Dr. Sharon Armon-Lotem
Early verbs
Early verbs – meaning and argument structure
Reading: Armon-Lotem, S. and R. A. Berman. 2003. The Emergence of Grammar: Early Verbs and Beyond. Journal of Child Language 30.4: 845-877
Verbs- “lexical items which express activities, changes of state, etc.”
Early Verbs- “verb forms that are the first to appear in speech of children at the one-word stage and in the transition to word-combinations.”
Questions?
· Why are verb acquired later than nouns?
· What can we learn from early verbs?
· How is verb meaning acquired?
· Does the acquisition of verbs follow the same word-meaning constraints as for nouns?
· How does the context influence the acquisition of verbs?
· How are the argument structure and selectional restrictions of a verb acquired?
· How is the verbal morphology acquired?
What can we learn from early verbs?
LIOR, girl, aged between 1;5 and 1;6, in interaction with her mother
Breakdown of 120 Verb Forms by Mood/Tense and Stripped Stem Forms [in percentages]
Stripped / Infinitive / Imperative / Present / Past / Future46 / 6 / 19 / 10 / 15 / 5
· Children have a large repertoire of verbs even before they start combining words, and these represent a range of predicate types: syntactically, both transitive and intransitive, and semantically, verbs which express activities, events, and states.
· These verbs: 1. Lack overt syntactic marking of argument structure.
2. Fail to make productive use of inflection.
3. Show no ability to use morphological verb-pattern alternation
Nonetheless, early verbs play an important role in children's early grammar and in the interface between the development of syntax, morphology, semantics and the lexicon.
How is verb meaning acquired? Syntactic vs. semantic bootstrapping
· How do children acquire words? The acquisition of a morpheme in the lexicon combines syntactic and semantic information.
· How do children pair concepts with their phonological realization? Nouns can be acquired by pairing strings of sounds with concepts inferred from the world using constraints on word learning – whole object, mutual exclusivity (Markman 1994), e.g., by the naming game.
· Bloom 1994 - nouns are acquired by syntax-semantics mapping.
· How are verbs acquired? Would it be in the same way as nouns?
How are verbs acquired?
Many verb meanings are compatible with most events
What is the meaning of bix?
1. Look at the bix.
2. a. The boy is bixing
b. The boy is bixing the flowers
c. The boy is bixing the flowers to the girl
3. The girl is bixing the flowers from the boy
What is acquired when verbs are acquired? Lexical entries for verbs
Fly, N, two winged insect,-human, +animate, +count … / Fly, Vi, move through air with wings
(About, away, off, out)
Vt, Control an aircraft, transport in air
craft
[NP] Patient
Want, Vt, desire, wish
[NP] Patient
[to VP]
[NP to VP]
Give, Vbt, transfer possession, hand over,
[NP, PP] Patient, Goal
[NP, NP] Goal, Patient / Donate, Vbt, present as a gift
[NP, PP] Patient, Goal
Gleitman (1994)
· Children infer verb meaning by using the syntactic arguments which appear in the sentence (subcategorization frames).
· Blind children learn verb meaning without “observing” the situation.
Syntactic bootstrapping (Fisher et al 1994): Structure dependent learning. Syntactic cueing of word meaning. Children rely on a verb’s syntactic subcategorization frames to learn its meaning. The learner represents the linguistic input that is to be paired with the extralinguistic input (perceptual, conceptual & pragmatic), acquired by inspecting ongoing events, by a parse tree within which previously learned nominals as well as the novel verb occur.
Semantic bootstrapping (Pinker 1994): Semantic cueing of syntax / word meaning. A verb’s content is learned from its situation of use. The semantics of at least some verb forms are acquired without syntax.
Lexical reconciliation (Grimshaw 1994)
Fundamental claim: There is a principled relation between the syntax and the semantics of a verb.
“The range of syntactic configurations associated with a verb is highly predictable from its semantics, once parametric syntactic variation is taken into account.”
Mapping of lexical semantic representation onto syntactic representation:
Lexical semantics of a predicate > argument structure
Argument structure + parametric properties of phrase structure > s-structure
Many verb meanings are compatible with most events
1. a. The boy is ----ing
b. The boy is ----ing the flowers to the girl
2. The girl is ----ing the flowers from the boy
There is no situation in which (2a) is true while (2b) is false
2. a. We killed the dragon
b. The dragon died
> Observation of the world is not enough
> Subcategorization frames contain critical information
3. a. He put the book in his room
b. He wrote the book in his room
4. a. He became a doctor
b. He shot a doctor
> Only a few cases of one-to-one mapping from subcategorization frames to meaning
> Analysis across sentence types is required for sets of frames
5. a. I know her
b. I know that she is here
6. a. ani makir ota
b. ani yodea she hi po
> Clustering of different senses under a single morpheme varies across languages.
> A predictable relationship does not hold
> Without knowing about meaning it is not possible to know how to group subcategorizations together
> We need a way to use the two sources of information > Reconciliation
Reconciliation
· Semantic-to-syntactic mapping provide predictive mechanism. Semantics predicts syntax where UG makes it possible
· The subcategorization structure provided by the syntax serves as a checking mechanism. Syntax eliminate wrong semantic candidates where possible
Acquiring verbs
Tomasello, M. & A. Kruger, 1992. Joint attention on actions: acquiring verbs in ostensive and non-ostensive contexts, Journal of Child Language 19, 311-333
· In order to acquire a word it should be mapped onto a referential context.
· How? Joint attention (an ostensive context)
· Can this work for action verbs?
· Can we point to actions? Do we point to actions?
Activities: Play session begins with child and mother sitting on floor in family's play area. Village mat is lying on the floor with a large sack of toys lying nearby. Activities include play with train, vehicles and road signs on playmat
*MOT: look what mummy found today.
*MOT: right. what other cars do you want ?
*MOT: do you want your Weetabix tanker ?
*MOT: you saw one when you were coming home did you ?
*MOT: and what was your tanker like coming home from school ?
*MOT: right here's the A_A man coming to mend the cars .
*MOT: okay where's he going ?
*MOT: He’s mending the points is he ?
*MOT: okay what else do you want ?
*MOT: lots of people getting on the bus ?
*MOT: right, what else do you want ?
*MOT: what have you tied them together for?
*MOT: I think that red one there belongs to the transporter really
*MOT: well that looks he's got BP, so that means petrol .
*MOT: so that they can go to the petrol station over there .
*MOT: have you got enough things out ?
*MOT: where are all these cars going ?
*MOT: is there none of them going to the station ?
*MOT: do you want some little cars to be parked at the station ?
*MOT: all right yes put it there if you want .
*MOT: any more space left in the car park ?
*MOT: are you going to park that one in the station then as well ?
*MOT: oh is he picking all his letters up from the station ?
*MOT: right now then you tell me what all these people are doing on this
*MOT: well where's the sign that's got children crossing the road ?
*MOT: put it there to say children crossing the road from school .
*MOT: what'll happen if they go near the river ?
· Adults usually use verbs in non-ostensive context.
· Naturalistic findings - 60% of verbs were used for impending actions. Children were most responsive to such verbs.
Experimental procedure with nonce verbs in impending, on going and completed modes, presented to three groups of children (ages 1;10-2;2).
· After presenting the verb to the child, he was tested on both comprehension and production.
o Production - impending context was best.
o Comprehension - impending and completed are equally good.
· Verbs are more easily acquired in non-ostensive context (when referring to impending and completed actions rather than ongoing actions). Why?
In class exercise - How are the argument structure and selectional restrictions of a verb acquired?
From: Sharon Armon-Lotem (1997). The minimalist child: Parameters and functional heads In the acquisition of Hebrew.
Table 1 - The development of argument structure for the verb ‘fall’
Hagar / Leor / Lior / SmadarNo argument / lipol
‘to-fall’ (1;10;01)
lo nafal
‘not fell’ (1;10;14) / nafal
‘fell’
(1;09;17) / (na)fal
‘fell’
(1;05;26) / (na)fal
‘fell’ (1;06;14)
oy! nafal!
‘oy! fell!’ (1;07;07)
NP subject / Hagar (na)fal
‘Hagar (fm) fell (ms)’ (1;07;14)
hagari nafla ‘Hagar (fm) fell (fm)’ (1;10;17) / oto nafal
‘car fell’
(1;11;22) / Pigi lala
‘Pigi fell’
(1;06;14)
dubi nafal
‘teddybear fell’
(1;07;07)
VS / (n)afal xaruzim
‘fell (sg) bids’
(1;07;17) / nafal ha-magevet
‘fell (ms) towel (fm).
(2;05;15) / (na)fal ze
‘fell this’
(1;07;00) / hala ze
‘fell this’
(1;07;05)
nafal Kushi
‘fell Kushi’
(1;07;07)
Location / nafal, la-mayim
‘fell, to-the-water’
(1;09;26) / nafal Bamba (me ha-)kos
‘fell (ms) Bamba (fm) (of the-)cup’ (1;07;11)
Dative clitic / nafal li
‘fell to-me’
(2;01;13)
nofel lax
‘fell to-you’
(2;02;13) / nafal li
‘fell to-me’ (2;04;22) / nafal li
‘fell to-me’ (1;08;02) / nafal li
‘fell to-me’ (2;01;08)
NP subject +location / Shauli nafal me ha-kise
‘Shauli fell of the-chair’
(2;01;19) / ze yipol al ha-shulxan
‘this will-fall on the-table’
(2;04;22)
hi nafla me ha-ofnoa
‘she fell of the-bikes’
(2;05;15)
NP subject + dative clitic / ze nofel lexa
‘it falls to-you’
(2;02;25) / ze nafal li
‘it fell to-me’
(1;11;13)
nafal li kcat ha-tush
‘fell to-me a-bit the-pen’
(2;02;10)
Table 2 - The development of argument structure for the verb 'want'
Hagar / Leor / SmadarNo arguments / lo roca
‘Don’t want’ (1;07;02) / lo roce
‘Dont want’ (1;09;04) / lo roca
‘Dont want’ (1;08;13)
Nonfinite verbal complement (modal-like use) / roce (l)e(hitla)besh
‘Want to dress up’
roce lishon
‘Want to sleep’ (1;07;22) / roce (leh)o(r)id
‘Want to take off’ (1;09;04) / (ro)ce (li)shon
‘Want to sleep’ (1;08;13)
NP complement / roca od
‘Want (some)more’
(1;08;01)
lo roca yadayim
‘Don’t want hands’
roca aba
‘Want Daddy’ (1;08;04) / roce musika?
‘Want music’ (1;09;00)
roce (et) ze
‘Want this’ (1;09;00) / roca (sa)kin
‘Want a knife’ (1;07;16)
NP subject / ani lo roca
‘I don’t want’ (1;08;09) / Liori lo roce
‘Liori doesn’t want’ (1;11;13) / ani roca
‘I want’ (1;11;13)
Subject and object / ani lo roca kola
‘I don’t want Kola’ (1;09;21) / ani roce kariyot
‘I want pillows’ (1;10;15) / ani roca et ha-harkava ha-zoti
‘I want this pazzle’ (2;01;26)
Non-NP complement / *ani roca same'ax
‘I want happy’ (1;10;01) / roce kan, roce sham
‘Want here, want there’ (1;09;24)
roce levad
‘Want on my own’ (1;11;13) / kodem be ze ani roca
‘First I want with this’ (2;00;07)
Nonfinite CP complement + Subject / ata roce le'exol?
‘You want to eat?’ (1;10;01)
ani roca letayel
‘I want to take a walk’ (1;10;03) / hu roce lesaxek
‘He wants to play (1;11;00) / buba roca (la)shevet (al ha-)sapa
‘Doll wants (to) sit (on the) cauch’ (1;07;07)
ata roce lir'ot?
‘You want to see?’ (1;11;13)
Finite CP complement without complementizer / roca loasim mastik
‘Want chewing chewing-gum’ (2;01;24) / rocim ti(f)tax or
‘Want turn on (the) light’ (1;09;11)
roce tavi'i
‘Want bring’ (1;09;11)
roce ani yasim ‘Want I put’ (1;11;04)
Finite CP complement with complementizer / ani roca she nelex lir'ot
‘I want that we go to see’ (1;11;18)
Table 3 - The development of argument structure for the verb 'give'
No Arguments / latet
‘to-give’
(1;07;24) / tni
‘give!’
(1;06;20)
Frozen dative or accusative / ten lax
‘give you’ (1;07;03) / ten lo
‘give him’ (1;09;04) / tni li
‘give me’ (1;07;02)
NP complement / aba, titen Dani
‘Daddy, give Dani’
(1;10;22) / sefer tni
‘book give’
(1;06;20)
ten od daf
‘give another page’ (1;07;05)
Frozen dative with NP complement / ten lax bakbuk
‘give you a-bottle’
(1;07;17) / od ten lo
‘more give him’
(1;09;17) / tni li od daf
‘give me another page’
(1;07;05)
Two NP complements / (x)a(r)uzim latet le (H)agari
‘bids to-give to Hagari’ (1;08;09) / savta Xana lo natna leRotem smalot
‘Granma Xana not gave to Rotem dresses’ (2;00;00)
Non-frozen dative clitic with NP complement / li titni gvina
‘to-me give cheese’
(1;09;06)
tni lo lexem
‘give him bread’
(2;00;03) / natan li srita
‘gave me a-scratch’
(2;05;06) / (na)tat li et ha-kaze?
‘you-gave me ACC the-like-this’
(1;11;13)
Nonfinite CP complement / tni li la'avor
‘let me to-pass’
(1;11;20) / ten li lir'ot
‘let me to-see’
(2;01;28) / savta Matilda natna li gam lishtot ba-mita
‘granma Matilda gave me also to-drink in-the-bed’
(2;02;10)
NP subject + / ani roca latet le Bindi
‘I want to-give to Bindi.’
(2;01;05)
aba natan li bananot
‘Daddy gave me bananas’ (2;01;23) / savta Xana natna lanu et ha-smalot ha-ele
‘granma Xana gave us ACC the-dresses these’
(2;00;00)
The acquisition of the verbal system in Hebrew
Early verb usage: unclear forms or aspectually limited use of particular verbs
• Over 80% look like IMPERATIVES or INFINITIVES without the infinitival morpheme le- 'to'.
• PAST forms are used only with a perfective actions, e.g., fal for nafal ‘fell’ and PRESENT tense forms are used for durative actions, e.g., xel for oxel ‘eating’.
• All non-infinitival forms are singular, with some nonproductive variation between masculine and feminine forms, e.g., kxi ‘you, feminine, take’ is used for a male addressee.