2.4.71 supo dhātuprātipadikayoḥ. At this juncture, 7.2.117 taddhiteṣv acām ādeḥ teaches that the first vowel of pañcāla undergoes vr̥ddhi given that the following affix is marked with Ñ. Upon applying this rule, we get: pāñcāla + aÑ. The a of pāñcāla is deleted by 6.4.148 which teaches that the final i or a of a bha item is deleted when it is followed by ī or a taddhita affix. Thus, we get our base pāñcāla.
At this point, the tradition prescribes the addition of the affix Jas to the base pāñcāla: pāñcāla + Jas. By 4.1.174 te tadrājāḥ, the taddhita affixes, including aÑ, which occur after a syntactically related genitive to indicate a ‘kṣatriya descendent of the kṣatriyas of a janapada’ take the technical designation tadrāja. Thus 2.4.62 tadrājasya bahuṣu tenaivāstriyām becomes applicable here: it teaches that LUK replaces a tadrāja affix introduced after a nominal stem when it denotes plurality if that plurality is expressed by the stem ending in that affix except when followed by a feminine affix.
If the derivation stops here, we get pāñcāl + Jas 🡪 *pāñcālaḥ, which is not the correct answer. On 2.4.62, the Bhaimī commentary on the Laghusiddhāntakaumudī suggests the metarule nimittāpāye naimittikasyāpy apāyaḥ, which we have discussed above, to solve this problem: because aÑ is deleted, the vr̥ddhi of the first vowel (cf. 7.2.117) and the deletion of the final a (6.4.148), which were caused by aÑ, also must be undone, thereby giving us the correct form:
99
pañcāla + Jas 🡪 pañcālāḥ. However, I do not accept this metarule, as stated above. I perform this derivation as follows.
[ p a ñcāl a + aÑ ]
7.2.117 6.4.148 4.1.2
7.2.117 taddhiteṣv acām ādeḥ: same as above.
6.4.148 yasyeti ca: same as above.
4.1.2 svaujasamauṭchaṣṭābhyāmbhisṅebhyāmbhyasṅasibhyāmbhyasṅasosāmṅyossup
This is a case of DOI. By my interpretation of 1.4.2, the right-most rule 4.1.2 applies and we get pañcāla + aÑ + Jas. Here multiple rules are applicable:
[ p a ñcāl a + aÑ ] + Jas
7.2.117 6.4.148 2.4.62
7.2.117 taddhiteṣv acām ādeḥ: same as above.
6.4.148 yasyeti ca: same as above.
2.4.62 tadrājasya bahuṣu tenaivāstriyām: same as above.
This is a case of DOI. By my interpretation of 1.4.2, the right-most rule 2.4.62 wins and we get: pañcāla + Jas 🡪 pañcālāḥ (6.1.102 prathamayoḥ pūrvasavarṇaḥ) which is the correct form. As in the previous example, at this point 7.2.117 and 6.4.148 can no longer apply. This shows that my solution is better than the traditional one.
gārgīyāḥ
(5) Now let us derive the nominative plural of ‘the student of gārgya’, or in other words, the student of the descendent of garga’. To derive this form, cha is added to [gārgya + Ṅas] by the following rule:
4.2.114 vr̥ddhāt chaḥ: affix cha is added to a syntactically related item termed vr̥ddha (cf. 1.1.73 vr̥ddhir yasyācām ādis tad vr̥ddham) in the remaining senses.
100In gārgya + Ṅas + cha, Ṅas is deleted by 2.4.71 supo dhātuprātipadikayoḥ and we get gārgya + cha.32 Let us look at my solution first. I will only highlight the cases of conflict here. The following rules are applicable:
[ gārgya + cha ]
7.1.2 4.1.2
7.1.2 āyaneyīnīyiyaḥ phaḍhakhacchaghāṁ pratyayādīnām: the sounds ph, ḍh, kh, ch and gh, when occurring at the beginning of the affix, are replaced with āyan, ey, īn, īy and iy respectively.
4.1.2 svaujasamauṭchaṣṭābhyāmbhisṅebhyāmbhyasṅasibhyāmbhyasṅasosāmṅyossup The two rules do not block each other.
By my interpretation of 1.4.2, we apply the RHS rule 4.1.2 and get: gārgya + cha + Jas. Here, multiple rules are applicable:
[ gārgya + cha ] + Jas
2.4.64 7.1.2
7.1.2 āyaneyīnīyiyaḥ phaḍhakhacchaghāṁ pratyayādīnām: same as above.
2.4.64 yañañoś ca: LUK replaces the gotra affixes yaÑ and aÑ introduced after a nominal stem when that nominal stem ending in these affixes itself denotes plurality and is not followed by a feminine affix.
If we apply 2.4.64 at this step, 7.1.2 will be applicable at the following step. If we apply 7.1.2 at this step, thereby replacing ch of cha with īy (which gives us gārgya + īya), then 4.1.89 gotre’lug aci comes into play:
101
4.1.89 gotre’lug aci: LUK does not replace a taddhita affix denoting a gotra descendant, when the following affix begins with a vowel and is introduced in the prāgdīvyatīya section33.
Therefore, 2.4.64, which teaches LUK, will not be applicable at the following step. 7.1.2 blocks 2.4.64. This is a case of unidirectional blocking, and thus of Type 2a (DOI conflict).
By my interpretation of 1.4.2, the RHS rule 7.1.2 wins and we get gārgya + īya + Jas. Here by applying 6.4.148 yasyeti ca, we get gārgy + īya + Jas. At this stage, 6.4.151 applies:
6.4.151 āpatyasya ca taddhite’nāti: a y which occurs after a consonant and is part of a taddhita affix signifying an apatya ‘off-spring’ which is in turn part of an item termed bha, is replaced with LOPA, when a taddhita affix not beginning with a, follows.
This gives us the correct form: gārgīyāḥ.
Even though Patañjali does discuss this derivation in his commentary on vt. 234 on 4.1.89 gotre’lug aci, he does not discuss this conflict.35
puṣyaḥ
(6) Let us now derive the nominative singular of puṣya ‘a moon (which is) in conjunction with the constellation Puṣya’ of the sentence adya puṣyaḥ ‘today the moon is in conjunction with constellation puṣya. We start by adding the affix aṆ to puṣya + Ṭā by 4.2.3:
4.2.3 nakṣatreṇa yuktaḥ kālaḥ: the taddhita affix aṆ is introduced after a nominal form which signifies a particular constellation (nakṣatra) and ends in tr̥tīyā ‘instrumental’, to denote the time when the moon is in conjunction with that constellation.
By 2.4.71 supo dhātuprātipadikayoḥ, Ṭā is deleted, leading to puṣya + aṆ. Here, the following rules are applicable:
[ p u ṣy a + aṆ ]
7.2.117 6.4.148 4.2.4 4.1.2
102
7.2.117 taddhiteṣv acām ādeḥ: same as above.
6.4.148 yasyeti ca: same as above.
4.2.4 lub aviśeṣe: a taddhita affix introduced after a nominal stem ending in tr̥tīyā and denoting a constellation is replaced with LUP when the time of conjunction is not qualified with specifications.36
4.1.2 svaujasamauṭchaṣṭābhyāmbhisṅebhyāmbhyasṅasibhyāmbhyasṅasosāmṅyossup
Let us look at the relationship of 4.2.4 with the two rules 6.4.148 and 7.2.117. If we apply 4.2.4 at this step, thereby deleting the affix which triggers rules 7.2.117 and 6.4.148, neither of these two rules will be applicable at the following step. However, if we apply any of these two rules at this step, 4.2.4 will still be applicable at the following step. So, 4.2.4 unidirectionally blocks 6.4.148 and 7.2.117 and is thus in conflict with both of them.
By my interpretation of 1.4.2, the right-most rule 4.1.2 applies and we get puṣya + aṆ + sU. Here multiple rules are applicable:
[ p u ṣy a + aṆ ] + sU
7.2.117 6.4.148 4.2.4
By my interpretation of 1.4.2, the right-most rule 4.2.4 applies and we get: puṣya + sU 🡪 puṣyaḥ, which is the correct form.
The Bhaimī commentary on the Laghusiddhāntakaumudī does not mention this conflict. However, after applying 4.2.4 at this step, it does say that by 1.1.63 na lumatāṅgasya, 7.2.117 and 6.4.148 fail to apply at the following step.
103
pañcamaḥ
(7) Let us now derive the nominative singular form of ‘fifth’. We add ḌaṬ to pañcan + Ṅas by the following rule:
5.2.48 tasya pūraṇe ḍaṭ: the taddhita affix ḌaṬ occurs to denote the sense of pūraṇa ‘that by which something is brought to completion, ordinal number’ after a syntactically related nominal stem which signifies number and ends in ṣaṣṭhī ‘genitive’.
In pañcan + Ṅas + ḌaṬ, Ṅas is deleted by 2.4.71 supo dhātuprātipadikayoḥ, so we get pañcan + ḌaṬ. Thereafter, the following rules become applicable:
[ pañc an + ḌaṬ ]
6.4.143 5.2.49 4.1.2 6.4.143 ṭeḥ: the ṭi (cf. 1.1.64 aco’ntyādi ṭi) of an item termed bha is replaced with LOPA when an affix marked with Ḍ follows.
5.2.49 nāntād asaṁkhyāder maṭ: the augment mAṬ is attached to the taddhita affix ḌaṬ when used to denote its ordinal, after a n-final nominal stem which ends in ṣaṣṭhī ‘genitive’ and does not have a number as its initial constituent.
4.1.2 svaujasamauṭchaṣṭābhyāmbhisṅebhyāmbhyasṅasibhyāmbhyasṅasosāmṅyossup 4.1.2 neither blocks nor is blocked by the other two rules.
Now let us look at the relationship between 5.2.49 and 6.4.143. If we apply 5.2.49 at this step, then ḌaṬ will take the augment mAṬ. As a result, it will begin with a consonant. This implies that pañcan is no longer followed by an affix beginning with a vowel or y, and therefore it cannot be called bha. Thus, 6.4.143, which applies only to items termed bha, will not be applicable to an at the following step.
If we apply 6.4.143 at this step, an gets deleted, so ḌaṬ will no longer be preceded by an item ending in n. Therefore, 5.2.49 will not be applicable at the following step.
Both rules block each other. This is a case of mutual blocking, and of Type 2a (DOI conflict).
By my interpretation of 1.4.2, we apply the right-most rule 4.1.2 and get pañcan + ḌaṬ + sU. Here two rules are applicable:
104
[ pañc an + ḌaṬ ] + sU
6.4.143 5.2.49
6.4.143 ṭeḥ: same as above.
5.2.49 nāntād asaṁkhyāder maṭ: same as above.
This is a case of DOI. By my interpretation of 1.4.2, we apply the RHS rule 5.2.49 and get: pañcan + ma + sU. By 1.4.17 svādiṣv asarvanāmasthāne37, pañcan takes the technical designation pada, and so by 8.2.7 nalopaḥ prātipadikāntasya, the n of pañcan gets deleted. Thus, we get the correct form: pañcamaḥ.
On 5.2.49, Nyāsa says that 5.2.49 is antaraṅga with respect to 6.4.143 and thus wins.
kālimmanyā
(8) Let us now look at the derivation of kālimmanyā ‘a woman who considers herself to be Kālī’. This is the feminine version of the upapada compound made of the two padas kālī and manya. manya is derived by adding KHaŚ to the verbal root man ‘to consider’ by the following rule:
3.2.83 ātmamāne khaś ca: affixes KHaŚ and ṆinI are added to the verbal root man when the root co-occurs with a pada which ends in a sUP and the derivate denotes ātmamāna ‘thinking about one’s own self’.
Now, because KHaŚ is marked with Ś, it is a sārvadhātuka affix by 3.4.113 tiṅśit sārvadhātukam. Here, 3.1.69 divādibhyaḥ ŚyaN instructs us to add the affix ŚyaN between the root man, which belongs to the fourth class of verbal roots, and KHaŚ, which is a sārvadhātuka affix used in the active sense. This gives us manya + a. By 6.1.97 ato guṇe, both a and the guṇa sound following it are replaced with the latter. This gives us manya.
105
Now let us build the compound: [kālī Ṅas manya]7. By 2.4.71 supo dhātuprātipadikayoḥ, Ṅas is deleted. Here two rules are simultaneously applicable:
kāl ī - manya
6.3.66 6.3.67
6.3.66 khity anavyayasya: the final vowel of the first member of a compound, except indeclinables, is replaced with a short vowel, when a constituent marked with KH combines to follow.
6.3.67 arurdviṣadajantasya mum: augment mUM is introduced to arus, dviṣat and word ending in a vowel, except indeclinables, when a constituent marked with KH combines to follow.
If we apply 6.3.66 at this step, 6.3.67 will be applicable at the following step. But if we apply 6.3.67 at this step, ī will no longer be the final sound of the pūrvapada. Thus, 6.3.66 will not be applicable at the following step.
This is a case of unidirectional blocking and thus of Type 2a (DOI conflict).
By my interpretation of 1.4.2, the RHS rule 6.3.67 wins, and we get: *kālīmmanya, which is the wrong form. To get the correct form, we first need to apply 6.3.66, thereby shortening ī or kālī, and only then introduce the augment mUM, which gives us the correct form: kālimmanya. After this we add the feminine affix ṬāP by 4.1.4 ajādyataṣ ṭāp to get kālimmanyā.
On this topic, the Kāśikā says: mumā hrasvo na bādhyate, anyathā hi hrasvaśāsanam anarthakaṁ syāt ‘shortening is not blocked by the mUM. Otherwise, the instruction about shortening would be futile’.
Coming back to the problem, how do we explain this anomaly? Notice that these rules have been taught one after another. Let us look at them along with words that have been continued into them by anuvr̥tti.
6.3.66 khity anavyayasya (uttarapade hrasvaḥ)
6.3.67 (khity anavyayasya uttarapade) arurdviṣadajantasya mum
106
While the tradition continues khity, anavyayasya and uttarapade from 6.3.66 into 6.3.67 by anuvr̥tti, it does not continue hrasvaḥ into 6.3.67 by anuvr̥tti. I think that Pāṇini intended for hrasvaḥ too to be continued into 6.3.67 by anuvr̥tti. To facilitate its case agreement with anavyayasya, hrasvaḥ should be read not in the nominative but instead in the genitive, as hrasvasya, in 6.3.67. This gives us the following meaning of 6.3.67: ‘augment mUM is introduced to arus, dviṣat and a word ending in a short vowel, except indeclinables, when a constituent marked with KH combines to follow.’
Let us see how the derivation proceeds if we accept my interpretation of 6.3.67. At the step kālī - manya, only 6.3.66 is applicable. Upon its application, we get: kāli - manya. Now, 6.3.67 applies and we get kālimmanya. Upon adding the feminine affix ṬāP, we get the correct form: kālimmanyā.
-
I use the ‘+’ sign between a base and an affix. Since ac is not an affix with respect to prati, I put a ‘– ’ instead of a ‘+’ between prati and ac. ↩︎
-
Among other things, this rule teaches that KvIN occurs after the root añcU ‘to bend’ when this root co-occurs with a pada ending in sUP. ↩︎
-
LOPA replaces the penultimate n of a verbal base ending in a consonant and not marked with I [in the Dhātupāṭha] when an affix marked with K or Ṅ follows. ↩︎
-
Affix vI unaccompanied [by any other sound] is replaced with LOPA. ↩︎
-
We know this because of a paribhāṣā it mentions, will I will discuss below. 16 Another version of this, which we occasionally find in paribhāṣā texts, is nimittāpāye naimittikasyāpy apāyaḥ. ↩︎
-
SK 417 (6.3.138 cau). ↩︎
-
One could argue that this should be [kālī am manya]. For a detailed discussion on this topic, see Scharf (2016). ↩︎