Syntax – Week 1, Lecture 2
1
Topic 1. Arguing for syntactic
categories and constituents
1. What role does syntax play in communication?
When two speakers communicate, they are sharing a view of the world.
Imagine a speaker who has some information they want to communicate
to a hearer such as “Hamlet killed Polonius”. For this information to be
communicated successfully, the hearer must end up sharing this view of
the world. How does this occur?
Step 1: Both speaker and hearer have shared knowledge about what
the words Hamlet, Polonius & kill mean.
However, it is not enough for the speaker to just list the event and
participants by saying: “Hamlet, Polonius, kill”. How do we know who hit
whom? There must be some rules, such as word order in English:
Step 2: (For English speakers) Word Order
i.e., Hamlet killed Polonius ≠ Polonius killed Hamlet
Syntax is the study of the rules which structure information in sentences.
At the outset, we know that there must be rules because:
• people can reliably understand one another.
• people know when a sentence is wrong
– e.g., * “Bruce hit cat the.”
Furthermore, there can’t be too many rules because a child can learn
nearly all of them by the time they are four.
Syntax – Week 1, Lecture 2
2
2. Overview of Topic 1
What are the patterns which underlie sentences in various languages? How
do we determine them? How do we describe these patterns? Can we create
a theory to account for them? We will study:
1. The possible range of what these rules (i.e., syntactic information)
are and how they are expressed in various languages.
2. The methods we can use as tools for analysing sets of linguistic data
so as to identify the syntactic information they contain.
3. Structuring Information
All languages have ways of structuring information.
3.1 Word order and phrases (e.g. English)
English uses a word order scheme to structure information.
For example, some words go together in phrases:
The deranged prince killed the old man.
The old man, the deranged prince killed.
*The old, the deranged prince killed man.
Traditional grammar designates the various words above into various
categories:
Nouns: words which denote people, places or things.
Adjectives: words which describe a property of a noun.
(We will expand on this, but if you’re not comfortable with these, do some
reading, in particular Radford.) To return to our two sentences – order
matters on other levels as well. In English, adjectives appear before the
noun they modify:
Syntax – Week 1, Lecture 2
3
*The prince deranged killed the man old.
Adjectives and nouns tend to occur in clusters. We can call these clusters
phrases – so when we refer to a phrase, we are talking about a cluster of
words which goes together.
In the cluster the deranged prince – we have an adjective and noun. The
noun is crucial to the phrase, and the adjective is only saying something
about the noun. You can see this:
The prince killed the old man.
*The deranged killed the old man.
Because the noun is crucial to this phrase, we can call it a noun phrase
(NP). We also have the word the in the NP. It also contributes to the
meaning, albeit in a more subtle way. These words are also part of the
noun phrase – not only do they say something about the noun, they can’t
be moved from it:
The old man, the deranged prince killed.
*The, the deranged prince killed old man.
What about kill? Kill is a verb which designates an event of killing. Part of
its meaning requires that there be a killer and a victim. The order of the
NPs w.r.t. the verb tells us who does the killing and who is killed:
The deranged prince killed the old man.
The old man killed the deranged prince.
So now, we can re-examine these two sentences using our terminology. We
have two NPs, and the linear order of these two NPs gives the different
interpretations of the sentence. In other words – we are describing English
syntax in terms of:
Syntax – Week 1, Lecture 2
4
Categories of words (e.g., N, Dtv, V and Adj).
Syntactic Constituents or phrases (e.g., NP → Dtv Adj N)
The order of phrases determines their grammatical function. For example,
the first NP in English is usually the subject, the thing which does the verb.
3.2 Morphology (e.g. Warlpiri)
However, there are other ways of structuring information. For example,
Warlpiri (Central Australia):
wita wajilipu-ngu maliki ngaya-ngku wiri-ngki
small chase-PAST dog cat-ERG large-ERG
‘Large cat chased small dog.’
wita wajilipu-ngu maliki-rli ngaya wiri-ngki
small chase-PAST dog-ERG cat large-ERG
‘Large dog chased small cat.’
The ERG suffix above can be thought of as indicating the thing which does
the verb. The unmarked nouns and adjectives are having the verb done to
them.
These sentences can have any word order. The words which go together
(such as large cat and small dog in the first sentence) don’t actually have
to appear next to each other – but instead, because of the suffixes, we can
work out what goes together semantically, and who is doing what to whom:
maliki-rli wita wajilipu-ngu wiri-ngki ngaya
ngaya wajilipu-ngu maliki-rli wita wiri-ngki
maliki-rli ngaya wajilipu-ngu wiri-ngki wita
maliki-rli wita ngaya wajilipu-ngu wiri-ngki
etc…
Syntax – Week 1, Lecture 2
5
Note that it would be neat if the ergative suffixes on the different nouns
and adjectives were all the same, but they are not – however, they mean
the same thing 🙁
There is a close conceptual relationship between the morphology in
Warlpiri and word order in English – they are talking about the same types
of information. Phrasal constituents exist in Warlpiri – but it is those words
which share the same case marking which indicate constituents, not word
order as in English.
Likewise, both languages indicate who did what to whom. Warlpiri
indicates this using morphology on the nouns, English uses word order, and
Icelandic uses both:
Icl. Hestur sá mann
horse.nom see.pst man.acc
“(A) horse saw (a) man”
Maður sá hest
Man.nom see.pst horse.acc
“(A) man saw (a) horse”
*Maður hest sá
Man.nom horse.acc see.pst
4. Review – How Languages Express Syntactic Information
Two languages which express syntactic information largely via linear order
of words and phrases are English and Scots Gaelic. In contrast, Warlpiri
expresses syntactic information via morphological marking on words, e.g.
case.
Both languages differentiate between Syntactic Category (different classes
of the words, e.g. nouns vs. verbs) and Syntactic Constituents (different
types of phrases, e.g. noun phrases, verb phrases) – but how do we
Syntax – Week 1, Lecture 2
6
determine these?
4.1 Determining Syntactic Category – Substitution
The huge cat chased a tiny dog
A huge dog chased a tiny cat
tiger
frog
*kill
*the
*deranged
*red
This test provides good evidence that tiger, dog, cat and frog belong to the
same word class. Without even speaking English, we can already predict
other grammatical sentences using the above words – what are they?
So, the substitution test tells us what kinds of things have the same
distribution – words with the same type of distribution can occur in the
same slot, or frame:
The huge ____ chased the tiny dog.
Caution: What if we look at a wider range of nouns – such as truth or
courage? Not all nouns can go in the frame above – you need to develop
an awareness of when this problem is syntactic and when it is semantic.
There are additional tests for determining word categories here which will
be helpful (see next week).
4.2 Tests Determining Constituency (that is, what words form
larger phrases)
1. Substitution – single word for a group of words
Syntax – Week 1, Lecture 2
7
Q. What happened to the tiny dog?
A. The huge cat chased it.
If a group of words can be replaced by a single word – then we have
evidence that that group is a unit/constituent/phrase.
2. Transportation – movement of a group of words
The tiny dog, a huge cat chased.
*The tiny, a huge cat chased dog.
Can the transportation test above be used to determine constituency for
Warlpiri?
For Warlpiri :
• Syntactic categories are defined in terms of morphological
properties rather than distributional ones.
• Phrasal constituents are defined in terms of groups of words which
share the same case marking.
• Who is ‘doing the verb’ and who/what is being affected by the verb
is indicated by case marking, not by word order.
5. Worked Example – Scots Gaelic
(Data is taken from Brown & Miller Ch 4)
1. Bha an cù dubh
2. Bha an cat bàn
3. Bha Calum mór
4. Bha an cù sgìth
5. Bha Calum sgìth
6. Bha Màiri beag
7. Bha an gille mór
8. Bha an cù beag
9. Bha Màiri bàn
10. Bha an gille beag
11. Bha an cat mór
12. *Bha an Calum sgìth
Syntax – Week 1, Lecture 2
8
13. *Bha cat dubh
14. *An cat dubh bha
15. *Bha cat an dubh
16. *Bha dubh an cat
Examining sentences (1) to (16) – we don’t know what any of the words
mean. Nevertheless, we can still determine the syntactic rules underlying
these sentences and so understand why some sentences are grammatical,
and others ungrammatical. This shows that in a very significant sense,
syntax must be independent of meaning.
How do we start? You must become familiar with the data – and the best
way to do this is to rewrite the data sets according to whatever arbitrary
categories you might devise. We will start with all the grammatical three-
word sentences, simply because they are likely to be the simplest. Note that
there are two minimal pairs – and they can be considered to be a frame:
3 Bha Calum mór
5 Bha Calum sgìth
For example, (3), (5) can be thought of as:
Bha Calum _____ where the blank can be filled by mór or sgìth.
Sgìth and mór then belong to the same syntactic category because they can
be swapped with one another. Likewise, beag and bàn also belong to the
same category:
6 Bha Màiri beag
9 Bha Màiri bàn
Now I am going to make an educated guess – I will consider Calum and
Màiri to belong to another category together. This is because they are both
capitalised and look similar to the English names Calum and Mary. If this is
the case, then the syntactic categories can be summarised as:
Syntax – Week 1, Lecture 2
9
A B C
Bha Calum
Màiri
mór
sgìth
beag
bàn
A three-word grammatical sentence can be then written as:
S → A B C
A = {Bha}
B = {Calum, Màiri}
C = {sgìth, mór, beag, bàn}
4-Word Sentences (listed as is convenient)
First let’s examine two more minimal pairs:
8 Bha an cù beag
10 Bha an gille beag
(8) and (10) tell us that cù and gille are in the same syntactic category.
Likewise cat and gille below in (11) and (7):
11 Bha an cat mór
7 Bha an gille mór
By extension then, cat, gille and cù must be in the same category. Having
made this connection, we can rewrite our data to take advantage of this
fact:
1 Bha an cù/cat/gille dubh
4 Bha an cù/cat/gille sgìth
8 Bha an cù/cat/gille beag
2 Bha an cù/cat/gille bàn
11 Bha an cù/cat/gille mór
Syntax – Week 1, Lecture 2
10
7 Bha an cù/cat/gille mór
10 Bha an cù/cat/gille beag
This new table clearly shows that dubh, sgìth, beag, bàn and mór all belong
to the same category. The syntactic categories can be summarised as:
D (= A) E F G (= C)
Bha an cù dubh
cat sgìth
gille beag
bàn
mór
We now have two rules which can account for all the data seen:
S → {
𝐴𝐵𝐶
𝐴𝐸𝐹𝐶
}
A = {bha}
B = {Calum, Màiri}
C = {mór, sgìth, beag, bàn, dubh}
E = {an}
F = {cat, gille, cù}
However, we are not yet done. Using substitution, we have identified all
the different categories in the sentences, as well as their arrangement.
However, we can also use substitution to identify syntactic constituents –
that is, groups of words which go together. Recall how the test works – if a
single word can be swapped for a group of words, that group of words is a
phrase or constituent:
Q. What happened to the tiny dog?
A. The huge cat chased it.
Consider:
Syntax – Week 1, Lecture 2
11
6 Bha Màiri beag
8 Bha an cù beag
The above shows that a class B word can substitute for a class E and class F
word. There is massive evidence for this – consider (among others):
6 Bha Màiri beag
10 Bha an gille beag
5 Bha Calum sgìth
4 Bha an cù sgìth
So, we can identify a constituent or phrase (call it XP) which can be formed
from either a B class word or a E and F class word:
XP → {
B
EF
}
This means that our rules for grammatical sentences can be rewritten:
S → A XP C
XP → {
B
EF
}
Now, to test our rules above, we check them with the ungrammatical
sentences – are they correctly predicted as being ungrammatical?
12. *Bha an Calum sgìth
Yes – Calum is a class B word, it is not allowed to appear with a class E word
(an). This sentence is correctly predicted as being ungrammatical. Note that
we know this without understanding a word of Gaelic. Checking the other
ungrammatical sentences is left to you as an exercise.
Finally, using our knowledge of the world, can we match any of the syntactic
categories we identified above with real ones (such as nouns, verbs,
adjectives etc.)?
Syntax – Week 1, Lecture 2
12
Hint: If we allow that B is a proper noun, then what might E and F be?
Final Phrase Structure Rules:
S → V NP Adj
NP → {
PN
Dtv N
}
V = {bha}
PN = {Calum, Màiri}
N = {cat, gille, cù}
Dtv = {an}
Adj = {mór, sgìth, beag, bàn, dubh}
Finally, the lexical items and translated sentences:
an the
bàn white
beag small
bha was
calum calum
cat cat
cù dog
dubh black
gille boy
Màiri Mairi
mór big
sgìth tired
1. The dog was black.
2. The cat was white.
3. Calum was big.
4. The dog was tired.
5. Calum was tired.
6. Mairi was small.
7. The boy was big.
8. The dog was small.
9. Mairi was white.
10. The boy was small.
11. The cat was big.
As an exercise for next lecture, work out the phrase structure rules (i.e.,
identify the order of the constituents and categories) for the following
sentences:
17. Chunnaic mi an cù I saw the dog
18. Bhuail e an gille He struck the boy
19. Ghlac Calum breac Calum caught a trout
Syntax – Week 1, Lecture 2
13
20. Reic Calum an cù Calum sold the dog
21. Sgrìobh mi litir I wrote a letter
22. Ghlac Calum an cù Calum caught the dog
23. Bhuail Tearlach an cù Charlie struck the dog
24. Chunnaic an cù Calum The dog saw Calum
25. Bhàsaich Tearlach Charlie died
26. Fhuair Tearlach cù Charlie got a dog
27. Sheinn mi I sang
28. Bhàsaich an cù The dog died
29. Chunnaic Calum cù Calum saw a dog
Be aware that some verbs are associated with two NPs (such as “struck”
above) while others are only associated with one NP (such as “died” above).
This means your phrase structure rules will need to have two classes of
verbs – call them Vi (intransitive – one noun) and Vt (transitive – two nouns)
You will have to work out what each word means – you do this using
minimal pairs again. For example:
28. Bhàsaich an cù The dog died
25. Bhàsaich Tearlach Charlie died
Clearly Bhàsaich must mean “died” and Tearlach means “Charlie”. You can
then use these two translations to help with other sentences such as (23)
and (26).
Syntax – Week 1, Lecture 2
14
The full list of Scots Gaelic sentences:
(Data is taken from Brown & Miller Ch 4)
1. Bha an cù dubh
2. Bha an cat bàn
3. Bha Calum mór
4. Bha an cù sgìth
5. Bha Calum sgìth
6. Bha Màiri beag
7. Bha an gille mór
8. Bha an cù beag
9. Bha Màiri bàn
10. Bha an gille beag
11. Bha an cat mór
12. *Bha an Calum sgìth
13. *Bha cat dubh
14. *An cat dubh bha
15. *Bha cat an dubh
16. *Bha dubh an cat
17. Chunnaic mi an cù I saw the dog
18. Bhuail e an gille He struck the boy
19. Ghlac Calum breac Calum caught a trout
20. Reic Calum an cù Calum sold the dog
21. Sgrìobh mi litir I wrote a letter
22. Ghlac Calum an cù Calum caught the dog
23. Bhuail Tearlach an cù Charlie struck the boy
24. Chunnaic an cù Calum The dog saw Calum
25. Bhàsaich Tearlach Charlie died
26. Fhuair Tearlach cù Charlie got a dog
27. Sheinn mi I sang
28. Bhàsaich an cù The dog died
29. Chunnaic Calum cù Calum saw a dog