FLA Revision

Vouloumanos & Werker (2007)

Newborns prefer to listen to speech over non-speech

Vouloumanos et al (2010)

Preference for human speech over primate vocalisation appears after 3 months

Ramus et al. (2000)

Newborns can discriminate languages but not when the recording is backwards

Jusczyk, Cutler & Redanz (1993)

American 9-month-olds prefer to listen to low-pass filtered strong/weak words over weak/strong words

Jusczyk et al. (1993)

American and dutch 9-month-olds listen longer to their native words that feature language-specific segmental phonotactics - but not if low-pass filtered.

Jusczyk, Luce & Charles-Luce (1994)

English-learning 9-month-olds listen longer to nonsense syllables with high phonotactic probabilities

Swingley & Aslin (2002)

14-month-olds look less at the corresponding picture if the word is 'mispronounced' by a single feature

Maye, Werker, & Gerken (2002)

6-month-olds can perform distributional learning from phonetic input

Aitchison & Chiat (1981)

4- to 9-year-olds produce 'consonant harmony' in recall of newly learned words. However, more likely to be a failure of memory rather than articulation.

Brent & Siskind (2001)

Only a small portion of utterances in infant-directed speech consist of a single word

Saffran, Aslin, & Newport (1996)

Infants looked longer at the side that was playing part words. Novelty effect.

Cutler & Carter (1987)

The most common prosodic word type in English is strong-weak disyllables.

Kelly & Martin (1994)

Strong syllable = word onset 95% of the time in IDS

Jusczyk, Houston & Newsome (1999)

7.5-month-old infants use predominant prosodic patterns to segment words

Bortfeld et al. (2005)

Infants can use familiar words for further segmentation

Casello et al. (1999)

Noun bias in early lexical development, possibly due to the whole object assumption

Landau, Smith & Jones (1988)

Shape bias makes children extend the referent of a noun based on its shape

Smith and Yu (2008)

Children can use co-occurrence probabilities to perform cross-situational learning

Tomasello & Barton (1994)

2-year-olds can determine the referent of an unknown word based on their understanding of speaker intention (theory of mind)

Akhtar et al. (1996)

2-year-olds can also determine the referent of an unknown word based on their understanding of the speaker's knowledge state.

Markman (1990)

Mechanisms behind word-learning are specific to language. Evidence: shape bias only observed when words are involved.

Bloom (2000)

Mechanisms behind word-learning apply generally. Evidence: children can fast-map word-meaning as well as facts. Mutual exclusivity works both for word-learning and fact-learning.

Dual Route Model

All inflected forms learnt through memory. The child over applies a default rule to irregulars. Then irregular forms are learned and block rule application.

Single Route Model

Stores some past-tense forms in memory. Forms with infrequent patterns adopt similar patterns due to analogy. The child recovers from over-regularisation through reinforced memory of irregular patterns.

Rumelhart & McLelland (1986)

Fed a computer verbs (10 high frequency for 10 trials, 410 medium for 190). Then tested network on present tenses and saw what past-tense form it supplied.

Marcus et al. (1992)

The emergence of -ed marking is sudden, and coincides with over-regularisation. This shows that -ed is acquired as a rule (which overapplies)

Marchman (1997)

Regular verbs are not entirely protected from frequency effects. Children sometimes over-irregularise when there are many similar-sounding irregulars.

Clahsen & Almazan (1998)

Children with WS show deficiency in irregular past-tense verbs, but not regular ones. Rule and memory are different components.

Gopnik & Crago (1991)

Children with SLI show deficiency in regular past-tense verbs but not in irregular ones. Rule and memory are different components.

McClelland & Patterson (2002)

A phonological deficit can explain some of the dissociation (consonant clusters). SR Model

Gerken & McIntosh (1993)

Children can find the correct picture better when they are told "Find the dog for me" than "Find was dog for me"

Dye et al. (2018)

10-month-olds prefer "the dog" over "dog the"

Constructivist accounts

Syntactic knowledge develops gradually from concrete item-based learning to abstract structural knowledge.

Generative approach

Abstract categories are innate (includes lexical and functional). Children are aware early on of the relationship between functional and lexical categories.

Pine and Lieven (1997)

No overlap in the use of DET with the same nouns (item-based)

Valian et al (2009)

Children's overlap in DET use is about 50% (same as adults)

Yang (2010)

Cannot expect high levels of overlap in spontaneous speech because we don't always recombine articles for every noun.

Semantic bootstrapping

Meaning-category correspondence guides the initial phase of category and relationship learning in syntax.

Pinker (1984)

Children are innately equipped with knowledge of the mapping between syntactic category types and functions on one hand and semantic types on the other.

Phonological/prosodic bootstrapping

Phonological and prosodic cues aid the learning of syntactic categories and phrases. Syntactic boundaries are marked by prosodic boundaries.

Shi et al. (1998)

Function words tend to have shorter vowel durations, weaker amplitude, and simpler syllable structure than lexical words.

Kelly (1988)

In English, disyllabic nouns tend to have initial stress while disyllabic verbs tend to have final stress.

Distributional learning

A word's syntactic category and its grammatical role in a sentence can be learned by looking at the words that surround it, its morphological marking, and its relative position in a sentence. It goes beyond the frequency of occurrence of a particular word.

Wexler (1998)

Very Early Parameter Setting. Head directionality is set very early.

Akhtar & Tomasello (1997)

2- to 3-year-olds' verb use closely parallels what they have heard. Even if that means accepting word order configurations inconsistent with the target language

Gertner et al. (2006)

Even 21-month-olds performed above chance with novel verbs in preferential looking procedure

Fernandes et al. (2006)

28- and 34-month-olds can generalise from neutral and intransitive sentences.

Savage et al. (2003)

Children's structural priming seems to be lexically based (at least before age 4)

Radford (1996)

Children's early grammars are truncated versions of the adult grammar (small clauses)

Rizzi (1993/4)

Children project all the necessary functional categories available in the adult grammar but children do not necessarily know the obligatoriness of these categories (Truncation Account)

ATOM: Wexler (1998)

Agreement/Tense Omission Model. Children's early grammars project all functional categories, parameters are set from the earliest stages. Children optionally produce them due to a maturational constraint.

Unique Checking Constraint (UCC)

Check the features of a verb against only one functional category -> an item can be checked at AGR or TNS but not both.

MOSAIC: Freudenthal et al. (2007)

Model of syntax acquisition in children. Empiricist/Usage-based approaches: Children learn strings of language directly from the input.

Freudenthal et al. (2007)

Recency effect: Good correlation between utterance-final non-finites in the input and RIs in children.

Legate & Yang (2007)

Variational model. Combines presence of categories (=/-TNS) with nature of the input. Utterances with (overtly marked) tense/agreement reward + TNS. Captures crosslinguistic data by combining categories with input data.

Nativism (G)

Most core aspects of syntactic knowledge are part of Universal Grammar and innate

Development (G)

Involves selecting the options set by UG and learning language-specific properties of syntax outside UG

Continuity (G)

What underlies children's sentence is the same kind of abstract grammar that adults have.

Non-adultlike production/comprehension (G)

Either they haven't set UG to the correct settings, they are still learning idiosyncratic properties of the language, or committing performance errors.

Emergentism (C)

Much of syntactic knowledge is learned from the input

Development (C)

Involves learning concrete exemplars of word combinations, and then gradually making generalisations over such exemplars.

Discontinuity (C)

What underlies young children's sentence is item-based and not the kind of abstract grammar that adults have.

Non-adultlike production.comprehension (C)

Because children start by learning fixed combinations, they sometimes overapply them to inappropriate contexts.

Goodluck & Tavakolian (1982)

Children have adultlike syntactic knowledge but are not good at processing sentences in experimental tasks. (G)

Diessel & Tomasello (2000)

Children's early relative clauses are just amalgams of a prefabricated schema and a bare untensed VP. (C)

Kidd et al. (2007)

Children are better at processing types of relative clauses that they hear or say the most (C)

Santelman et al. (2002)

Do-support is an idiosyncratic feature of English (outside UG) that just needs to be learned (G)

Guasti et al. (1995)

English-learning children mis-set the negation parameter to the 'do not raise' setting (G)

Rowland & Pine (2000)

Children are accurate in production of items that are frequent in the input, and inaccurate for those that aren't

Ambridge & Rowland (2009)

Children who have have the 'What does [THING] [PROCESS]" schema produce doubling errors, children who say "What she does like?", don't. (C)

Trueswell & Gleitman (2007)

Children become sensitive to phrasal ordering predictors (word order), lexcially specific predictors, and less reliable semantic, contextual and referential predictors.

Trueswell et al. (1999)

Put the frog on the napkin in the box. The Kindergarten Path Effect.

Late closure

When possible, attach incoming lexical items into the clause or phrase currently being processed

Minimal attachment

Attach incoming lexical material into the phrase marker being constructed with the fewest nodes consistent with the well-formedness rules of the language

Kirkham et al. (2002); Fiser & Aslin (2002)

People are about as successful with learning the "words" when it is a visual stimuli, rather than an auditory one. Not just applicable to language.

Marcus et al (1999)

9-month-olds can only learn when AAB patterns are varied. Children need variation not repetition to learn abstract patterns.

Non-adjacent dependencies

Children track non-adjacent co-occurrence probabilities only if adjacent transitions are highly unpredictable.

Gomez (2002)

Infants can pick out frames that are distributionally like functional elements. Distributional information is key to category learning more generally.

Braine (1987)

Tested adults' ability to learn MN/PQ systems under different conditions. Adding a meaning cue helps to solve the MN/PQ problem. Many natural languages have semantic cues to noun categories (grammatical gender/noun class systems)

Gerken et al. (2005)

Tested children's ability to learn a complex gender system without semantic cues. Children can learn a complex morphological pattern based on phonological cues, but they have to be substantial cues. Semantic cues aren't necessary but some additional distributional cue must be present. If cues are around, then infants can learn grammatical categories through statistical learning.

Reeder et al. (2013)

Tested adult's generalisation of words to new contexts under different conditions. If they hear all A&B, or there's partial overlap, they learn the system. Partial overlap, 3x Exposure, they only learn heard forms.

Saffran and Thiessen (2003)

Tested 9-month-old infants' ability to learn a novel phonotactic constraint (voiceless stops in onset and voiced stops in coda, and vice versa). Phonotactic rules with fewer features are easier to learn.

Culbertson & Newport (2015)

Tested 4- to 7-year-olds' ability to learn a novel syntactic rule. Simpler syntactic rules are easier to learn.

Hudson Kam & Newport (2009)

Test whether children reduce unconditioned variation by regularising. They do.

Soderstrom (2007)

Properties of IDS: diminutives, lower type:token ratio, lower MLU, higher, more variable pitch, exaggeration of prosodic cues, hyperarticulation of sounds.

Newport et al. (1977)

Properties of CDS: Short MLU, well-formed, structurally simple, lots of deixis, lots of repetitions

Hurtado et al. (2008)

More maternal (and paternal) input at 18mo = larger vocabulary size at 24mo

Specific disorders

There is a deficit in one or a small number of skills, with typical functioning in other areas, e.g., SLI/DLD

General disorders

Impairments in most, if not all, cognitive functions, e.g. Down syndrome

Normal IQ

85-115

Moderate learning difficulties IQ

50-70

Severe learning difficulties IQ

Bishop (2016/17)

There are a group of children who have difficulty learning language not caused by autism, general mental delay, severe emotional disorder, acquired childhood aphasia, or peripheral impairments. They have DLD.

Developmental Language Disorder

A developmental disorder affecting primarily oral language, more specifically grammatical abilities. Affects 7% of school population. Asymmetry between cognitive and language abilities

Exclusion criteria for SLI

Normal hearing, no known history of recurrent otitis media, no emotional or behavioural problems, performance IQ 85 or above, normal neurological status.

Language profile for DLD

Omits grammatical morphemes or use wrong ones, difficulty producing or comprehending complex syntactic structures, difficulty producing cohesive narratives.

Rice & Wexler (2001)

Development of English tense in children with DLD is much slower than their TD peers.

van der Lely & Battell (2003)

Wh-questions in DLD children. No do-support, and gap-filling.

Gathercole & Baddeley (1990); Orsolini et al (2001)

DLD children have problems with phonological processing.

Kail (1994); Montgomery & Leonard (1998); Tallal et al. (1996)

DLD children have longer latencies when performing both language-mediated and non-language-mediated activities