5 words
These
"atoms" of meaning are used to create an unlimited range of meanings by their
contribution to words, phrases, and sentences. (cf. "combinatorial semantics")
In
English we have stems and inflections. There are words that can "stand alone"
like cat and dog as well as inflections -- morphemes that can
only be used when attached to other morphemes. THese are sometimes known as
"bound" morphemes, e.g. /-s/ or /-ed/ in contrast to the "free" morphemes like
cat and dog.
Words
have hierarchical structure analogous to phrase and sentence structure. For
example nouns (N) can be formed by taking a Nstem and inflecting it. Thus
dog+pl is dog+/s/ using the rule N-->Nstem + Ninf. (see Pinker 128-9). (The
arrow --> is read as "consists of" just as in syntactic rules defining
phrases and sentences, e.g.
NP-->det+N+(S). as in (the boy (that played the violin)).
Many languages, eg. Turkish or the Eskimo in the video, have very long words
corresponding to sentences built up out of many stems and inflections. English
does little of this but remember our discussion of the sentence (and VP and V)
"Stop!"
While
the morphological rules determine how to create new words from existing
morphemes, the phonology of the language determines how those new words are
pronounced. Compare forming the plural of cat and that of dog
and fish. (The rule takes the voicing from the final consonant of the
stem and modifies the pronunciation of the plural morpheme /-s/, carrying over
the voicing quality. If the consonant is voiced (short VOT[1]), the plural morpheme is voiced [-z]; if the consonant is
unvoiced (long VOT), the plural is not voiced [-s}. If the consonant is a
sibilant like -sh, or -dg, then a reduced vowel is appended followed by the
voiced plural, e.g. fishes, judges.
These
morphemes are "stored" in the lexicon or mental dictionary. For each there is
information about its meaning, syntax, and pronunciation. There may be other
information available including something about the contexts of use of that
word, associations with that word, and frequency of use information.
The
phonology of a language in a sense defines the possible words of the language.
We all know that while "stick" and "zarf" is a possible word in English,
"ftick" or "zraf" are not. Moreover we know the plurals "sticks" and "zarfs"
even though we may not know the meanings or syntactic role (part of speech).
Linguists talk about "morpheme structure" rules -- the structure of possible
morphemes and phonological rules that blend together strings of morphemes as in
the plural rule -- which we can see also applies to formation of possessives
and third person singular present tense on verbs, e.g. its the fish's tail, and
"He fishes for carp."
English, for example, has lots of CVC words but few native CVCV ones -unlike
many if not most languages. We also have lots of consonants clustered
together, eg. strip. Words borrowed into English remain identifiable as
"foreigners" because they do not follow typical English morpheme structure
patterns acquired by children long before they have much vocabulary.
See
our project results, Pinker, and my notes on dictionaries. Fluent speakers
use tens of thousands of words and probably know in some sense several more
tens of thousands in their passive or "recognition" vocabularies.
While
there are many classifications, two major types are of great importance --
closed class or function words and open class or content words.
Some
words and inflection morphemes are known by every speaker of English, e.g.
I, the, of, and, and the inflections /-ed/
and /-s/. There are several hundred of these, usually acquired in early
childhood and used very frequently by all speakers of English. These are
sometimes referred to as "closed class" morphemes because no new function words
have come into English for perhaps a thousand years -- quite different from the
content words.
Other
words like dog, fish, and happy -- carry much of the semantic content of our
linguistic expressions. There are tens of thousands of these content words and
new ones are always coming into the language, e.g. internet. In contrast to
the function words, no speaker knows all the content words of their language.
Pinker
talks about "listemes" -- irregular forms or strings of words whose meaning is
not derivable from their individual meanings. Expressions like "kick the
bucket" have a taken on a new meaning.
One way a new word is formed is in place of a lengthy expression, e.g.
computers that are networked --> internet. This is known as
lexicalization and is continually occurring as cultures change.
Generally a single form replaces the phrase but sometimes the phrase remains
but takes on a global meaning.
Psycholinguists
debate to what extent derived words may actually become established in users
lexicon. Thus while employer is derivable from "employ + -er", it may be seen
as a wholistic form by many speakers unlike for example, "ungluer" or some such
"agent" noun formed from a verb form. (See Pinker for examples.)
These
are qualitatively different from content words in a number of ways including
phonology and as "pure referring expression." See the Semenza and Zeitlin
reading in regards to Frege; also Burke et al (1990) on "tip of the tongue"
phenomena.
These
get their referential values from the expressions and contexts in which they
are used, e.g. Look at that!, I am typing very quickly now.
We
continue to acquire new words as adults, e.g. in taking new courses, reading
about new topics. Children however learn at an astounding rate -- maybe 5 to
10 words a day with no overt teaching. They must extract the meanings and
syntactic info from their personal experience with each word, often on the
first encounter.
This rate is a striking difference from the gradual rate apes, etc. seem to
acquire human based symbols or signs. The most anyone has claimed for apes is
about 500 symbols -- about a tenth that of a four or five year old human
without any instruction or spoonfuls of jam!
How does this happen? One factor is phonology which essentially lays out the
structure of morphemes (see above.) Thus even though you may not have
encountered zarf until a few days ago, you had a word-form schema for it
all ready to be filled in.
Another
seems to be the instinctive ability of children to "guess" what a speaker is
talking about. Quine (in Pinker, video) calls this the "gavagai" problem,
Being
a fluent speaker of English is more than just having a good vocabulary. WE
have to use it rapidly and accurately. We talk/listen at rates of 100 to more
than 200 words per minute. Imagine how fast our thoughts form, are encoded
into words and phrases, and our body moved. Conversely the listener must
extract the word and phrase and sentence structure from those movements,
meaning "computed" and evaluated in context -- i.e. comprehension.
Psychologists study these lexical processes using priming, associations, naming
time and lexical decision tasks, and other methods including collecting speech
errors.
One
notorious problem in using the lexicon is polysemy -- the fact that many forms
have more than one distinct meaning. English -- because of its minimal
inflections -- also has ambiguity of syntactic role -- what part of speech is
"Fish?"
Just
as commonly used phrases may take on a life of their own as lexical items in
the mental dictionary, commonly used words tend to get shorter -- television
> teevee, internet > net. Presumably this makes communication more
efficient.
(to
be discussed soon)
[1] Voice onset time (VOT) is the time
between the release of the stopped airflow as in [pat] or [bat] and the startup
(onset) of the vibrations of the vocal folds in the larynx (voicing).
Consonants with a relatively long delay between release and onset are said to
be "unvoiced" consonants (p,t, k, f, s) whereas those with a very short or no
delay are said to be "voiced" (b, d, g, v, z). Try these yourself, comparing
when you normally begin phonating (vibrating your larynx.) See acoustics notes.