Microsoft Word – Wk2Tute_Mbabaram_KatesAnswerGuide
Wk2 Tutorial 5/8/13 & 7/8/13
Page 1 of 7
Mbabaram dataset – Kate’s answer guide
Step 1 – Establishing presumptions
Roughly 80% of the world’s languages have some kind of dominant word order, by which we
mean the relative order of verb, subject and object in a basic sentence. In English, for
example, the word order in a basic transitive sentence is SVO (i.e subject verb object):
The dog (SUBJ) chased (V) Chomsky (OBJ).
It is this word order that tells us it is the dog that is doing the chasing, not Chomsky. So
considering that about 80% of languages have a fixed word order, it can be useful to assume
that word order will be present in an unfamiliar dataset when we have no prior knowledge
of the language’s syntactic/morphological features or those of the neighbouring languages.
• Presumption 1: Word order is fixed
Another presumption that is useful to establish before beginning an analysis of an
unfamiliar dataset is whether or not we expect the expression of core grammatical functions
(i.e. subject (SUBJ), direct object (DO), indirect object (IO)) as an overt word/phrase to be
obligatory. In about 20% of the world’s languages, including English, subjects pretty much
always need to be overtly expressed in this way:
Chomsky chased the dog.
He chased the dog.
*Chased the dog.
In about 60% of languages, subjects can be expressed through an affix on the verb, which
often means that they do not necessarily need to be expressed as independent
words/phrases. In Wubuy, which is the language that I’m working on, you can actually do his
for both subjects and objects:
ngu-rulbu-wanagayn
3MSG/FEMa-back-hold.PP
‘He held it (the python) by the back’
In about 9% of the world’s languages, overt expression of subjects is entirely optional if the
reference is already clear from the context. For example, in Japanese:
tegami o yon-da
letter OBJ read-PST
‘I/we/he/she/they read the letter.’
As the number of languages that allow this is relatively low, we’re going to assume for now
that Mbabaram is not one of them. Also, if we consider the simplest looking example from
the dataset, (3), we can speculate that Mbabaram is not a language that is marking core
grammatical functions on the verb through affixation:
Wk2 Tutorial 5/8/13 & 7/8/13
Page 2 of 7
(3) dog lob
In (3), there are two words, one of which must be our verb. Both of these words are
monosyllabic, suggesting that they are not very morphologically complex. That is why we
will begin by assuming that in Mbabaram, core arguments must be expressed by an overt
word/phrase.
• Presumption 2: Overt expression of verbs and core arguments (i.e. SUBJ, DO, IO) as
a word/phrase is obligatory
Although these presumptions don’t necessarily hold true across all languages, they are a
useful starting point, because it would be incredibly difficult (and probably impossible) to
analyse an entirely unfamiliar dataset without making certain assumptions beforehand. It is
also useful to explicitly state what these assumptions are when possible, because if your
analysis doesn’t end up working, it makes it easier to understand why this may be the case
and to go back and revise what you’ve assumed about the language.
Step 2 – Two word sentences
Based on the presumptions we have just made, what can we assume about two word
sentences?
• All our two word sentences must consist of an intransitive V and a noun (N) which is
the SUBJ.
• All words in initial position belong to the same word class, and all words in final
position belong to the same word class.
Now we can draw up a substitution table for 2 word sentences:
Table 1. Two word intransitive sentences
Class A Class B
3 dog lob
5 mog lonuŋ
9 dog lonǝ
At this point, we know that we have two different word classes, where one is an intransitive
V and one is a N, but we can’t really tell which is which. However, we might speculate that
those in Class B are the same intransitive verb with different tense inflections (i.e. they all
begin with lo-), which potentially leaves Class A as our nouns.
Step 3 – Three word sentences
Based on the presumptions made in step 1, what can we assume about three word
sentences?
• All our three word sentences must consist of either:
– a transitive V and two Ns (SUBJ & DO)
– an intransitive V, a N (SUBJ), and some kind of non-obligatory modifier (e.g.
adjective (ADJ)/adverb (ADV))
Wk2 Tutorial 5/8/13 & 7/8/13
Page 3 of 7
In Step 2, we established a set of words belonging to Class A and Class B in intransitive
sentences. Now, we can look for intransitive examples of three word sentences. These will
be the three word sentences that have both a word belonging to Class A and a word
belonging to Class B from Table 1.
As we already know that these sentences are intransitive, then the additional word that
doesn’t belong to Class A or B must belong to some kind of class of non-obligatory modifiers
(e.g. ADJ/ADV), but we can’t yet tell exactly what kind of modifiers they are.
Table 2. Three word intransitive sentences
Class A Class C Class B Class D
4 mog lonǝ alŋgi
7 mog lob anmɨn
8 dog lim lob
Now that we’ve established which three word sentences are intransitive, we can move on to
looking at the transitive ones, and this is where we can start drawing some conclusions
about which class is the verb class and which is the noun class. The three word sentences
we’re left with are:
1. dogul mog njab
2. mogul dog njarǝ
11. mogul dogul njab
What’s interesting here is that in each example, we have two words from Class A, where the
first one takes the suffix -ul (in (11) the second word also takes this suffix, which is puzzling).
As we have two instances of Class A words in each of these examples, then we can assume
that words belonging to Class A must be nouns (where -ul is probably some kind of
inflectional suffix, most likely a case marker), and the sentence final words are transitive
verbs (Class Bii).
Table 3. Three word transitive sentences
Class A
(N+suffix)
Class A
(N)
Class Bii
(Trans V)
1 dog-ul mog njab
2 mog-ul dog njarǝ
11 mog-ul dog-ul njab
Following from this, we can also identify the Class B words from Step 2 as our intransitive
verbs (Class Bi).
Wk2 Tutorial 5/8/13 & 7/8/13
Page 4 of 7
Step 4 – Four word sentences
Based on the presumptions made in step 1, what can we assume about four word
sentences?
• All our four word sentences must consist of either:
– A ditransitive V and three Ns (SUBJ + DO + IO)
– A transitive V, two Ns (SUBJ + DO) and some kind of adjunct (e.g. ADJ, ADV)
– An intransitive V, a N (SUBJ) and two adjuncts (e.g. ADJ, ADV)
Beginning with possible intransitive sentences, we can identify the verb lob in sentence (13),
which we have already established as being an intransitive verb in the previous steps. We
have also already identified mog as belonging to the class of nouns (so this would be our
subject) and lim and anmɨn as being some kind of non-obligatory modifiers:
Table 4. Four word intransitive sentences
Class A
(N)
Class C
(modifier)
Class Bi
(Vi)
Class D
(modifier)
13 mog lim lob anmɨn
Moving on to possible transitive sentences, we can identify the transitive verbs njarǝ in (6) and
njab in (12), both of which have two words from the class of nouns (dog-ul and mog) as well
some kind of modifier from Class D (alŋgi and anmɨn).
Now we’re left with one more sentence to analyse, (10). First of all, we can identify two
words from our class of nouns: mog-ul and dog. We can also find lim in this sentence, which
we have already categorised as some kind of modifier, but interestingly it is taking the -ul
suffix, which we have so far only seen on nouns and which we had speculated as being a
case suffix. Essentially, this suggests that lim is a nominal modifier (e.g. ADJ) that is taking
the suffix -ul so that it agrees with the noun that it is modifying (i.e. mog-ul). Finally, our last
word, njaruŋ must be our verb, and as there are only two nouns present, it must be a
transitive verb according to the presumptions we made earlier.
Table 5. Four word transitive sentences
Class A
(N+suffix)
Class C
(N modifier)
Class A
(N)
Class Bii
(Vt)
Class D
(modifier)
6 dog-ul mog njarǝ alŋgi
12 dog-ul mog njab anmɨn
10 mog-ul lim-ul dog njaruŋ
The analysis of njaruŋ as a transitive verb is possibly supported by a potential morphological
breakdown of our verbs, such that we have two verb stems, lo (intransitive) and nja
(transitive), and three suffixes (possibly tense markers), -b, -nǝ/-rǝ and -nuŋ/-ruŋ:
Wk2 Tutorial 5/8/13 & 7/8/13
Page 5 of 7
Table 6. Possible morphological breakdown of verbs
Class Bi Class Bii
Tense A lo-b nja-b
Tense B lo-nǝ nja-rǝ
Tense C lo-nuŋ nja-ruŋ
Step 5 – Answers to tutorial questions
Now that we’ve analysed our dataset, we’re ready to answer the tutorial questions:
Part A
• Study these sentences, find the regularities in their structure, and on this basis try to
identify the incorrect sentence.
• Correct it by modifying or replacing the incorrect word.
• State the reasons which have led you to your conclusion.
The incorrect sentence is (11):
*mog-ul dog-ul njab.
In all the other sentences in this data set, when there are two nouns present, only the first
one takes the -ul suffix. Therefore, we can correct the sentence by removing the suffix on
the second noun:
mog-ul dog njab.
Part B
• Now consider what the most probable grammatical structure is for the sentences.
• Can you identify, even tentatively, any parts of speech?
• Can you say anything about the role of what look like grammatical morphemes?
Parts of Speech
We’ve already established the following:
Class A (nouns) = {mog, dog}
Class Bi (intransitive verbs) = {lob, lonə, lonuŋ}
Class Bii (transitive verbs) = {njab, njarə, njaruŋ}
Class C (nominal modifiers) = {lim}
Now we are left with Class D, with the members alŋgi and anmɨn. These are not likely to be
N modifiers, as discontinuous noun phrases are not common in languages that have fixed
word order. What I mean by a ‘discontinuous noun phrase’ is one in which a head N and a N
modifier that belong to the same NP are not adjacent to each other (i.e. there is something
that doesn’t belong to the NP, such as a V, in between the two). In sentence (6), for
example, we would have a discontinuous NP if we decided that Class D are N modifiers.
Instead, it is much more likely that Class D are our adverbs:
Class D (adverbs) = {alŋgi, anmɨn}
Wk2 Tutorial 5/8/13 & 7/8/13
Page 6 of 7
Grammatical morphemes:
-ul suffix on nouns and nominal modifiers, which is likely to be a case suffix.
There are two possible analyses for what kind of case this could be marking, each of which
makes a different assumption about what the basic word order in Mbabaram is.
Analysis 1:
-ul marks transitive subjects (i.e ergative case), which would mean that the word order is
SOV (as in 41% of languages).
Analysis 2:
-ul marks direct objects (i.e accusative case), which would mean that the word order is OSV
(as in 0.3% of languages).
Analysis 1 is much more likely, and -ul does in fact mark ergative case in Mbabaram.
Grammatical structure of sentences:
To figure out the grammatical structure of sentences in Mbabaram, we can begin by writing
out ‘Phrase Structure Rules’ (PSRs) for each of the grammatical sentence patterns we have
in our data set:
Initial PSRs:
S ! A Bi (3, 9, 5)
S ! A C Bi (8)
S ! A Bi D (4, 7)
S ! A C Bi D (13)
S ! A-ul A Bii (1, 2, 11)
S ! A-ul A Bii D (6, 12)
S ! A-ul C-ul A Bii (10)
Then, we want to start thinking about how these rules can be further collapsed or
abbreviated. First of all, we have figured out that members of Class C modify members of
Class A, which suggests that together, they form a single constituent, an AP. This is
supported when we compare sentence (7) and sentence (13):
Class A Class C Class Bi Class D
7 mog lob anmɨn
13 mog lim lob anmɨn
Wk2 Tutorial 5/8/13 & 7/8/13
Page 7 of 7
Sentences (7) and (13) are exactly the same, except for the additional presence of lim in
(13). This suggests that mog lim can be substituted for just mog (comparable to Chomsky
likes [universals] vs. Chomsky likes [some strange universals] in English).
The other thing we noticed in our analysis is that Class C and Class D are optional, so we can
put brackets around these in our phrase structure rules to indicate this.
Revised PSRs:
S ! APi Bi (D)
S ! APii APi Bii (D)
APi ! A (C)
APii ! A-ul (C-ul)