Luna Qian
Student number: 1009179817
LINC02
Feb 8, 2026


The /m/ Infix in Four Western Austronesian
Languages

Introduction
This paper examines the /m/ infix in four Western Austronesian languages. In all four
languages, the infix is inserted after the first consonant of the base. At first glance, this looks
straightforward. But once we compare the surface forms more closely, we notice that the
languages react differently to the same infix. Some delete segments, some assimilate, and
some insert vowels. So clearly the infix itself is not the whole story.
In this paper, I’ll use Optimality Theory to explore why these differences emerge. Rather than
assuming each language just has its own separate rules, OT gives a way to compare them by
looking at how the constraints are ranked. When I put the four languages side by side, I started
to notice that they are dealing with very similar problems. The infix is inserted in the same
position in each language, but the outcomes are not the same. This made me think that the
difference probably does not come from the infix itself, but from how each language ranks its
constraints.
Sarangani Blaan
In Sarangani Blaan, the infix normally surfaces without much complication:
● tiis → tmiis

● saloʔ → smaloʔ

● dado → dmado

Clusters like /tm/ and /sm/ are allowed. However, in the form:
● bunal → munal

We do not get “bmunal”. Instead, the initial /b/ deletes. At first I thought maybe the language just
avoids clusters in general. But that does not really work, because other clusters like /tm/ and /sm/ are
clearly allowed. So it cannot simply be about clusters as such. There has to be something more specific
going on here. So the issue must be more specific.The problematic cluster would be “bm”, which
contains two labial consonants. It seems reasonable that the language avoids adjacent consonants with
the same place of articulation. A constraint like *LAB-LAB (or OCP-Place) could capture this.If
*LAB-LAB outranks MAX-IO, deletion will occur.
Ranking: *LAB-LAB >> MAX-IO

/bmunal/ *LAB-LAB MAX-IO
☞ munal *
bmunal *!

The faithful candidate fatally violates the markedness constraint. Even though deletion violates
MAX-IO, it is tolerated. It is also worth considering assimilation as an alternative. For example,
/bmunal/ could hypothetically become *mmunal. But this would still involve adjacent labials and
would not actually solve the markedness problem.

/bmunal/ *LAB-LAB MAX-IO
mmunal *! *
☞munal *

I also wondered whether assimilation could work—say, turning */bmunal/ into something like
*mmunal. But that would still leave two labials side by side, which doesn’t really fix the
problem. So in the end, deletion turns out to be the only repair that actually works here.

Timugon Murut
In Timugon Murut, the infix also appears after the first consonant:
● tuun → tumuun

● gajo → gumajo

But in:
● patoj → matoj

● bigod → migod

Here, instead of deletion, the nasal changes to match the following consonant. This looks like place
assimilation. Compared to Sarangani Blaan, this language does not delete segments. Instead, it
modifies features. This suggests a different ranking.
If the language prefers adjacent consonants to agree in place, then:AGREE-PLACE >> IDENT-PLACE

/pmatoj/ AGREE-PLACE IDENT-PLACE
☞ matoj *
pmatoj *!

The faithful candidate violates agreement. The assimilated form violates IDENT-PLACE but
wins. What’s interesting here is how Timugon Murut chooses feature change over deletion.
Compared to Sarangani Blaan, which just removes the segment, this language tweaks the
features instead—almost like it’s trying to preserve material while still avoiding a bad cluster.
Kulalao Piiwan
Kulalao Piiwan initially looks similar:
● tulək → tmulək

● kan → kman

But in:
● pili → pnili

● burəs → bnurəs

The clusters appear somewhat unstable. It is not entirely clear whether the key issue is place agreement
or cluster complexity. At first I considered whether assimilation might explain it, but the data do not
consistently show clear place matching. Instead, the cluster structure itself seems marked.
So a constraint like *COMPLEXONSET may be relevant here. If *COMPLEXONSET outranks
DEP-IO, the language will avoid certain complex onsets even if that means altering the output.
Ranking: *COMPLEXONSET >> DEP-IO

/pmili/ *COMPLEX DEP-IO
☞ pnili *
pmili *!

Deletion is another logical possibility, so we should consider it.
/pmili/ *COMPLEX MAX-IO
☞ pnili *
p ili *!

Deletion would violate MAX-IO. Since deletion does not occur, MAX-IO must outrank the deletion
option. I’m not fully certain whether place features alone explain these forms. It is still not completely
clear to me whether the main issue here is place agreement or just general cluster complexity. The
patterns seem to point toward syllable structure being important, but I am not entirely certain that this
captures everything in the data.
Tjuabar Paiwan
In Tjuabar Paiwan, we see vowel insertion:
● təkəɭ → təməkəɭ

● pajsu → pənajsu

Here, the language takes a different route: it inserts a vowel. That means it’d rather add
something new than delete or alter what’s already there—which tells me something about how
it prioritizes syllable structure. This suggests that ONSET (or related syllable well-formedness
constraints) outrank DEP-IO.
Ranking: ONSET >> DEP-IO

/tməkəɭ/ ONSET DEP-IO
☞ təməkəɭ *
tməkəɭ *!

Insertion violates DEP-IO but avoids a more serious structural violation.

Discussion
All four languages insert the same /m/ infix in the same position. But the repair strategies differ:

Language High Constrain Strategy
Sarangani Blaan *LAB-LAB Deletion
Timugon Murut AGREE-PLACE Assimilation
Kulalao Piiwan *COMPLEX Cluster repair
Tjuabar Paiwan ONSET Insertion

Looking across the four systems, it seems that markedness constraints are generally ranked above
faithfulness in some way. But the exact constraint that drives the repair is not the same in each
language. That is probably why we see deletion in one case, assimilation in another, and vowel
insertion elsewhere. At first these patterns look unrelated, but once the rankings are compared more
carefully, they do not seem random. I initially expected the four languages to behave more similarly,
but the data suggest that the interaction between markedness and faithfulness can produce quite
different outcomes.

Conclusion
The /m/ infix behaves differently across the four languages because each language ranks constraints
differently. Overall, this comparison shows that even small changes in constraint ranking can affect the
surface forms quite noticeably. Even though the /m/ infix is placed in the same position, each language
repairs the resulting structure differently. There may be additional constraints involved that are not
fully explored here, but the ranking differences seem to account for the main patterns in the data.

References
McCarthy, J. J. (2008). Doing optimality theory: Applying theory to data. Blackwell Publishing.
Prince, A., & Smolensky, P. (1993/2004). Optimality theory: Constraint interaction in
generative grammar. Blackwell.
Zec, D. (2007). The syllable. In P. de Lacy (Ed.), The Cambridge handbook of phonology (pp.
161–194). Cambridge University Press.

学霸联盟