LINGUIST List 23.1266|
Tue Mar 13 2012
Sum: Suppletion Cross-linguistically Responses
Editor for this issue: Zac Smith
To post to LINGUIST, use our convenient web form at http://linguistlist.org/LL/posttolinguist.cfm.
From: Piers Kelly <Piers.Kellyanu.edu.au>
Subject: Suppletion Cross-linguistically Responses
E-mail this message to a friend
Query for this summary posted in LINGUIST Issue:
Many thanks for taking the time to respond to my query on suppletion.
This email is my attempt to summarise the responses. Some quick
My dissertation is on an auxiliary language of the Philippines called
Eskayan which was created, according to folklore, by a pre-Hispanic
ancestor and rediscovered in the early 20th century by a Messianic
rebel soldier. On the face of it, Eskayan is a fairly straightforward
relexification of Cebuano, a language spoken widely in central and
southern Philippines. So, in other words, for every lexeme in Cebuano
there is a corresponding lexeme encoded in Eskayan. This goes for
verbal affixes and nominalising morphology too.
But this is an oversimplification. It turns out that many of the Eskayan
verbs and nominalisations show what looks like suppletion. In effect,
these are words that are relexifications of Cebuano terms but with
Eg, the Cebuano realis perfective affix mi- is encoded in Eskayan as
chdin-. But in some realis-perfective Eskayan forms, this morphology
cannot be detected or analysed – it is simply understood by
convention. Eg, the Eskayan word bintaal corresponds to Visayan
migamit (‘used’). No part of the word bintaal can shown to be doing the
inflection. This happens all over the place and not just in common
Eskayan is mostly a written language and has its own traditional
literature which I’ve been using as a corpus. I wanted to get a sense of
how suppletive Eskayan really was in relation to other languages. I am
making the argument that Eskayan is a reflection of what it's creator
understood languages to be like. It looks like he wanted to bring in
suppletion to represent irregularity as a descriptive fact about (his)
Summary of responses:
• Anie Thompson suggested looking at Corbett 2007
(http://epubs.surrey.ac.uk/1313/) who discusses how to quantify
suppletions, treating it as a sliding scale of canonicity.
• Several of you suggested looking at the work of Llubja
Veselinova, including her PhD dissertation which later appeared as
Veselinova, L (2006) Suppletion in verb paradigms. John Benjamins. ;
and to: Veselinova, L. (2011). Suppletion According to Tense and
Aspect. In: Dryer, Matthew S. & Haspelmath, Martin (eds.) The World
Atlas of Language Structures Online. Munich: Max Planck Digital
Library, capítulo 79, Map 79; Veselinova, L (2011) Verbal Number and
Suppletion. In: Dryer, Matthew S. & Haspelmath, Martin (eds.) The
World Atlas of Language Structures Online. Munich: Max Planck Digital
Library, Map 80
• Calle Borstell pointed me towards these two theses: http://su.diva-
searchId=1&pid=diva2:373138, and the suppletion database from the
Surrey morphology group: http://www.smg.surrey.ac.uk/
• Minna Persson pointed me to this: http://wals.info/chapter/21
• Daniel Ross, a grad student from the University of Illinois had
some feedback on the nature of suppletion generally (reproduced
• Many of you provided examples of languages with lots of
suppletion and I’ve included these below for reference.
Tlingit has suppletion of verb roots that has to be lexically
specified. In verbs of motion there is a distinction between the
movement of a single entity and the movement of plural entities
represented by different verb roots. This is despite the existence of
singular versus plural marking in the agreement prefixes of the verb,
which occurs with all verbs. So:
‘I went there’
‘we went there’
This suppletion also interacts with the noun classification system
which has its own system of suppletion. In Tlingit, like in other
Na-Dene languages, noun classes are expressed through the use of
different verbs of handling, which is partly done with verb root
suppletion and partly through different ‘qualifier’ prefixes. So a
verb of handling has one form for a small round object, another form
for a flat flexible object, and so forth. These different categories
are partly expressed through suppletion of the verb root, and some of
this suppletion is due to whether the verb describes a single entity
or plural entities.
I don't know if it helps you, there are a good couple of languages with
verbs that show suppletion (or stem-alternation) according to the
number (sing - plur) of the most affected participant (the patient). I
found this phenomenon in the next languages: Georgian, Yurok,
Southern Paiute, Ainu,Sandawe, Koasati and Sumerian. I can also give
you the list of literature I used in collecting my examples, if you need it.
Hope it helps!
I am writing about your query to the Linguist List. Irish is a language
with much suppletion. You will find many interesting details by looking
at the dissertation detailed at 1061-3.htm>.
the Tibetan languages have a 'standard' suppletion of stems in the
case of the word go
(I = the so-called 'present' or 'imperfective' stem, II: the so-called 'past'
or 'perfective' stem, IV: the so-called 'imperative' stem; none of these
labels really fit, but that should not be of much concern for you):
Classical Tibetan: I: 'gro, II: song, IV: song
Lhasa (written equivalent) I: 'gro, II: phyin, IV: song
Ladakhi I: cha, II: song; IV: song
In most varieties, this is the only suppletive form.
Classical Tibetan has about 1450 verbs (doublets with spelling
alternations not yet sorted out), the modern varieties usually maximal
800 to 900 verbs.
Some varieties also have a suppletive stem IV for the verb 'come'
I: yong/'ong, II: yong(s)/'ong(s), IV: shog
some Amdo Tibetan varietieshave one or two more such verbs
(please have a look at Roland Bielmeier 2004: “Shafer’s proto-West
Bodish hypothesis and the formation of the Tibetan verb paradigms”.
In: Anju Saxena (ed.), Himalayan Languages. Past and present.
(Trends in Linguistics, 149.) Berlin: Mouton: 395-412.
A Valency Dictionary of Ladakhi Verbs
You probably already have loads of responses already, but I thought
I would mention that the Yuman language family of California is famous
its extensive use of suppletive verb forms -- I've often wondered how
learned the languages for so long, but now there is no way to observe,
all of the languages are moribund.
Navajo/Apache also has a number of different verb bases for use in
aspects -- this is a characteristic of the Athabaskan family generally,
surface phonological changes make the differences appear more
extreme than they
University of Arizona
I'm a graduate student at the University of Illinois with an interest in
That's an interesting question. I think it borders on discussions of
complexity in general, or at least has some of the same methodological
concerns: is it really practical to count how many such forms a
Here's some info on an upcoming conference (including two relevant
reference works) that might be of interest to you:
(I'll be there, but probably not presenting, as my paper was selected as
In my personal opinion, ''suppletion'' is a misused term (not by you, but
in general). It basically means ''it's weird, and we don't know''.
Suppletionoften refers to several types, as long as each type is itself
unusual or unexplained by normal means. And it's used differently by
For example, one analysis (presented in one of my linguistics classes,
''Typology'' actually) claimed a case of suppletion in Arabic. Arabic
nouns and adjectives take a final /-a/ suffix in the feminine form, at
least they usually do. There are exceptions that are feminine without it,
and some masculine words with it. It's probably something like 99%
reliable, though, and it's entirely productive.
The complexity enters in that this /-a/ magically becomes /-at-/ when
any material is added after it. To skip the details, I'll just say that this
can be based on register (very formal Arabic pronounces this more
often) and is obligatory when the noun/adjective is followed by certain
case markers (which also are only used in a high register), and a
possessive ending (which is not necessarily formal) also makes it
So to make this explicit:
sadiiq = friend.MASC
sadiiq-a = friend.FEM
sadiiq-at-ii = friend.FEM.1Spossessive - 'my [female] friend'
So where did the /t/ come from? It certainly didn't just appear randomly.
Other words ending in /-a/ don't get magical /t/ epenthesis. And it can't
be the underlying form either, because other words ending in /-at/ don't
lose the /t/ in most cases.
This mystery lead some researchers to the conclusion that it must be
''suppletion'', or basically that they don't know what's going on.
My argument for it is that there is a feature on certain words (almost
always overlaps with the [feminine] feature) for this. There's really no
other way around it. You could alternatively claim some sort of
complicated morphology where this particular morpheme has its own
phonological rules, I suppose. Either way, suppletion was a lot less
explanatory than a real answer, whatever that may be (or however
hard it may be to figure out).
In the case of canonical suppletion like go/went, I think you'll probably
find more about that just calling them ''irregular verbs''.
But then comes the next problem: how are you defining (that is,
limiting) suppletion? Are all irregular verbs suppletive? ride=>rode,
As for answering your question about specific languages, I've studied
about 12 languages (not that I'm fluent in them) and each has some
suppletive forms, or just ''irregular verbs''. They're called different
things, like ''stem changing verbs'' (Spanish) or ''strong verbs''
(German), etc. Something that is very interesting, though, is that these
aren't just random-- a lot of the forms can be somewhat predicted.
Consider the English verbs quit/hit or rise/ride (which have parallel past
tense forms), or perhaps even buy/bring => bought/brought. Which of
these is ''suppletive'' or ''irregular''?
An old idea I've always found intriguing is from Pānini (if you want a
citation I can find one, but it's fairly well known), stating basically that
there are no exceptions. Instead, there are just many rules, some of
which apply more specifically than others. So the past tense for all
English verbs is -ed, but for a few there's another rule of changing the
stem vowel to -o- from -i-, or whatever it may be. And we get a few odd
cases of lexical items that just have their own rule, eat=>ate. Looking at
it that way, suppletion is even less clear: is it just the times when a rule
isn't universal in a language? Or when it's particularly infrequent?
Getting back to the languages I mentioned, here are some thoughts:
All languages I've seen have irregular verbs in one way or another. In
Indo-European languages they often come from two original systems,
one of vowel change (umlaut, etc) and another with a suffix. These
both survived in the Germanic languages at least, and to some minor
extent (just echoes) in Latin. Other irregularities come from innovations
and sound changes, such as in Spanish when only an unstressed
vowel become a diphthong, so the infinitive remained as it was, but the
conjugated forms changed, such as ''pod-er'' (can.INF) and ''pued-o''
Basque has layers of regularity (classes of verbs that work differently)
and then the most common verbs (at least) that are irregular.
Arabic has few irregular verbs in some sense because it follows a
pretty strict root-and-pattern system, but not everything is predictable
(such as what vowels may fill the patterns), and there are probably
some exceptions, not that I can cite them at the moment.
Japanese is pretty regular, but a couple verbs have unpredictable
Swahili is incredibly regular, with almost no exceptions. But at least the
verb ''to be'' in the present tense is randomly a different form, ''ni''
which doesn't vary by person or number, and none of the normal
affixes attach to it. And several other verbs have their own unique
properties, although I'd submit Swahili as an example of a language
with almost no irregularity outside the most common verbs. (Lots of
affixes to remember, but mostly regular usage.)
There is a definite trend that irregular patterns are in common verbs.
So if you look at ''to be'' across languages, you're almost guaranteed
that in at least 9/10 it will be irregular. In fact, I don't know of a single
case where there isn't something odd with that verb. Swedish comes
close to regularity, with är in the present tense for all persons and
numbers (verbs don't agree in Swedish), but that's irrelevant when you
consider the infinitive 'vara', etc.
In fact, aside from analogy (eg, ''copywrote''?) infrequent words should
never be irregular (at least not for long)-- it just won't be preserved.
And that's only verbs. What about plurals, or cases for nouns? There
are other topics too.
But to attempt an answer to your question, I think English is probably
going to be a ''very irregular'' language, at least when compared to
most languages in the world (that it will be somewhat near the top).
The reason is simple: English is in an odd place between logical
morphology and isolating words. You probably won't find much
irregularity in an agglutinative language because it can't afford the
extra uncertainty-- with that many affixes, everything needs to be
regular or it won't be retained (and this is more true for polysynthetic
languages). And in the case of an isolating language (Chinese for
example) irregularity doesn't seem to hold much meaning, since most
words don't change shape. I suppose you could have a language that
has no morphology aside from suppletion, but I don't know of anything
quite like that. Right in the middle we get inflectional or fusional
languages like English, Latin, etc., but specifically at the English end
(toward isolating, away from agglutination) is where you're likely to find
a mess of morphology... again, just my instincts on that one (but true as
far as I know).
If you're interested in finding the rules within the chaos (or under it),
then comparing to work done on German would be a good idea (there
has been a lot done) because that can show you how to extract
patterns from a seemingly random system. There isn't anything that
thorough for English, at least nothing I've seen.
For historical data, nothing will be better than the development of the
Romance languages from Vulgar Latin. A lot has been written about
Spanish (among others), and I'm actually taking a class about this now.
We can trace all of the words back to Vulgar Latin (and many to
Classical Latin) and figure out why there are now irregular forms.
Analyses of other languages exist, but the historical records are just
much better for these languages than others. (Maybe if you wanted to
look at Sanskrit, or Greek you could find similarly detailed descriptions.)
Again, I think you're asking an interesting question. I hope these (long)
thoughts help you some. Let me know if you have other thoughts on it
or some examples of the data-- I'd be happy to try to think of a similar
case in another language that I've seen.
My response was quite long, but you're welcome to share it. One
interesting point that I thought of since I last emailed was that there is
partialsuppletion (unexplained morphemes, spelling changes, etc.), and
also complete suppletion, such as go->went. On that topic, I think you
might find the verb ''to go'' interesting in the Romance languages. In
each, there is a weird mix of at least two verbs, usually an infinitive like
''ir'' (Spanish) or ''andare'' (Italian) and conjugated forms like
''voy''/''vado'', that appear to come from some earlier form *vadire, but I
really have no idea what that is or what is meant, if it existed (although
it did come from somewhere, surely). I don't know of too many verbs
that have completely irregular forms, though.
Have you found mostly irregularities within the mostly regular system or
have you found some completely irregular forms? In that case, from
what I know, those tend to be just two verbs collapsing.
For example, ''wend'' apparently is an old verb in English, with a past
tense ''went''. It means to walk/wander, or something like that. And
that's where we got ''went'' for ''go''.
There's also a really weird case in Latin: fero (1S.PRES), ferro (inf.),
tuli (perf.), latus (past participle). (It means ''carry'', as in ''transfer''--
Your research sounds interesting, so, sure, send me some info when
you get a chance. My area is definitely not the languages you're
studying, so I probably won't follow all of the details, but I'm sure I'll find
the basics interesting.
Read more issues|LINGUIST home page|Top of issue
Page Updated: 13-Mar-2012
While the LINGUIST List makes every effort to ensure the linguistic relevance of sites listed
on its pages, it cannot vouch for their contents.