EGH215 Weeks 6-11

6 Characteristics of early word learning

Average American child at 12 months understands about 50 words (Fenson et al. 1994) and can learn a new word every two hours (Pinker 1994)

From as early as 7 months, children begin to comprehend words; comprehension precedes production

After the initial 50 words learned, there is an explosion in word learning: a vocabulary burst

The criterion for a spurt is 10 or more words in two and a half weeks

Vocabulary bursts are cross-linguistic incl. sign language

In English, Bates et al. (1994) found that nouns increase when vocabulary grows to between 50-100 words, verbs and adjectives at 100-400 words and function words after 400 words

Children's receptive vocabulary is always much larger than their productive vocabulary

Receptive vocabulary can be measured by indexes like the MacArthur-Bates Communicative Development Inventory (MBCDI)

Receptive vocabulary is seen to contain many more verbs (Gentner, 1978), particularly 'light verbs' (used in a lot of environments e.g. make; go)

Fast mapping/rapid learning studies show that children can acquire a word based on one or very few exposures (Carey 1978) (see study on teaching pre-schoolers a new colour term)

6 The logical problem or word

The reference problem: how does a child map the word onto the correct aspect of the visual scene?

For object class words, there are two parts to the word learning problem: the mapping problem (mapping the word onto its referent); the categorisation problem (extending the word to other members of the category, defining class inclusion)

6 Principles that guide word learning

The whole object constraint: a novel label explicitly applied to an object is more likely to refer to the whole object than its parts, substance, colour, motion, temporary state or other property

The taxonomic constraint: a novel label applied to an object can be extended to objects of the same kind rather than objects that are associated in some spatial, causal, or other thematic relation

The mutual exclusivity constraint: children assume each object has one and only one label based on a preference for simple one-to-one mapping

This principle guides initial mapping

This principle guides categorisation, biasing taxonomic relations over thematic relations

Despite the fact that children are inherently interested in thematic relationships, they will not extend a label on the basis of a thematic relationship

This principle overrides the whole object constraint to allow children to learn words that refer to parts of objects, materials and class inclusion terms

It hypothesises that if children already know the name of an object, they will not take a novel term to refer to the same object, but instead to a part of the object or its material etc.

6 Syntactic bootstrapping

Syntactic bootstrapping refers to using knowledge of syntax to learn the meaning of words

If the correlation between meaning and syntactic structure (e.g. transitive vs. distransitive) hold, this can be used as a tool for learning verb meaning

7 Word order

Languages differ according to whether the head precedes or follows the complement

Infants are sensitive to word order and can use word order to understand grammatical relations

7 Early syntax

Holophrastic speech (~12 months)

Children have around 50 words (productively) before they begin to combine them; combinations appear around 15 months

Holophrase: one word sentence i.e. one word is used to convey complex meaning

Telegraphic speech (~18-26 months)

Language consists mainly of nouns, verbs and adjectives

Function words (e.g. complementizers, tense, determine heads and number marking) are optionally omitted

Children filter out the elements of speech that their grammar cannot analyse

~ 2 years

More complex syntax, including embedding and movement; morphosyntax continues to develop

~ 3 years

Language-specific properties continue to develop; evidence of fundamental syntax of a grammar

7 Root infinitives

Finite verb forms: inflected for tense and person/number; anchored in time e.g. she cooked pasta

Non-finite verb forms: do not mark tense or person/number; not anchored in time e.g. I want Jane to cook pasta

Some languages have non-finite inflection e.g. French -er marks infinitives

English does not mark the infinitive with a pronounced inflection (it has an impoverished inflectional morphology system)

English acquiring children therefore produce bare verbs and bare participles

Root infinitives: the use of infinitives in a matrix clause (not acceptable in the adult language)

Between ages 2-3 children will produce infinitive forms of verbs in main clauses

RIs are very robust in the languages that have them

RIs in French

All verbs in main clauses must be tensed

Finite verbs always occur before negation, and children have some sensitivity to this when producing root infinitives; French 2-year-olds place finite verbs above negation as adult speakers would

Note that English acquiring children never make the mistake of raising the verb above negation

RIs in Germanic languages

Germanic languages follow the V2 rule, meaning that finite verbs must occur in the second position of matrix clauses, while infinitive verbs occur finally (SOV)

This pattern is evident in child grammar

7 Small Clause Hypothesis (Radford, 1986)

Children's grammars (to ~2 years) have the structure of adult small clauses i.e. subject + predicate with no overt tense

E.g. adult: I heard [John sing]

SCH predicts that children should entirely lack all functional material, only producing lexical categories

It posits that functional categories either
(1) mature at a later point
(2) are projected when children learn the function words, which they don't learn initially as they are not phonetically or semantically salient

Pros:

Explains the omission of functional morphology and is a simple hypothesis (no abstract structures)

Cons:

  • Posits a gross discontinuity with adult grammar
  • Assumes maturation of functional material, or lexical learning
  • There is evidence for the presence of functional categories, and when they are used they are used correctly

Truncation Hypothesis (Rizzi, 1994; 2004)

This posits that children can optionally truncate their syntactic representations i.e. trees, and explains the omission of function words since functional material is the highest part of the tree

E.g. root infinitives are a VP truncation

Pros:

  • Explains why the omission of functional material is optional and variable because specification of the top node is variable
  • It accounts for the omission of subjects if children are said to truncate below the TP
  • It helps to explain correlations in child productions

Cons:

  • Can't explain why English children produce RIs in wh-questions
  • Can't explain why RIs don't occur in all languages e.g. Italian

8 Null Subject (NS) Stage

Some languages permit subjects to be null (e.g. Italian, Spanish), while other languages do not (e.g. English, French)

Null subjects mean that the subject can be silent and optionally overt; non-null subject means that the subject is obligatorily overt

During the telegraphic stage, children acquiring non-null subject languages optionally omit subjects

Example of null subject utterances in English:
Read bear book
Outside cold

Null subject languages do not have expletives e.g. it, there
Italian: piove (it is raining)

A study of French children (Kraemer, 1994) found a correlation between finiteness and subjects whereby children allow null subjects to occur more freely with non-finite verbs, and supply the subjects more reliably with finite verbs

8 Grammatical contingencies

The distribution of elements during the telegraphic stage is not random, but exhibits patterns and dependencies

Non-nominative subjects are restricted to non-finite contexts
E.g. him do it; *him does it

Inflected be always occurs with an overt subject
E.g. I am going; *am going

Modals only occur with overt subjects
E.g. I can go; *can go

Null subjects only occur with non-finite verbs
E.g. where go?; *where goes?

8 Null subject parameter

Hyams (1983, 1986) proposes a mis-setting of the NS parameter, meaning that all children are born with the +NS option for the NS parameter, and around age 3, children acquiring non-NS languages switch to the -NS option

This seeks to address the developmental problem of language acquisition

However, there is substantial empirical evidence that children acquiring non-NS languages don't behave like +NS children (e.g. Italian)

E.g. Italian children produce NSs 70% of the time compared to 30% for English children
NSs appear in finite embedded clauses in Italian, but not English, Dutch or German
NSs with modals in Italian, but not in English
NSs in finite wh- questions in Italian but not in English, Dutch or French

8 Root null subject parameter

Rizzi (1994) suggests that in some languages/registers of speech (such as diary drop), null subjects are licensed in the specifier of the highest phrase of the sentence

If the truncation account is correct, this would mean that children in the NS stage temporarily have a non-adult grammar; therefore they should not only produce NS sentences, but also be able to interpret NS sentences as grammatical declaratives

Adult grammar only allows the tree to be truncated in specific environments: newspaper headlines, diary drop, and imperatives

9 Childhood bilingualism

Bilingualism: knowing more than one language

Simultaneous bilingual development

Characterised by children having very early (argued within first month or first two years), regular, and continued exposure to more than one language; there is Language A and Language Alpha

Examples are caregivers with different L1s, or caregivers who have the same L1, but this differs to the surrounding community e.g. daycare

Successive/sequential bilingual development

Characterised by the L1 being essentially acquired before the learning of L2 commences; NB that the age at which a child is considered successive rather than simultaneous is subject to discussion

Example contexts include immigration and learning a language in a school/educational context

Bilingual children are valued in language acquisition studies as they are considered a 'perfect matched pair' and allow non-linguistic factors e.g. maturity/ cognitive development to be factored out

Simultaneous bilingual development follows the same pattern and milestones as monolingual development in the early stages e.g. babbling and the telegraphic stage

Note that first words may be slightly delayed, although still within the expected age range. On the other hand, they might not be delayed, but result from differences in the lexicons, or preferences for a word in one language than another, for example

Bilingual development differs from that of monolingual children because they display code switching and language mixing

Code switching: a speaker alternates between two or more languages in a conversation or discourse, typically at the sentence or discourse level. It involves a more deliberate and conscious choice (not confusion) to switch between languages based on factors like social context, audience, or emphasis

Language mixing: a speaker alternates between two or more languages within a single sentence or utterance, incorporating elements from one language into another. This mixing may occur because of a lack of vocabulary in one language, cultural influence, or stylistic purposes

9 Single System Hypothesis (Leopold 1939, Volterra & Taeschner 1978

Suggest that children initially develop a single grammar and lexicon, gradually transitioning through three stages to attain an adult-like separation between their languages

Stage one: up to about 1;8, characterised by one lexicon and one primitive set of rules, evidenced by the lack of translation equivalents (only one word is used for one thing/concept) and the mutual exclusivity constraint (avoids synonyms with children using one term if languages aren't perceived to be separate)

Stage two: from 1;8-2;8, characterised by two lexicons and one set of syntactic rules. This is evidenced by the presence of TEs, using different words in language A and language Alpha. One grammatical system is used across languages

Stage three: from 2;9, characterised by two lexicons and two sets of syntactic rules

Problems with the Single-System Hypothesis:

  • TEs can appear in early bilingual utterances
  • Utterances can be produced with target-like word order despite reverse ordering occurring between languages
  • Different languages have different developmental paths e.g. RIs

9 Separate-Systems Hypothesis (Meisel 1989)

Posits that children have two separate grammars and lexicons from the beginning of development

9 General differences/similarities with monolingual development

Simultaneous bilingual children seem to follow roughly the same development sequence in each language(s) as monolingual speakers do

Bilingual children construct grammars in the same way as monolingual children

Early separation of languages is evident in the production of subjects and RIs; children produce more subjects in their non-null subject language than in the null subject language, and they produce RIs in tandem with monolinguals

There is no specific delay in acquiring two languages over acquiring one language, except with respect to lexicon (total number of words across two languages is comparable to monolinguals in early stages)

Code-switching suggests that although the bilingual's language systems are separate, there is some interconnection between the two systems

9 Childhood L2

Successive/sequential bilingualism: the second language is introduced when the first language acquisition is already well underway (after around 3 years)

It takes a child much less time to acquire L2 than L1; children reach the level of their monolingual peers after around one year (3-7 years-old)

There is a transfer from the L1 to L2, meaning that L2 acquisition is influenced by the first

Interlanguage grammars: second language learners (children and adults) construct intermediate grammars of the target second language

Instead of trying to imitate native speakers, L2 learners try to uncover the rules of the target language

Older children may have some advantages in learning an L2 and progress more quickly than an L1-acquiring infant

However, being older, L2 learners may be disadvantaged to attaining native-like proficiency such as pronunciation and inflectional morphology

E.g. vocabulary learning in older L2 children benefits from greater cognitive skills and pre-existing labels in L1. Older children may also start learning an L2 using more complex and longer sentences

9 Heritage Language Learners

Young children may speak a language in the home which they do not use in the wider community (e.g. in instances of immigration), and may not encounter the majority language until schooling at around age 5

In cases of heritage language learners, bilingualism is not guaranteed an the L1 can undergo attrition

The L1 can undergo attrition: the gradual decline in native language proficiency

This may be due to exposure to the dominant language in the community (e.g. via media), or language used by peers

Attrition is characterised by a loss of vocabulary, complex syntactic structures and production ability. Despite this, comprehension is usually retained (passive/ receptive bilingual: comprehension but no production)

Any language acquired as a child, or later, has the potential to be lost and regained throughout the lifespan

A heritage language learner is someone who, as an adolescent or adult, starts to re-aquire or improve L1 language skills