Transcript Prescriptivism and Descriptivism

Prescriptivism and Descriptivism
September 19, 2012
Current Work with Bonobos
• After Nim Chimpsky, funding for primate language
studies mostly dried up.
• …although a few experiments went on.
• One project involves bonobos, a sub-species of
chimpanzees.
• Bonobos Sherman and
Austin have also been
trained to use lexigrams.
• Kanzi learned just by
watching Sherman and
Austin’s training!
• But the Bonobo project is now in trouble—check out:
http://news.iowapublicradio.org/post/bonobo-hope-great-ape-trust-sanctuary
Mission Objectives
1. Wrap up Prescriptivism ~ Descriptivism
2. Try to figure out how language can be creative.
•
The previous problems with prescriptivism:
1. Confusion about application of prescriptive rules
•
(they’re not natural)
•
Hypercorrection
2. Standards can shift over time
3. Prescriptive rules form a poor understanding of natural
language.
Problem #3: Missing Patterns
• Prescriptivist rules do a poor job of accounting for many
of the patterns we find in natural language.
• Here’s one prescriptive rule which misses a consistent
pattern:
• “Incorrect”: I feel bad (about the accident).
• “Correct”: I feel badly (about the accident).
• Why? The verb “feel” should be modified by an adverb
(“badly”), not an adjective (“bad”).
• But is bad/badly modifying the verb or the subject of the
sentence?
Linking Verbs
• How about these examples?
• Bob is happy.
(*Bob is happily.)
• Susie looks hot.
(*Susie looks hotly.)
• The water seems fine.
(*The water seems finely.)
• I feel sleepy.
(*I feel sleepily.)
• James Brown feels good. (*James Brown feels well.)
• The verbs in these sentences are known as linking
verbs.
• They connect the subject to some property
describing the subject.
• (They do not modify the verb itself.)
Different Standards
• Rules for a standard form of a language…
• Normally describe the variety of language used by
the group in power.
• Other forms of the language are non-standard.
• And are often identified with social, regional or ethnic
groups.
• Linguists have discovered that all forms of language
(standard or not) are rule-based and orderly.
•  Non-standard forms of the language are not simply
mistake-ridden versions of the standard form.
•  There is no linguistic reason to consider one variety of
language superior to another.
Quick Write:
Appalachian English
• Appalachian English is a
variety of English
traditionally spoken in the
Appalachian mountains.
• Developed (and
maintained) unique
features due to isolation
from outside communities.
• One interesting feature:
• a-prefixing…
“a” Prefixing Summary
• [a-] form cannot be a noun (#1 and #5)
• [a-] form cannot be an adjective (#2 and #6)
• [a-] form cannot be preceded by a preposition
(#3 and #7)
• first syllable of [a-] form must be stressed (#4 and #8)
• Note: people often consider speakers of Appalachian
English to be unsophisticated
• …but the proper use of the [a-] prefix involves a
relatively complex set of conditions.
AAVE
• Another variety of English that has (traditionally) been
low on the prestige scale is African-American Vernacular
English (AAVE).
• a.k.a. Black Vernacular English (BVE), Ebonics
• Predominantly spoken by African-Americans
• but not all African-Americans…
• and some others, as well.
• AAVE has a variety of interesting features...
• some familiar: multiple negation, ain’t as an auxiliary
• others are less familiar…
AAVE Verbs
• Verb conjugation: third personal singular verbs lack an [-s]
marker.
• Ex: He look, it do, she have
• “Paradigm leveling”
• = making a set of related forms more uniform
• (similar to “he don’t”/”she don’t”)
• Under certain conditions, the verb “to be” can be deleted.
• Ex: you so crazy, she workin’, he lucky
• In the same conditions, “to be” can be contracted in
standard English:
• You’re so crazy, she’s working, he’s lucky…
To Be Deletion
• What are the right conditions for deletion/contraction?
AAVE
Standard English
You so crazy.
You’re so crazy.
*He as nice as he say he. *He’s as nice as he says he’s.
*Here I.
*Here I’m.
They mine.
They’re mine.
*How beautiful you.
*How beautiful you’re.
• The verb needs to link the subject to something after it.
AAVE: Habitual Be
• AAVE also has a form of “to be” that standard English
does not.
• “habitual” be
• Habitual be expresses something that the subject does
on a regular basis.
• Examples:
• He be working at Tim Horton’s.
• She be late. (= She is usually late.)
• She late. (= She’s late (right now).)
• Do you be tired? (=Are you often tired?)
Descriptive Benefits
• Language tends to operate in patterns, even if they are
non-standard.
• Important: Appalachian English and AAVE speakers
are not just speaking English with mistakes.
• Descriptive linguistics enables us to understand how
those patterns work.
• Even if you want to change the world, you’re better off
understanding how it works to begin with.
• History of economics analogy.
To Be Fair
• Standards are useful because they provide a single form
of the language to teach to non-native speakers.
• They help establish uniformity in the written language.
• They can help clear up confusions.
• for instance: supposably
• They also help to distinguish those who have mastered
the arbitrary rules from those who haven’t.
• (for better or worse)
• Otherwise:
• They are not useful for (scientific) linguistic analysis.
Linguistic Creativity (again)
• One of the crucial design features of language was creativity
(or productivity).
• Charles Hockett:
“Language users can create and understand completely
novel messages.”
“In a language, new messages are freely coined by
blending, analogizing from, or transforming old ones. This
says that every language has grammatical patterning.”
“In a language, either new or old elements are freely
assigned new semantic loads by circumstances and
context. This says that in every language new idioms
constantly come into existence.”
• How is it possible for human beings to do this?
To Infinity and Beyond
• Last week, we found out that honeybees can produce
a variety of different “dance messages”.
= “Food source beyond 65 feet, fly at 0
degree angle with the sun.”
= “Food source beyond 65 feet, fly at 45
degree angle with the sun.”
To Infinity and Beyond
• The number of different messages the bees can produce is
limited only by the number of angles they can differentiate:
• “Food source beyond 65 feet, fly at 1 degree angle with the
sun.”
• “Food source beyond 65 feet, fly at 2 degree angle with the
sun.”
……………
• “Food source beyond 65 feet, fly at 359 degree angle with
the sun.”
• Q: Can the bees dance at angles they haven’t seen before?
• If so, how?
Different Infinities
• What kind of infinities exist in human language?
• Note that we can say (translations of) everything the
bees can say:
Fly at a 1 degree angle with the sun.
Fly at a 2 degree angle with the sun.
……………
Fly at a 359 degree angle with the sun.
• We can get as detailed as we want to about it, too:
Fly at a 45 degree, 13 minute, 27.6685 second angle
with the sun.
Infinity + 1
• In addition to the infinity of things the bees can say, we can
say other things, too.
• Examples (borrowed from Ray Jackendoff):
A numeral is not a numbskull.
A numeral is not a nun.
A numeral is not a nunnery.
……………
A nun is not a nursery.
……………
An oboe is not an octopus.
Linguistic Infinities
• These are uninteresting, but novel sentences.
• In order to understand them, you must know the rule by
which they are constructed.
• Rule:
[Sentence] = A X is not a Y.
• Point:
• Knowledge of rules is more abstract than just
knowledge of sentences.
Language Model #1
• In this model, all we
A nun is not a
nursery.
Fly at a 45 degree
angle with the sun.
I like linguistics.
“know” are the individual
sentences we can use in
language.
• (no rules)
• This is a good enough
model to describe the
vervets’ (or prairie dogs’)
“language”.
Language Model #2
A X is not a Y.
X at a Y degree
angle with the Z.
X likes Y.
• In this model, we
“know” all the rules we
can use to combine
words to form sentences
in a language.
• This is a good enough
model to describe the
bees’ “language”.
• Is it good enough for
human language?
What do you think?
• No. There are even bigger infinities.
• Check out these sentences:
Bill thinks that Beth is a genius.
Sue suspects that Bill thinks that Beth is a genius.
Charlie said that Sue suspects that Bill thinks that Beth is a
genius.
Jean knows that Charlie said that Sue suspects that Bill
thinks that Beth is a genius.
ad infinitum...
• Some “real” examples:
How many rules do we need?
1. X verbs that Y is a Z.
2. W verbs that X verbs that Y is a Z.
3. V verbs that W verbs that X verbs that Y is a Z.
•
and so on…
•
Q: Can we store all these patterns in our heads?
•
A: No, because no matter how many we store, there is
always a longer one…
•
So how do we know all of these sentences?
Language Model #3
S = X likes Y.
• Jackendoff: “We know
not just patterns of
words, but patterns of
patterns.”
S = A X is not a Y.
S = X verbs that S.
• This is how we can be
infinitely creative with a
finite set of rules.
Check it out
• Included among the infinite number of things we can say is
a lot of complete nonsense.
• Examples (from Chomsky and Lewis Carroll):
• Colorless green ideas sleep furiously.
• I’m memorizing the score of the sonata I hope to
compose someday.
• ‘Twas brillig, and the slithy toves
Did gyre and gimble in the wabe…
• Check out the postmodernism generator:
• http://www.elsewhere.org/pomo/
What’s the difference?
• Nonsense sentences work because they fit in with the
patterns formed by the sentences that actually do make
sense.
• (and that we use every day)
• Compare with the following:
• Large green lizards sleep soundly.
• I’m memorizing the score of the sonata I hope to
perform someday.
• ‘Twas evening, and the slimy toads
Did squirm and wiggle in the cage…
What’s the difference? (part 2)
• But the following sentences don’t work at all:
• Green sleep ideas furiously colorless.
• I’m memorizing the perform of the score I sonata to
hope someday.
• Brillig and, slithy and the toves
Wabe gimble in the gyre and did…
• Note: just because we can say an infinite number of
things, we can’t just say anything…
Technical Terminology
• The set of rules that we know for creating sentences in a
language is the grammar of that language.
• The rules of grammar that we know are very abstract.
(patterns of patterns)
• Strings of words which do not adhere to these rules are
ungrammatical.
• Q: If these rules are so abstract, how did we figure out
what they are?
• How do we learn language?
Beneath the Surface
• Note: we learn the language that we hear as we grow up,
but…
• We never hear the rules.
• We can only learn from examples.
• Our knowledge of language is sub-conscious.
• Analogy: driving a car.
• This knowledge is difficult to characterize.
• (It is not explicitly taught to us.)
How is that possible?
• Theory: language acquisition is so hard that we can’t do it
by just observing other language users.
• (we need help)
• Claim: every human being has a “Language Acquisition
Device” (LAD)
• LAD = innate knowledge of language.
• The LAD helps us learn language as we grow up.
• Interacts with experience.
Predictions
•
The LAD theory makes some important predictions.
1. Universal Grammar (UG)
• All languages should share certain features in
common
• …due to the workings of LAD.
• A basic example:
• All languages have nouns and verbs.
2. Poverty of the Stimulus
• There should be properties of language that people
“know” without ever having experienced them.
A More Complicated Example
• How do you turn the following sentence into a yes/no
question?
• The boy who is sleeping is dreaming of a new car.
• = Is the boy who is sleeping dreaming of a new car?
• Not: *Is the boy who sleeping is dreaming of a new car?
• “The boy” is linked to the second “is”.
• Kids understand this connection without ever being
taught about the link.
• They never form the question the wrong way.
• Think: baby turtles crawling towards the ocean.
Recursion
• Recursion = another universal property of language?
• which is unique to humans?
• (Noam Chomsky thinks so.)
• Remember, recursion =
• involving a procedure that can refer to itself.
• Ex: an English sentence may consist of:
• [Noun] [verbs] that [sentence].
• With this rule, we can make sentences like:
• Jean knows that Charlie said that Sue suspects that
Bill thinks that Beth is a genius.
• Sentences like this could be infinitely long…
Limited Infinities
• However: there are limitations on how much we can
remember.
• This means that a sentence like: “I don’t know if Ross
suspects that Monika thinks that Chandler hopes that Joey
supposably believes that Phoebe heard that…”
• couldn’t really go on forever.
• Check out another kind of recursion:
• The boy scared Mary.
• The boy that the dog bit scared Mary.
• How about:
• The boy that the dog that the cat scratched bit scared
Mary. (?!?)
Competence vs. Performance
• An important distinction:
• Linguistic Competence:
• What a (native) speaker knows about a language.
• Linguistic Performance:
• How language is actually used in speech production
and comprehension.
• Word strings that are ungrammatical violate the rules of
linguistic competence.
• Other strings are impossible to say (or understand)
because of performance limitations.
Performance Problems
• Note: it is not impossible for native speakers of a
language to make mistakes.
• Ex.: slips of the tongue.
• You have hissed all my mystery lectures.
• = You have missed all my history lectures.
• My wife made me some banana bed yesterday.
• = My wife made me some banana bread yesterday.
• Stammering, pauses, hesitations.
• What matters (for grammar) is not what you actually do so
much as what you think about what you do.