Irregular Verb Forms for Patient Focus (<ən) and Actor Focus (<um>)

First of all, lets define the aspects for Central Philippine languages as we’ll be using these terms down below:

1. Imperfective are actions that are viewed internally as ongoing.
2. Perfective are actions that are viewed externally as a whole unit, irrespective of whether its ongoing or not.

NAF <EN PATIENT AFFIX

I have always been struck by the irregularity of the conjugation of the <ən as the Begun forms do not show the infix <in>, unlike the other NAF affixes i>, <an, and the AF affix <um>. Fortunately, I stumbled upon Sagart’s “PAN MORPHOLOGY IN PHYLOGENETIC PERSPECTIVE” that mentions 4 languages still retaining the Austronesian original patient focus forms. I quote: “Yet in the An world the perfective aspect marker (PERF) *<in> and patient focus marker (PF) *-en are found attached to the same verb stem only in four West coast Formosan languages: Saisiat (-ən or -in), Pazeh (-en), Thao (-in) and Siraya (-ən). Elsewhere such forms are not found.”  And he gave a Saisiat examples from Zeitoun et al. 1996, like:

hiza	ʔalaw	maʔan	minayʔangsow	s<in>iʔael-ən
that	fish	1S.Gen	left	<PRF>-eat-PF
That fish is what I have left of my eating

If this is retained in Bikol, the verb kakan ‘eat’ should have forms like the following:

Focus Affix	Base+Focus Affix	Base+ Imperfective(CA> Redup)+Focus Affix	Base+Begun+ Focus Affix	Base+Begun+ Imperfective(CA>Redup)+Focus Affix
<um>	kumakan	*kuma:kakan > ma:kakan	kuminakan	*kumina:kakan  > mina:kakan
<ən	kakanon	ka:kakanon	*kinakanon > (kinakan)	*kina:kakanon > (kina:kakan)
i>	ikakan	ika:kakan	ikinakan	ikina:kakan
<an	kakanan	ka:kakanan	kinakanan	kina:kakanan

Or for the verb lu:toɥ ‘cook’:

Focus Affix	Base+Focus  Affix	Base+ Imperfective(CA> Redup)+Focus Affix	Base+Begun+ Focus Affix	Base+Begun+ Imperfective(CA> Redup)+Focus Affix
<um>	lumu:toɥ	*lumulu:toɥ > ma:lu:toɥ	luminu:toɥ	*luminulu:toɥ  > mina:lu:toɥ
<ən	luto:ɥon	luluto:ɥon	*linuto:ɥon > (linu:toɥ)	*linuluto:ɥon > (linulu:toɥ)
i>	ilu:toɥ	ilulu:toɥ	ilinu:toɥ	ilinulu:toɥ
<an	luto:ɥan	luluto:ɥan	linuto:ɥan	linuluto:ɥan

The bold forms kinakanon / linuto:ɥon and kina:kakanon / linuluto:ɥon should have been the forms if the forms were not simplified, as Sagart explained further as the reason for dropping the <ən. The fact that it still exists in those 4 languages is a pretty good evidence why the conjugation should be brought back in the Filipino language for symmetry’s sake.

AF <UM> AFFIX

The other Bikol AF affix with irregular forms is <um> , which should have the bold forms kuma:kakan / lumulu:toɥ and kumina:kakan / luminulu:toɥ as indicated above. So how did we arrived at this reconstructed forms for the <um> infix?

Let’s first compare some Philippine-type languages’ conjugation of this affix based on aspect“:

		Bikol	Tagalog	Old Tagalog	Cebuano	Waray	Palawano	Agutaynun
Mood/Aspect	base	kakan	bili	gawa	palit	sakay	surong	kuran
Unbegun Perfective	<um>+base	kumakan	bumili	gumawa	palit	nasakay	sumurong	kumuran
Unbegun Imperfective	<um>+<RDP>+base	ma:kakan	bi:bili	ga:gawa	mupalit	masakay	susurong	kumuran
Begun Perfective	<um>+<in>+base	kuminakan	bumili	gungmawa	mipalit/nipalit	sinmakay	suminurong	kiminuran
Begun Imperfective	<um>+<in>+<RDP>+base	mina:kakan	bumi:bili	gungma:gawa	mipalit/nipalit	nasakay	sumusurong	kukuran

As can be seen from the table, Begun is signalled through the <in> infix, Aspect by reduplication. <RDP> means the prefixation of the reduplicated CV: of the word base, with a long vowel in the reduplicated CV. Note that in Waray and Hiligaynon (not shown), the order of the infixes is <in>+<um> rather than <um>+<in> as in Bikol. Also note that the Bisayan languages has dropped the <RDP>. The Old Tagalog form <uŋm> was originally <umin>, then through vowel loss (elision) become <umn> and by dissimilation becomes <uŋm>. As for the Unbegun Perfective, I quote Guglielmo Cinque here which says that “As Will (1989) notes, “the perfective aspect is possible only with verbs in the past or in the imperative mood” (p.142).” on footnote 36.

From the illustration above, we can see that Bikol has irregular forms for Unbegun Imperfective and Begun Imperfective, or in other words in the Imperfective aspect . Tagalog has irregular forms in Unbegun Imperfective, Begun Perfective and Begun Imperfective, and  the Begun Imperfective form should have been the Unbegun Imperfective form. The old Tagalog forms are only irregular in the Unbegun Imperfective, and the irregularity in the Begun forms in current Tagalog was brought by the simplification of the <ungm> infix. Cebuano has all forms irregular. Waray and Palawano  are regular only in Begun Perfective. Palawano’s Begun Imperfective should have been the Unbegun Imperfective form. The right form for the Unbegun Imperfective is present in current Tagalog and Palawano, but in both languages these forms function as Begun Imperfectives.

Although Ilokano tense/aspect does not correspond with the Central Philippine languages tense/aspect semantically, it has regular forms and with an additional Uncompleted Perfective aspect  which is not regularly formed with an <um>. Check out the table below:

		Ilokano	Pangasinan	Kapampangan
Aspect	base	gatang	inom
Neutral (Infinitive/Imperative)	<um>+base	gumatang
Uncompleted Perfective		(?)gatangto	oninom	muran
Uncompleted Imperfective	<um>+<RDP>+base	gumatgatang	oniinom	mumuran
Completed Perfective	<in>+<um>+base	gimmatang	inoninom	minuran
Completed Imperfective	<in>+<um>+<RDP>+base	gimmatgatang

I will fill the slots for Kapampangan and Pangasinan later. Note that Ilokano’s <RDP> is only CV:- if the base word has a medial glottal stop and regardless of stress patter of the base word, otherwise it is CVC- (according to Reid), and the order of affixation between <um> and <in> is reversed. What I find in the Ilokano forms is that the Uncompleted Imperfective and the Completed Imperfective forms are regularly formed, corresponding to Unbegun Imperfective and Begun Imperfective, respectively in Central Philippine languages.

Which Aspect Forms Are More Original?

One thing that we can notice is that in the Ilokano system which we call here as General Philippine system, the infix <in> is not added to the Imperfective, only to the Perfective. Also, the infix <in> meaning is “Completed”, unlike “Begun” in the Central Philippine languages (Tagalog, Bikol, Bisayan, Mansakan, etc.) . Laurence Reid in his “ON THE DEVELOPMENT OF THE ASPECT SYSTEM IN SOME PHILIPPINE LANGUAGES” compared the the aspect system between General Philippine and Central Philippine systems and concluded that they have “little systematic correspondence”  and that  the original aspect and voice system in the General Philippine system was the more original for these reasons:

1. It is found in the other different groups, apart from Ilokano, like Bashiic, Cordilleran, Sambalic, Manobo and Danao languages.
2. Some Bisayan languages exhibit a General Philippine system in restricted parts of the verb (Potential Mode) and not in the “General Mode” which has the innovations.

Data from Formosan languages seems to agree with him, as another author, Dorinda Tsai-hsiu Liu, who wrote “Tense and Aspect System in Kanakanavu” said: “The tense and aspect system in Formosan languages is generally understood to demonstrate a two-way contrast of future vs. non-future (also known as realis vs. irrealis) (see Ogawa & Asai 1935; Tsuchida 1976; Huang 1995; Zeitoun et al. 1996; among many others). For instance, a verb affixed with an AT or non-actor trigger (NAT) marking (the so-called ―neutral‖ form) is mostly indicative of a past or present action/situation in Atayal, Bunun, Paiwan, Puyuma and Saisiyat when no tense and/or aspect marker is present in the clause (Zeitoun et al. 1996:24).   Meanwhile, the future tense is normally lexically or morphologically marked in these languages.”  In the same way, the Ilokano system has a separate future marker. Also according to Reid, <in> marks the completive aspect in Proto-Extra-Formosan.

Other Issues discussed by Reid

Reid also discussed what is the original order between <um> and <in>: whether aspect affixation first before voice affixation ( and would result in ɥi-C<in>V / C<um><in>V- forms ) or voice affixation first before aspect affixation (which would result in ɥ<in>iCV / C<in><um >V- forms ) and concluded that the original order  is <um><in>  (like the Bikol form) rather than <in><um> based on the following in spite of languages with different patterns in the same subgroups in the different groups:

1. languages that have ɥ<in>iCV / C<in><um >V- have frozen forms with ɥi-C<in>V / C<um><in>V-, (no examples given).
2. languages that have ɥ<in>iCV / C<in><um >V-  have ɥi-C<in>V / C<um><in>V- in restricted forms, like Kalinga in the <um>, Balangaw for <um> with non-syncopated initial base word vowel and Sarangani Manobo for verbs with initial labial or glottal stops.
3. languages where the original forms of nag> is minag>, like Northern and Southern Alta, and Casiguran Dumagat.

I don’t think the last evidence that Reid supplied is acceptable, because this implies that voice was affixed first, before the aspect. If its the other way around, the form would have been  magC<in>V- and not m<in>ag>CV-. Although if the source of mag> is  pag><um>, then he might be right: Pag> pinag> puminag> minag. He quotes in footnote 10: “De Guzman (1978 : 150) analyzes the mag- prefix as a combination of a derivational prefix pag- plus an m- inflectional prefix, an analysis which neatly captures the historical development of the prefix. Wolff (1973: 74, 84) also notes the similarity in form and function of (um) and mag- and considers mag- to be the surface form of a deep structure pag- + (um).” In Footnote 26, I also quote: “From an earlier *p(um) (in)a- sequence. In Proto-Extra Formosan (if not in Proto-Austronesian), when (um) was infixed into words with initial bilabial stops, the first two segments of the infixed word were deleted. This rule is still present in languages such as Tagbanwa, Palawano, some Manobo languages, and Blaan (see also Wolff 1973: 84). Frozen forms such as matay die (from p(um)atay) occur in various other languages, such as Tagalog.”

Reid also discussed which of the <RDP> form is the more original, concluding that it’s the CVC- (like the Ilokano form) rather than the CV:- since:

1. It is necessary to reconstruct CV- reduplication to mark noun plurality, as this is reflected widely in the family.
2. The Cordilleran languages, like Ilokano, distinguish between the form CV- for noun plurality and CVC- for continuative verb forms.
3. It is probable that the CV:- form in Central Philippines arose out of similar constraint of disallowing  –ɥC- in Cordilleran languages, where in the Central Philippine languages, the loss of the glottal stop –ɥ– resulted in vowel lengthening to finally have CV:- and generalization of this form to all consonant position.

I partially do not agree with the last reason as:

1. Not all Central Philippine languages disallow –ɥC- implying that their parent language does not disallow –ɥC- either. Example is Bikol which has such words, for example with –ɥC- : (1) yaɥpit “narrowness”, (2) baɥlak “split”  (3) buɥbuɥ “let the contents out” and (4) kiɥlay “walk with a limp”.
2. In Tagalog since –ɥC- is disallowed, the –ɥ– should not be there in the base word, so nothing to disallow. Example are : “new” : Tagalog /ba:go/, Bikol /baɥgo/,
3. In other languages like Bisayan languages, there is metathesis, so –ɥC- becomes -Cɥ– , so also no –ɥ– to disallow. Example: “new” : Cebuano /bagɥo/

But I do agree that the vowel length signals that some other consonant was removed from there and generalized, maybe for ease of pronunciation, as some Central Philippine languages has compensatory vowel lengthening when a consonant is deleted. Tagalog for example, shows vowel length where there is –l- in Bikol. Example: “sun” : Tagalog /ɥa:raw/, Bikol /ɥaldaw/. Also as Reid also mentioned, Bontok generalized the CVC form to have geminate consonants for those with medial glottal stop, glides and liquids. Overall, I would have to agree that the original form would have been CVC- with –ɥC-allowed.

The discussion above indicate the possibility of the conjugation in the proto-language as follows (using the Bikol word base but not implying that its the proto-form):

Mood/Aspect	base	kakan	kakan	luto:ɥ	luto:ɥ
Irrealis Perfective (Infinitive/Imperative)	<um>+base	kumakan	magkakan	lumuto:ɥ	magluto:ɥ
Irrealis Imperfective	<um>+<RDP>+base	kumakkakan	magkakkakan	lumutluto:ɥ	maglutluto:ɥ
Realis Perfective	<um>+<in>+base	kuminakan	minagkakan	luminuto:ɥ	minagluto:ɥ
Realis Imperfective	<um>+<in>+<RDP>+base	kuminakkakan	minagkakkakan	luminutluto:ɥ	minaglutluto:ɥ

The above forms easily explains why the <RDP> is CV:-: Central Philippine languages does not allow geminate consonants and possibly that’s the underlying cause of the simplification to CV:-. As a matter of fact, some of Pangasinan’s vowel length is derived from loss of geminate consonant, as described by Zorc in “On the Development of Contrastive  Word Accent: Pangasinan, a Case in Point”.

Other Referrences:

1. Daniel Kaufman, “Interpreting the geography of TAM marking across Indonesia”.

2. Henry Chang, “Rethinking the Tsouic Subgroup Hypothesis: A Morphosyntactic Perspective”

3. JONATHON ERIC CIHLAR, “Compositional Interaction of Sub-Event Aspectual Markers, –in- and Reduplication, in Tagalog“