1 Week 2 Tutorial – Mbabaram – Jake’s answer guide
1.1 Preliminary Assumptions
1.1.1 Fixed word order
For this language, we’re assuming that word order is fixed, and so that a word’s position in the sentence
is related to how it should be interpreted.
As an example, English’s word order is fixed:
(1) a. The cat jumped the dog
b. The dog jumped the cat
By switching around the cat and the dog, we end up with different sentences that mean different things
— the subject and the object of the sentence in English is dependent on the position of that noun in the
sentence. Furthermore, since we know the cat and the dog can be substituted for each other in the same
position, then we know that they’re of the same category.
1.1.2 Overt expression of a verb and its arguments
We’ll also be assuming that a sentence needs to express a verb and any arguments it has.
As an example, English obligatorily expresses the verb and its argument:
(2) a. (i) John jumped
(ii) *jumped
(iii) *John
b. (i) The man patted the cat
(ii) *The man patted
(iii) *patted the cat
In these, the examples in (ii) and (iii) can’t mean the same as the respective sentence in (i). This is
because the verb and its objects are obligatorily expressed.
1.2 Analysing the data
To begin, we’ll pair all sentences of equal length to make the data easier to analyse.
3 dog lob
5 dog lon@
9 mog lonuN
1 dogul mog njab
2 mogul dog njar@
4 mog lon@ alngi
7 mog lob anm1n
8 dog lim lob
11 mogul dogul njab
10 mogul limul dog njaruN
12 dogul mog njab anm1n
13 mog lim lob anm1n
We’ll start with the two word sentences first.
1
1.2.1 Two word sentences
Looking at the two words table first, we can see that since 5 and 3 form a minimal pair, we know that lon@
and lob are of the same category.
(3) Observation 1: lob and lon@ are of the same category
Remember also that we’re assuming fixed word order, so a word’s position in the sentence corresponds to
the same category across sentences of the same type. Looking at the same data, we can hypothesize that
lonuN is of the same category as lob and lon@ too, since it’s in the second position of a two word
sentence. Similarly, this would lead us to suggest that mog and dog are part of the same category as
well.
(4) Hypothesis 1 — there are two categories
a. Category A = {mog, dog}
b. Category B = {lob, lon@, lonuN}
We’ve also assuming that the simplest sentence minimally contains a noun and a verb, so this means that
of the two word sentence, one must be a noun and one must be a verb. Hence, Category A or Category
B are nouns or verbs, though we don’t know which yet just from looking at the data in the two words
table alone. To find this out, we need to look at the rest of the data.
1.2.2 Three word and four word sentences
One important thing to note when making these tables is to look for regularities in the data. While the
two word table looks regular (that is, the sentences all appear to be of the same type), the three word
and four word tables appear to have irregularities, which suggests that we’re missing something. When
doing this sort of analysis, it’s always useful to put sentences that look similar together, so that we can
try and see if there’s any patterns to them.
It’s also useful to keep in mind our previous hypothesis based on the two word table — based on our
knowledge that lon@ and lob are of the same category, we can group the three word and four word
sentences into the following columns:
Three word sentences (revised):
4 mog lon@ alngi
7 mog lob anm1n
8 dog lim lob
1 dogul mog njab
2 mogul dog njar@
11 mogul dogul njab
Four word sentences (revised):
10 mogul limul dog njaruN
13 mog lim lob anm1n
12 dogul mog njab anm1n
At this point, we can see that there is something strange about one of these sentences — while the
sentences 1 and 2 both have the presence of the -ul marker on dog or mog (depending on which is first
in the sentence – a critical fact given thet assumption of fixed word order), sentence 11 has it twice on
both mogul and dogul. Since we know that only one of these sentences is wrong, and that 1 and 2 both
only have one instance of -ul, then 11 is a strong candidate for being the incorrect sentence due to the
presence of a second -ul marker.
However, sentence 10 also contains two -ul markers:
10 mogul limul dog njaruN
At the moment, these are the only two sentences that contain two -ul markers. We have good reason
to believe mog and dog are of the same category, and our other data never shows two cases of the -ul
2
marker with them both in the same sentence. But although they may be the same category, can we say
the same for lim? Let’s have a look at the contexts in which they appear together:
Contexts with just lim or mog or dog
3 dog lob
4 mog lon@ alNgi
5 mog lonuN
7 mog lob anm1n
9 dog lon@
Contexts with lim and mog or dog, or mog and dog
1 dogul mog njab
2 mogul dog njar@
6 dogul mog njar@ alNgi
11 mogul dogul njab
12 dogul mog njab anm1n
8 dog lim lob
13 mog lim lob anm1n
Contexts with lim, mog, and dog
10 mogul limul dog njaru
There are two important things we can see in this data. The first is that lim never shows up by itself —
it is always in the presence of mog or dog. This could just be an accident of the data, but as far as we
can observe, lim cannot appear by itself, and appears to be optional.
The second is that the -ul marker only appears when dog and mog are in the same sentence — it doesn’t
appear when lim is with just dog or mog. However, when lim is in the same sentence with dog and mog,
it appears to take the -ul marker.
If lim was of the same category as mog and dog, we would expect it to behave the same way when
there’s two of that category in a sentence. Considering sentence 10 and 11 again, if lim was of a different
category, that would allow us to say that sentence 11 is the incorrect sentence, which fits with the general
pattern seen in sentences 1 and 2. If we were to say that lim was of the same category and that sentence
10 was the incorrect sentence, then we’d fail to explain that piece of data.
As such, the better supported hypothesis is that lim isn’t the same category as mog and dog
(5) Observation 2: lim is of a different category, and appears to be optional
Returning to the hypothesis that dog and mog are the same category, and that a sentence must minimally
have a verb and a noun, then we’re left with either the case that dog and mog are both verbs, or that
they’re both nouns. While it’s pretty safe to assume at the outset that they’re both nouns given their
behaviour, let’s have a look how we might assure ourselves that they are.
1.2.3 Determining nouns and verbs
If dog and mog were nouns, then that would make Category B – lob, lon@, lonuN – intransitive verbs, and
njab, njar@ transitive verbs (and, given the similarity, we would expect njaruN to be a transitive verb
too).
So, by hypothesis:
(6) Hypothesis 1 (revised) — there are two categories
a. Category A (nouns) = {mog, dog}
b. Category B1 (intransitive verbs) = {lob, lon@, lonuN}
Category B2 (transitive verbs) = {njab, njar@, njaruN}
Given the distribution of these categories, this would appear to be correct, and it captures the fact
that Category B2 only shows up when dog and mog are both in a sentence. The other hypothesis —
that Category A are verbs — fails to capture this fact about Category B2, and so the hypothesis that
Category A are nouns is better supported by the data.
3
As such, we can take a sentence like 1 — dogul mog njab — as being a sentence with two nouns and a
transitive verb. From this, we might analyse -ul as a subject or object marker.
To recap so far, we’ve decided that there are at least three categories:
(7) Categories:
a. Category A (nouns) = {mog, dog}
b. Category B1 (intransitive verbs) = {lob, lon@, lonuN}
Category B2 (transitive verbs) = {njab, njar@, njaruN}
c. Category C = {lim}
1.2.4 Classifying lim, anm1n, and alNgi
We’ve almost analysed all the data for this set, however we haven’t classified lim yet, or the remaining
words anm1n, and alNgi. Considering their optionality, it could be an adjective/nominal modifier or an
adverbial. To decide what they are, let’s take a look at some near minimal pairs for the remaining words
we haven’t classified:
1 dogul mog njab
12 dogul mog njab anm1n
8 dog lim lob
13 mog lim lob anm1n
9 dog lon@
4 mog lon@ alNgi
2 mogul dog njar@
6 dogul mog njar@ alNgi
These sentence pairs appear to show that anm1n and alNgi are optional, like lim. At first, we might
be tempted to say that they form part of Category C with lim. However, crucially, if we look at their
distribution across the sentence pairs, we see that they don’t change their form (while lim does, taking
the marker -ul in sentence 10) and they don’t appear in the same position. Given that lim appears to
agree with the -ul marked noun, then it is likely to be an adjective or nominal modifier. If we analyse
lim as a nominal modifier, we might suggest that anm1n and alNgi are adverbials of some sort, given
that they appear to be optional, but don’t take agreement marking like lim.
(8) Hypothesis 2:
a. Category C (nominal modifiers) = {lim}
b. Category D (adverbials) = {anm1n, alNgi}
A quick check of sentence pairs with this hypothesis in hand lends support to it — if we take the
optionality of this class, then sentence pairs 2 & 6, 3 & 7, 4 & 9, and 8 & 13 show that mog and dog
can be substituted for each other, and hence are of the same category, which is in line with our previous
hypotheses and observations.
As an example, given the apparent optionality of anm1n, we can compare sentence pairs 3 & 7 and 8 &
13:
3 dog lob
7 mog lob anm1n
8 dog lim lob
13 mog lim lob anm1n
As noted, these minimal pairs show that dog and mog can be substituted for one another.
4
1.2.5 Summing up
(9) Final analysis:
a. Category A (nouns) = {mog, dog}
b. Category B1 (intransitive verbs) = {lob, lon@, lonuN}
Category B2 (transitive verbs) = {njab, njar@, njaruN}
c. Category C (nominal modifier) = {lim}
d. Category D (adverbials) = {anm1n, alNgi}
This would mean that sentence 11 is indeed the incorrect sentence, and we can correct it to a grammatical
one:
(10) Sentence 11 (corrected): mogul dog njab
2 The -ul marker
We saw that some sentences in Mbabaram had nouns marked with -ul. If Mbabaram were like English,
we might expect this to be similar to Accusative marking in English transitive sentences:
(11) a. She[NOM ] shook
b. She[NOM ] shook him[ACC]
In English, the argument of an intransitive verb has nominative case, like the agent of a transitive verb.
The object or patient of a transitive verb receives accusative case.
This would mean that Mbabaram would have the word order OSV. This is a possibility, but OSV
languages make up about 0.5% of the world’s languages.
An alternative is that Mbabaram marks the agent of transitive verbs, which would make it an SOV
language, of which is the most common word order for languages, with about 45% languages being SOV.
This would make Mbabaram an ergative language, which it happens to be.
As opposed to nominative-accusative languages, ergative-absolutive languages treat the object of tran-
sitive verbs and the argument of intransitive verbs as the same, whereas the agent of transitive verbs is
marked differently.
If we pretended English had ergative-absolutive marking, what it may look like is:
(12) a. She[ABS] shook
b. Her[ERG] shook he[ABS]
5
3 Important lessons
Figuring out your preliminary assumptions: For this problem set, we assumed that this language had
a fixed word order, and that every sentence need to have at least one noun and one verb. This quickly
allowed us to identify words as belonging to the same category.
Minimal pairs: Always start analysing the data set for minimal pairs. If one word can be substituted
for another, then they’re overwhelmingly likely to be of the same category.
Laying out the data: Drawing up tables and putting data in a graphic format makes the data much more
manageable, and allows you to quickly compare things.
Look for regularities: Organising data and looking for regularities can tell you a lot about how a lan-
guage works. It’s best to look for the most general hypothesis that accounts for the largest amount of
data.
Make hypotheses and test them: It’s always okay to make an initial hypothesis which you may end up
revising, or even completely abandoning.
6