CV-Truncation or CV-Deletion
In Robert Blust’s “Austronesian Nasal Substitution: A Survey“(2003), he identified a process called CV-Truncation or CV-Deletion which is being confused with real Nasal Substitution, so he gave it an alternative name Pseudo-Nasal Substitution (PNS). This is a process where Western Malayo-Polynesian languages disfavored dissimilar labial consonants in successive syllables, such that the sequences *pVm- or *bVm- are absent in these languages. I quote:
“Although *p-um- and *b-um- apparently also occurred in PAN, in many attested languages such sequences of dissimilar labial consonants are excluded across a morpheme boundary just as they are within a morpheme. To avoid such sequences where they have arisen through infixation with *-um- ‘Actor Focus’, many languages drop the initial CV- (or, in some cases, just the initial C) of the affixed word, producing a result that is superficially similar to NS: base*pVCVC : infixed form *p-um-VCVC > base pVCVC : apparently prefixed form /mVCVC/.”
Using the Thao language from Taiwan, he gave examples of bases that starts with p, and other bases that starts with consonants other than p to illustrate this process. The below table have some of the examples he gave:
|Root||Gloss||Expected Word||Word Surface Realization|
|pilalaha||spread, of the legs||p‹um›ilalaha||milalaha|
|pulhbuz||to sink something, make something sink||p‹um›ulhbuz||mulhbuz
Verbs that start with a non-labial consonants (k› for example) result in the infix being “transparent” . In labial-initial roots, there is substitution of the initial labial with m, except those in f› initial bases, where the infix ‹um› has a zero allomorph. He concluded that “Surface forms such as matash, then, are assumed to derive from earlier *p-um-atash, with loss of the initial CV-.”
He mentioned that Van der Turk noticed this in Toba Batak as well, a language in Indonesia, where he quoted “stems infixed with -um- lose the initial syllable if and only if they begin with a labial consonant: “Where an m occurs in place of the beginning of the stem – word truncation must be presumed, e.g., mate instead of pumate, its truncated form being due to the dislike of the language to having each of the first two syllables in succession in a trisyllabic word beginning with a labial.””
Blust enumerated languages where this process is still or were active, which I list below:
|Range of Truncation||Consonants Affected||Languages|
|C-||p,b||Old Javanese, Bolaang Mongondow|
|CV-||p,b||Toba Batak, Palawan Batak, Sarangani Manobo, Tausug, Kiput, Mukah Melanau|
|CV-||p (but not f)||Thao|
|CV-||p,f,w (but not b, bh)||Muna|
|CV-||all consonants||Squliq Atayal|
In each of these instances, the apparent operation of NS is an illusion: labial-initial bases infixed with *-um- have simply dropped the initial syllable (or consonant), leaving the nasal of the infix as an apparent nasal substitute of the deleted base-initial consonant. A priori these observations might be taken as evidence that PNS was active in PAN, and some writers have reached this conclusion (e.g.,Wolff 1973:73,Mills1975:[1i 146). However, other observations suggest that PNS has arisen repeatedly through independent change. First, some languages lack PNS [Ilokano, Tagalog, Chamorro and Bontok]. If we assume that PNS was active in PAN, we must account for its absence in many languages by an analogical extension of the pattern found in bases that do not begin with a labial consonant. But such analogical extensions would not have shared a common history, because their distribution does not follow accepted subgroup boundaries (e.g.,Tagalog and Bontok lack PNS, yet Tagalog subgroups more closely with Palawan Batak and Sarangani Manobo,which have it). It is thus simplest to assume that PNS has developed independently in the languages that share it.
Second, some languages show PNS for a labial stop that has remained unchanged, but not for one that has been altered to a segment with less constriction. Thao, as already seen, has PNS in bases that begin with p-, but not in those that begin with f (< *b). If this process had been inherited from PAN, we would expect f initial bases to contain active or fossilized instances of Actor Focus forms with m-, but none are known, thus suggesting that CV-deletion took place in Thao after the sound change *b> f. In summary, PNS is a process of CV-truncation that is motivated by strong but not invariant constraints against the surface sequences bVm- and pVm-. It is thus confined to a highly specific phonological environment. Because many languages fail to truncate the initial CV- of a labial-initial base that is infixed with a reflex of *-um-, it must be assumed that this widespread tendency to eliminate surface sequences of pVm- or bVm- across a morpheme boundary has arisen repeatedly through convergent change that was in turn driven by inherited structural pressures.”
Blust identified two scenarios: (a) PNS originated and was active in PAN and daughter languages inherited active or fossilized forms, and (b) PNS arose independently in the daughter languages. However, there is another scenario: that it originated in PAN and truncation was inherited by the daughter languages, some of which lost it and some retained it. In this third scenario, the daughter languages inherited not the active or frozen forms, but truncation itself as a word formation strategy, so that once a daughter language lose this strategy, it just resumes to not truncating the forms. This is what happened in Tagalog and Bikol, as described below.
Truncation of roots as a process in Tagalog and Bikol
We will not be using Blust’s term Pseudo-Nasal Substitution as the more salient characteristics is not the substitution itself but the truncation. It’s also actually not true that Tagalog lacks CV-truncation, and not true that it only affects labial-initial bases.
In Frank Blake’s “A Grammar of the Tagalog Language” (1925, p. 56), he mentioned that
“Disyllabic roots beginning with b and p, besides having the regular active forms, may change b or p to m for the modal, to n for the preterite, and the reduplicate the first syllable of the preterite for the present.”
He also mentioned that other roots beginning with k, t and vowels undergoing the same process, such as kaon, kuha, tuka, tukso, akyat, ayaw, igib, ihi, inum, una, urung, osos, otot, uwi. I tabulate below the forms which he wrote:
This is confirmed by an earlier work, that of W.E.W. MacKinlay’s “A Handbook and Grammar of the Tagalog Language” (1905, p. 173). He wrote:
“Bisyllabic (two-syllabled) roots commencing with b, k, p, t, or a vowel, generally admit of a similar irregularity in the imperative, past, and present tenses; n being prefixed to vowel roots for the past and present tenses and m for the imperative, while the initial letter of b, k, p and t roots changes to n for the past and present tenses and to m for the imperative.”
One of the examples used by MacKinlay was kuha:
I suppose all these examples are derived from infixation then truncation, like so:
|Modal (or Subjunctive)||Preterite (or Past)|
|b‹um›ása → (bu)mása → mása||b‹um›‹in›ása → (bumi)nása → nása|
|p‹um›ások → (pu)mások → mások||p‹um›‹in›ások → (pumi)nások → nások|
|t‹um›okso → (tu)mokso → mokso||t‹um›‹in›okso → (tumi)nokso → nokso|
|‹um›alis → (u)malis → malis||‹um›‹in›alis → (umi)nalis → nalis|
|k‹um›uha → (ku)muha → muha||k‹um›‹in›uha → k(umi)nuha → nuha|
If we remove the truncation, the current formation of the regular “modal” is actually as above, and the formation of the current “preterite” is the regular formation of such in Bikol. The present is derived by reduplicating the first syllable of the preterite/past. So we can see in the above table that this process used to be active in root bases in Tagalog but not anymore now, and not just restricted to labial-initial bases.
This was also present in Bikol. In Andres de San Agustin’s “Arte De la Lengua Bicol” (1795, p.49-50):
“La formacion del Preterito perfecto, se hace con modos diferentes, porque si la Raiz del Verbo comienza por Vocal, se le antepone esta Particula Imin; â la primera Vocal de ella Y.g. Arog, imitar, anteponiendo el, imin, forma Iminarog, el que imitò, y de Apod, iminapod, el que llamò; pero si la Raiz comenzare en consonante, se ha de formar este Preterito , peniendo de despues de la primera letra consonante, y la segunda Vocal el imin, como holit, enseñar, poniendo en medio de la letra, h, y la Vocal,o, el imin, dice Himinulit, el que enseñò. Mas si la consonante con que comenzare la Raiz fuere P, O, B, se convierten en esta Particula min y se liga con lo restante de la Raiz, como esta Raiz. Pili, que significa escoger que forma minili, el que escogiò, Basa, leer, minasa, el que leyò.
La formacion de el Imperativo……
NOTA. que si la primera consonante de la Raiz fuere P, O, B, la tal [??] P, O , B, se convierte en M. V.g. Pirit, forzar, formar Mirit, fuerza. Basa, leer. Masa, lee. Pero se debe notar que en todos los Presentes de Imperativo, siempre se ha de poner el nombre de la persona, o personas, a quien se manda. V.g. Humulit ca, enseña tu. Masa camo, leed vosotros.”
This process is also not anymore happening in Bikol, but good to know that it used to be operative, at least until the 18th century.
Jason Lobel, in his dissertation “Philippine and Northern Bornean Languages: Issues in Description, Subgrouping, and Reconstruction” (p.298), came across the same information, and stated that Maranaw also has this process currently.
“For the infinitive, is infixed to most roots. For vowel-initial roots, this is reduced to a prefixed m-. For roots with an initial bilabial /b p/, the first syllable is dropped, leaving what appears to be a replacive m-. This latter process, called Pseudo Nasal Substitution (cf. Blust 2004), was once widespread even in Central Philippine languages like Old Tagalog, Old Bikol, Old Waray, etc., as outlined in Lobel (2004).”
I’ve read somewhere that this is also present in Kapampangan, but I still need to obtain a Kapampangan grammar to have an example.
Truncation of derivatives as a process in Philippine languages
Furthermore in Tagalog, Blake has identified another set of roots that undergo a similar process (p.57):
“A large number of polysyllabic roots beginning with pa form their active tense stems by changing p to m in the modal and future, to n in the preterite and present, and reduplicating the second syllable of the root in the future and present.”
|administer or receive the Holy Communion||pakinabang||makinabang||makikinabang||nakinabang||nakikinabang|
And MacKinlay too (p175):
“Some sixty-six polysyllabic verbal roots commencing with pa replace the first syllable with na in the past and present and with ma in the imperative and future. In the present and future tenses the second syllable of the root is reduplicated and not the first.”
Now this type of truncation on pa› prefix is still active to this day, not just in Tagalog but also in Bikol, and most likely in other Philippine languages.
John Wolff, in “Similarities Between Indonesian and Tagalog and their historical Basis” (1981) extended this to also cover pag› and pang› prefixes:
“The active affix -um- added to bases pang- and pag- forms man- and mag- respectively (I.e. we consider the affixes mang- and mag- of Tagalog to consist of -um- plus pang- and pag- respectively, and there is fairly good descriptive evidence to make this a valid analysis.)”
It seems John Wolff elaborated on this in his earlier work “Verbal inflection in Proto-Austronesian” (1973) but I can’t find a online version of this work. However, Robert Blust, in his “Three Notes on Early Austronesian Morphology” (p.455) commented:
“In his pioneering exploration of PAn verbal morphology, Wolff (1973:72) suggests that “the prefixes that in many of the languages of the Philippines show up as mag-, maN-, maka-, and ma- (and the various other tense-aspect forms with which these affixes are in a paradigm) or their cognates in other Austronesian languages are, in their deep structure, composed of one or more inflectional morphemes plus a derivational morpheme. The inflectional and derivational morphemes [that] underlie these verbal affixes of modern Austronesian languages reflect the inflectional and derivational affixes [that] we ascribe to the protolanguage.” He illustrates these relationships with the active verb prefix *paR- (and, in principle, *paN-), as shown in table 3.”
Wolff’s formation of the past derivative is incorrect, as Reid’s “The evolution of focus in Austronesian” (1981, p.451) has a better derivation:
“The Tagalog derivational prefixes can be reconstructed in their m- and non-m forms as *paN-, *maN, *paR-, and *maR-, with the corresponding perfective forms *p-in-aN-, *(mi)naN-, *p-in-at-, and *(mi)naR- respectively.”
Reid has better support for this reconstruction (see my earlier post Realis Mood Infix Marker ), based on Northern Luzon languages he worked on, especially Casiguran Dumagat’s formation below.
In this case, the affixes are derived as follows in most Philippine languages:
|Affix Surface Form||Derivation|
|ma›||p‹um›a› → (pu)ma›|
|na›||p‹um›‹in›a› → (pu)(mi)na›|
|mag›||p‹um›ag› → (pu)mag›|
|nag›||p‹um›‹in›ag› → (pu)(mi)nag›|
|mang›||p‹um›ang› → (pu)mang›|
|nang›||p‹um›‹in›ang› → (pu)(mi)nang›|
Malcolm Ross’s “Proto-Austronesian verbal morphology: Evidence from Taiwan” and Stanley Starosta’s “A grammatical subgrouping of Formosan languages” extended this as well to the ka› prefix, as commented on by Blust (p. 440):
“Starosta (1995) proposes a number of PAn affixes, some of them not previously reconstructed, but without supplying fully the comparative evidence needed to support his inferences. These affixes include the four focus markers….plus the following:….8. inflectional affix: realis prefix *m- on a limited subset of intransitive verbs, but omitted in irrealis (future and negative) contexts, …. 17. *pa- “derived causative verbs from nonstative verbs,” 18. *m- “stative prefix that derived *ma-initial stative verbs from *ka-initial inchoatives and *mu-initial potential verbs from *ku-initial middle verbs. In this process as well as in *m- inflected realis forms, a following stem-initial *k- or *p- was lost after *m-,…
In the same volume Ross (1995) suggests that the PAn voice/focus system was expressed in indicative and nonindicative moods through the following affixational patterns:…..In addition he noted (740) that “there appear to have been four major formal classes of verb in PAn: (1) those that took Actor Pivot infixation of *-um- into the root, (2) a small class of verbs whose AP (and sometimes other pivot) forms had no affixes, (3) verbs whose roots began with *pa- and whose AP forms began with *ma-, derived historically from *-um- + *pa-, as *maCay, from *paCay. Many of these verbs are complex roots formed with the prefix *pa- ‘causative’, (4) verbs similar to those in (3), but whose root began with *ka-, and whose AP forms began with *ma-, derived historically from *-um- + *ka-. Many of these verbs are complex roots formed with the prefix *ka- (perhaps ‘inchoative’).”
Perhaps the most noteworthy claim made here is that the paradigmatic relationship of *ma- with *ka- ‘stative’, observed in many of the An languages of Taiwan, the Philippines, and western Indonesia, results from a process of infixing *ka- “perhaps inchoative” with *-um- ‘Actor Focus’ and subsequent loss of the initial CV-: *k-um-a- > ma- ‘stative’. Despite this attempt to establish a derivational link between *ka- and the stative prefix *ma-, Ross does not comment on the morphology of the causative prefix *paka-…
Starosta (1995) and Ross (1995) both suggest that *ma- ‘stative’ is derived from *ka ‘inchoative’ by affixation (*m-ka and *k-um-a respectively) but neither of these papers mentions *ka- ‘stative’, and in neither of them is the relationship between *pa- and *paka- spelled out clearly…as suggested by Ross (1995) and demonstrated by Himmelmann and Wolff (1999), *ma- ‘stative’ probably occurred in this form in PAn, but was a surface realization of *ka- ‘stative’ + *-um- (*k-um-a- > *ma-).
Blust added in a note that ka› is not covered anymore by his earlier PNS:
Note, however, that the loss of the initial syllable in *k-um-a- > *ma- ‘stative’ and other infixed stems that do not begin with a labial consonant cannot be motivated by any known phonotactic constraint.
The consonants that are involved are p, b, t, k, and glottal stop (ч). This process would therefore be different from the PNS that Blust described, or even operating on top of it.
From the above, we can say that subtraction (apheresis) is active in Philippine languages. This is on top of having other methods of forming words (morphological process) like affixation (prefixation, suffixation, infixation), reduplication, modification (length), compounding and possibly conversion.
This actually opens the affixation system to a lot of avenues for research:
- Can we present a more detailed description that supports derivation of the below affixes?
|Affix Surface Form||Derivation|
|ma›||ka›+ ‹um› → k‹um›a› → (ku)ma›|
|ma›||pa›+ ‹um› → p‹um›a› → (pu)ma›|
|mag›||pag›+ ‹um› → p‹um›ag› → (pu)mag›|
|mang›||pang›+ ‹um› → p‹um›ang› → (pu)mang›|
- Can all ma› affixes be shown to be a complex of affixes with ‹um› as a component? This would reduce AF affixes to just one affix.
- Is the nominal affix mag› derived from pag›, or is it from kag›?
- Are adjectives actually verbs? This is because Tagalog, Bikol and Bisayan adjectives are marked with ma›, so could be ka› bases with ‹um› infix, and Ilokano adjectives have na›, so could be originally with ‹um›‹in› affixes (or just ‹in›). Is this the reason why adjectives have reduplication, as reduplication is supposed to be a verbal marker (aspect)?
- Do manga, the plural marker, and may, the existential/possessive marker, have an ‹um› affix? This would make nouns also verbs and would show stacking of ‹um›. And what is its connection with Ilokano plural nouns that has reduplication, as reduplication is supposed to be a verbal marker (aspect)?
I would like to work on these in the coming days.
Note to myself: I need to find these works:
- Ross, Malcolm D. 1995. Proto-Austronesian verbal morphology: Evidence from Taiwan. In Austronesian studies relating to Taiwan, ed. by Paul Jen-kuei Li, Cheng-we Tsang, Dah-an Ho, Ying-kuei Huang, and Chiu-yu Tseng, 727–791. Symposium Series of the Institute of History and Philology, no. 3. Taipei: Academia Sinica.
- Starosta, Stanley. 1995. A grammatical subgrouping of Formosan languages. In Austronesian studies relating to Taiwan, ed. by Paul Jen-kuei Li, Cheng-we Tsang, Dahan Ho, Ying-kuei Huang, and Chiu-yu Tseng, 683–726. Symposium Series of the Institute of History and Philology, no. 3. Taipei: Academia Sinica.
- Svelmoe, Gordon, and Thelma Svelmoe. 1974. Notes on Mansaka grammar. Language Data, Asian-Paci²c Series No. 6. Huntington Beach, California: Summer Institute of Linguistics.
- Wolff, John U. 1973. Verbal in³ection in Proto-Austronesian. In Parangal kay Cecilio Lopez: Essays in honor of Cecilio Lopez on his seventy-²fth birthday, ed. by Andrew B. Gonzalez, 71–91. Philippine Journal of Linguistics Special Monograph Issue No. 4. Quezon City: Linguistic Society of the Philippines.
- Blust, Robert. 1999a. Subgrouping, circularity and extinction: Some issues in Austronesian comparative linguistics. In Selected Papers from the Eighth International Conference on Austronesian Linguistics, ed. by Elizabeth Zeitoun and Paul Jen-kuei Li,
476 oceanic linguistics, vol. 42, no. 2 31–94. Symposium Series of the Institute of Linguistics (Preparatory Of²ce), Academia Sinica, No. 1.
- Blust, Robert . 1999b. Notes on Pazeh phonology and morphology. Oceanic Linguistics 38:321–365.
- Huang, Lillian M. 2000. Verb classification in Mayrinax Atayal. Oceanic Linguistics 39:364–390.
- Zeitoun, Elizabeth, and Lillian M. Huang. 2000. Concerning ka-, an overlooked marker of verbal derivation in Formosan languages. Oceanic Linguistics 39:391–414.
- Robert Blust, Elizabeth Nielsen Avoidance of Dissimilar Labial Onsets: The Case of Subanon