The Structure of Language

1
The Structure of Language
The study of phonetics is part of the larger study of
language.
Purpose: To show how phonetics fits into the language
system.
Language: Term is used in to related but different ways:
1. A specific language – French, Portuguese, Farsi, Urdu
2. Much broader sense – the general design plan that is
common to all languages. All human languages are built on
the same underlying design plan, but differ in many details.
Many differences in detail across different dog breeds and
different individual dogs. BUT – the deeper truth is that they
are all built on the same body plan and have far more in
common than the superficial differences might suggest.
What are some features common to all dogs?
1. Social animals
2. Territorial
3. Omnivorous but with a strong preference for meat
4. Same basic configuration of skeleton
5. Same number and basic shape of teeth
6. Any dog will breed with any other dog, regardless of
large differences in size and general appearance
7. … (a large number of other features)
The Structure of Language
Language works this way as well. All human languages are
built on the same basic design plan. The broad design
features that all languages have in common run deeper
and are far more important than the differences in
details, as large as those difference may at first appear.
Languages are defined by their grammars – a set of rules
that allows a speaker (or signer in the case of a sign
language) to generate permissible, well-formed
utterances (and the knowledge that allows one to recognize
“broken” utterances when they are encountered).
Your knowledge of English grammar allows you to figure out:
John went to the concert. Good utterance
*John the concert to went. Not so good
2
Notes:
(1) A sentence can be perfectly meaningful but still be
ungrammatical :
*This is a four doors car.
*He drove a red big car.
It’s perfectly clear what these sentences mean, but they’re
busted; i.e., ungrammatical.
(2) The word grammatical here does not mean the same thing
that it meant in grade school.
She ain’t got no crayons.
Where were you at, John. We was waitin’ on you.
These sentences conform to the grammar of some dialects of
English; they happen not to conform to the dialect of
standard English.
There are two very different uses of the term grammar: (1)
descriptive grammar: rules real speakers actually use, (2)
prescriptive grammar: rules that English teachers (and
other “experts”) believe speakers ought to use.
Examples of prescriptive grammar rules:
Don’t end a sentence with a preposition
Don’t split infinitives
Don’t use ‘like’ like this: “So I was, like, ‘Calm down, man;
you’re getting all agitated’.
Say “Betty and I went to the party,” not “Betty and me went
to the party.”
Violations of descriptive grammar:
*That book looks alike.
*I threw out it.
*Frank seems sleeping.
*Bag garbage no good. Ski good.
*I did not know how should I dress.
Linguistics as a discipline is concerned almost exclusively with
descriptive grammar, not prescriptive grammar. More
about this later when we discuss dialect.
One more point: grammar is sometimes used to refer
specifically to syntax (word-order rules), but more recently it
refers to all of the rules of the language, including syntax,
semantics (meaning), and phonology (sound pattern
rules). More later.
Now, finally, back to the two uses of the word language:
• a specific language (English, Dutch, Hungarian, etc.)
• the general design structure of all human languages
The 1st meaning is simple and obvious, but what about the
2nd? What features do all human languages have in
common? Called the Universal Grammar – it’s a huge,
gimongous list (and incomplete).
Here a just a few:
1. Rules are always structure dependent. E.g., English
question formation:
John will run.
Will John run?
3
Question is formed by reversing the order of the subject & the
verb – subject & verb being structural properties.
One more:
Hanley is the most stubborn person in the department.
Is Hanley is the most stubborn person in the department?
How about this hypothetical rule: Form a question by reversing
the order of the last 2 words in the sentence.
Hanley is the most stubborn person in the department.
Hanley is the most stubborn person in department the?
John will pitch on Thursday.
John will pitch Thursday on?
(1) Not a rule of English; (2) not a structure-dependent rule;
most important: (3) no rule remotely like this in any
language, yet this rule would work just fine.
The soldier that is ill is in the hospital.
How do we make a question of this? Which of the 2 instances of
is gets moved?
*Is the soldier that ill is in the hospital? (move the 1st one – not so good)
Is the soldier that is ill in the hospital?
It’s the 2nd instance of is that gets moved, but why? Does the
rule say – move the 2nd instance of the verb if there are 2?
No. The 1st instance of is gets passed over because it’s buried
inside of a NP that is treated as a unit – the NP being a
structural property.
No language uses a rule that says, “move the 2nd instance
of the verb if there are 2”, or “move the 1st instance of
the verb”
No child ever makes the mistake of getting mixed up about
which verb to move. Why? Because they come into the world
knowing that rules are structure dependent and not dependent
on something like serial position – though serial-position rules
would work fine.
More on this point soon.
2. Nearly all languages have agreement rules.
The box is in my office.
The boxes are in my office.
Subject and verb agree for number (plural vs. singular).
Languages vary a lot in what kinds of things there needs to be
agreement on. Not all languages enforce agreement on
number, but nearly all languages incorporate lots of
agreement rules.
Spanish (and French and many other languages) enforce
agreement on gender:
los perros (dogs), las casas (houses), los árboles (trees),
las tablas (tables), las flores (flowers), las montañas
(mountains)
4
Important: Agreement is not a necessary feature.
• It’s quite easy to imagine a language without agreement.
• In English, number is already specified by the noun. Why
give exactly the same information by supplying a plural verb
to go along with a noun that you already know is plural?
• Not all languages have this particular form of agreement,
but nearly all languages have agreement rules.
• Is subject-verb number agreement part of the universal
grammar? How about agreement for gender as in French
and Spanish?
3. Phonological rules: All languages incorporate soundpattern
rules called phonological rules.
• beed – beat
• bid – bit
• league – leak
• cub – cup
• cab – cap
• lag – lack
What do you notice about the lengths of the vowels on the left
vs. those on the right? Rule: Vowels are lengthened when
they precede voiced consonants.
Not all languages have this particular rule. All languages
have large numbers of sound-pattern rules like this one.
A similar rule from English. Look at these plural forms:
walks
lips
rats
tracks
————–
labs
dogs
awards
doors
What sound is added to form the plural in the 1st group vs. the
2nd group? Orthographically, it’s always an ‘s’, but what
sound is it? (Note: /lAbs/ — as opposed to /lAbz/ — is not
impossible to pronounce.)
Languages don’t necessarily need to incorporate phonological
rules – though all of them do.
They can’t be essential to communication – nearly every
language besides English gets by fine without the vowellengthening
rule.
Similarly, English gets by fine without the very different set of
sound pattern rules in Spanish, Tamil, Hindi, Korean, etc.
All languages incorporate phonological rules.
Are phonological rules part of the universal grammar?
Is the vowel lengthening rule (e.g., “cab” vs. “cap”) part of the
universal grammar?
5
4. Head First/Head Last
Phrases in all languages contain a special “boss” word called
the head. The head controls grammatical features of other
words in the phrase.
The fox in sox is in the yard.
*The fox in sox are in the yard.
“fox” is singular, “sox” is plural. Why is it that the verb agrees
with the “fox rather than “sox”? Because it’s the “boss” word;
i.e., the head of the noun phrase “fox in sox.”
Flying out of Kalamazoo on small planes is scary.
*Flying out of Kalamazoo on small planes are scary.
“Flying” here is the head of the phrase because the phrase is
mainly about flying, not planes, so the verb agrees with the
singular “flying”, not the plural “planes”.
English is a head-first language – the head precedes all other
words in the phrase. Many other languages reverse this.
English: Kazu ate sushi. (Kazu=NP; ate sushi=VP; ate=head)
Japanese: Kazu sushi ate.
So, Japanese is a head-last language. So, big deal? Here’s
the big deal: Every head-first language applies the headfirst
rule to all of its phrases: NPs, VPs, PPs. Everything.
Similarly, every head-last language applies the head-last
rule to all of its phrases: NPs, VPs, PPs. Everything.
English: to Tokyo (preposition)
Japanese: Tokyo to (postposition)
There are no languages that mix these up – e.g., head-first for
NPs, head-last for VPs and PPs. Also, no “head-middle”
languages.
There is no reason that languages have to behave this way.
It is easy to imagine a language that uses Japanese-like headlast
NPs along with English-like PPs:
Kazu sushi ate at home. (head-last NP, head-1st PP)
Or the other way around:
Kazu ate sushi home at. (head-1st NP, head-last PP)
These mixed rule systems don’t happen. Ever. Why?
• It’s more logical?
• I say
• You say (formerly Thou sayest)
• He says
Anything logical about this system?
• By coincidence, 6,000 separate human languages
happened to adopt this regularity – without benefit of
committee meetings.
• Neural circuitry incorporating this (& many other
grammatical regularities) are built into the brain – just like
the neural circuitry that allows a bat to convert sonar into an
image is wired into bat brains, or the circuitry that allows
spiders to know how to spin a web is wired into spider
brains.
6
Why is the concept of a universal grammar important?
Current thinking among most linguists:
When children acquire language they do not learn the
universal grammar at all. They already know it.
Children do not need to learn that there is agreement; they
need to learn what those specific agreement rules are.
They do not need to learn that rules are structure dependent;
they need to learn what those structure-dependent rules
are.
They do not need to learn that there are sound-pattern rules;
they need to learn which particular sound pattern rules apply
to the language they are learning.
They do not need to learn about the concept of a head, but
they do need to learn whether their language is head-1st or
head-last.
The Modularity of Language
Central feature of language: It is a layered or modularized
system – the neural substrate for language is not a blob of
brain tissue that “knows” about language, but a collection of
interconnected specialists that know only know an awful lot
about just one thing.
This is true of all complex systems. A car is not a mass of
metal and plastic that knows how to go. Cars have
specialized modules:
• fuel delivery system (carburetor/fuel injector)
• combustion chambers
• transmission
• suspension/steering system
• exhaust system
• etc.
The human body is modularized. It’s not a blob of protoplasm
that knows how to live – it’s a highly interconnected system
of specialists that each handle just one kind of job:
• Circulatory system (pump, veins, arteries)
• Waste management system (kidneys, liver)
• Central control system (brain, spinal cord, …)
• Musculo-skeletal system
• Sensors (visual, auditory, tactile, …)
• etc.
Language is modularized – it’s a highly interconnected
collection of experts, each of which handles just one kind of
analysis.
Major modules of the language system:
• Semantics (meaning)
• Syntax (structural rules governing word order)
• Lexicon (mental dictionary)
• Morphology (word-making rules – walk, walked, walking, …)
• Phonology (sound-pattern rules)
• Phonetics (articulation/sound patterns)
7
The layers of the language system are interconnected but
DISTINCT – i.e., different from one another.
A few examples:
Syntax and semantics are not the same thing.
*Colorless green ideas sleep furiously. (Chomsky)
• syntactically well formed but semantically anomalous (i.e.,
all messed up).
• conforms perfectly to English syntax but violates semantic
rules. Your syntax module reports that it checks out;
your semantics module reports that it’s busted.
*We threwed out it yesterday. The lawn we throwed it onto.
Semantics? OK. Syntax? Nope.
*You no stupid computers.
Phonology (sound pattern rules) and the lexicon (mental
dictionary) are distinct from one another; e.g.,
brick – blick – bnick
• brick: a word
• blick: a non-word, but conforms to English phonological
rules that constrain word shapes
• bnick: a non-word that violates an English phonological
rules that constrains sound sequences
What’s happening here? These examples prove
specialization, or modularity:
“brick”: Your lexicon specialist tells you that this is a word.
“blick”: Your lexicon specialist tells you that this is not a
word, but your phonology specialist tells you that it
could be.
“bnick”: Lexicon specialist: not a word; phonology
specialist: not a permissible word.
Morphology: Rules for word formation (e.g., dog -> dogs; walk
-> walking)
If boof were a word, what kind would boofable be (noun, verb,
adjective, etc.)?
How can the word understand (verb) be turned into noun?
He used to live in Pakistan, so he has a good _________ of
cricket.
What form of understand goes in the blank?
How about making an adjective?
Some of the concepts were unfamiliar to me, but the teacher
made these ideas ________________.
What form of understand goes in the blank?
What module of the language system are you relying on to
answer these questions?
You are applying your knowledge of morphological rules –
rules for forming words out of smaller units called
morphemes.
Morpheme: The smallest unit of language that has meaning.
Some examples – 1 word in all cases, but not always 1
morpheme:
dog
dogs
able
unable
believe
believable
unbelievable
8
One more point of terminology about morphemes: Morphemes
can be either bound or unbound (also called free).
Unbound: Can stand alone as a separate word (e.g., dog,
walk, park, …). Unbound morphemes are also called free.
Bound: Must appear in combination with 1 or more other
morphemes; e.g., suffixes like -s, –ed, -able; prefixes such as
pre-, post-, un-, etc.
premature: pre=bound, mature=unbound
blindness: blind=unbound, ness=bound
Check out the exercise on my web page called Counting
Morphemes.
One final point of terminology about morphemes: The concept
of a morpheme is pretty straightforward, but counting them is
not always so simple. A few examples:
How many morphemes in these?
uniform Do speakers realize this word is derived
from morphemes meaning one form?
agnostic The word gnostic does exist, but do most
speakers know this; i.e., do speakers treat
this as meaning not gnostic?
atheist Do speakers understand this to mean not
theist?
Phonology and phonetics are not the same thing; for
example, note the /p/ in the following:
[thAp] (“top”; released)
[thAp|] (“top”, unreleased)
[phAt] (“pot”, aspirated)
[spAt] (“pot”, unaspirated)
These realizations are phonetically distinct but phonemically
or phonologically equivalent; i.e., they are members of the
same broad phonemic category /p/.
Released and unreleased /p/: allophones or allophonic
variants.
Aspirated and unaspirated /p/ are allophones or allophonic
variants
Other examples:
/g/: “geese” vs. “gone”
/t/: “tap” “kitten” “button” “eighth” “fatty”
/E/: “tomato” vs. “potato”
The /g/ of “geese” and the /g/ of “gone” are allophones: Same
phonemic/phonological/linguistic category; different phonetic
realizations of the category.
Compare /l/ of “Lee” vs. “law”.
These distinctions vary across languages. Differences
which are allophonic in one language may be phonemic in
another, and vice versa.
9
How to tell whether two speech sounds are members of
the same or different phoneme class
Are /p/ and /b/ allophones of the same phoneme class, or are
they members of different phoneme classes?
Different. Here’s the test: Can we find a pair or words with
different meanings, where this difference in meaning is
conveyed by the /p/-b/ difference? Yes. Many.
pin-bin, pat-bat, pan-ban, pill-bill, pace-base, peek-beak …
So, /p/ and /b/ are different phonemes, not allophones of the
same phoneme category.
What about [ph] vs. [p]; i.e., the aspirated /p/ in “pot” versus the
unaspirated /p/ in “spot”. Can we find a pair of words in which
an aspirated /p/ means one thing while an otherwise identical
word with an unaspirated /p/ means something else?
The fact that a sound occurs in a language does not mean
that it has the status of a phoneme. Vowel nasalization:
compare the vowels in “pad” and “man” (and notice what you’re
velum is doing): [pQd] [mQ)n] (tilde = nasalized)
But, vowel nasalization is predictable in English – it occurs
whenever a vowel precedes a nasal consonant. The
presence vs. absence of nasalization never signals a
difference in meaning; i.e., it is not contrastive.
So, in English, [Q] and [Q]) are allophones of one another.
Not true in all languages; e.g., in French, Portuguese, & a few
other languages, differences in word meaning can be
signaled based on whether the vowel is nasalized or not
– just as in pin vs. bin in English.
French: “beau” (good looking; [bo]) vs. “bon” (good; [bo]) )
So, in French, are [o] and [o]) allophonic variants of one
another?
In English, are [o] and [o]) allophonic variants of one another?
Central idea: Contrast. Does a distinction serve a contrastive
function? If the answer is yes, then it’s phonemic.
Otherwise, we’re talking about allophonic variation.
Terminology
In English, are [o] and [o]) :
1. allophonic variants or allophones of one another
2. phonetically different but phonemically/
phonologically/linguistically equivalent
In French, are [o] and [o]) both phonetically distinct and
phonemically/phonologically/linguistically distinct.
10
Summary
A phonemic or phonological type is an abstract linguistic
category that can be phonetically realized in different ways.
These phonetically different but phonologically/phonemically/
linguistically equivalent realizations of phonemes are
called allophones of the phoneme category.
Dog
(abstract category, analogous to a phoneme)
Physically distinct but equivalent members of the abstract
category dog. These are analogous to allophones of a
phoneme category.
/t/
(abstract phoneme type)
Physically distinct realizations of the abstract category /t/.
These are analogous to physically distinct but equivalent
members of the category dog. These are allophones of /t/.
[phAt] [phAt|] [thAp] [stAp] [kIRi] [b/n]`
“pot” “pot” “top” “stop” “kitty” “button”
(released) (unreleased) (aspirated) (unaspirated) (flap) (glottal stop,
nasal release)
Last point: Sound types that are allophonic variants of a
single phoneme class in one language may be separate
phoneme categories in another language.
English French
/o/ /o/ /o/)
[o] [o]) Two distinct phoneme classes
“boat” “moan”
Nasalized/non-nasalized
allophones of /o/
11
Pragmatics
The module referred to as pragmatics may or may not be properly viewed
as part of the linguistic system, but it clearly plays a major role in language
comprehension.
Customer: Is my prescription ready?
Pharmacist: Yes.
Customer: Can you get it for me?
Pharmacist: Yes.
Customer: Will you get it for me?
Pharmacist: Yes.
Customer: I have a baseball bat. I’ll use it.
Pharmacist: I didn’t know that.
Customer: Get it for me. Now.
Pharmacist: OK. Why didn’t you say so when you first came in?
• What aspect of language is the dense pharmacist having difficulty
with?
• Phonology? Syntax? Semantics?
• Grammatically, what type of sentences is the first utterance (i.e.,
declarative, interrogative, imperative, etc.)?
A Short Story
Janie heard the jingling of the ice cream truck. She ran
upstairs to get her piggy bank. She shook it till some money
came out.
• Roughly how old is Janie?
• Does the money consist of coins or paper currency?
• What is Janie likely to do with the money?
Where in the language of the story do we find the answers to
these questions?
If we don’t get this information entirely out of the language,
where does it come from?
A Shorter Story
Tyler brought a six pack to the party. His mother found out
about it.
• Roughly how would you guess Tyler is?
• Six pack of what?
• What do you think the mother’s reaction was?
A Really Short Story
Bill: I’m leaving you.
Louise: Who is she?
• What is the story underlying this conversation?
• What do you think Bill means by “leaving”? Running out to
gas up his car? Going out to pick up the dry cleaning?
• How are you able to reconstruct a story based on two 3- to
4-word sentences?
• Is it your linguistic knowledge that allows understand what
is going on here?
12
Language vs. Speech
Last point: I’ve been talking about language and speech as
though they were the same thing. Not.
All speech is language, but not all language is speech. Two
major counterexamples:
• Written language (different in some important ways from
spoken language, though it’s still language)
• Sign language (e.g., American Sign Language)
Sign language is not a stripped down, impoverished
version of spoken language. It conforms to the universal
grammar and contains all same elements as spoken language:
structure-dependent rules, agreement, head-first/head last
constraints, even “movement” analogs of phonological rules.