717 
 
 C74-3 
 
 UC-NRLF 
 
 $C 23M 20=1 
 
E>; CHANGE 
 DEC 11 WIS 
 
 Ube mniversit^ ot Cbtcago 
 
 THE PEPET LAW IN PHILIPPINE 
 LANGUAGES 
 
 A DISSERTATION 
 
 SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF ARTS 
 
 AND LITERATURE IN CANDIDACY FOR THE DEGREE 
 
 OF DOCTOR OF PHILOSOPHY 
 
 (DEPARTMENT OF SANSKRIT AND COMPARATIVE PHILOLOGY) 
 
 BY 
 
 CARLOS EVERETT CONANT 
 
 CHICAGO 
 
 1913 
 
TLhc xnniverstts of Cbicaao 
 
 THE PEPET LAW IN PHILIPPINE 
 LANGUAGES 
 
 A DISSERTATION 
 
 SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF ARTS 
 
 AND LITERATURE IN CANDIDACY FOR THE DEGREE 
 
 OF DOCTOR OF PHILOSOPHY 
 
 (DEPARTMENT OF SANSKRIT AND COMPARATIVE PHILOLOGY) 
 
 BY 
 
 CARLOS EVERETT CONANT 
 
 CHICAGO 
 
 1913 
 
The Pepet Law in Philippine Languages. 
 
 By Carlos Everett Conant, Lecturer of Indonesian Languages, University of Chicago. 
 
 In the vocalism of Indonesian languages the original indifferent vowel 
 plays an important role. Resembling the Hebrew shewa, and the obscure vo ve j 
 of many Indo-European languages, it was so colorless and indefinite in pn. 
 nunciation that it developed differently in different speech groups. In somt 
 languages it remained practically unchanged, as in Javanese, where it is called 
 pepet 1 , while in others it evolved into various and more or less definite vowel 
 sounds, e. g., IN 2 atep roof became Jav. atep, Mai. atap, Tag. dtip, and Bis. atup. 
 
 Brandstetter 3 gives the following concise statement of the varied re- 
 presentation of pepet in several of the more important languages of Indonesia: 
 
 "The Pepet Law: Where the IN parent speech (Ursprache) had an e 
 (called pepet in Javanese), OJav., Tontb., Bug. and Karo also have e, Mkb. 
 and Mak. a, Bis. and Toba o, Tag. /, Day. e, Mai. in final syllable a, in the 
 penultimate syllable <?, Mlg. in accented syllable e, in a syllable following the 
 tone, /." 
 
 The following table will illustrate the above law: 
 
 rice 
 
 sugar cane 
 
 roof 
 
 hear six 
 
 Jav. 
 
 tebu 
 
 atep 
 
 defter enem 
 
 Mak. 
 
 tabu 
 
 ata 
 
 laftere anaft 
 
 Bis. bugds 
 
 tubo* 
 
 atup 
 
 liufidg unom 
 
 Toba boras 
 
 tobu 
 
 — 
 
 onom 
 
 Tag. bigds 
 
 tubd 
 
 dtip 
 
 dinig anim 
 
 Day. behas 
 
 tewu 
 
 atep door 
 
 — — 
 
 Mai. beras 
 
 tebu 
 
 atap 
 
 deftat enam 
 
 Mlg 
 
 1 thrughout the article to desij 
 
 rent enim. 
 
 ' Pepet or' e will be employee 
 
 gnate the original indifl 
 
 2 The list of abbreviations: 
 
 
 
 
 Bgb. Bagobo 
 
 IN Indonesian 
 
 
 OMlg. Old Malagasi 
 
 Ban. Banawi 
 
 Inb. Inibaloi 
 
 
 Pamp. Pampanga 
 
 Bat. Batan 
 
 Isn. Isinai 
 
 
 Pang. Pangasinan 
 
 Bil. Bilan 
 
 Itw. Itawi 
 
 
 Phil. Philippine 
 
 Bkl. Bikol 
 
 Jav. Javanese 
 
 
 Sml. Samal 
 
 Bis. Bisaya 
 
 Kim. Kalamian 
 
 
 Sbl. Sambal 
 
 Bol. Bolinao 
 
 Knk. Kankanai 
 
 
 Sng. Sangir 
 
 Bon. Bontok 
 
 Kuy. Kuyunon 
 
 
 Sund. Sunda 
 
 Bug. Bugis 
 
 Lep. Lepanto 
 
 
 Tgk. Tagakaolo 
 
 Chro. Chamorro 
 
 Mgd. Magindanau 
 
 Tag. Tagalog 
 
 Day. Dayak 
 
 Mak. Makassar 
 
 
 Tgb. Tagbamv.i 
 
 Gad. Gaddang 
 
 Mai. Malay 
 
 
 Ting. Tingyan 
 
 Har. Haraya 
 
 Mlg. Malagasi 
 
 
 Tir. Tirurai 
 
 Hlg. Hiligaina 
 
 Mnb. Manobo 
 
 
 Tontb. Tontemboan 
 
 Ibg. Ibanag 
 
 Mkb. Minankabau 
 
 Yog. Yogad 
 
 Ilk. Iloko 
 
 OJav. Old Javanese 
 
 
 Bis., when not qualified, will be here intended to include the three great dialects, Hili- 
 gaina, Cebuan, and the Samar-Leyte dialect. 
 5 "Mata-Hari", Luzern 1908, p. 52. 
 * Unless otherwise indicated, u and o are interchangeable in Philippine languages. 
 
The Pcpet Law in Philippine Languages. 
 
 921 
 
 For a of Tag. tubo see below, p. 933. 
 
 It is the purpose of the present study to trace the evolution of this 
 indifferent vowel (pepet) thru several of the Philippine languages and dialects. 
 The material will be treated under seven classes or types, as follows: 
 
 I. the a/7-class, represented by Phil, atep roof, i. e. words having a in 
 the first of two syllables the second of which has pepet; 
 
 II. the /?#-class: Phil, begas rice; 
 
 III. the //7-class: Phil, hipen tooth; 
 
 IV. the /?/-class: Phil, bell to buy; 
 
 V. the M/7-class: Phil, pused navel; 
 VI. the /?«-class: Phil, penu full; 
 
 VII. the /7/7-class: Phil, lebeh to excavate. ■ 
 
 Following out this classification, it will be convenient to limit the study 
 at first to eleven of the more conspicuous speech groups, viz.: Tag., Pang., 
 Ilk., Mgd., Tir., Pamp., Ibg., Bkl, Bis., Bgb., and Sulu. The first comparative 
 table will give a general view of the phenomena of the seven classes in the 
 eleven languages named. This will be followed by a series of seven tables, 
 each illustrating a single class, and arranged in the order given above. After 
 studying the phenomena of the pepet law as shown by the material thus 
 presented, other languages and dialects will be examined according to the 
 same classification, tho v less formally and completely, owing to their greater 
 scarcity of available material. 
 
 Class 
 
 I ap 
 
 \l p a 
 
 III ip 
 
 IV pi 
 
 V up 
 
 VI pit 
 
 VII pp 
 
 Phil. 
 
 atep 
 
 begas 
 
 hipen 
 
 beli 
 
 pused 
 
 penu 
 
 lebeh 
 
 Tag. 
 
 diip 
 
 bigds 
 
 nipin 
 
 bili 
 
 pusud 
 
 puno 
 
 libih 
 
 Pang. 
 
 ate > 
 
 belds 
 
 hipen 
 
 bili 
 
 pus eg 
 
 pdnu 
 
 — 
 
 Ilk. 
 
 atep 
 
 • bagds 
 
 hipen 
 
 — 
 
 piiseg 
 
 punno 
 
 — 
 
 Mgd. 
 
 atep 1 
 
 begds 
 
 hipen 
 
 — 
 
 puset 
 
 penu 
 
 lebeh 
 
 Tir. 
 
 atef 
 
 begds 
 
 kifen 
 
 betlei 
 
 fused 
 
 fend 
 
 lebeh 
 
 Pamp. 
 
 atdp 
 
 abyds 
 
 ipan 
 
 abli 
 
 pusad 
 
 apnii 
 
 albdh 
 
 Ibg. 
 
 at dp 
 
 baggd' 
 
 hipan 
 
 balli 
 
 futdd 
 
 pannii 
 
 labbdh 
 
 Bkl. 
 
 at up 
 
 bagds 
 
 hipon 
 
 bili 
 
 pusod 
 
 pand 
 
 lubdh 
 
 Bis. 
 
 atiip 
 
 bugds 
 
 hipon 
 
 bili 
 
 piisod 
 
 pund 
 
 luboh 
 
 Bgb. 
 
 atop 
 
 buggds 
 
 hipon 
 
 balli 
 
 pusod 
 
 punno 
 
 lubboh 
 
 Sulu 
 
 a tup 
 
 bugas 
 
 ipun 
 
 bi 
 
 pusud 
 
 — 
 
 lubah. 
 
 An examination of the above table with reference to the individual lan- 
 guages shows that pepet regularly becomes i in Tag., e in Pang., Ilk., Mgd. 
 and Tir., a in Pamp. and Ibg. and u in Bkl., Bis., Bgb. and Sulu. Languages 
 which, like Tag., regularly show i for original pepet, may be spoken of as 
 /-languages, those of the Pang, type, as ^-languages and those of the Pamp. 
 and Bgb. types, as a-languages and ^-languages, respectively. 
 
 Before proceeding to our comparative and analytic study of the pepet 
 vocalism, attention should be called to certain other phonological peculiarities 
 
 1 The sources for Mgd., Tir., Bgb. and Sulu do not, as a general rule, indicate the stress 
 accent. Wherever ascertainable, the stressed syllable will be marked in this paper by the acute 
 accent ('). 
 
 270919 
 
922 Carlos Everett Conant, 
 
 of the languages examined. Most, if not all, of these peculiarities will be found 
 to exist to a greater or less extent in other Indonesian speech groups, and 
 a general comparative study of any one of them, metathesis, for example, 
 would be worthy of separate treatment in a copious article. For our present 
 purpose, however, it will be sufficient to note such secondary phonetic changes 
 as must be taken into account in order to recognize the original phonetic 
 equivalence of words so dissimilar in appearance as Pamp. abyds and Ibg. 
 baggd 1 , both accurately representing Phil, begas rice, according to individual 
 phonetic laws of the two languages. Thus, while both Pamp. and Ibg. are 
 a-languages regularly showing a for pepet in the penult, the Pamp. abyds 
 has metathesis of the first syllable and y for the usual Phil, g of the RGH 
 series, neither of which phenomena is shared by Ibg. baggd 1 , which doubles 
 the Phil, g and represents Phil, final s, as regularly, by an Ibg. t that has 
 degenerated to the glottal top (hamza), tho it is retained with full pronun- 
 ciation in the Ibg. dialects, Gad., Itw. and Yog. baggdt Compare here Ibg. 
 appd' four, beside Gad., Itw., Yog. appdt 
 
 Consonant gemination. — Several Phil, languages and dialects 
 double a single intervocalic consonant under certain conditions. 
 
 The languages of the above table which show this doubling are Ilk., 
 Ibg. and Bgb., the examples being Ilk. punno, Ibg. baggd', balli, pannii, labbdn, 
 and Bgb. buggds, balli, punno, lubbon. 
 
 Other speech groups showing gemination of consonants are the Ibanag 
 dialects called Gaddang, Itawi, and Yogad, and the Igorot dialect, Inibaloi. 
 The following brief table will illustrate the more common cases: 
 
 Phil. Ilk. Ibg. Gad. Itw. Yog. Inb. Bgb. 
 
 four epat uppdt appd' appdt appdt appdt appat appat 
 
 six enem innem anndm annem ennSm annSm annim annam 
 
 seven pitu pito pitu pitu pitu pitu pitto pitto. 
 
 In all these languages the gemination is real, that is, the two consonants 
 are distinctly pronounced, e. g., the pp of the word for "four" is sounded as 
 in Ital. Giuseppe, and not as in Eng. upper. 
 
 The first two of the three examples follow the law of gemination of a 
 single consonant following a pepet vowel (see below, pp. 927 ff.). But it is to 
 be noted that the t of Phil, pitu, where the preceding vowel is not originally 
 pepet, but /, is doubled only in Inibaloi and Bagobo, an indication that these 
 two languages have a stronger tendency to gemination than the others, tho 
 in this instance it is quite possible that the phenomenon is due to analogy 
 with the gemination of the other numerals. This latter explanation is further 
 borne out by the persistence of the single t of Phil, batii "stone" in all the 
 geminating languages here enumerated 1 . 
 
 1 Blake, "Contributions to Philippine Grammar", Jour. Am. Or. Soc, vol. 27, New Haven 
 1907, p. 336, has noticed the doubling of single consonants in Ilk. and Ibg., but one of the 
 two examples given for Ibg., namely, battu, is erroneous, the correct form being batii. In the 
 same article (pp. 331 and 332) attention is called to the varied vocalism seen in Tag. bigas, 
 difttg, silid, the suffix-m, and anirn, and their cognates in Bis. Bkl. Ilk. Pang. Mgd. Ibg. and 
 Pamp. It is then stated "quite possible that this varied vocalism is the representation of a fourth 
 
The Pepet Law in Philippine Languages. 923 
 
 The rr written by Padre BennAsar in Tir. words is not a case of gemi- 
 nation, but is the Spanish mode of representing a single r sharply trilled. In 
 certain languages, notably, Ibg. and its dialects, double consonants are often 
 the result of assimilation rather than gemination (see below, under consonant 
 assimilation). 
 
 Consonant assimilation. — Cases of both partial and total assimi- 
 lation 1 are to be found in abundance in certain Philippine speech groups. 
 
 The most common illustration of partial assimilation is that of a nasal 
 conforming to the class of the following consonant, a common example being 
 the variants Tag., Pamp., Mgd., Sulu, Tir., Bgb., Kuy. kambih goat, and Bkl. 
 Bis. kandin, Gad., Itw". gandin. The Ibg. word kazzih shows total assimilation. 
 The most striking example of partial assimilation presented by the material 
 to be examined in this paper is that of the Pamp. change of a stop con- 
 sonant to the class of the consonant immediately following. The consonants 
 in question are most commonly brot into contact with each other as a result 
 of metathesis, e. g. Pamp. abpd fathom from Phil, depa, where, after meta- 
 thesis, the dental sonant d becomes the labial sonant b before the labial surd 
 p. In the same manner labial-to-palatal assimilation is shown by Pamp. agkds, 
 from. Phil, bekas to shoot an arrow, and the labial p of Phil, apdu gall 
 becomes the dental t before d in Pamp. atdii. This partial assimilation of 
 stops is, however, very limited and of exceptional occurrence, even in Pamp., 
 as is shown by Pamp. atbti (Phil, tebu), atbiis (Phil, tebus), abldk (Phil, betak), 
 akbdg (Phil, kabag), akddl (Phil. kadSl), agtdl (Phil, getel), apdd (contrasted 
 with atdu for apdd), and the Pamp. variants agpdh and abpdh rule, standard. 
 
 Total assimilation is a characteristic of some languages, notably Ibg. and 
 its dialects, e. g. Ibg. dggu gall (Phil, apdu), dggau day (Ilk. &c. aldau), the 
 consonant of the RLD series becoming g in Ibg. as in iguh nose and 'piga 
 how much?, Ibg. Mug egg (Phil, Mug), Ibg., Itw. uffu, Gad. uffu, beside 
 Pang, ulpo thigh. The Ilk. equivalent luppo shows metathesis and gemination. 
 Assimilation follows metathesis in Ibg. appd (Phil, depa) fathom (see below, 
 table II). The case of Ibg. tallu &c. will be treated below (p. 935). The Ibg. 
 assimilation of a final consonant to a following initial consonant does not 
 concern us here. 
 
 Metathesis. — This, perhaps the most striking characteristic of the In- 
 donesian languages, shows a high degree of development in Philippine speech, 
 where its manifestations are exceedingly varied and often so complex as to 
 render their classification difficult. 
 
 primitive Philippine vowel, an indistinct vowel like the Indo-European shewa (Cf. Brandstetter, 
 "Tag. u. Mad.", p. 34), which in a similar way is represented by several different vowels in 
 the various Indo-European languages (Cf. Brugmann, "GrundriS &c", zweite Bearb., Strafiburg 
 1897, Bd. 1, p. 170)". The existence of the pepet vowel in the IN parent speech had years 
 before been established by the Dutch scholars and Brandstetter, who had identified this 
 obscure vowel with the prototype of the i : u correspondence of Tag. bigds and Bis bugds. My 
 own study of the pepet vocalism of Phil, languges was begun in the Philippine Islands in 1901 
 and was suggested by Brandstetter's treatment of the IN obscure vowel in his "Die Beziehungen 
 des Malagasy zum Malaischen", Luzern 1893, pp. 21, 22, 23, et passim. 
 
 1 Cf. SlEVERS, "Grundzuge der Phonetik", 5 th ed., Leipzig 1901, p. 277. 
 
924 Carlos Everett Conant, 
 
 A case commonly noted is that of the metathesis of two consonants 
 thrown together by the syncopation of an intervening vowel from which the 
 stress has been removed by the addition of a formative suffix, e. g. Tag. aptdn 
 from atip, Bis. imnon for inumon from iniitn; but the cases of metathesis 
 appearing in the material collected for the present study are mostly of a 
 different character, in which the transposition is not of concurrent consonants, 
 but of a consonant and an adjacent vowel or of two consonants more or less 
 widely separated. Metathesis of a consonant and adjacent vowel is seen in 
 Pamp. altdu, abyds, atyds, abpd (table II below), Ibg. appd (for adpa < dapa), 
 Mgd. alpd or arpd (beside lepd, repd), and Ilk. luppo beside Pang, ulpo (see 
 above). 
 
 Metathesis of consonants separated by a vowel is seen in Bkl. gabdt 
 (Phil, begat) weight, Ilk. gasiit (Phil, gatus) hundred, Ilk. gessdt (Phil, getas) 
 to cut or break thread. Initial and final consonants exchange places in Ilk. 
 sagdt (Phil, tegas) hard, Ilk. subbut (Phil, tebus) to redeem, the Ilk. variants 
 gorrood and dolloog thunder, and Pang, sennit, Ilk. sam'it beside Tag., Bis. 
 tam'is sweet. 
 
 Loss of intervocalic /. — Several languages show, with greater or less 
 regularity, loss of an / between vowels, sometimes with, and sometimes without, 
 resulting contraction. Sulu always drops / between two like vowels, which 
 are then contracted, e. g. dan (Phil, dalan) way, bl (for bill, Phil, bill) to 
 buy, o (for olo, Phil, ulu) head. The / is retained in Sulu wain eight but lost 
 in kauhan (Cebu Bis. kaluha'dn) twenty. The loss is less regular in Tag. 
 where no resulting contraction takes place, e. g. ddan way, but dalan to sow; 
 bili to buy; pdo or pdwo ten (Phil, pulu), but dlo head. In Bontok "ten" is 
 (sim)po'o, ' while three and eight are told and walo, respectively. Kankanai 
 and Tingyan also have tula (told), but wd'o (Phil, wala), and (sim)po ten, 
 tho / reappears in Ting, duapulu twenty. In Isn. the Phil, numerals tela, wala 
 and pulu become tin, weu and piu, respectively, while / remains in Isn. sala 
 sin, and tulid straight. 
 
 RGH and RLD laws. — The phenomena of these laws, even within the 
 limits of Philippine territory, are too varied and complex to permit of detailed 
 study here 1 . While the consonant of the RGH series appears in most Phil, 
 languages as g, as contrasted with the r of Toba and Mai. and the h of Day. 
 and Sangir, there are several of them in which it is represented by other 
 sounds, notably r, I und y, tho the Phil, g often appears in the same lan- 
 guages alongside the other representatives. The following table, showing 
 examples for the RGH consonant in initial, medial and final position, will 
 present the more common cases: 
 
 
 night 
 
 hundred 
 
 vein 
 
 rice 
 
 lip 
 
 Phil. 
 
 gabi 
 
 gatus 
 
 ugat 
 
 begas 
 
 bibig 
 
 Ilk. 
 
 rabii 
 
 gasiit 
 
 urdt 
 
 bagds 
 
 bibir and bibig 
 
 Tir. 
 
 — 
 
 ratus 
 
 urrat 
 
 begds 
 
 b6w£r 
 
 Pang. 
 
 Idbi 
 
 lasus 
 
 uldt 
 
 belds 
 
 bibil 
 
 1 For a more extensive study of these laws see "RGH Law in Philippine languages", 
 JAOS vol. XXXI, pp. 70—85. 
 
The Pepet Law in Philippine Languages. 
 
 925 
 
 
 night 
 
 hundred 
 
 vein 
 
 rice 
 
 Inb. 
 
 (ka)M(an) 
 
 dasus 
 
 ulat 
 
 bekds 
 
 Kim. 
 
 labii 
 
 — 
 
 — 
 
 — 
 
 Knk. 
 
 labi 
 
 gasut 
 
 uwat 
 
 — 
 
 Bon. 
 
 lafi 
 
 lasdt 
 
 wath 1 
 
 — 
 
 Lep. 
 
 labi 
 
 — 
 
 uat 
 
 — 
 
 Ban. 
 
 — 
 
 — 
 
 ulot 
 
 — 
 
 Ting. 
 
 labi 
 
 kdsut 
 
 — 
 
 bogas 
 
 Pamp. 
 
 — 
 
 gat us 
 
 uydt 
 
 abyds 
 
 Bat. 
 
 — 
 
 yatus 
 
 uyat 
 
 — 
 
 lp 
 
 bibil 
 
 bibi. 
 
 Ilk. and Tir. are the /--languages, the r being more regularly found in 
 the latter. The /-languages are Pang., Kim. and the Inb., Knk., Bon., Lep., 
 Ban. and Ting. The ^/-languages are Pamp. and Bat. It is to be noted that 
 most of these languages have also g in some of the examples, this g in a 
 few cases being changed to the corresponding surd k. 
 
 Cases of apparent irregularity in the representation of the RLD con- 
 sonant will be treated as they appear in the tables that follow. 
 
 Other phonetic peculiarities appearing in the material to be examined 
 will be given attention only when deemed necessary in order to identify a 
 word with its cognates in other languages. 
 
 We now proceed to the study of the pepet law by examining the material 
 classified as outlined above (p. 921). 
 
 
 
 
 Ta 
 
 ble I: the 
 
 ap-class. 
 within, 
 
 
 
 
 roof 
 
 plant 
 
 grasp 
 
 sharp 
 
 under 
 
 great, much 
 
 leech 
 
 Phil. 
 
 atep 
 
 tan em 
 
 dakep 
 
 tarem 
 
 dalem 
 
 dake-l-a 
 
 limatek 
 
 Tag. 
 
 dtip 
 
 tanim 
 
 dakip 
 
 talim 
 
 lalim 
 
 dakild, malaki 
 
 limdtik 
 
 Pang. 
 
 atep 
 
 tanem 
 
 dakep 
 
 tarem 
 
 dalem 
 
 dakel 
 
 — 
 
 Ilk. 
 
 atep 
 
 tanim 
 
 dakep 
 
 tadem 
 
 addlem 
 
 dakkel 
 
 alimdtek 
 
 Mgd. 
 
 atep 
 
 — 
 
 dakep 
 
 tarem 
 
 idalem 
 
 dakel 
 
 limatek 
 
 Tir. 
 
 atef 
 
 — 
 
 — 
 
 tarrem 
 
 dalem 
 
 dakel 
 
 limetek 
 
 Pamp. 
 
 at dp 
 
 tandm 
 
 dakdp 
 
 tardm 
 
 lalam 
 
 dakdl 
 
 limdtak 
 
 Ibg. 
 
 atdv 
 
 tandm 
 
 dakdi' 
 
 tardm 
 
 araldm 
 
 dakdl 
 
 alimatd k 
 
 Bkl. 
 
 atup 
 
 tanum 
 
 dakup 
 
 tardm 
 
 irdrum 
 
 dakul, dakuld 
 
 limdtuk 
 
 Bis. 
 
 at up 
 
 tanum 
 
 dakup 
 
 talum 
 
 ddlum 
 
 daku 
 
 limdtuk 
 
 Bgb. 
 
 atop 
 
 — 
 
 dakop 
 
 — 
 
 tadalom 
 
 ddkol 
 
 limatok 
 
 Sulu 
 
 atup 
 
 tanam 
 
 dakup 
 
 — 
 
 ha-lum 
 
 dakola 
 
 limatok. 
 
 In the a/7-class the operation of the pepet law is remarkably uniform. 
 If we disregard Sulu tanam, which may have been borrowed from Mai., the 
 examples in the above table show no exception to the rule that pepet becomes 
 i in Tag., e in Pang., Ilk., Mgd. and Tir., a in Pamp. and Ibg., and u (or o) 
 in Bkl., Bis., Bgb. and Sulu. The o of Ibg. ato p , dako p , alimato k is no excep- 
 tion, as it regularly stands for an Ibg. a representing Phil, e when followed 
 by a final glottal stop (hamza) which represents one of the surd stops, k, t 
 or p. This o has an open sound as in Ital. pud, and is entirely distinct from 
 the Ibg. a. When a suffix is added to the root, the surd stop is restored and 
 
 1 The Bon. examples in this paper are taken from Jenks, "The Bontoc Igorot", Manila 1906. 
 
926 Carlos Everett Conant, 
 
 the Ibg. a reappears, e. g., alo p , with the locative suffix -an, becomes atappdn 
 place of roof(ing) with Ibg. doubling of original p. Phil, a regularly remains 
 unchanged in Ibg., e. g. baggd 1 , taggd 1 (table II below). 
 
 The consonant of the RLD series is given in the hypothetical Phil, words 
 heading the tables as r when medial (tarem), and as d when initial (dalem) 
 or final (pused, table V). This r is here used merely as a convenient symbol 
 and is not to be considered as in any way indicative of the original character 
 of the RLD consonant, which in the majority of Phil, languages appears as / 
 when intervocative. I have chosen r in order to differentiate the RLD con- 
 sonant from an original /. 
 
 The discussion of prefixed elements, as seen in the case of Phil. dal£m 
 and limatek, where the identity of the examples is evident, is here unnecessary. 
 
 Sulu ha-lum is for ha-lalum (<ha-dalum) with loss of intervocalic / and 
 resultant contraction. The Phil, words for "great, much" show three variations: 
 dake, dakel and dakela. The first variation is shown by Bis. daku and Tag. 
 malaki (for tna-daki). The intermediate dakel is the prototype of the majority 
 of the examples, including, besides those here given, Batan rakah, where h 
 represents Phil. / (see below, p. 939). The third variation appears in Tag. dakild, 
 Bkl. dakuld (great beside dakiil much), and Sulu dakola, and probably in 
 Kim. dakolo and Chro. ddnkulo 1 . 
 
 Table II: the pa-class. 
 
 
 
 
 
 to 
 
 
 
 erupt. 
 
 
 rice 
 
 hard 
 
 fathom 
 
 chew 
 
 demolish 
 
 weight 
 
 float 
 
 Phil. 
 
 begas 
 
 tegas 
 
 depa 
 
 sepa 
 
 geba 
 
 begat 
 
 letau 
 
 Tag. 
 
 bigds 
 
 tigds 
 
 dipd 
 
 sapd 
 
 gibd 
 
 big'dt 
 
 litdu 
 
 Pang. 
 
 belds 
 
 segdt 
 
 depa 
 
 sepd 
 
 gebd 
 
 beldt 
 
 letdu 
 
 Ilk. 
 
 bagds 
 
 sagdt 
 
 deppd 
 
 sapd 
 
 rebbd 
 
 — 
 
 lettdu 
 
 Mgd. 
 
 begds 
 
 tegds 
 
 lepd, repd, 
 alpd, arpd 
 
 sepd 
 
 gebd 
 
 begat 
 
 letau 
 
 Tir. 
 
 begds 
 
 tegds 
 
 — 
 
 sefd 
 
 gebd 
 
 begat 
 
 letau 
 
 Pamp. 
 
 abyds 
 
 atyds 
 
 abpd 
 
 sapd 
 
 — 
 
 bdyat 
 
 altdu 
 
 Ibg. 
 
 baggd' 
 
 taggd* 
 
 appd 
 
 sapd 
 
 — 
 
 — 
 
 Idtau 
 
 Bkl. 
 
 bagds 
 
 tagds 
 
 dupd 
 
 sdpa 
 
 gabd 
 
 gabdt 
 
 latdu 
 
 Bis. 
 
 bugds 
 
 tugds 
 
 dupd 
 
 supd 
 
 gubd 
 
 bug'dt 
 
 lutdu 
 
 Bgb. 
 
 bugds 
 
 tuggds 
 
 duppd 
 
 suppa 
 
 gubbd 
 
 — 
 
 luttau 
 
 Sulu 
 
 bugas 
 
 — 
 
 dupa 
 
 sopah 
 
 — 
 
 bog at 
 
 — 
 
 With the exception of the Ilk. and Bkl. examples, this class shows a 
 uniform and undisturbed operation of the pepet law. Tag. sapd is an isolated 
 exception for which I have found no parallel among the thirty words of this 
 class which I have examined. The Mgd. variants for Phil, depa are interesting 
 as showing the unstable representation of RLD in that language, the con- 
 sonant occurring indifferently, as r, I or d (cf. further Mgd. rugu, lugu or 
 dugu blood), and a peculiar metathesis in which le, re become, with change 
 of vowel, al, ar. 
 
 1 But see my paper "Consonant Changes and Vowel Harmony in Chamorro", "Anthropos", 
 vol. VI (1911), pp. 136—146. 
 
The Pepet Law in Philippine Languages. 927 
 
 It is especially worthy of note that the geminating languages, Ilk., Ibg. 
 and Bgb., show frequent doubling in this class, that is, of a consonant follo- 
 wing the pepet vowel, while table I shows only one case of gemination (Ilk. 
 dakkel) before the pepet vowel. Likewise in the following tables, as a general 
 rule classes IV, VI and VII show doubling of the medial consonant in the 
 geminating languages, as contrasted with classes III and V, in which no examples 
 of gemination appear. Ibg. tdddag is not an example of gemination, but of 
 assimilation (Phil, tindeg). It may therefore be set down as a working rule that!- 
 Those Philippine languages and dialects which permit of consonant 
 gemination double a single intervocalic consonant preceded by a 
 vowel representing original pepet, whatever be the nature of the 
 following vowel; but this gemination does not take place between 
 two vowels of different origin, the second of which is a pepet vowel. 
 Ilk. bagds, sagdt and sapd, contrary to the above rule, show no gemi- 
 nation, and at the same time have a instead of e in the first syllable, while 
 the other Ilk. examples have the regular doubling and the e representation 
 of pepet. The bagds type, tho constituting half the examples here given, is 
 exceptional (see additional list of pa-class examples given below), but the 
 regular coincidence of the a vocalism and the single consonant is significant, 
 and not to be regarded as merely accidental. The identity of the three words 
 with those listed as their cognates in other Phil, languages is unquestionable, 
 from both the semantic and the phonetic standpoint, the metathesized sagdt 
 being supported by Pang, segdt and the similarly metathesized Ilk. subbiit 
 (Phil, tebus) to redeem (table VI), and the pepet origin of the first a of sapd, 
 in spite of the isolated Tag. sapd, being further vouched for by Mai., Jav. 
 sapah and Toba sopa. Thus we are prepared to treat these three words as 
 forming a category of Ilk. words in the pa-class showing at the same time 
 the ungeminated consonant and the a vocalism of pepet. The explanation of 
 this striking phenomenon is simply that the pepet vowel is assimilated to the 
 a of the following syllable when only a single consonant intervenes, while 
 the attractive force of the a of the second syllable is not sufficient to affect 
 the pepet vowel of the preceding syllable when the two vowels are further 
 separated by gemination. The vowel u (o), on the other hand, has a stronger 
 influence in Ilk. over the pepet vowel of the preceding syllable, as shown by 
 Ilk. punno (Phil, penu), tubbo (Phil, tebu), &c. of table VI, where the regres- 
 sive vocalic assimilation takes place in spite of the intervening gemination. 
 The following additional examples are given as further illustration of the pepet 
 vocalism of the pa-type in Ilk. and Bkl.: 
 
 Phil. Ilk. Bkl. Tag. 
 
 bekas bekkds bukds bikds 
 
 betak bettdk batdk bitdk 
 
 beak — badk bidk 
 
 deg'as deg'ds — dig'ds 
 
 getas gessdt gatds — gutas 
 
 Ilk. deg'ds is no exception to the rule for gemination nor to the law of 
 assimilation just stated, since the hamza takes the place of, and is equivalent 
 to, an additional consonant. 
 
 Bis. 
 
 Phil. 
 
 Ilk. 
 
 Bkl. 
 
 Tag. 
 
 Bis. 
 
 bukds 
 
 left a 
 
 left/id 
 
 laftd 
 
 Una 
 
 lurid 
 
 butak 
 
 letak 
 
 lettdk 
 
 latdk 
 
 litdk 
 
 lutak 
 
 bu'ak 
 
 pesa 
 
 pessd 
 
 pasd 
 
 pisd 
 
 pusd 
 
 dug'ds 
 
 tena 
 
 tennd 
 
 taftd 
 
 Una 
 
 tufta. 
 
928 
 
 Carlos Everett Conant, 
 
 Turning our attention now to the Bkl. words of this class, we find that 
 they regularly show a for original pepet, instead of the u of the ap-class 
 (table I), only two out of the fifteen examples cited showing u, viz., dupd 
 and bukds. Is this Bkl. a to be explained also as due to vocalic assimilation 
 as in the case of the Ilk. bagds-iype? Our answer depends upon an exami- 
 nation of the Bkl. examples of the other classes where regressive vocalic 
 assimilation could affect the pepet vowel, that is to say, those having the 
 pepet vowel in the first syllable. In all of these classes there are Bkl. examples 
 showing a for pepet, apparently without regard to the quality of the vowel 
 in the next syllable, e. g. Bkl. sapi (Phil, sepi), pano (Phil, penu), danug 
 (Phil, deneg), lunud or lamld (Phil, lened). On the other hand, Bkl. always 
 shows u (o) for pepet in a final syllable (cf. tables I, III, V and VII). We thus 
 discover that Bkl. has a tendency to represent pepet in the penultimate syl- 
 lable by a, and is therefore not a w-language exclusively, but also an a-lan- 
 guage in so far as the natural representation of penultimate pepet is con- 
 cerned. We are prepared to say, then, that the first a of Bkl. bagds is not 
 to be explained in the same manner as that of tfai Ilk. bagds, namely, as a 
 case of assimilation, but the regular Bkl. vocalism of penultimate pepet. Such 
 occasional exceptions as dupd and bukds are probably due to the working 
 of analogy. For the metathesis of Bkl. gabdt and Ilk. gessdt, see above (p. 924). 
 Worthy of note is the variety of the RGH consonant in Pang., Ilk. and Pamp., 
 while the Tir. examples in table II show only g (see above, p. 926). 
 
 
 
 Tabl 
 
 e III: the ip- 
 
 ■class. 
 
 
 
 
 tooth 
 
 slave 
 
 cockroach 
 
 desire 
 
 to stand 
 
 black 
 
 Phil. 
 
 fiipen 
 
 (e)ripen 
 
 ipes 
 
 ibeg 
 
 tindeg 
 
 item 
 
 Tag. 
 
 hipin 
 
 alipin 
 
 ipis 
 
 ibig 
 
 tindig 
 
 itim 
 
 Pang. 
 
 fiipen 
 
 aripen 
 
 ipes 
 
 ibeg 
 
 talindeg 
 
 — 
 
 Ilk. 
 
 fiipen 
 
 adipen 
 
 ipes 
 
 — 
 
 — 
 
 — 
 
 Mgd. 
 
 fiipen 
 
 uripen 
 
 ipes 
 
 — 
 
 tindeg 
 
 item 
 
 Tir. 
 
 kifen 
 
 rifen 
 
 ifes 
 
 ibeg 
 
 tindeg 
 
 itam 
 
 Pamp. 
 
 ipan 
 
 alipan 
 
 ipds 
 
 — 
 
 tindig 
 
 — 
 
 Ibg. 
 
 nipan 
 
 aripan 
 
 ipd 1 
 
 — 
 
 tdddag 
 
 — 
 
 Bkl. 
 
 nipon 
 
 oripun 
 
 — 
 
 ibug 
 
 tindug 
 
 itum 
 
 Bis. 
 
 fiipon 
 
 ulipon 
 
 ipus 
 
 ibug 
 
 ttndug 
 
 ittim 
 
 Bgb. 
 
 nipon 
 
 — 
 
 ipiis 
 
 ibug 
 
 tindug 
 
 itum 
 
 Sulu 
 
 ipun 
 
 ipun 
 
 — 
 
 — 
 
 tindog 
 
 itum 
 
 The jp-class, like the ap-class, shows remarkable uniformity of the pepet 
 vocalism. The i of the first syllable evidently exerts no influence over the 
 pepet vowel. Pamp. tindig and tinig (Phil, tinig, Tag. tihig. Bis. tihog voice) 
 are exceptions for which there is at present no explanation. The possibility 
 that they are borrowed from the neighboring language, Tagalog, is remote 
 owing to their primitive meaning. Tir. itam is a Mai. loan word. The o of 
 Ibg. ipo', of which there are several examples in the tables that follow, is 
 for Ibg. a as explained above ( p. 927 ). In Ibg. tdddag n is assimilated to the 
 following d and the a of the first syllable is due either to an exceptional 
 assimilation to the following a, or to analogy. The initial (S) of (i)ripin will 
 be discussed below (pp. 934 ff.). 
 
The Pepet Law in Philippine Languages. 929 
 
 Table IV: the pi- class, 
 
 to buy strip off seed to buy strip off seed 
 
 Phil. beli sepi ben'i Pamp. ablt aspi bini 
 
 Tag. 
 
 bili 
 
 sipi 
 
 binhi 
 
 V Ib §- 
 
 balli 
 
 tap pi 
 
 bini 
 
 Pang. 
 
 bili 
 
 sipi 
 
 bini 
 
 ^Bkl. 
 1 Bis. 
 
 bili 
 
 sapi 
 
 banhi 
 
 Ilk. 
 
 — 
 
 — 
 
 bin'i 
 
 bili 
 
 sipi 
 
 binhi 
 
 Mgd. 
 
 — 
 
 — 
 
 benih (Mai.) 
 
 Bgb. 
 
 balli 
 
 — 
 
 binni 
 
 Tir. 
 
 betlei 
 
 sefe 
 
 bene 
 
 Sulu 
 
 bi 
 
 — 
 
 — 
 
 Unquestioned examples of the /?/-class are few. Total regressive assimi- 
 lation of vowels has here been more extensive than in any of the other 
 classes, tho its operation has in some cases been so erratic as to defy clas- 
 sification. 
 
 So great is the apparent irregularity of vocalism here that the investigator 
 is strongly tempted to set up a variable prototype, e. g., beli: bili, pending 
 the identification of further material for comparison. Most of the examples, 
 however, are readily explained as due to the law of vocalic assimilation, 
 restricted by an intervening consonantal increment, such as that produced by 
 gemination, between the pepet vowel and the following i. 
 
 Before proceeding with the study of the Phil, examples, let us further 
 justify their classification by reference to their cognates in some of the other 
 related languages. The e of Phil, beli is represented regularly according to 
 the general pepet law (p. 920) in OJav. well, Mai. Bali beli, Toba boll, Mak. 
 balli. Cam blei shows loss of pepet as in brah (Phil, begas, Mai. beras). 
 But Day. bili and Mlg. vidi, in both of which we should expect e, show 
 assimilation to the following i. For sepi I have traced no cognates outside 
 of Phil, territory. Cognate with Phil, ben'i are Mai. be'nih, Toba boni, which 
 show the regular vocalism, and OJav. winih, Sund. binih, Mak. bine, Bug. wine, 
 and Day. binyi, which show assimilation. The h of Tag. binhi, Bid. banhi, 
 and Bis. binhi takes the place of the hamza, as often in the //-languages 
 (cf. Tag. Bkl. Bis. Bgb. Sulu dahun, Phil, da'un leaf). It is quite possible that 
 the same holds true for the y of Day. banyi 1 . 
 
 Returning to the Phil, material in table IV, we find in the Tag. examples 
 neither difficulty nor assistance, since Tag. is an /-language. The Pang. Ilk. 
 Bis. and Sulu examples all suffer assimilation of the pepet vowel to the follo- 
 wing /. The Tir. words have regularly e. The t of Tir. betlei is obscure, but 
 there can be little doubt of the identity of this word, as it is the only Tir. 
 term for barter (buy or sell), and offers no other phonetic difficulty, the final 
 ei being practically the same sound as the final S of sefe and bene", and re- 
 presenting original /, just as en in Tir. bateu stands for original u. Pamp. has 
 assimilation in bini, but not in abli and aspi, where it is prevented by the 
 intervention of two consonants brot together by metathesis. Likewise in Ibg. 
 the pepet vowel is assimilated in bini, where only a single consonant sepa- 
 rates it from the attracting vowel, but not in balli and tappi, where gemination 
 
 1 Brandstetter, "Mata-Hari", Luzern 1908, p. 24, considers the phonetic interrelation 
 of these cognates "vielfach unklar". The Ilk. variant beni given in this citation is not found in 
 the Lopez-Carro Iloko dictionary, nor have I found it in other sources. 
 
930 
 
 Carlos Everett Conant, 
 
 takes place. Ibg. tappi has t for Phil, s regularly as in taki' (Phil, sakif) pain. 
 Bkl. here wavers between the regular a and assimilation. Most peculiar and 
 inconsistent of all the examples studied for any class are the Bgb. balli and 
 binni, especially when we compare here Bgb. palli (IN pili) to choose, showing 
 a for an unmistakably original /. 
 
 Mai. sepit, Sund. jepit, Jav. sapil (with exceptional a) pinchers must be 
 referred to a prototype having pepet in the first syllable, which suffers assi- 
 milation in Mak., Bug. sipi. The Phil, cognates have i in both syllables in all 
 the eleven languages of the classified tables: Tag., Pang., Ilk., Mgd., Pamp., Bkl., 
 Bis., Bgb. sipit (with varying accent), Ibg. sipi', Tir. sifit, Sulu gipit (if g can 
 be explained). Whether assimilation of the pepet vowel has here acted inde- 
 pendently in the various Phil, languages or had already taken place in the 
 Phil, prototype is an open question; but that the IN prototype was a word 
 showing pepet, and that the penultimate i of the non-/ languages is a result 
 of assimilation at some stage of IN speech evolution is, in my opinion, beyond 
 doubt. A good example of the /7j'-class outside of Phil, territory is IN tepi 
 edge, border, which shows the regular vocalism in Mai. and Jav. tepi, Toba 
 topi, Mak. tappi, Bug. teppi. It seems probable that assimilation is prevented 
 in the Mak. and Bug. examples by the intervening consonant gemination as 
 in Mak. balli, while it appears in Mak. bine, sipi and Bug. wine, sipi. If this 
 is true, we have in Mak. and Bug. an exact parallel to the Phil, law of vocalic 
 assimilation. The only possible Phil, cognates of Jav. tepi &c. which I have 
 been able to trace are Bkl. tapi to lack little of, Bis. tapi edge of boat, Pang. 
 tdpi board, and Ilk. tappi to fill to the edge, run over. If these are to be 
 with the non-Philippine words, which to me seems more than probable, the 
 uniform Phil, a is very obscure. 
 
 Table V: the «/?-class. 
 
 
 navel 
 
 brain 
 
 hair 
 
 knee 
 
 worm 
 
 snake 
 
 yes 
 
 Phil. 
 
 pused 
 
 utek 
 
 buek 
 
 tued 
 
 uled 
 
 uWg 
 
 ue(n) 
 
 Tag. 
 
 pusud 
 
 utak 
 
 buhok 
 
 tuhod 
 
 6od, 6wod, dhod 
 
 — 
 
 do 
 
 Pang. 
 
 puse'g 
 
 utdk 
 
 buek 
 
 — 
 
 — 
 
 uleg 
 
 on 
 
 Ilk. 
 
 piiseg 
 
 utek 
 
 bodk 
 
 — 
 
 — 
 
 uleg 
 
 wen 
 
 Mgd. 
 
 puset 
 
 utek 
 
 buk 
 
 
 uled 
 
 ular (Mai.) 
 
 wai 
 
 Tir. 
 
 fused 
 
 utek 
 
 ebuk 
 
 etur 
 
 — — 
 
 utrar 
 
 hoo, hee 
 
 Pamp. 
 
 piisad 
 
 utak 
 
 budk 
 
 tud 
 
 uldd 
 
 — 
 
 dwa 
 
 Ibg. 
 
 futdd 
 
 ut6 k 
 
 vu k , vu' x 
 
 tudd 
 
 — 
 
 uldg mouse 
 
 uwdn 
 
 Bkl. 
 
 pusdd 
 
 hutuk 
 
 biihuk 
 
 tuhud 
 
 lilud 
 
 — 
 
 6ho 
 
 Bis. 
 
 piisod 
 
 lit ok 
 
 buhdk 
 
 tuhud 
 
 lilud 
 
 — 
 
 do 
 
 Bgb. 
 
 pusod 
 
 utuk 
 
 — 
 
 — 
 
 olod 
 
 
 00 
 
 Sulu 
 
 pusud 
 
 utuk 
 
 buhok 
 
 tuhud 
 
 ud 
 
 
 hit. 
 
 1 The correct Ibg. word is vu k , and not vu', tho the latter is the only form given by 
 Pavo in his "Diccionario Espaiiol-Ibanag" (sic), Manila 1867. Both forms are given in the older 
 work of Bugari'n, "Diccionario Ibanag-Espaiiol", Manila 1854, and tw A only in a still older MS 
 Ibg.-Span. dictionary in my possession. The k form is also supported by the testimony of the 
 Ibg. dialects Gad. and Itw., which have buk and ahiik, respectively, the final k being here 
 fully pronounced. As the original final surd stops are not distinguished in pronunciation in Ibg., 
 where they pass into hamza, except when supported by a suffix, they are often confused with 
 each other in that language. This is especially frequent with words, which, like bu k , have no 
 
The Pepet Law in Philippine Languages. 931 
 
 With the exception of Tag., the languages here present the regular pepet 
 vocalism, unless pepet is lost entirely as occurs in several cases, where the 
 two vowels are concurrent, e. g., Pang, on, Mgd. buk, Tir. ebuk, elur, Pamp. 
 tad, Ibg. vu' e . 
 
 This is probably best explained as an absorption of the weak pepet 
 vowel by its neighbor. Sulu ud and ha are examples of contraction of two 
 concurrent similar vowels, the former after loss of the intervening / (see above, 
 p. 924). In some words, however, Sulu, being an /^-language, bridges the hiatus 
 between two vowels by the intercalation of h. Phil, buek and tued show this 
 h uniformly in Tag., Bkl., Bis. and Sulu. buhuk, tuhud. 
 
 Tag., which has hitherto shown regularly i for pepet, here deviates from 
 its regular vocalism, and, with the single exception of the isolated liiak, sub- 
 stitutes u(o). We evidently have before us an example of progressive 
 vocalic assimilation, which is not paralleled elsewhere in the material of the 
 eleven languages here tabulated, unless the isolated Ilk. book is thus to be 
 explained. The assimilation with which we have to do in the other classes 
 is regressive. Progressive assimilation is likewise doubtless the explanation 
 of the second u of Sund. bank hair and iuur knee 1 , and of Batan badk, tuud, 
 Kim. food, Kuy. book. That this assimilation in Tag. is prevented by the 
 intervention of more than one consonant is indicated by the regular i of 
 Tag. batlig wart, whose pepet origin is vouched for by the cognates, Pang. 
 but/eg, Bkl., Bis. butlog, Pamp. batlig. We are already prepared for this case 
 of prevented assimilation by our study of the same phenomena in the pa- 
 class and the /?/-class. 
 
 The a of Tag. dtak is an exception to which I know no parallel in 
 Tag. Brandstetter (Prodromus, p. 51) sets up the variant series utak, atek, 
 atok to account for the vocalic variety seen in Tag. dtak, Jav. utak and atek, 
 and Mak. otoq (<7 = hamza). I am, however, strongly of the conviction that 
 further investigation of the laws of pepet evolution as affected by the more 
 powerful action of the laws of assimilation and analogy will establish the 
 original identity of these various forms, and the preponderance of examples 
 showing a vowel of unmistakable pepet origin, together with those whose 
 
 forms with suffix to preserve the character of the original stop. The erroneous / written by Payo 
 and by Bugarin (or one of his numerous revisers) in addition to the correct A-form, is doubtless 
 due to popular analogy with other Ibg. words of more or less similar meaning properly ending 
 in original t, e. g. kuW kinky hair, gunu' hair of the wild palm, duddil* hair of the body. 
 The Ibg. wWi, therefore, does not belong with Mai. rambut, as suggested by Brandstetter 
 ("Prodromus", p. 42), but with Toba buk, Sund. buuk and the Phil, words in k. For my ortho- 
 graphy v instead of the t of the Spanish dictionaries and grammars, cf. my paper "F and V in 
 Philippine Languages", p. 139. 
 
 1 Brandstetter, "Prodomus", p. 41, instead of setting up a dissyllabic IN prototype 
 buek, tued, from which both the dissyllabic and the monosyllabic forms are eastly derived as 
 explained above, considers the monosyllabic type the original one and then attempts to explain 
 the longer forms Tag. buhok, tuhod, Sund. buuk, tuur as extensions of this prototype, admitting, 
 however, that the extension (Zerdehnung) seen in Bis. and Tag. buhok is "ratselhaft". In the 
 same work (p. 48) the author shows how simply and naturally the dissyllabic prototypes with 
 pepet : teras, terab, berat degenerate, thru OJav. twas, twab, bwat, to New Jav. tos, a-tob, bot. 
 
932 Carlos Everett Conant, 
 
 vowel could have developed from either pepet or another vowel, makes it 
 certain that when such identity is established it will be on the basis of original 
 pepet. That the positing of variant forms in general in the study of IN pho- 
 nology and morphology is merely a convenient makeshift for a temporary 
 classification of phenomena not yet sufficiently investigated, and hence does 
 not commit the investigator to any theory that may later be proved untenable, 
 is distinctly stated by Brandstetter ("Mata-Hari", p. 53, par. 96), and this is 
 the only possible method of proceeding to a scientific arrangement of many 
 phonetic phenomena presenting problems awaiting solution in this compara- 
 tively new field of research. 
 
 The similarity, both in form and meaning, of Phil, tiled and uleg, together 
 with the fact that most Phil, languages having the one word do not have the 
 other, would at first sight suggest a confusion here of the final RLD and RGH 
 consonants, and lead one to the conclusion that the word with a final con- 
 sonant wavering between that of the RLD and the RGH series has come to 
 have the exclusive meaning "worm" in some languages and that of "snake" 
 in others. But in spite of the physical similarity of the two objects, worm 
 and snake, they seem never to have been confused by the primitive Indo- 
 nesian, certainly not by the Filipino, all of whose languages, so far as I have 
 been able to examine them lexically, have distinct words for the two ideas. 
 Thus the blank spaces under "worm" and "snake" in the above table (V) 
 may be semantically filled out as follows; "worm": Pang, bigis, Ilk. egges, 
 Tir. sofot, Ibg. tuggit; "snake": Tag. dhas, Mgd. nipai, Pamp. ubitian, Ibg. 
 irdu, Bkl., Bis. hdlas, Bgb. bakossan, Sulu has (the Tag., Bkl., Bis. and Sulu 
 words being, of course, identical). Under none of the definitions given is 
 there any suggestion of confusion between the concepts "worm" and "snake". 
 Add to this the uniform RLD consonant in the words having the former 
 meaning and the equally uniform RGH consonant of the others, as well as 
 the existence in Mai. of both ulat (hulat) worm and ular snake 1 , and the 
 probability of confusion of the two prototypes disappears. 
 
 The exceptional a of Tir. tirrar may be due to the influence of the 
 adjacent r sounds. The w of Ilk. wen and Mgd. wai represents the original 
 a which, after loss of accent, has weakened to a semivowel; that of Pamp. 
 owa and Ibg. uwdn is a semivocalic glide developed between the two vowels, 
 while in Bkl. oho the h is inserted a in bdhok, tuhod. The vocalism of Mgd. 
 wai, where we should expect we, is unclear. The Tir. variants hoo, hcc 
 exemplify a law of vocalic interchange peculiar to Tir., and as yet little under- 
 stood, but paralleled by the Tir. pluralizing variants de, do, da. 
 
 Table VI: the /?«-class. 
 
 
 full 
 
 sugar cane 
 
 redeem 
 
 pedere 
 
 sound 
 
 to sate 
 
 to boil 
 
 Phil. 
 
 penu 
 
 tebu 
 
 tebus 
 
 etut 
 
 tenug 
 
 oesug 
 
 scbu 
 
 Tag. 
 
 pund 
 
 lubo 
 
 tubds 
 
 a tot 
 
 tunog 
 
 buS6g 
 
 subd 
 
 Pang. 
 
 pdnu 
 
 tabu 
 
 - 
 
 atot 
 
 tandl 
 
 — 
 
 sabd 
 
 ' The / of Mai. ulat is for the sonant d of the RLD series by the law of final stop con- 
 sonants, the r of ular reguiarly for RGH. 
 
The Pepet Law in Philippine Languages. 
 
 933 
 
 
 full 
 
 sugar cane 
 
 redeem 
 
 pedere 
 
 sound 
 
 to sate 
 
 to boL 
 
 Ilk. 
 
 punnd 
 
 tubbo 
 
 subbut 
 
 uttdt 
 
 — 
 
 bussiig 
 
 sobbd 
 
 Mgd. 
 
 penu 
 
 tebu 
 
 tebus 
 
 tud 
 
 tanuk 
 
 — 
 
 — 
 
 Tir. 
 
 fend 
 
 — 
 
 tebus 
 
 etiit 
 
 — 
 
 besor 
 
 — 
 
 Patnp. 
 
 apnti 
 
 atbu 
 
 atbus 
 
 atut 
 
 atni 
 
 dbsi 
 
 asbd 
 
 Ibg. 
 
 pannu 
 
 tavvu 
 
 tavvu* 
 
 attii' 
 
 tannug 
 
 battug 
 
 tavvu 
 
 Bkl. 
 
 pand 
 
 tubu 
 
 tubus 
 
 atdt 
 
 tandg 
 
 basdg 
 
 sabd 
 
 Bis. 
 
 pund 
 
 tubd 
 
 tubds 
 
 utdt 
 
 tundg 
 
 busog 
 
 subd 
 
 Bgb. 
 
 punnd 
 
 tubbo 
 
 tubbos 
 
 uttdt 
 
 — 
 
 bussog 
 
 — 
 
 Sulu 
 
 — 
 
 tubu 
 
 — 
 
 utut 
 
 — 
 
 — 
 
 — 
 
 Tag. shows u for pepet as in the preceding class, but here by regres- 
 sive assimilation. Ilk. likewise has u for the same reason (cf. above, p. 931). 
 Pang, shows a instead of the regular e. This is evidently a case of partial 
 assimilation to the following u. The geminating languages here show doubling 
 of the medial consonant in all the examples (see above, consonant doubling), 
 and the Pamp. examples, excepting atut, show metathesis in the first syllable. 
 Ilk. subbut is an example of metathesis of initial and final consonants not 
 infrequent in that language (see above, under metathesis). Mgd. has an ex- 
 ceptional a in tanuk, which shows the surd k finally for the sonant g, as 
 Mgd. puset for Phil, pused (table V). A comparison of the final consonantism 
 with that seen in Mgd. uled (Phil. ulSd) and Mgd. tud (Phil, etuf) would 
 indicate that final stops in Mgd. waver between surd and sonant, whatever 
 may have been the original sound. 
 
 In Pamp. atni and absi we have a peculiar treatment of original final 
 ug, the g of the RGH series becoming the semivowel y or i, which with the 
 preceding u first forms a diphthong ui, from which the labial element is later 
 lost after shifting of accent to the final component i. 
 
 The Bkl. words have the regular a for pepet in the penult in five of the 
 seven examples here tabulated, in spite of the following u (cf. above, p. 931). 
 
 
 
 li 
 
 tble VII: 
 
 the pp-cl 
 cleave, 
 
 ass. 
 
 to close 
 
 
 
 excavate 
 
 pole 
 
 hear 
 
 stick 
 
 thorn 
 
 (hand) 
 
 six 
 
 Phil. 
 
 lebefi 
 
 teken 
 
 deheg 
 
 deket 
 
 tenek 
 
 kemkem 
 
 enem 
 
 Tag. 
 
 libih 
 
 tikin 
 
 dinig 
 
 dikit 
 
 tinik 
 
 kimkim 
 
 anim 
 
 Pang. 
 
 — 
 
 teken 
 
 denel 
 
 — 
 
 tenek 
 
 kemkem 
 
 anem 
 
 Ilk. 
 
 — 
 
 tekken 
 
 define g 
 
 rekket 
 
 tennek 
 
 kemkem 
 
 innem 
 
 Mgd. 
 
 leben 
 
 teken 
 
 — 
 
 deket 
 
 tenek 
 
 — 
 
 anem 
 
 Tir. 
 
 lebeh 
 
 — 
 
 — 
 
 deket 
 
 — 
 
 kemkem 
 
 enem 
 
 Pamp. 
 
 albdh 
 
 atkdn 
 
 — 
 
 — 
 
 — 
 
 kamkdm 
 
 andm 
 
 Ibg. 
 
 labbdh 
 
 takkdn 
 
 — 
 
 dakkd' 
 
 tannd k 
 
 — 
 
 anndm 
 
 Bkl. 
 
 lubun 
 
 tukun 
 
 dahug 
 
 dokdt 
 
 tiinok 
 
 komkdm 
 
 anum 
 
 Bis. 
 
 lubiin 
 
 tukun 
 
 duhiig 
 
 dukiit 
 
 tuntik 
 
 kumkiim 
 
 unum 
 
 Bgb. 
 
 lubbun 
 
 — 
 
 ' — 
 
 ddkkot 
 
 — 
 
 komkom 
 
 anndm 
 
 Sulu 
 
 luban 
 
 — 
 
 duhuk 
 
 — 
 
 tunok 
 
 kumkum 
 
 unom 
 
 This class, like the a/7-class (table I), is remarkably uniform in its pepet 
 vocalism. This is evidently due to the fact that the development of the pepet 
 vowel is here undisturbed by the attractive influence of a neighboring dis- 
 similar vowel. 
 
934 
 
 Carlos Everett Conant, 
 
 Aside from the exceptional a in the ultima of Bgb. anndm and Sulu 
 luban, for which no explanation can here be offered, and the peculiar vocalism 
 of the first syllable of Phil, enim to be treated below, the Bkl. shows the 
 only peculiarity requiring special comment. It will be noted that in this class 
 Bkl. has a for penultimate pepet only two of the seven examples listed, the 
 other five showing u. The following additional list of words in the /7/7-class 
 shows nearly the same proportion of Bkl. examples having a in the penult: 
 
 Phil. 
 
 Bkl. 
 
 Tag. 
 
 Other 
 
 languages 
 
 begkes 
 
 bugkiis 
 
 bigkis 
 
 Bis. bugkus 
 
 Mai. berkas 
 
 betek 
 
 butdk 
 
 bitik 
 
 Pang, betek 
 
 Ilk. bettek 
 
 deles 
 
 dolds 
 
 dills 
 
 Pang, deles 
 
 Ibg. dallo 1 
 
 getel 
 
 giitul 
 
 gitil 
 
 Pang, getel 
 
 Pamp. agtdl 
 
 neknek 
 
 nokndk 
 
 niknik 
 
 
 
 petes 
 
 put us 
 
 pitis 
 
 Bis. putus 
 
 Pamp. aptds 
 
 seged 
 
 sogdd 
 
 sigtd 
 
 Tir. seged 
 
 Pamp. asydd 
 
 tegeb 
 
 tagob 
 
 tigib 
 
 Mgd. tegeb 
 
 Pamp. atydb 
 
 terek 
 
 tdrok 
 
 tldik 
 
 Ilk. teddek 
 
 Pamp. atddk 
 
 teres 
 
 tados 
 
 tiris 
 
 Pang, seret 
 
 Pamp. at das 
 
 Ibg. dallo' has an original final t for Phil. 5 regularly as in tabbu' (Phil. 
 tSbus), and Pang, seret has metathesis of initial and final consonants. 
 
 Penultimate a and u interchange in the Bkl. variants lamid, laniid to sink 
 in water (cf. the cognates Pang, lened, Ilk. lenned, Bis., Sulu laniid). Thus we 
 see that while the penultimate pepet vowel in Bkl. here, as in other classes, 
 wavers between a and u, the latter predominates to such an extent that it may, 
 for our purpose be considered the rule, and a the exception. If we contrast with 
 this case that of the pu-dass (table VI), where Bkl. regularly shows a in spite of 
 an original a of the following syllable, we are forced to the conclusion that 
 vocalic assimilation cannot explain the phenomena before us. In fact, the only 
 certain case of vocalic assimilation in che Bkl. material studied is that of bill 
 in the /?/-class, beside the unassimilated pepet vowel of sapi and banhi. 
 
 Now a review of all the Bkl. material we have collected shows that pepet 
 invariably becomes a (0) in a final syllable, and, as a rule, becomes a in the 
 penult of all classes except the /7/7-class, where it regularly becomes u. Of 
 the two pepet vowels we have seen that u is the more stable, and hence, if 
 there are two pepets in the same word, naturally evolving like sounds, as is 
 seen in all the other languages of the /7/7-class, the two vowels will naturally 
 be u rather than a. The exceptions like daniig are explained as due to analogy, 
 the a-u succession following that of the /?#-class and the large number of 
 other Bkl. words of the same vocalism. 
 
 A pepet vowel in initial position develops peculiarly in certain languages. 
 This vowel most commonly appears as a weak, colorless a. Its most con- 
 spicuous examples are the IN numerals enem six and epat four, to which may 
 be added Phil. (S)ripSn (table III). 
 
 The same a appears in the reduplicated syllable of Phil, tetieiln three, 
 which has become stereotyped in some languages while others show the 
 simple IN form tilu 1 . 
 
 1 Cf. Blake, "Contributions to Philippine Grammar" in J. of the Am. Or. Soc., vol. 28, p. 204. 
 
The Pepet Law in Philippine Languages. 935 
 
 Phil. 
 
 Tag. 
 
 Pang. 
 
 Ilk. 
 
 Mgd. 
 
 Hlg. 
 
 Bgb. 
 
 Kuy. 
 
 Inb. 
 
 epat 
 
 apdt 
 
 apdt 
 
 uppdt 
 
 apat 
 
 apdt 
 
 appdt 
 
 apdt 
 
 dppat 
 
 'enem 
 
 anim 
 
 anem 
 
 inn em 
 
 anem 
 
 aniim 
 
 anndm 
 
 ane'm 
 
 annim 
 
 (e) ripen 
 
 alipin 
 
 aripen 
 
 adipen 
 
 uripen 
 
 ulipon 
 
 -— ' 
 
 — 
 
 — 
 
 tet(e)lu 
 
 tat 16 
 
 (talo) 
 
 (talld) 
 
 (tela) 
 
 tatld 
 
 tatld 
 
 tatld 
 
 (tdddu, 
 
 The parenthesized talo and telu are from the unreduplicated prototype 
 tela. Ilk. talld and Inb. tdddu, as well as Ibg., Gad., Itw., Yog. tallu, may 
 phonetically represent either the simple or the reduplicated form. If they are 
 from the simple tela, the double consonant is the result of gemination (see 
 above, p. 922), but if from the reduplicated tet(e)lu, it is the result of regressive 
 assimilation following syncopation of the intervening weak <?. It is probable, 
 however, that the Ibg. talld and Ilk. talld are, notwithstanding their similarity 
 of appearance, from different prototypes, the former being from Phil. tet(e)lu 
 and the latter from Phil. tela. The unassimilated t of Ilk. itldg egg stands as 
 evidence against the reduplicated prototype for Ilk. and the same may be said 
 of the Inb. tdddu in view of Inb. exduk 1 egg, where assimilation does not 
 take place. In Ibg., on the other hand, Phil, itlug appears as Mug. With this 
 evidence alone, Ibg. tallu could be referred to tetfejla as well as to telu, and 
 the former prototype is suggested as the more probable by the Bat. tatdd 2 
 (Phil, tetfejla). 
 
 Ilk. talld, appdt, innem show an exceptional variety of pepet represen- 
 tation which is difficult to account for. It is possible that the quality of the 
 following consonant has here affected thaf of the weak vowel in question. 
 Hlg. has, beside the forms here given, told, updt, unum, like the other Bis. 
 dialects (except Kuy.). 
 
 The a vocalism of the original unaccented pepet which appears consi- 
 stently in tetfejlu, epat and enem, in several languages and dialects instead 
 of the regular pepet vowel peculiar to those languages is doubtless due pri- 
 marily to the influence of the original a in the last syllable of all but one 
 of the first five IN cardinals, esa, dua, tela, epat, lima. Thus the original 
 final a of dua would tend to give an a coloring, especially in counting, to 
 the weak unaccented pepet of the following telu, and this influence would 
 have a still greater effect upon the more isolated pepet of the reduplicated 
 
 1 Ortography of Scheerer, "The Nabaloi Dialect", Manila 1905, p. 103; x represents the 
 sound of ch in the Scotch word loch. 
 
 2 Tatdu is doubless the correct form. It is taken from a word list given to me orally by 
 a Batan servant boy at Aparri (north coast of Luzon) in October 1904. This boy had recently 
 arrived from his native island, Batan, and I hence consider his word list moje reliable than 
 that which I took one year later af Claveria (North Luzon) from a native of Batan who had 
 many years before migrated with his family to Luzon. The latter gave tdddu, which, if correct, 
 shows assimilation of t to the following d, since Bat. does not double a single consonant. He 
 similarly gave the word for goat as kaddin, which the servant boy had pronounced kanclin, 
 Furthermore, tatdo is the form found both in the Batan Catecismo of Padre Rodriguez (reprinted 
 by Retana in his "Archivo del Bibliofilo Filipino", vol. 2, Madrid 1896), p. 13, et passim, and 
 in the Batan "devocionario" entitled "Nu Napia Amigo", Manila 1901, p. 62, et passim. I would 
 therefore discard both tdddu and the by-form tatlo given by Scheerer, "The Batan Dialect as 
 a Member of the Philippine Group of Languages", Manila 1908, Plate I. 
 
936 Carlos Everett Conant, 
 
 form t$t(g)lu. The initial a of apdt would in turn be due, either to the assi- 
 milative influence of the original a of the final syllable, or to analogy with 
 the already established a of the first syllable of its predecessor, tatlii, or more 
 probably, to both these influences combined. Finally, the a of lima exerts 
 its influence upon the initial unaccented pepet of its successor enem, and the 
 a thus resulting is now amply fortified by the analogy of the penultimate a 
 of tatlii and apdt. The a's thus arising then become still more firmly esta- 
 blished by mutual support under the natural operation of the laws of analogy. 
 
 A striking example of analogy is furnished by the Pamp. numerals adwd 
 and apuld, which have prefixed an a owing to the initial a of atlu, apdt, 
 andm. For the i of isd one in non-i languages there seems to be no satis- 
 factory explanation, unless we posit the variants esa, isa. Tag. and Bon. isd 
 could be referred to either of these variants, but the following forms must 
 go back to Ssa: Knk. esa, Kal. eta, Bat. asd 1 , Bis., Isn. usd. Tgb. and Hlg. 
 have both usa and isa. Other examples of the isa type are Bkl. 2 , Pang., 
 Pamp., Sulu, Mgd., Tgk. isd, Ilk., Ting, maisa (for ma + isa), Itw. isa, Ibg., 
 Gad. itte (t regularly for Phil, s and e as in Ibg. due two beside dua), Kuy. 
 isard (lit. "one only"). The proclitic form sa seen in Bkl. sard, Inb. saxei, 
 Ibg. tdddai, Bgb. sabbad, Mnb. sabad, Tir. seba'an one, Tag. sanpwwo, Bkl. 
 sampdlo, Mgd., Bgb. sapulu, Gad. tdfulu (with secondary Gad. accent and 
 regular t for s), Pang, samplo ten (lit. "one ten"), and Mai., Jav., Sund. sa 
 one and sapuloh ten may be explained either as a third variant beside 8sa, 
 isa, or as esa with loss of the initial pepet when the word becomes proclitic. 
 Cam sa may represent either sa or esa, it being a peculiarity of that language 
 to suppress a penultimate pepet vowel, e. g., Cam brah (IN biras) rice, kldu 
 (IN telii) three, pak (IN epat), nam (IN enem). An IN u is thus suppressed 
 in Gam sa pluh (IN pulu) ten. 
 
 It is evident from the above examination of the pepet vocalism of the 
 numerals, that in several languages they form a distinct category subject to 
 a special secondary influence, namely, the combined operation of assimilation 
 and analogy, and hence may be set aside as not belonging to the general 
 phenomena of the pepet law. Excluding, then, this peculiar vocalism of un- 
 accented pepet in the numerals and in the first syllable of the exceptional 
 (e)ripen, we may now proceed to a more concise statement of the evolution 
 of the indifferent vowel in each of the eleven languages above tabulated. 
 
 Tagalog: Pepet regularly becomes i; but when the vowel of an adjacent 
 syllable of the same root word is an original u (o), pepet is assimilated to 
 this vowel, becoming a (o), but not to a preceding u (o) if more than a single 
 consonant intervenes, e. g. pusud (Phil, pusid), puno (Phil, pinu), but butlig 
 (Phil, butlig). 
 
 1 Blake, op. cit., p. 203, explains the initial a of asd as prefix. I quote his explanation 
 without comment: "Batan asa is probably the root particle sa which is found in the majority 
 of the forms of one, with a prefix a probably identical with the a of Tagalog ang, just as the 
 i of iisa is identical with the / of Pampangan ing", 
 
 3 Bkl. isd is the form used in counting, cf. Marcos DE Lisboa, "Vocabulario de la Lengua 
 Bicol", Manila 1865, s. v. isd. 
 
The Pepet Law in Philippine Languages. 937 
 
 Pangasinan: Pepet regularly becomes e, exceptionally / or a. It appears 
 as / by assimilation to an original i of the following syllable in the same 
 root word (bill: Phil, bell), and as a by partial assimilation to an original 
 u (o) of the following syllable (pdnu: Phil. penu). 
 
 Iloko: Pepet regularly becomes e, exceptionally a, i or u. It appears 
 as a by assimilation to an original a of the following syllable in the same 
 word when only a single consonant intervenes (bagds: Phil, begas), and as e 
 or u by assimilation to an original i or u, respectively, of the following syl- 
 lable of the same word (bin'i: Phil, bent, punno: Phil. penu). 
 
 Magindanau: Pepet appears regularly as e, exceptionally as i. It be- 
 comes / by attraction to an original i in the following syllable of the same 
 root word (sipit: IN sepit). 
 
 Tirurai: Pepet becomes uniformly e, except in sifit: IN sepit, where 
 it becomes i Ibg. assimilation to the i of the following syllable. 
 
 Pampanga: Pepet regularly becomes a, exceptionally / by assimilation 
 to an original i of the following syllable in the same root word when only 
 a single consonant intervenes (bini: Phil, beni, but abli: Phil. belt). 
 
 Ibanag: Pepet regularly becomes a, exceptionally i by assimilation to 
 an original i of the following syllable in the same root word when only a 
 single consonant intervenes (bini: Phil, beni, but balli: Phil. belt). 
 
 Bikol: Pepet regularly becomes u(o) in a final syllable and a in the 
 penult; but if the original vowel of both syllables is pepet, it becomes u(o) 
 in both. Before an original i of the following syllable in the same root word, 
 it is sometimes assimilated, becoming i (bill: Phil, beli), and sometimes be- 
 comes the regular penultimate a (sapi: Phil. sepi). 
 
 Bisaya: Pepet regularly becomes u(o), exceptionally I by assimilation 
 to an original i of the following syllable in the same root word (bill: Phil. beli). 
 
 Bagobo: Pepet becomes u(o) eycept when followed by an original i 
 in the next syllable of the same root word, when it either it assimilated, be- 
 coming i (binni: Phil, beni), or becomes a (balli: Phil. beli). 
 
 Sulu: Pepet regularly becomes a(o), exceptionally i by assimilation to 
 an original i of the following syllable of the same root word (bi < bii < bill 
 < Phil. beli). 
 
 Other languages and dialects. We now continue our study by 
 examining the pepet vocalism of the following languages and dialects, one 
 of which, Chamorro, tho not within Philippine territory, is conveniently classi- 
 fied here: 
 
 1. Ata (near Mt. Apo, S. Mindanao) 1 . 
 
 2. Banawi (mountains of N. Luzon). 
 
 3. Batan (Batan Islands, to N. of Luzon). 
 
 4. Bilan (mountains of S. Mindanao, S. of Ata territory). 
 
 5. Bontok (Igorots of Lepanto-Bontok province, N. Luzon). 
 
 ' For more detailed geographical information cf. Scheerer's sketch map in his work, 
 "The Batan Dialect &c", p. 17, and, for the Luzon territory, Worcester's authoritative work, 
 "The Non-Christian Tribes of Northern Luzon", in the Philippine Journal of Science, vol I, 
 No. 8, Manila 1906. 
 
938 Carlos Everett Conant, 
 
 6. Chamorro (Marianne Islands). 
 
 7. Gaddang (Ibanag dialect, N. Luzon). 
 
 8. Inibaloi (Igorots, Benguet province, N. Luzon). 
 
 9. Isinai (mountains of Nueva Vizcaya province, N. Luzon). 
 
 10. Itawi (Ibanag dialect, N. Luzon). 
 
 11. Kalamian (N. Palawan). 
 
 12. Kankanai (Igorots of N. Benguet, N. Luzon). 
 
 13. Kuyunon (Bisaya dialect, Cuyo Islands, between Panay and Palawan). 
 
 14. Lepanto (mountains of N. Luzon). 
 
 15. Manobo (mountains E. of Gulf of Davao, S. Mindanao). 
 
 16. Samal (Samal I. Gulf of Davao, S. Mindanao). 
 
 17. Sambal (Zambales province, W. Luzon). 
 
 18. Tagakaolo (Apo range, W. of Gulf of Davao, S. Mindanao). 
 
 19. Tagbanwa (Palawan I.). 
 
 20. Tingyan (mountains of N. Luzon). 
 
 21. Yogad (Ibanag dialect of N. Luzon). 
 
 ^-languages. 
 
 Lepanto: Pepet becomes e: zelok (Tag. etc. itlug, Mai. telof) egg, 
 oeg (uleg 1 ) snake, tined (tehed) back of neck. The last example shows i in 
 the penult. For loss of / in oeg, cf. Lep. uat (ugat) vein, where the consonant 
 of the RGH series, appearing secondarily as /, is lost in intervocalic position, 
 and buan (bulan) moon, where the / is original ; but is seems not to be lost 
 before o(u), zelok, olo (ula) head (cf. above, p. 924). The discrepancy between 
 the final consonant sounds of zelok and oeg is more apparent than real, since 
 final stops are generally not exploded in Philippine languages, and it is pro- 
 bable that to the German ear of Schadenberg, from whose list the above 
 examples are taken, the same consonant appeared, now as k, now as g. 
 
 Kankanai: Pepet becomes regularly e, exceptionally u(o): esd (esa) 
 one, epdt (Spat) four, enem (enem) six, eweg (uleg) snake, told (tela) three, 
 bii'ok (buek) hair, the o of the last two examples being the result of assimi- 
 lation to a neighboring original u(o). In this last respect, Knk. is sharply 
 differentiated from the neighboring dialect Inb., which has bu'ek. The first e 
 of eweg, on the other hand seems to be case of assimilation of an original u 
 to an accented pepet vowel. 
 
 Inibaloi: Pepet regularly becomes £, exceptionally i and a: atep (atgp) 
 roof, acdlem (Ilk. addlem, Tag. lalitri) deep, acaxel (daktt) much, bekds (begas) 
 rice, utek (utek) brain, piiseg (pused) navel, bu'ek (bu£k) hair, illeg (ulgg) 
 snake; atiit (etut) pedere tdddo (telu) three, dppat (epat) four, annim (enim) 
 six, macim (marem) afternoon. Where a occurs it is in the penult, and the 
 two examples of i are in an accented final syllable ending in m. The c (=di 
 in church) of acdlem, acaxel, and macim is the regular representative in Inb. 
 of the RLD consonant, and x (= ch in Scotch loch) is for intervocalic k, cf. also 
 Inb. koxo (kuku) finger nail. 
 
 Kuyunon: Pepet regularly becomes e, exceptionally a and u: iddlem 
 (iralem, Bis. idlum) below, rdet (Bis. da'ut, Tag. la'it) bad, lieg (Bis. li'ug, 
 Tag. Wig) neck, ibeg (ibeg) desire, love, ipen hipen tooth (with loss of initial 
 
 When not otherwise indicated, the parenthesized form is to be understood as Phil. 
 
The Pepet Law in Philippine Languages. 939 
 
 n as in Pamp. ipan and Sulu ipun, table III); anem (enem); tatlo (tetfejlu), 
 apdt epat; puno (penu) full, tubus (tebus) redeem, book (buek). The numerals 
 show the penultimate a (cf. pp. 935 and ff.), and the u (o) of the last three 
 examples is the result of a-assimilation as in Tag. 
 
 Kalamian: Pepet regularly becomes e, exceptionally u (o): eta (esa), 
 epat (epat), enem (enem), kentii (Bat. anal, Chro. -unai, OJav. heni) sand, 
 kiripen 1 (eripenj slave, bitonken (bituen) star, kuled (uled) worm; tolo (telu), 
 dakolo (Chro. ddnkulo, cf. above, p. 926) great, tood (tued) knee. The first o 
 of tolo and dakolo is the result of assimilation to the o of the following 
 syllable, and the second o of tood shows assimilation to an original u (o) 
 immediately preceding, as contrasted with bitonken and kuled, where pro- 
 gressive assimilation is prevented by intervening consonants. Kim. eta has t 
 for Phil, s like the Ibg. dialects, e. g. Kim. katawa, Ibg. atawa (asawa) 
 spouse, Kim. toto, Ibg. tutu (susu) uber. One of the chief characteristics of 
 Kim. is a parasitic k, which is seen most commonly prefixed to an initial 
 vowel, as in katawa, keuai, kiripen, kuled, sometimes in the interior of a 
 word, as in bitonken and takon (Tag., Pang, taon, Mai. tahun) year, and 
 sometimes finally, as in lotok (Bis. hito, Tag. Iulo) to cook 2 , polok (IN pulu) ten. 
 
 Languages showing both e and a. 
 Batan: Pepet regularly becomes e in a final root syllable, and a in a 
 penultimate syllable: hipen (hipen) tooth, ipwes (ipes) roach, puseg (pused) 
 navel, uhed (uled) worm, rahet (daet, seeunder Kuy) bad, bituhen {bituen) 
 star, anem (enem) six, labeh (lebeh) to bury, adheyen (deheg-en) hear (imv.), 
 asd (esa) one, tatdu (tetlu) three, dpat (epat) four, atiit (etut) pedere; it 
 becomes u by progressive assimilation in tiiud (tued) knee, and buuk (buek) 
 hair, where no consonant intervenes, contrast puseg bituhen. Bat. has both 
 dadake and rakuh meaning great, the former being Phil, dake (Bis. daku, 
 Tag. malaki) with reduplication, and the latter the extended form dakel (see 
 above table I). For the exceptional u instead of e in rakuh there is no satis- 
 factory explanation. In sehseh 3 (Tag. silsil,, Ilk. Pang, selsel, Pamp. salsdl, 
 Ibg. tattdl (Bis., Bkl. sulsiil), both syllables show e for pepet contrary to the 
 rule for penultimate a. We have here the reduplication of a monosyllabic 
 root as in the case of Phil, kemkem (table VII, p. 933), a type that is very 
 common in all Philippine languages, and in such forms pepet seems always 
 
 1 Written quiripuen in Padre Jeronimo's Vocabulario, the u being written after the labial 
 p to indicate the obscure sound of e, cf. Padre Cosgaya's Spanish orthography of the Pang, 
 cognate aripuen (pronounced aripen, with e as in Ger. sageri). 
 
 2 Padre Jeronimo's coser (p. 17 of the "Vocabulario Castellano-Calamiano") is evidently 
 erroneously written for cocer. This is indicated, not only by the phonetic correspondence of the 
 Phil, words, but by the meaning of the words in the list immediately preceding and following 
 coser, the order being comido, crudo, coser, serveza for ceivezd), vino &c, where serveza 
 shows the same error. 
 
 3 From the form written ipanejsej repent (imperative) in "Nu Nap:a Amigo, p. 421, et 
 passim. The Phil, root selsel, which develops different shades of meaning in different languages, 
 based on the general idea "to crush, blunt, rivet", has often in the reflexive and passive the 
 derived meaning "be sorry, repent". 
 
940 Carlos Everett Conant, 
 
 to develop the same vowel in the two syllables, as tho they were treated as 
 separate words. 
 
 The h of sehseh, like that of iihed and rakuh is the regular represen- 
 tation of Phil. / in Bat., cf. also Bat. uho (ulu) head, hahyit (laii.it) sky, rahan 
 (dalan) way. It is strongly aspirated, approaching the spirant sound of g in 
 Ger. "Ziege", and hence is represented in the Span, orthography of Nu Mapia 
 Amigo by /. When taking the word lists from natives of Batan Island I obser- 
 ved the close similarity of this sound to that Inb. x (see above, p. 938). The 
 h of bituhen and rahet is parasitic, as often in the ^-languages, Tag., Bis., 
 Bkl., Sulu (cf. the medial parasitic k of Kalamian, p. 939), where it bridges 
 the hiatus between two vowels, i. e., replaces intervocalic hamza. 
 
 The retention of Phil. / in Bat. labeh is an exception for which I have 
 found no parallel. Bat. adheyen (deheg-en) shows metathesis of the first syl- 
 lable and y for Phil, g (RGH) as regularly (see above, p. 924). 
 
 Togad. Gaddang and Itawi: These are dialects of Ibg. and, like that 
 language, regularly represent pepet by a, which, however, is sometimes modi- 
 fied, becoming obscure e of the ^-languages: Gad., Yog., Itw. baggd (begas) 
 rics, tdllti (tetlu) three, appdt (epat) four, Gad. filtad (pused) navel, Gad., 
 Yog. utak (utek) brain, Gad. pannu (penu) full, Gad. attii, Itw. dttu (etutj 
 pedere, Gad., Yog., Itw. annem (enem) six, Gad. hipen, Yog. hipen, Itw. hipan 
 (hipen) tooth. This e may be considered as a secondary modification of a 
 since original full vowels are often thus obscured in these dialects, e. g. Gad. 
 maple (Ilk. mappyd) good, Gad. kdmet (Tag., Bis. kamot) hand, Itw. memmemd 
 (ma-mama, IN mama) to chew, but Gad. mdmmama. 
 
 Like Ibg., these dialects are characterized by consonantal gemination 
 (baggdt, appdt), total regressive assimilation of consonants (tdllu, see discussion 
 of Tag. tallu, pp. 935 ff.), / for Phil, p before u (Gad. fiitag, Itw. fiitad, Yog. 
 td-fulu, Phil, pulu, but Gad., Yog., Itw. appdt, pitii seven, and t for Phil, s 
 except before i (Gad., Yog., Itw. baggdt, Gad. futag, Itw. futad, Yog. tagatut 
 [sagatus] one hundred, but Ibg. siku, Gad. siku, Phil, siku elbow); but they 
 differ from Ibg. in retaining unchanged the final surd stops, k, t, and p (baggdt, 
 utak) and the pepet a preceding such final surd stop (utak, but leg. uto k . 
 
 U-\ angtiagcs. 
 
 Tingyan: Pepet regularly becomes u (o), exceptionally a, e or /: ddkon 
 (dake-n) great, bogds (begas) rice, tiilu (telu) three, updt (epat) four, book 
 (buek) hair; andm (enem) six; beken (Day., Tir., Mgd. beken, Har., Sulu bukiui) 
 not so; nebin (hipen) tooth, maisa, for ma + isa (esa or isa) one. The material 
 is not sufficient for an analysis of the exceptional vocalism, a, e and i. The 
 n of ddkon is doubtless a connective (the ligazon of the Spanish grammarians) 
 like the n of Bis. dakiih balai large house. 
 
 Isinai: Pepet becomes uniformly u (o): osa or ossa (esa) one, opat 
 (epat) four, onom (Snem) six, lubu (leben) bury, dnon (kan-£n, Tag. kdn'iti, 
 Bis. kdn'on) food. 
 
 The change of IN k to hamza seen in anon is a peculiarity of Isn. and 
 may occur in any position, initial, medial, or final, e. g. a (ka) you (sing.) 
 
The Pepet Law in Philippine Languages. 941 
 
 in Christiano a? 1 Are you a Christian? a(ak) I in the answer to the above 
 question, O Ama, christiano a Yes, Father, I am a Christian, ana' (anak) off- 
 spring, le'ai (laki) male. The loss of n in lubu has, so far as I know, no 
 parallel in Isn. 
 
 Tagbanwa: Pepet regularly becomes u (o), exceptionally i or a: bugas 
 (begas) rice, itom {item) black, usa (esa) beside isa (isa) one (see above, 
 p. 936), tulo (tela) three, unon (enem) six, bo'ok (buek) hair, madlom (OJav. 
 malem, Mai. ma lam, Gam mo'lam, Ilk. malem, Chro. macum) night, ka'un 
 (Bis. kd'on, Tag. kd'in) eat; nipin (nipen) tooth; bituan (bituen) star. Tgb. 
 nipin has initial /z for IN zz. 
 
 With the final n in zz/zo/z for IN /zz, compare Tgb. /m<? (lima) five, but 
 the m is retained unchanged in itom and madlom. 
 
 Chamorro: Pepet becomes u (o), exceptionally e: dtof (atep) roof, 
 tdnum (tanem) plant, pugas (begas) rice, lotsa (Tag. lisd, Bis. taz) nit, ta/o 
 (tela) three, gunurn (enem) six, huhug (deneg) hear; /?//<£# (nipen) tooth. 
 
 The £ of «//«?« is due to the / of the preceding syllable 2 . 
 
 The consonantal peculiarities of Chro. illustrated by the above examples 
 are the following: IN p becomes / (dtof, nijen); IN b becomes p (pugas, cf. 
 piilan, IN bulan moon); the consonant of the RLD series becomes h initially 
 (hiinug, cf. hdnum, Phil, danum water); a parasitic g is developed (gunum, 
 cf. ginem, IN inum drink) and IN initial n is simplified to n (ntfen, cf. naan, 
 Phil, nalan name). 
 
 Unclassified languages. 
 
 Banawi: Pepet becomes i in olig (uleg) snake. 
 
 Bon to k: In this dialect, the development of the pepet vowel, like that 
 of other sounds, notably original b 3 , seems to be of a variable and uncertain 
 character. It appears as i in isa (esa or isa), ipdt (epat) four, inim (enem) six, 
 piisig (pused) navel; asa light e in the last syllable of utek (utek) brain, and 
 cuh-nen (den(e)g-en) hear; as the sound of u in Eng. but* in the first syllable 
 of the last example, and in owug (uleg) snake ; and as o in fook (buek) hair 
 and told (telu) three. In Bon. cukcdki great is seen the Phil, dake (table I) 
 with reduplication of all except the pepet vowel, the original a being 
 obscured to u. 
 
 With Bon. c for Phil, d (RLD) in cuh-n&n and cukcuki, compare ciiwa 
 (dua) two. The w of owug is a labial glide after loss of intervocalic / (see 
 above, p. 924). With the / of fook, compare Bon. lifo (ribu) thousand. 
 
 In view of the vocalism of the numerals and pusig and cukcdki, it is 
 probable that Bon. should be classified among the /-languages, the other 
 vowels being explained as cases of vocalic assimilation, total (fook, told) or 
 partial (utek, owdg). 
 
 1 This question and the following answer are copied literally from Padre Alarcon's 
 "Catecismo", p. 32. 22, where d is written for 'a, and d for a'. 
 
 2 Cf. my paper "Consonant Changes and Vowel Harmony in Chamorro", "Anthropos", 
 vol. VI (1911), pp. 136—146. 
 
 3 Cf. Jenks, "The Bontoc Igorot", Manila 1905, p. 229, and my paper, "F and V in 
 Philippine Languages", Manila 1908, in Division of Ethnology Publications, vol. V, part II, p. 137. 
 
 4 Cf. Jenks, op. cit., p. 228 : u = u in but. 
 
942 Carlos Everett Cojnant, 
 
 Sambal 1 : Pepet becomes i, e, a or u(o): alaki (dake) large (Bol.); 
 nipen (nipen) tooth, kdnen (kan-en) food, cooked rice, dnem (enem) six, dpat 
 (epat) four, tdro (Bol.) tolo (lba) three, buyas (begas) rice, bituun (bituen) star. 
 
 Sbl. nipen shows n for original n as in several other speech groups, 
 and y for the consonant of the RGH series in buyas, as in Pamp. and Bat. 
 (see above p. 925). 
 
 Ata: This is probably an rt-language : tatlo (tet(e)lu) three, appdt {epat) 
 four, anndm (enem) six. In pound (penu) full we evidently have a case of 
 regressive assimilation of which this same root has furnished so many examples 
 in other languages. Ata, like Bgb., is a geminating language, as is apparent 
 from the examples here given. 
 
 Bilan: This, like the preceding, is probably an ^-language: atlo (tela) 
 three, faai {epat) four, agtiaman (enem) six. Bil., like Tir. and Tgk., is an 
 /-language. With faat compare Bil. fito (pita) seven. Metathesis is especially 
 frequent and varied in character in this language. It is seen in the first syllable 
 of atlo and faat, for tald and a fat, and in the last syllable of the root in 
 gasfaolan (for gas-falo-an, IN pulu) ten, beside atlo falo (ielu piilu) thirty. 
 The peculiar form aguaman (agwaman) may be easily explained en the basis 
 of Phil, enem and comparison with the other numerals in Bil. To aman for 
 anam by metathesis, is prefixed the parasitic gw from gwalo (walu) eight 
 (cf. Inb. gwalo, Chro. gwdlog), the g analogy being assisted by the g of 
 nagfitd seven, gasium nine, and gasfaolan ten, and to this increment is further 
 added the initial a of the first three Bil. cardinals anisu, aluii, atlo. 
 
 Manobo: Pepet becomes u (o) in upat (epat) four, ika-unum (enem) 
 six, and pond (p£nu) full. Mnb. ikaltd three (properly third) is formed from 
 the ordinal prefix ika and Phil, tela with syncopation of pepet and metathesis 
 of the consonants thus brot together. 
 
 Tagakaolo: Pepet becomes u (o) in ufat (epat) four and ka-nuon (enem) 
 six. The latter form, if correctly written, is difficult of analysis. Tgk. is an 
 /-language: with ufat compare fito seven and folo ten. 
 
 Samal: Pepet becomes u (o) in too (telu) three, upat (epat) four, and 
 
 ika-unum (enem) six. One is isa in Sml. Syncopation of / appears in too and 
 
 wad (walu) eight. 
 
 Recapitulation. 
 
 /-languages: Ta^alog, probably Bontok, and perhaps Banawi. 
 
 ^-languages: Iloko, Inibaloi, Kankanai, Kalamian, Kuyunon, Lepanto, 
 Magindanau, Pangasinan, Tirurai, Batan (a in penult). 
 
 ^-languages: Pampanga, Ibanag, and its dialects, Gaddang, Itawi, and 
 Yogad, and probacly Ata~and Bilan. 
 
 (7-languages: Bagobo, Bisaya, Chamorro, Isinai, Sulu, Tagbanwa, 
 Tingyan, and probably Manobo, Samal, and Tagakaolo. Bikol has u (o) in 
 the ultima, and a in the penult except in the /?/?-class where it lias //. 
 
 1 The examples arc taken from the lists in Appendix B of Rrkd's "Negritos of Zambales" 
 in , .Ethnological Survey Publications", vol. II, part. I, Manila 1901. The words were selected 
 from the two colums headed "Zambal of Bolinao" and "Zambal of lba". Where the Bol. and 
 lba forms differ, 1 have so indicated in parenthesis. 
 
The Pepet Law in Philippine Languages. 943 
 
 Doubtful: Sambal. 
 
 It is seen from the above recapitulation that languages of the same 
 class are often widely separated geographically, and conversely, that several 
 classes may have representatives with in a comparatively small area. In fact 
 the different classes are so universally commingled geographically, that no 
 given territory can be said to favor any one of the different vowels evolved 
 from original pepet. 
 
 From a review of the classified phenomena of pepet vocalism as a whole, 
 it is evident that deviations from the normal development of the indifferent 
 vowel according to the regular operation of the pepet law are due almost 
 exclusively to the interference of the laws of vocalic assimilation and analogy. 
 
 Our study has also developed the fact that some Philippine languages, 
 like Malay and Malagasi, show a double pepet vocalism, one vowel regularly 
 appearing in the penult and another in the ultima of the IN prototype, and 
 that the Philippine languages, wherever they show this double vocalism, have 
 a in the penult. This is the case of Batan and Bikol, and doubtless of the 
 exceptional a seen in the penult of some Bagobo words, and sporadically 
 elsewhere. That this penultimate a is not the result of assimilation or analogy, 
 but an undisturbed and natural development of the indifferent vowel in a 
 special position, has appeared from a searching investigation of the Bikol 
 material, and the testimony of the Batan. 
 
 We have also found that several languages have a special penultimate 
 a in the Phil, numerals telu, epat and enem, due to the combined action of 
 assimilation and analogy. 
 
 Bibliography. 
 
 1. Sources consulted for the Philippine languages and Chamorro. 
 
 Ata. 
 
 Montano J. Rapport a M. le ministre de Pinstruction publique sur une Mission aux lies Philip- 
 pines et en Malaisie (1879—1881). Paris 1885. 
 
 Bagobo. 
 GlSBERT Mateo. Diccionario Espanol-Bagobo. Manila 1892. (The dictionary proper is preceded 
 
 by a grammatical sketch of the Bagobo language, pp. IX XVI.) 
 - Diccionario Bagobo-Espanol. Manila 1892. (Not printed with the foregoing.) 
 
 Banawi. 
 
 Schadenberg Alex. Beitrage zur Kenntnis der im Innern Nordluzons lebenden Stamme. (An 
 article pub. in "Verhandlungen der Berliner Gesellschaft fiir Anthropologic, Ethnologie 
 und Urgeschichte", Nov. 16, 1889, pp. 649—727.) 
 
 Batan. 
 
 Conant C. E. A list of Batan words (MS) orally taken from two natives of Batan Island, one 
 of whom, a servant boy recently arrived from his native island, gave his list a Aparri, 
 north coast of Luzon, Oct. 7, 1904, and the other, an elderly settler in Claveria, a small 
 town on the northwest coast of Luzon, gave his list at the latter town, Nov. 9, 1905 
 About 200 words. 
 
 Dominican missionary or missionaries (not named). Nu Napia Amigo anmana Devocionario du 
 chirin nu Ibatan a ichasantos nu cristiano: a pinarin da ni Padres Misioneros du Islas 
 Batanes. Manila 1901. (Bound together with the following, which is also bound separately.) 
 
 — Visitas du Santisimo cani Santa Maria a pinayapu ni S. Alfonso Ligorio. Manila 1901. . 
 
944 Carlos Everett Conant, 
 
 Paula Franco de and Castano NicolAs. Diccionario Espafiol y Batan. (Date and place un- 
 certain. About 200 items of this work have been copied by Retana, Archivo del Biblid- 
 filo Filipino, vol. II, Madrid 1896, Prologo, pp. XIII— XIX.) 
 
 Rodriguez Jose. Catecismo dc la Doctrina Christiana. Manila 1834. (Reprinted by Retana, 
 Archivo del Bib. Fil., vol. II, Madrid 1896, pp. 260—306.) 
 
 SCHEERER Otto. The Batan Dialect as a Member of the Philippine Group of Languages of 
 Ethnology Publications, vol. V, part I, Manila 1008. 
 
 Bikol. 
 
 -f Lisboa Marcos de. Vocabulario de la lengua Bicol. 2 d ed., Manila 1865. (The only dictionary. 
 A folio vol. in 2 parts, Bkl.-Sp. and Sp.-Bkl., 417 and 104 pp.) 
 Herrejon Santos. Lecciones de Gramatica Bicol-Hispana. Binondo 1882. 
 Vera Roman Maria de. Gramatica Hispano-Bicol. Manila 1904. 
 
 Bilan. 
 
 Montano J. Raport &c. (see Ata). 
 
 Bisaya. (Cebuan.) 
 
 Conant C. E. A Bisaya-English Dictionary (MS) prepared at Cebii, P. I., 1906, with collaboration 
 
 of Vicente Sotto and Juan Villagonzalo. About 5500 words. 
 Encarnacion Juan Felix de la. Diccionario Bisaya-Espanol, 3' 1 ed. Manila 1885, 437 pp. fol. 
 
 bound together with the following. 
 Diccionario Espafiol-Bisaya, 3 d ed. Manila 1885, 349 pp. fol. (The standard work for the 
 
 Cebuan dialect. It also contains many words from other Bis. dialects.) 
 Encina Francisco. Gramatica Bisaya-Cebuana del P. Fr. En. Augustino Calzado reformada 
 
 por el M. R. P. Fr. Nicolas Gonzalez, Manila 1885. 
 Guillen Felix. Gramatica Bisaya para facilitar el estudio del Dialecto Bisaya Cebuano. Malabon 1898. 
 I Zueco de S. Joaquin Ramon. Metodo del Dr. Ollendorf para aprender a leer, hablar y 
 
 escribir un idioma cualquiera, adaptado al Visaya. 2 l ed. Manila 1884. 
 
 Bisaya. (Hiligaina. Spoken in Panay and Occidental Negros.) 
 
 Mentrida Alfonso de. Diccionario de la lengua Bisaya, Hiligucina y Haraya de la Isla de 
 Panay. Manila 1851. (This work comprises the first 460 pp. of a vol. of 827 folio pp. the 
 second part which is the following item.) 
 
 Martin Julian. Diccionario Hispano-Bisaya. Manila 1842. (The only large dictionary of the 
 Panayan dialects. My access to this rare work was due to the courtesy of the Hon. T. 
 H. Pardo de Tavera, who kindly allowed me to consult his copy in his library at Manila.) 
 
 Lozano Raymundo. Cursos de lengua Pannyana. Manila 1876. (Contains a Bisaya-Spanish dic- 
 tionary, pp. 91—231.) 
 
 Mentrida Alfonso. Arte de la lengua Bisaya-Hiligayna de' la Isla de Panay . . . corregido y 
 aumentado por el M. R. P. Jose Aparicio. Tambobong 1894. 
 
 Bisaya. (Samar and Leyte.) 
 3 SAnchez DE la Rosa Antonio. Diccionario Hispano-Bisaya para las provincias de Samar y 
 Leyte. Manila 1895. (480 folio pages, bound together with the following item.) 
 — Diccionario Bisaya-Espanol . . • para las provincias de Samar y Leyte. Manila 1895. (332 pp. fol.) 
 Gramatica Hispano Visaya . . . de las provincias de Leyte y Samar. Manila 1887. (Contains 
 a most instructive Compendio Visaya, pp. 298—327. Note that the author wrote Bisaya 
 [not Visaya] in his later work just listed.) 
 Imgueroa Antonio. Arte del Idioma Visaya de Samar y Leite. 2' ed. Biondo 1872. 
 
 Bisaya. (Masbate and Tikau.) 
 Rosa Pablo de la. Vocabulario Visaya-Ingles Metodo practice) snn^ ingles agud mapag- 
 adalan sang mga taga isla sang Masbate cag Ticao. Manila 1905. 
 
 Bisaya (of Cuyo Kuyunon). 
 Augustinian Friar An. Adalan sa mga Christianos nga insultat sa cuyunon ig sa isarang Padre 
 Augustino Recoleto. 2' 1 ed. Manila 1904. (A pamphlet of 72 pp. containing the Chri- 
 stian doctrine in Kuyunon.) 
 
The Pepet Law in Philippine Languages. 945 
 
 Conant A. E. A word list of the Kuyunon dialect (SM) furnished orally by a native of Cuyo. 
 Manila 1904. Contains the numerals and names of parts of the body (ca. 50 words). 
 
 Bontok. 
 
 Jenks A. E. The Bontoc Igorot. Ethnological Survey Publications, vol. I, Manila 1905. 
 
 Chamorro. 
 
 Conant C. E. Consonant Changes and Vowel Harmony in Chamorro. Publ. in "Anthropos", 
 
 vol. VI, pp. 136—146. 
 Fritz Georo. Chamorro-Grammatik. In "Mitteilung des Seminars fur orientalische Sprachen an 
 
 der kdniglichen Friedrich Wilhelms-Universitat zu Berlin, Jahrgang VI, erste Abteilung: 
 
 Ostasiatische Studien". Berlin 1903, pp. 1—27. 
 — Chamorro- Wdrteibuqh in zwei Teilen : Deutsch-Chamorro und Chamorro-Deutsch, auf der Insel 
 
 Saipan, Marianen, gesammelt von G. F., kaiserl. Bezirksamtmann auf Saipan. Berlin 1903. 
 Ibanez del Carmen Aniceto. Diccionario Espanol-Chamorro que dedica a las escuelas de 
 
 Marianas el P. Fr. A. lb. del C. Cura Parroco de Agaiia. Manila 1865. 
 Safford W. E. The Chamorro Language of Guam. Published serially in vols. V, VI and VII 
 
 of the "American Anthropologist", Washington, D. C. 1903, 1904, 1905. 
 
 Gaddang. 
 
 Conant C. E. A Brief Comparative Word List of the Yogad, Gaddang, and Itawi Dialects (MS). 
 
 A list of 75 English words with their equivalents in the three dialects taken orally from 
 
 several natives in N. Luzon, 1904 and 1905. 
 Rodriguez Jose. Catecismo de la Doctrina Christiana en lengua Ga-dang impreso por primera 
 
 vez el ano 1833. 2 d ed. Manila 1897. (173 pp.) 
 
 Haraya. 
 
 (See above, Bisaya Hiligaina. Padre Pedro Chirino gives the Haraya version of the "Ave Maria" 
 in his famous work "Relacion de las Islas Filipinas", 2 l ed. Manila 1890, p. 54. Mentrida, 
 in his Hiligaina grammar [pp. 18 — 20] gives the Haraya "declinacion" of the articles 
 and pronouns. Padre Hilario Santaren wrote a "Catecismo Historico" entirely in the 
 Haraya dialect, printed by the Colegio de Santo Tomas, Manila 1877.) 
 
 Hiligaina. 
 
 (See Bisaya Hiligaina.) 
 
 Ibanag. 
 
 Bugarin Jose. Diccionario Ibanag-Espafiol compuesto en lo antiguo por el R. P. Fr. J. B., 
 reducido a mejor forma por el R. P. Fr. Anronio Lobato de Sto. Tomas; compendiado 
 por el R. P. Pr. Julian Velichon, Vicario Provincial; reducido a metodo mas claro, con 
 un suplemento, y dado a luz por el R. P. Fr Ramon Rodriguez, antiguo ministro en 
 Cagayan y actuel Procurador General de la Provincia del Santisimo Rosario, de orden de 
 N. M. R. P. Provincial Fr. Antonio Carrillo. Manila 1854. (The only large Ibg.-Span. 
 diet, published. It is a folio vol. containing 12 -|— 280 -j- 76 pages.) 
 
 MS Ibanag-Spanish dictionary (tille page lacking) containing 174 folios (348 pp. and an "Indice 
 de las raices anticuadas estrahidas de cuerpo del Diccionario para mayor comodidad" 
 containing 16 folios to "quippal". It is very closely but clearly written and contains a 
 large number of words and definitions not found in the foregoing. 
 
 Payo Pedro. Diccionario Espanol-Ibanag (sic). Manila 1867. 
 
 Fausto de Duevas Jose Maria. Arte Nuevo de la Lengua Ybanag. 2' 1 ed. Manila 1854. 
 
 (The so-called grammars of Nolasco de Medio and Nepomuceno are manuals in Ibg. for the 
 study of Spanish.) 
 
 Iloko. 
 
 Carro Andres. Vocabulario Iloco-Espanol. 2' 1 ed. Manila 1888. (The only large Ilk.-Span. diet., 
 
 a folio vol. containing XII -f- 295 pp.) 
 Floresca Romual do. Vocabulary English-Ilocano. Vigan 1904. (237 pp. small 8 vol.) 
 Lopez and Carro. Gramatica Ilocana. 3' 1 ed. Malabon 1895. 
 
946 Carlos Everett Conant, 
 
 Inibaloi. 
 
 Scheerer otto. The Nabaloi Dialect. Ethnological Survey Pub. vol. II, Manila 1905. (Corrected 
 and checked by my own lists taken in Baguio, Bcnguet 1903, 1905, and 1906.) 
 
 Isinai. 
 Rocamora Francisco. Catecismo de la Doctrina Crlstiana en la lengua dc Isinay 6 Inmeas. 
 
 Manila 176. (176 pages.) 
 Conant C. E. Isinai-English Word List compiled from the foregoing. Baguio, Benguet, P. I., 
 
 1907 (MS). 
 
 Itawi. 
 (See Gaddang.) / 
 
 Kalamian. v 
 
 Jeronimo de la Viroen de Monserrate. Vocabulario Castcllano-Calamiano. MS of 1789 
 published by W. E. Retana, Arch, del Bib. Fil. vol. II, pp. 207-224. 
 
 Kankanai. 
 
 Conant C. E. Kankanai Word Lists (MS) taken orally from eight Kankanai boys, each being 
 questioned privately, at Baguio, Benguet 1903. Fifty words, chiefly numerals and parts 
 of the body. 
 X Lagsaca M. (Quoted by Scheerer, "Batan Dialect", p. 20 et passim.) 
 
 Kuyunon. 
 
 (Sec Bisaya of Cuyo.) 
 
 Lepanto. 
 Schadenberq Alex. op. cit., for Banawi, q. v. 
 
 Magindanau 
 
 Juanmarti Jacinto. Diccionario Moro-Maguindanao-Espafiol and Diccionario Espafiol Moro- 
 Maguidanao. Manila 1892. (The two parts bound in one 4 to vol. 272 and 242 pp.) 
 Gramatica de la Lengua de Maguindanao segun se habla en el centro y en la costa sur de 
 la Isla de Mindanao. Manila 1892. 
 SwriH Capt. C. C. A Grammar of the Maguindanao Tongue. Washington, D. C, 1906. (A trans- 
 lation of the foregoing.) 
 
 Manobo.'' 
 A Montano J. Rapport &c. (sec Ata.) , 
 
 Pampanga. ,/ 
 \ Berqano Dieoo. Vocabulario de la lengua Pampanga en Romance. Reimpreso: Manila 1860. 
 (The only large dictionary, 343 pp. fol., of which pp. 279—343 are taken up with a 
 "Diccionario de Romance en Pampango".) 
 Arte de la lengua Pampanga. Manila 1729. 
 Fernandez Eligio. Nuevo Vocabulario Espafiol, Tagalo y Pampango. 4"' ed. Manila 1896. 
 Parker Luther. An English-Spanish-Pampango Dictionary. Manila 1905. 
 
 Pangasinan. ^ 
 
 Cosgaya Lorenzo Fernandez. Diccionario Pangasinan-Espanol. Manila 1865. (in two parts: 
 "Diccionario de la lengua Pangasinana" and "Vocabulario Hispano-Pangasinan", 330-f-121 
 pp. fol. The only large dictionary and exceedingly race. I secured my copy privately in 
 Manila after searching for a copy for two years in vain.) 
 
 Pellicer Mariano. Arte dc la lengua Pangasinan 6 Caboloan. 3' ed. Manila 1904. 
 
 Samal. \S 
 
 ' i Montano J. Rapport &c. (see Ata). 
 
 Sambal. v 
 REED W. A. Negritos of Zambales. Ethnological Survey Pub. vol!. II, part I, Manila 1904. 
 
 Sulu. * 
 
 CowiE Andson. English-Sulu-Malay Vocabulary. London 1893. 
 
 Havnes T. H. English, Sulu, and Malay Vocabulary. Published in Journal of the Straits Branch 
 Roy. As. Soc, Dec. 1885 and Dec. 1886. 
 
 Tagakaolo. S 
 
 Montano J. Rapport &c. (see Ata). 
 
The Pepet Law in Philippine Languages. 947 
 
 Tagalog. 
 
 Noceda Juan de and Sani.ucar Pedro de. Vocabulario de la lengua Tagala. Reimpreso en 
 Manila 1860. (642 pp. fol. in two parts: "Vocabulario Hispano-Tagalog (sic)" pp. 423—642. 
 The most complete and reliable dictionary.) 
 
 niqg Charles. A Tagalog English and English Tagalog Dictionary, Manila 1904. (360 pp. 8 vol.) 
 
 Minguella de las Mercedes Toribio. Ensayo de Gramatica Hispano-Tagala. Manila 1878. 
 
 San Josef Francisco de. Arte y Reglas de la lengua Tagala. Manila 1832. 
 
 S. Augustin Gaspar de. Compendio del Arte de la lengua Tagala 3 d ed. Manila 1979. 
 
 Totanes Sebastian de. Arte de la lengua Tagala, y Manual Tagalog (sic). Binondo 1865. 
 
 Tagbanwa. 
 
 Everett A. Hart. Word list incorporated in Swettenham's "Comparative Vocabulary of the 
 Dialects of some of the Wild Tribes inhabiting the Malayan Peninsula, Borneo, &c", 
 Journal of the Straits Branch of the R. A. S., June 1880. 
 
 Marcilla v. Martin Cipriano. Estudio de los Alfabetos Filipinos. Malabon 1895. 
 
 Tingyan. 
 
 Meyer Hans. Eine Weltreise. Leipzig 1884. 
 
 Tigurai. 
 
 Bennasar Guillermo. Diccionario Tiruray-Espafiol. Manila 1892. 
 — Diccionario Espanol-Tiruray. Manila 1893. 
 
 Un P. misionero de la Compania de Jesus (known to be Padre F. BennAsar). Observaciones 
 gramaticales sobre la lengua Tiruray. Manila 1892. 
 
 Yogad. 
 
 (See above, Gaddang.) 
 
 2. Other works which have been especially useful in the preparation of this article : 
 
 Set in Adriani N. Sangireesche Spraakkunst. Leiden 1893. 
 
 Set in Brandstetter Renward. Malayo-polynesische Forschungen. Erste Reihe: II. Die Be- 
 ziehungen des Malagasy zum Malaiischen. Luzern 1893. — Zweite Reihe: II. Tagalen 
 und Madegassen. Luzern 1902. III. Ein Prodromus zu einem vergleichenden Worterbuch 
 der malaio-polynesischen Sprachen. Luzern 1906. IV. Mata-Hari, oder Wanderungen eines 
 indonesischen Sprachforschers durch die drei Reiche der Natur. Luzern 1908. 
 
 Cabaton Antoine. Dix dialectes indochinois recueillis par Prosper Odend'hal, administrateur 
 des services civils de l'Indochine; etude linguistique par A. C. Reprint from the March- 
 April, 1905, number of "Journal Asiatique", Paris 1905. 
 
 Favre L'Abbe P. Dictionnaire Malais-Francais. Vienna 1875, 2 vols. 8vo., and Dictionnaire 
 Francais-Malais. Vienna 1880, 2 vols. 8vo. 
 
 Ferrand Gabriel. Un texte arabico-malgache du XVI 9 siecle, transcrit, traduit et annote d'apres 
 les mss. 7 et 8 de la Bibliotheque nationale, being an extract from Notices et Extraits, 
 vol. XXXVIII, pp. 449—576. Paris 1904. 
 
 Ferrand's great work on the Malagasi dialects, entitled "Essai de phonetique comparee du 
 malais et des dialectes malgaches", Paris 1909, did not come to hand in time to be 
 consulted for the present study. 
 
 Hardeland Aug. Dajacksch-Deutsches Worterbuch. Amsterdam 1859. 
 
 Kern H. Taalvergelijkende verhandeling over het Aneityumsch met een Aanhangsel over het 
 Eromanga. Amsterdam 1906. 
 
 Schmidt P. W. Die Mon-Khmer-Volker, ein Bindeglied zwischen Volkern Zentralasiens und 
 Austronesiens. Braunschweig 1906. 
 
 Sievers Eduard. Grundzuge der Phonetik, 5"' ed. Leipzig 1901. 
 
 Worcester Dean C. The Non-Christian Tribes of Northern Luzon, published in The Philippine 
 of Science, vol. I, no. 8. Manila, October, 1906. 
 
ERRATA 
 
 For some reason, not yet known to the author, his proof corrections, sent from America to Vienna, 
 failed to be made by the printer, hence the necessity of indicating them here. 
 
 A few entirely obvious misprints, e. g., thaf (for that), are passed without notice, 
 p. 920, line 2, for: Lecturer of, read: Lecturer in 
 
 922, line 14, for: top, read: stop 
 
 923, line 19, omit period after from 
 line 27, for: 'piga, read: piga 
 
 926, line 9, for: intervocative, read: intervocalic 
 928, line 4 from end, for: 927, read: 925 
 
 930, 3d line before Table V, read: If these are to be connected with 
 
 931, footnote 1, line 2, for: eastly, read: easily 
 
 932, 4th line before Table VI, for: inserted a, read: inserted as 
 
 933, 2d line following table at top of page, for: 931. read: 927 
 last line preceding Table VII, for: 931, read: 927 
 
 line 3 from end of page, lor: Xapia, read: Mapia 
 936, footnote 1, line 4, for: iisa, read: isa 
 >r : and as e, read : and as i 
 line 14, for: Ibg., read: by 
 
 line 3 from end, for: Bis ialum, read: Bis! ilalum 
 line 2 from end, parenthesize nipen 
 
 939, line 2, parenthesize epat 
 
 line 16, for: keuai, read: kenai 
 
 line 4 from end of text, for: tattal (Bis., read: tattal, Bis., 
 
 footnote 2, line 4, read: serveza (for cervez 
 
 footnote 3, line 1, for: Nap:a, read: Mapia 
 
 940, line 8, for: that Inb.. read: that of Inb. 
 line 15, for: Togad, read: Yogad 
 
 line 17. for: bagga, read: haggat 
 line 18, for: rics, read: rice 
 
 line before "U-languages," for: leg., read: Ibg. 
 lines 4 and 2 from end, for: anon, read: 'anon 
 last line, for: a (ka), read: 'a (ka) 
 
 941, line 1, for: a (ak), read: a' (ak) 
 
 943, second heading under "Bibliography," for: Ragobo, read: Bagobo. 
 line 4 from end, for: Xapia, read: Mapia 
 
 944, lnu- 6, for: Languages of, read: Languages. Division of 
 
 7, for: 1008, read: 1908 
 7th line under heading "Bisaya (Hiligayna," etc., for: Pannyana, read: Panayana 
 line 3 from end of page, for: insultat, read: insulat 
 
 945, kne 1, for: A. E., read: C. E., and for: (SM), read: (MS) 
 
 >nd line under heading "Ibanag," for: Anronio, read: Antonio 
 3d line under heading "Ibanag," for: Pr. Julian Velichon, read: Fr. Julian Velinchon 
 line 2 from end, read: Romualdo, and: 8vo 
 
 946, line 2 under heading "Linai." for: Manila 176, read: Manila 1876 
 under heading "Magindanau", for: SWITH, read: SMITH 
 under heading "Sulu." for: 1IAVXES, read: HAYNES 
 
 947, line 3, insert after: in two parts: the following: "Vocabulario de la lengua Tagala," pp. 1-422, and 
 line 7, for: 1979. read: 1879 
 
 under heading "Tagbanwa," for: MARCILLA V MARTIN, read: MARCILLA Y MARTIN 
 for the heading: Tigurai, read: Tirurai 
 
 at center of page, omit the words: Set in, before the names ADRIANI and BRANDSTETTER 
 last two lines of page, for: Philippine of Science, read: Philippine Journal of Science. 
 
 XOTE: As the manuscript of this study has been in the hands of the publisher since October, 1909, no 
 ■ nee is made to contributions to Indonesian philology appearing since that date, with the exception 
 he author's own articles, already prepared and awaiting publication, the proper citations having been 
 quently communicated to the publisher. 
 
 Chattanooga, Tennessee 
 March 1, 1913 
 
.1 
 
OAYLORO BROS. 
 
 MAKERS 
 
 SYRACUSE. - NY. 
 
 4rVi,;i>frif»'' 
 
 UNIVERSITY OF CALIFORNIA LIBRARY 
 BERKELEY 
 
 Return to desk from which borrowed. 
 This book is DUE on the last date stamped below. 
 
 f *fi 24 1948 
 
 
 4Feh' l| DECEIVED 
 Mo V 2249HJtlN 8 '68 -£ PM 
 
 2Kf LOAN D ^ T 
 
 *. 
 
 W6l 6 1955 LU 
 
 AUG 17 '67 -2 PM 
 
 AUG 2 1969 06 
 fiECD LD SEP 
 
 23 '69 -11AM 
 
\