Syntax - First Things First

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Syntax – First things first

Haj Ross
Program in Linguistics
College of Information, UNT
[email protected]
http://www.umich.edu/~jlawler/hajpapers.html
Seminar at the National University of Modern Languages, Islamabad 9.VI.2015.
I. The discovery of syntax, semantics, and universal grammar by Zellig Harris
around 1950.
His three fundamental observations
A. Big sentences are made up of little sentences (called “kernel,” or “atomic”
sentences)
B. The meanings of big sentences are made up of the meanings of the
kernel sentences.
C. Languages are more similar in their kernel structures than in their surface
structures.
II. An example of the surface structure of a big sentence (bold-facing
indicates a kernel sentence in hiding):
The unfortunate decision by Mike to run for reelection to the
presidency made him an unpopular candidate.
Kernel 1: Kernel 2 made Kernel 3
Kernel 2: Kernel 4 was unfortunate.
Kernel 4: Mike decided on Kernel 5.
Kernel 5: Mike ran for Kernel 6.
Kernel 6: Someone reelected Mike to Kernel 7.
Kernel 7: Mike was president.
Kernel 3: Mike was a candidate.
Kernel 8: The candidate was unpopular.
We notice that there are many changes between the order of the words in the
kernel structure and that of their order in the surface structure. These changes are
effected by a variety of transformations, examples of which we will look at in
detail in §8 below.

III. To be able to see the beauty of syntax, we must know:


1. How to tell the parts of speech of words. Some tips:
A. If you can pluralize a word, it’s a noun. cat ≠ cats – cat is a Noun.
B. If you can change it from the present (Today I itch) to the past (Yesterday, I
itched), itch is a Verb.
C. If you can find it before a Noun (heavy rock) and after a linking Verb (like be,
become, grow) – The rock was/became heavy – then heavy is an Adjective.
D. If you can modify a Verb on either side of the Verb (We softly wept / We wept
softly), then softly is an Adverb.
E. If it’s a word that occurs between two nouns, and tells what their spatial
relationship is (rocks IN boxes, rocks ON tables, fleas BEHIND trees) then it’s a
Preposition.
F. If it precedes nouns and indicates number of quantity (five men, many bumps,
much wine, all wheat) it’s a Quantifier.
Warnings!
A’. Mass nouns have no plurals (*2 dusts/dirts/goos/mucks/rubbles)
B’. In tenseless contexts, verbs can’t inflect (I commanded that yesterday he
clean*(ed) his room.)
C’. Some adjectives can’t occur prenominally: *the afraid cat; some sound bad
after linking verbs: ?the order was presidential.
D’. Some adverbs only occur postverbally (*I fast ran); some only occur
Preverbally (* I think hardly that’s funny.)
E’. Not all prepositions have spatial meanings (*save, for, without, except).
F’. Some quantifiers can occur after linking verbs: The reasons are ten/many.

2. How to make phrases from each of the above parts of speech.


Attaching a modifier to a phrase of type X makes an XP (an X phrase). A
word of type X followed by a possible object of X is an XP.
A. fat cats is an NP; so is cats from LA.
B. softly weep and weep softly are VP’s. So are eat pizza and give me ten bucks.
C. very heavy is an AP; so is fond of donuts
D. extremely often is an AdvP; so is independently of me
E. straight into the pool is a PP; so is into the pool.
F. In quite many cats, quite many is a QP; so is almost ten in The reasons are
almost ten.
Note that modifiers of Nouns are called Adjectives; modifiers of all the other
parts of speech are all called Adverbs. The Latin prefix ad- means “to;” the
morpheme(s?) ject come from the Latin verb jacere, “to throw.”
3. A Clause consists of an NP followed by a fully specified VP (plus various
optional modifiers).
4. Valence.
Each lexical item must be marked for its valence, which specifies how many
arguments the lexical item in question requires to convey “a complete thought.”
The arguments of a lexical item can be NP’s, PP’s, AP’s, and AdvP’s
Verbs Adjectives Nouns
Valence 1 sleep be rich cat
Valence 2 tickle NP, consist PP be fond PP brother PP
Valence 3 give NP PP, speak PP PP (be suggestive, indicative) letter PP PP
Valence 4 obtain NP PP PP ——— (sale) PP PP PP
Valence 5 buy him cars from Ed for lira
5. A complement of a verb, adjective or preposition is a clause functioning as
an NP. Complements have the following tree structure: NP
|
S (= sentence)
where “S” represents a clause with one of the following complementizers.
Complementizers are grammatical morphemes that indicate that the clause in which
they appear is a subordinate clause. The “complementized” clause can appear as a
NP object clause, or as an object clause of a PP, or as a sentential subject clause, or,
for bisentential verbs, as both subject and object:
Tensed that-clause: a. I know that you are a Venusian.
b. That I am a Galactic Warrior is obvious.
c. That she screeched indicates to me that she wants rum.
Tenseless that-clause a. I demand that it free my boss.
b. That she be released by 5 is imperative.
Embedded wh-clause a. I asked about who mailed what to whom when.
b. What was done to whom where is a mystery.
Acc Ing a. I object to him being photographed.
b. Them being color photos is particularly odious.
Poss Ing a. The police mentioned his patting our cat repeatedly.
b. Their having demanded attention was no secret.
Action Nominal a. We protest your repeated stroking of your skunk.
b. Your infrequent feeding of my guppies is criminal.
Derived Nominal a. We lament your gift to your toddler of a cobra.
( = nominalized sentences) b. His refusal of the crown caused a riot.
6. Types of subordinate clause – all of these can be preposed, some with more
difficulty than others:

We discovered soon that Tom cheated ¨ That Tom cheated we discovered soon.
A. Complement clauses
1. The clauses discussed immediately above are NP’s. They can be
referred to by proforms, such as it, that, and which.
We discovered soon [that Tom cheated]i, but Ed doesn’t know [iti /thati].
whichi Ed doesn’t know.
2. Noun complements: [NP]S.
These are like the “objects” of abstract nouns, such as fact, idea, notion,
hypothesis, scheme, trick, etc., as in sentences like
The idea that eggs are healthy has been disproven.
But we see that deeply, the object that-clause of idea derives from an of-
phrase which has a regular [NP]S complement as the object of the of :
He said [that eggs are healthy]i, but the idea of iti has been disproven.
B. Relative clauses (Harris called these “noun-sharing clauses.”) So in the
following sentence
The painting which I looked at was a forgery.
the underlined relative clause derives from Kernel 2 – I looked at the painting.
Deeply, Kernel 1, the main clause of this sentence, namely The painting was a
forgery, has Kernel 2 following its subject NP, as in the following structure:
The paintingHead S was a forgery
I looked at the paintingTarget

In the English process of making such embedded kernels into relative


clauses, the target NP is converted into the relative pronoun which and
moved to the clause-initial position. Other languages, such as Japanese (and
perhaps Urdu as well?), the only operation necessary is for the Target NP to
be deleted in its place in the embedded clause, which immediately preceded
the NPHead.
C. Adverbial clauses of location (where-clauses), of time (when-clauses, and
clauses starting with temporal prepositions like before, after, since, until, and
also the word while), and various clauses of condition (if, unless), of cause
(because, since, as), of concession ((al)though) and other even less well
understood ones (lest, as if, etc.) require more advanced discussion than would
be appropriate here.
7. Coordinate clauses – those introduced by conjunctions like and, (either . . . ) or,
(neither. . . ) nor and for. These clauses, unlike subordinate clauses, can never be
preposed. We will not discuss them further today.
8. Unary Transformations.
Harris divided transformations into unary transformations, which operated
on a single deep kernel. Some of these unary transformations follow below.
a. POST-VERBAL PP SCRAMBLING:
The deep kernel Haj talked to Pip about golf on Friday can surface in any
on of the following reorderings:
i. Haj talked about golf to Pip on Friday.
ii. Haj talked on Friday to Pip about golf.
iii. Haj talked to Pip on Friday about golf.
iv. Haj talked on Friday about golf to Pip.
v. Haj talked about golf on Friday to Pip.
All of these are grammatical in English; their choice in particular
situations depends on such issues as information flow, which we will
not have time to look at here.
b. PP FRONTING
To Pip Haj talked about golf on Friday.
About golf Haj talked to Pip on Friday.
On Friday Pip Haj talked to Pip about golf.
Again, these variants involve only the change in order of one PP.
Sometimes more than one PP can be fronted, especially if one of the
frontees is an adverbial PP:
?On Friday to Pip Haj talked about golf.
c. The Grand-daddy of all transformations: PASSIVIZATION
This transformation not only converts one of the NP objects of the
PP’s in the valence of talk into the subject of the passive verb complex
(be talked), it also demotes the deep subject Haj to a by-phrase:

i. Pip was talked to by Haj about golf on Friday. ≥


ii. ?Golf was talked about to Pip by Haj on Friday.
We notice that PASSIVIZATION is rendered either impossible, or at
best highly marked, if the NP that is removed from a PP that is not
immediately after the verb of the clause. Thus from this variant of the
original kernel
Haj talked about golf to Pip on Friday.
It is possible, for me at least, to form the passive clause in cii. But if
we try to make the derived subject Pip, we get ciii, which I would be
surprised to find anyone who completely accepts it:
iii. ?*Golf was talked about to Pip by Haj on Friday.
And worse yet, if we even think of removing Friday, even if the On
Friday phrase immediately followed talk:
iv. *Friday was talked on to Pip about golf.
Even more unthinkable:
v. **Friday was talked to Pip about golf on.
So we see that is a process which involves promoting an NP which
immediately follows the verb into the subject of the derived clause, a
fact which suggests an explanation for a fact noticed by Joan Bresnan:
vi. Horace took advantage of Mike repeatedly.
vii. Advantage was taken of Mike repeatedly by Horace.
viii. Mike was taken advantage of repeatedly by Horace.
ix. Horace took repeated advantage of Mike.
x. Repeated advantage of Mike was taken by Horace. >
xi. ? Mike was taken repeated advantage of by Horace.
Bresnan suggests that the deadverbial modifier repeated augments the
noun-phrasiness of advantage, making it probably almost universally
impossible to feel that take repeated advantage of is a verb whose object
is Mike, a reanalysis which is generally possible for the string take
advantage of.
d. PREPOSITION DELETION

i. Haj talked about golf to Pip. ¨


ii. Haj talked golf to Pip.
iii. *Golf was talked to Pip by Haj.
There are few verbs which allow an immediately post-verbal
preposition to be deleted, as has happened in the transformation of di
to dii, and I know of no verb which allows this process to subsequently
allow the new “object” – i.e., the immediately following NP – to
become a passive subject.
e. DATIVE
We note that English allows, for certain verbs, the indirect object to
deprepositionalize and to immediately follow its verb:
i. Ida gave the book to Kim.
ii. Ida gave Kim the book.
The passive versions of ei and eii are, respectively, eiii and eiv:
iii. The book was given to Kim by Ida.
iv. Kim was given the book by Ida.
There are, however, complex reasons, which I cannot discuss here, for
believing that the original assumption that ei can be transformed to eii
are incorrect (Cf. Paul Postal, Edge-Based Clausal Syntax).
f. GOAL TO DIRECT OBJECT
It does appear, though, that it is correct to assert that fi can be
transformed to fii.
i. They sprayed paint on(to) the wall.
ii. They sprayed the wall with paint.
The verb spray is a motional one; the Theme which follows it as a
direct object can be followed by a full Path, as in fiii:
iii. They sprayed paint [[from the kitchen]Source [through the dining
room]Trajectory [on*(to) the wall]Goal]Path.
And when only the Goal phrase of the Path follows the Theme, as is
the case in fi, then it can advance to become the direct object, as is the
case in fii. And as a direct object, it can become a passive subject, as
we see in fiv:
iv. The wall was sprayed with paint by them.
In my discussion of some unary transformations, I have left
unmentioned many important issues, in the interest of demonstrating
some of the possibilities of transformation of a single kernel clause. I
will close this discussion of unaries by citing, without the requisite
discussion or many of the problems raised in each case, only to suggest
the amount of flexibility that can be found in a single clause.
g. UNSPECIFIED NP DELETION:

i. Hans is eating something ¨ Hans is eatin.


ii. Lee was attacked by someone. ¨ Lee was attacked.

h. PARTICLE MOVEMENT:
i. We ran down the cyclist.
ii. We ran the cyclist down.
i. PATIENT SUBJECT:
i. Any sophomore can readily translate this passage.
ii. This passage translates readily for any sophomore.
DIRECTIONAL PARTICLE FRONTING:
j. i. The cops came in.
ii. In came the cops.

9. Binary transformations
We have already mentioned in passing one binary transformation – the
one which takes as input two kernels which share a NP, and makes one of
them into a relative clause. I will mention only a few more in passing, for I
want to reserve the last section of my discussion to that which involves
general, perhaps universal, restrictions on the types of transformational
operations which can be performed.
a. EQUI NP DELETION (also known as CONTROL)

i. Art ate breakfast (Kernel 1) after Art got up (Kernel 2) ¨


ii. Art ate breakfast after getting up.
b. GAPPING
i. Tom ordered wine (Kernel 1) and Hank ordered beer (Kernel 2).
ii. Tom ordered wine and Han beer.
c. SLIFTING
i. I think (Kernel 1) that Pourri is a moron (Kernel 2)
ii. Pourri is a moron, I think.

You might also like