Lecture 12: Universal Grammar
In looking at English data, we have been building a picture of the grammar that underlies this
This grammar is made up of the following components:
The store house of all the idiosyncratic aspects of the language How words are pronounced What words mean What category words belong to What their subcategory is
What arguments are associated with predicates What categorial restrictions they place on their
arguments
The rules which tell us about the basic syntactic arrangements of words into phrases: X’ X YP XP YP X’ Xn Xn, Ym – where m = 1 if n = 1, 2
otherwise
The theta criterion There is a one to one relationship between theta
roles assigned by a predicate and arguments that bare them
The Universal Theta Assignment Hypothesis Theta roles are assigned to a uniform position in
all constructions Theme = specifier of thematic VP Oblique (PP arguments such as locative, instrument,
etc.) = complement of thematic verb Agent = specifier of agentive verb
There is a very general movement rule which simply allows movement without further specification Move
Move anything anywhere
What actually moves and to where in any particular construction is determined by the interaction of all other grammatical principles
One principle that directly limits movement concerns bounding Movements have to be as short as possible
There have been several ideas of how to achieve this Subjacency
Movement allowed over only one bounding node Relativised Minimality
The movement of an element of type X must be to the nearest possible position relevant for X
The Case filter All overt DPs must sit in Case positions
Case is assigned by certain heads Finite I nominative Agentive V accusative P accusative ‘for’ complementiser accusative
Case is assigned locally To complement position To specifier position To specifier of complement
Controls the use of different types of pronoun Principle A
Controls the use of reflexive pronouns (anaphors) They must be bound in their smallest binding domain They only appear with reflexive verbs
Principle B Controls the use of personal pronouns
(pronominals) They must be free in their smallest binding domain They cannot mark a reflexive verb
These various grammatical components, although they deal with specific phenomena, interact with each other to produce a complex analysis of all of the structures of a language
They fit together as follows:
So far we have looked at these grammatical principles as though they describe English
But they are meant to be able to describe all languages
Therefore this is a theory not just of English grammar, but of Universal Grammar
Human languages differ from each other, but not indefinitely There are universal truths about human
language which would be unexpected if there were no limits
Human languages are translatable into other human languages If there were no limits to human language
we would expect there should be things can could be expressed in one language but not another
Human children learn human languages, no other species does If language acquisition were just a matter of
learning complicated rules, we would expect other species to be able to do it
Rats can learn complicated rules about travelling a maze It seems that human languages are hard wired into
human brains But it is clear that it is not the case the only English
is hard wired into English children and Chinese into Chinese children, etc.
So what is hard wired must be universal to all languages
Speakers of a language know things that they could not have possibly learned This knowledge must come from somewhere If it isn’t learned, it must be innate Again, innate linguistic knowledge cannot be
language specific Innate linguistic knowledge must be of a
universal nature
Human children learn human languages easily Far more easily and thoroughly than adults
can a foreign language Far more easily and thoroughly than
linguists can describe any human language An innate knowledge of Universal
Grammar would explain this
Clearly, languages (such as English) are not innate
There is no one human language Children are not born speaking a language
There is some process of acquisition We suppose therefore that Universal Grammar
is made up from two parts Principles: general and universal rules common to
all languages and so don’t have to be learned Parameters: varying aspects of language which
allow individual language to differ and which must be learned
It has been claimed that the rules of X-bar theory restrict all languages In all languages
Phrases have heads Heads take complements Phrases have specifiers
But languages differ in where these elements are placed
The simplest way languages differ is in terms of whether the head precedes its complement or follows it:
Head initial Head final
All heads precede their complements in English Inflections precede VP
may [VP go]
Complementisers precede IP if [IP he may go]
Determiners precede NP the [NP man]
Prepositions precede DP through [DP the tunnel]
All heads in Japanese follow their complement: Complementisers follow IP
[IP nihongo-ga muzukasii] to Japanese-nom difficult that‘that Japanese is difficult’
Postpositions [DP densha] de
train by‘by train’
In German, some heads precede and some heads follow their complements Complementisers precede IP
dass [IP Hans oft Kürbissuppe isst]that Hans often pumpkin soup eats
Determiners precede NP die [NP Brücke]
the bridge Prepositions
durch [DP die Stadt]through the city
Postpositions [DP meiner Meinung] nach
my opinion according to‘in my opinion’
Some languages allow more word order variation than others János Marit szereti János szereti Marit Marit János szereti Marit szereti János Szereti János Marit Szereti Marit János
This might be a problem for the claim that X-bar theory is universal
But such languages might allow more movement than those with stricter word order
The position immediately before the verb in Hungarian is the focus position János elment János ment el
Anything that moves to this position is interpreted as focus János leszállt a villamosról János szállt le a villamosról János a villamosról szállt le
The position in front of the focus is the topic Anything which moves to this position is
interpreted as topic a villamosról szállt le János János a villamosról szállt le János szállt le a villamosról a villamosról János szállt le
We therefore might assume Hungarian is
basically verb initial
Things move in front of the verb for specific reasons
Principles XP X’, YP (comma indicates that
no X’ X, YP order is specified)
Parameters Head parameter
a) head is first b) head is last
Specifier parameter a) specifier is first b) specifier is last
Some have argued that the differences in languages mean that there are some things which can be expressed in one language that cannot be in another
Claim Eskimo has over 100 words for snow, so the
English sentence ‘snow is falling’ does not translate the differences that Eskimo can make
This is nonsense Eskimo actually has only a few words for snow (so
does English: snow, sleet, hail, drift) Eskimo is a highly agglutinative language, which
means that sentences can often consist of one word
But because a one word Eskimo sentence cannot be translated into a one word English sentence does not mean that Eskimo can express things English cannot
Languages have different numbers of basic terms for colour
Basic term = Not compound (light blue) Frequent (ultramarine) Not seen as ‘a kind of’ (scarlet is ‘a kind of red)
English black, white, grey, red, green, yellow, blue, brown, orange,
pink, purple = 11 Hungarian
Similar number to English, but Not orange (narancs is a kind of sárga) Piros vs. Vörös
Dani Mili vs. mola
Again, though, just because a language has more or less basic colour terms does not mean to say that the same distinctions cannot be made in one language as opposed to another
Hungarian can distinguish between orange and yellow and English can distinguish between piros and vörös
The most we can say is that some languages express certain things more economically/elegantly than others
Words in one language may have to be translated into more than one word, or even whole sentences
Perhaps certain concepts are more evident or prominent in one community than another So a single word can trigger a whole cultural
experience which would need to be explained to another community
Other species do not have the vocal equipment to produce speech But no one ever thought that parrots, which can
imitate human speech, can speak Experiments have been carried out to teach
Chimpanzees and Gorillas sign language They have been spectacularly unsuccessful,
despite popular myths about them The best thing we can conclude from these
experiments is that they demonstrate that human language is a uniquely human ability
Only the possession of a human mind provides the ability to learn and use human language
Consider: Who did you think that he saw Who did you think he saw Who did you think saw him * who did you think that saw him
All English speakers agree that the last sentence is ungrammatical
How do they know this?
Perhaps someone told them Given that most English speakers find it hard to even
describe the generalisation these data demonstrate, let alone explain it, it is highly unlikely that anyone ever told them about it
Perhaps they were corrected as children This is not the kind of error children make, so it is
unlikely anyone ever corrected them on this Perhaps they worked it out on the basis of similar
phenomena It is hard to think of anything similar to these
observations, so it is unlikely that they worked it out by analogy
It must be something that follows from our linguistic knowledge – which was not learned
Children learn the majority of their language by about the age of 5 They spend the first year not leaning much
language, so it takes about 4 years This is not a lot of time, considering what
else they are doing
Children learn language despite what their parents do rather than because of it Parents vary radically in what they do to ‘teach’
language Parents are unaware of their own grammar and
so don’t make ideal teachers They are apt to tell children rather inaccurate
prescriptive things (“there ain’t no such word as ‘ain’t!”)
They tend to correct factual errors rather than grammatical ones
Child: “daddy gone” Mother: “no he hasn’t, he’s in the kitchen”
Children tend to disregard corrections parents provide Child: “nobody don’t like me” Father: no, it’s “nobody likes me” Child: “nobody don’t like me” ... Several repetitions Father: “no, listen! – nobody LIKES me” Child: “oh! Nobody don’t LIKES me”
This indicates that children learn from positive data only They work out the grammar of their
language from hearing grammatical sentences and not from being told what is ungrammatical
Moreover, parents don’t always speak grammatically We all make mistakes How do children know which sentences to
attend to and which to ignore?
From this, it seems obvious that children should not be able to learn language from scratch The data they have access to is too
problematic Yet they do learn language
Not every child learns language in the same order But not every child grows teeth in the
same order, and no one thinks that that is not an innate process
However there is a good deal of regularity Children go through distinct phases which
happen at certain ages (±2 months)
Babbling Production of random sounds, usually CV
Reduplicative (6 to 9 months) Repetitive CV sequences with monotonous
intonation Bababababa, dadadadad, etc.
Non-reduplicative ( jargon 9 to 12 months) Varied sequences with varied intonation
Even deaf babies babble Babies with tracheotomies (so they can’t
babble) still develop normal language after the tracheotomy is reversed
So it isn’t clear what the function of babbling is
First words start at about 1 year and the list grows slowly at first (until about 50 words)
Mostly nouns, some verbs Child uses ‘one word sentences’ Suddenly (about 18 months) the child
goes through a ‘vocabulary spurt’ and the next stage begins
Children’s first combinations of words start at about the same time as the vocabulary spurt
Number of verbs and adjectives increase Two word utterances can look like subject-
predicate structures Daddy gone
But can also be other relations Mummy sock Big ball Give ball
Towards the end of this stage three and four word sentences may be produced
The syntax spurt happens when different kinds of sentences suddenly appear Passives, interrogatives, subordinate clauses,
etc. At the same time functional categories start
to appear (determiners, auxiliaries, complementisers)
After this, the system is refined for the next 3 years and is virtually in place by 5 years of age
If Universal Grammar is an innate human capacity, it allows us to explain how children appear to do the impossible
Moreover, if Universal Grammar is made up of principles and parameters, it also provides us with a detailed theory of how language acquisition should take place Only parameter settings need to be learned
E.g. Is the language head initial or head final?
Principles and Parameters theory does not tell us why children seem to suddenly undergo rapid development at the age of 2
This development links two things Diverse syntactic structures The use of functional categories
Some have suggested that this link is not random Functional categories are the main syntactic
words without which many syntactic processes cannot take place
One theory of language acquisition which can account for the syntax spurt is that certain linguistic concepts mature in the brain, similar to how physical things mature in the body E.g teeth, puberty
One idea is that the notion of a functional category undergoes maturation It is not available before 2 years When it becomes available, the child
undergoes the syntax spurt
The notion of Universal Grammar helps us to explain a number of mysterious facts about language
It assumes that the general structure of the linguistic system is innate and basic to all languages UG + parameter settings = specific
languages
Therefore: Language is specific to humans All languages share a common basis and
demonstrate universal phenomena Language acquisition is a matter of setting
parameters There may be certain parts of the innate
system which mature and therefore set the time for certain aspects of language acquisition