Morphology Morphology
LIGN171: Child Language Acquisition LIGN171: Child Language Acquisition http://ling.ucsd.edu/courses/lign171 http://ling.ucsd.edu/courses/lign171
Morphology Morphology Morphology yields words with Morphology - - PowerPoint PPT Presentation
LIGN171: Child Language Acquisition http://ling.ucsd.edu/courses/lign171 http://ling.ucsd.edu/courses/lign171 LIGN171: Child Language Acquisition Morphology Morphology Morphology yields words with Morphology yields words with
LIGN171: Child Language Acquisition LIGN171: Child Language Acquisition http://ling.ucsd.edu/courses/lign171 http://ling.ucsd.edu/courses/lign171
Inflectional morphology
Changes words to fit specific context
English (regular) past tense (stem + -
ed)
walked, walks, walking
Some morphological affixes are productive
Apply to new forms entering the language
ed: faxed, emailed, : faxed, emailed, blick blick-
?
Others are less productive
Certain irregular past tenses
sing sang; ring sang; ring rang; spring rang; spring sprang; bring sprang; bring brought brought ( (brang brang); ); spling spling
(only unpredictable words are listed)
Bloomfield 1931
Chomsky 1965
walk walked walked
play played played
go go go-
ed? ?
“went went” ” blocks blocks “ “go go-
ed” ”
Irregular form blocks new, regular form
“flew flew” ” does not block does not block “ “flied flied” ”
*The pinch hitter flew out to center field
Doublet forms
dive -
dove or dived
Both forms need to be in the lexicon
Phonologically similar to real irregulars
squeeze -
squeezed or squoze squoze? ?
froze
glide -
glided or glid glid? or ? or glod glod? ?
hid or ride -
rode
Certain regular forms need to be stored,
2 mechanisms for language
Production by rule/grammar
Memorization in lexicon
How do you know what was used?
If the forms are different, it’ ’s not too hard s not too hard
If the forms are the same, it can be difficult
Individual differences
For different words
For different people
Redundancy provides easy means for compensation compensation
Damage to grammar -
Could compensate by memorizing in lexicon
Damage to lexicon -
Could compensate with grammar
Great if forms are the same (output is the same)
Problematic if forms are different
Regulars might be lost following damage to grammar
Irregulars might be lost following damage to lexicon
What would you produce instead?
Irregular past tense
dive -
dove
teach -
taught
Idiom
‘kick the bucket kick the bucket’ ’ = die = die
Some regulars
dive – – dived dived
squeeze -
squeezed
Regular past tense
live -
lived
teach -
teached
Literal meaning
kick the bucket
If produced by grammar
dive – – dived dived
squeeze -
squeezed
One possibility:
children's past tense errors (i.e. overregularizations
; goed goed) are most frequent at earliest stages of learning ) are most frequent at earliest stages of learning
Errors become gradually less frequent over time
What really happens (U-
shaped learning curve):
stage 1: perfection (almost)
stage 2: over-
regularization (not all the time, but a good deal more frequent than before) good deal more frequent than before)
stage 3: perfection (gradual transition from stage 2 to stage 3) stage 3)
Initially, children treat all all verbs as if they are irregular verbs as if they are irregular verbs, whose past tense forms must be memorized. verbs, whose past tense forms must be memorized.
They memorize the forms of the small number of verbs that they use well, and so they make few errors. use well, and so they make few errors.
Later, they learn that there is a regular past tense form, which is supplied by a rule which is supplied by a rule
at this point, there is a tendency for this rule to be overapplied
, leading to over leading to over-
regularizations.
Subsequently, the correct division of labor between regular and irregular past tense forms is established. regular and irregular past tense forms is established.
Dog dogs; horse dogs; horse horses; cat horses; cat cats cats
children children
knife knives knives
goose geese geese
deer deer; fish deer; fish fish fish
Emphasis on characteristic features features
Frequent features
Easily accessible features
Perceptual features
Common facts
No sets of necessary and sufficient conditions conditions
Graded structure – – best example is best example is ‘ ‘prototype prototype’ ’
Fuzzy boundaries
Where do ‘ ‘fruits fruits’ ’ end and end and ‘ ‘vegetables vegetables’ ’ begin? begin?
Morphological representations and representations and lexical forms are not lexical forms are not separate separate
Simple and complex forms comprise forms comprise Schemata Schemata
Probabilistic, prototype structure structure
/s/, /z/, /Iz Iz/ /
/z/ is very rare in singular forms (lens)
/Iz Iz/ is even more rare (kermes) / is even more rare (kermes)
/s/ is fairly common (fox, box, boss, etc.)
High cue strength: interpret singular form as already plural
Cue strength = sum effect of four cues (each vaguely divided into high, mid, low strength) divided into high, mid, low strength)
Salience
How acoustically prominent is the marker
/s/, /z/, /Iz Iz/ all highly salient / all highly salient
Type frequency
Number of nouns that take a particular form
/s/: mid; /z/: high; /Iz Iz/: low /: low
Cue validity
How reliably does the sound indicate plurality?
/z/: mid; /Iz Iz/: high; /s/: low /: high; /s/: low
Iconicity
Additive morphemes (especially syllabic) are highly identifiable
/Iz Iz/: high; /s/: mid; /z/: mid /: high; /s/: mid; /z/: mid
Symbolic rules
Phonology of stem is irrelevant irrelevant – – rule rule applies applies
Schema
Phonology of stem is relevant relevant – – rule may not rule may not apply? apply?
6-
7 year olds
Produced “ “wugs wugs” ” 97% 97%
67 67 33 33 / /-
Iz/ / niz niz 59 59 36 36 / /-
Iz/ / kazh kazh 52 52 38 38 / /-
Iz/ / gutch gutch 61 61 39 39 / /-
Iz/ / tass tass 14 14 86 86 / /-
z/ cra cra 16 16 80 80 / /-
z/, /-
s/ heaf heaf 10 10 90 90 / /-
z/ tor tor 8 8 92 92 / /-
z/ lun lun 3 3 97 97 / /-
z/ wug wug
0 % Correct % Correct % Expected Expected Word Word
Performance was worse when /Iz Iz/ was expected / was expected because / because /Iz Iz/ is harder than /s/ or /z/ / is harder than /s/ or /z/
Rule: “ “a final sibilant makes a word plural a final sibilant makes a word plural” ”
Repeated Morph Constraint
Languages do not repeat morphemes
*dogses dogses; * ; *horseses horseses; * ; *catses catses
Novel forms that appear to already be plural (yes or no) won (yes or no) won’ ’t add the affix t add the affix
The greater the degree of similarity to a plural form, the more likely it will be left plural form, the more likely it will be left unchanged unchanged
/z/ = plural allomorph with highest cue strength strength
/s/ = plural allomorph with lower cue strength strength
{ch ch}, { }, {zh zh} similar to plural allomorphs: } similar to plural allomorphs: fricative, continuant, sibilant ( fricative, continuant, sibilant (“ “hissing hissing” ”) )
/f/ is a fricative, and continuant, but not a sibilant sibilant
/a/ and /r/ are continuant, but not fricative
/n/ and /g/ negative on all three counts
67 67 niz niz 61 61 tass tass 59 59 kazh kazh 52 52 gutch gutch 16 16 heaf heaf 14 14 cra cra 10 10 tor tor 8 8 lun lun 3 3 wug wug
0 % Word Word
It is added by rule
Whether or not affix is added
(avoid violating repeated morph constraint)
German has many ways to form a plural
Stem + -
e
Fisch Fische Fische (fish) (fish)
Stem + -
(e)n e)n
Bauer Bauern Bauern (farmer); T (farmer); Tϋ ϋr r T Tϋ ϋren ren (door) (door)
Stem + -
er
Geist Geister Geister (ghost) (ghost)
Stem + -
s (claimed default – – “ “minority minority” ” default) default)
Park Parks (park) Parks (park)
Stem + -
Adler Adler (eagle) Adler (eagle)
Stem Umlaut
Vater V Vä äter ter (father); (father); Tochter Tochter T Tö öchter chter (daughter) (daughter)
Stem Umlaut + -
e; stem umlaut + -
er
Kuh K Kϋ ϋhe (cow); Wald he (cow); Wald W Wä älder lder (wood) (wood)
Articles also signal plurality
Der SING SING Die PLURAL ( Die PLURAL (masc masc) )
Die SING Die PLURAL (fem) Die PLURAL (fem)
Das SING Die PLURAL (neuter) Die PLURAL (neuter)
l l m m l l l l Umlaut Umlaut h h l l l l h h
er h h l l m m h h
e m m h h l l h h
s h h h h h h h h
(e)n e)n Iconicity Iconicity Cue Cue Validity Validity (Type) (Type) Frequency Frequency Salience Salience Marker Marker
If similar to existing plural, add -
0, else
Add default /-
s/ marker to all forms?
Add appropriate marker depending on stem form?
Graded effects consistent with schema/prototypes in both English and schema/prototypes in both English and German German
2 4 6 8 10 12 14 16 18 20 50-59 60-69 70-79 80-89 90-99
Mean: 82 Standard deviation: 11.6 Mean: 82 Standard deviation: 11.6
11.
If I take the nonsense words “ “bidaku bidaku” ”, , “ “padoti padoti” ”, and , and “ “tupiro tupiro” ” – – which of the following which of the following sequences of syllables has the highest transitional probability sequences of syllables has the highest transitional probability within the string: within the string: “ “bidakupadotitupirobidakutupiropadoti bidakupadotitupirobidakutupiropadoti” ”? ? a. a. bi bi-
da b. b. ku ku-
pa c. c. ro ro-
pa d. d. both b) and c) both b) and c) 12.
If word learning depends on a dedicated system, which of the following should be true? lowing should be true? a. a. fast fast-
mapping should not apply to facts or words b. b. fast fast-
mapping should apply to words and facts, but not dance routines c. c. fast fast-
mapping should apply only to facts, not words d. d. fast fast-
mapping should apply only to words, not facts 13.
The assumption that children can’ ’t understand relational terms until they understand the t understand relational terms until they understand the things being related is made by proponents of: things being related is made by proponents of: a. a. perceptual bootstrapping perceptual bootstrapping b. b. logical bootstrapping logical bootstrapping c. c. syntactic bootstrapping syntactic bootstrapping d. d. phonetic bootstrapping phonetic bootstrapping
a.
children exploit sentential information to learn about the meaning ng
b.
grammatical function words are difficult to perceive c.
lexical and grammatical development displays a non-
linear growth curve growth curve d.
both a) and b)
with the CDI have shown that: CDI have shown that: a.
verbs are acquired in that language earlier than nouns b.
verbs are acquired in that language earlier than function words c.
development of nouns grows faster than development of verbs (for the first 50 words) (for the first 50 words) d.
both b) and c)
Acquisition
Language learning is just like learning anything else like learning anything else – – it depends on intelligence it depends on intelligence and an ability to solve and an ability to solve problems problems
Development
The brain is programmed for language, just like for language, just like learning to walk learning to walk – – it is like a it is like a form of physical growth form of physical growth
Compromise? How much is learned and how much built in? Compromise? How much is learned and how much built in?
Lexical Constraints Hypothesis
Cognitive processes that constrain meaning
Mutual exclusivity
In a given language an object cannot have more than one name name
A child will not expect synonyms (car, auto)
Fast mapping
Novel words map onto objects which do not already have a name name
Whole object constraint
A novel word refers to a whole object, not its parts or features (color, shape, texture, etc.) (color, shape, texture, etc.)
Taxonomic constraint
A word refers to a member of a category