Lecture: Psycholinguistics Professor Dr. Neal R. Norrick

_____________________________________

Psycholinguistics

Universität des SaarlandesDept. 4.3: English Linguistics

SS 2009

A. Coordinate: L1 and L2 acquired

in separate contexts– each system is complete in

itself– person functions as

monolingual in both communities

7.4 Two languages in one brain7.4.1 Types of bilinguals

Weinreich (1953) distinguished three kinds of bilingualism

B. Compound: L1 and L2 acquired in same context

• the two systems are merged• person doesn't function as monolingual in either community• person may experience interference from L1 to L2 and from L2 to L1

C. Subordinate: L2 acquired based on L1

– only one system

– person functions as monolingual only in L1

– person experiences interference only from

L1 to L2

Notice that Weinreich’s typology works only at the lexical level, but bilinguals may experience interference at all levels from phonetics up to semantics.

7.4.2 Bilingual meaning systems

According to Macnamara (1970): • subordinate bilinguals function appropriately in L1, but inappropriately L2

• compound bilinguals function inappropriately in both languages• though coordinate bilinguals function appropriately in L1 & L2 they must experience confusion in their internal thought

But this assumes that word meaning and natural language semantics correspond directly to mental concepts.

By contrast, Paradis (1979, 1985) argues that both language systems are connected to a conceptual-experiential level of cognition

In fact, the situation is probably a mixture of these two positions:

• WATs and other tests show concrete concepts like tree and table seem to be shared, as in ‘compound’ diagram B above• but abstract concepts like freedom and justice are language-specific, as in ‘coordinate’ diagram A above

words identical in meaning and similar in form seem to share a single ‘lexical entry’

die Karotte carrot la carottedie Adresse address l'address

but the systematic semantics of the individual languages may still differ, thus German has rough synonyms

Karotte Mohrrübe Adresse Anschrift

probably semantic systems overlap with some areas shared and others distinct, e.g.

English ball spheric, bouncy, for playFrench balle spheric, bouncy, for play, small

given French ballon for larger, inflatable spheres, while these features are irrelevant for English ball

7.4.3 Bilingual phonology and syntax

Extended system hypothesis:

phonemes of L2 are processed as allophones of

L1 phonemes

Dual system hypothesis:

separate phonemic systems for L1 & L2

Tripartite system hypothesis:

shared phonemes in one system with separate

phonemes in separate systems

Stop consonants p t k, b d g could be shared in bilingual German-English system

but English fricatives in then and thin, and German fricatives in ich and ach must occur in separate systems

Similarly:• syntactic structures of L2 could be processed in

accordance with L1 syntax• L1 & L2 could have separate syntactic systems

• shared structures could be processed the same while separate structures would require separate processing

e.g. German & English NPs could be processed

similarly with special processing for German

preposed participles like:

das von der Kandidatin gewählte Thema

7.4.4 Language processing in the bilingual brain

Depending how they're acquired, L1 & L2 may even

be lateralized differently in brain:

• L2 lateralized in right hemisphere

• L2 less lateralized than L1

• L1 & L2 both less lateralized than in monolinguals

evidence from aphasia indicates that languages are

separately organized in brain, but not necessary

lateralized separately

As Paradis (1979, 1985) shows, bilinguals comes in many types;

Bilinguals may differ with regard to:• manner of acquisition (formal, informal)• mode of acquisition (oral, written)• method of acquisition

(deductive, inductive, analytic, global)• age of acquisition (during or after critical period)• stage of acquisition• degree of proficiency

• frequency and modes of use• language-specific features of L1 & L2 • sharing features and rules at various levels

on every linguistic level, structures might be shared or separate

e.g. if L1 speaker produces L2 perfectly, except for phonetics, i.e. has lots of interference from L1 to L2

at the level of phonetics, we could model thesituation as follows:

L1 L2

conceptual level single system

semantics x -- y

syntax x -- y

morphology x -- y

lexis x -- y

phonology x -- y

and if L1 speaker produces phonetically correct L2, but makes lots of interference errors in grammar and word choice, we could model the situation as follows:

L1 L2

conceptual level single systemsemantics x -- ysyntax x -- y morphology x -- y lexis x -- y phonology x -- y

Of course, some languages may naturally sharestructures at certain levels:

English-German bilinguals probably have a single set of stop consonants for both languages, but German speakers need to add the fricatives in then and thin, and English speakers need to add the fricatives in ich and ach and so on

In the simplest model, the concepts of experience run through a set of pipes and come out as either L1 or L2

(in the model Spanish and English)

The next model ignores the concepts and begins with separate tanks for the words of L1 & L2; again pipes run down, and one language spills out.

(This second model corresponds to Weinreich’s “coordinate bilingual”)

In third model, the concepts of experience run through pipes representing L1 & L2, they are assigned appropriate words from either L1 or L2, and they flow into another set of pipes, representing the grammar and phonology, and finally flow out as either L1 or L2.

But, as in Weinreich, there’s no way in these models to account for interference

Since there's interference between the systems, some pipes may be playing a role in both L1 & L2 systems, and the pipes must be leaky; since we can code-switch and translate, there must be leakage in both directions

It’s probably necessary to complicate the third model

The tanks of words from L1 or L2, need valves to turn them on or shut them off, representing the decision to

speak either L1 or L2 and block out the other

As we saw above, the words must flow into separate sets of pipes, representing the grammar, morphology

and phonology of either L1 or L2 as well; but some

pipes serve both L1 & L2 systems to some extent, to account for interference

At all levels, we must allow leakage to explain how

we can code-switch from L1 to L2

also possible:

comprehension is a single system for L1 & L2,

while production of L1 & L2 remains separate, because:

• comprehension precedes production in acquisition • comprehension more advanced than production at

all stages • though we can choose not to speak L1 or L2,

we can't choose not to comprehend • production is lost before comprehension in aphasia • comprehension returns before production in aphasia

again according to Paradis, we can envision: • single coherent underlying conceptual system• two cognitively separate systems - with some

shared areas in semantics, syntax, phonology

one system is suppressed due to context, frequencyof contact etcbut word/phrase from suppressed system may intrude, especially during word searchthere may be differences in processing due to acquisition history, strategies etc

8. Language comprehension

means understanding what we hear and read

comprehension as active search for coherence and sense based on expectations arising from context,

not a passive item-by-item recording and analysis of words in a linear sequence.

meaning and real-world expectations play a moreimportant role than grammar

top-down versus bottom-up processing

Until the age of four, kids interpret a-d the same way; even adults require longer to respond to c, d:

a. The cat chased the mouse.b. The mouse was chased by the cat.c. The mouse chased the cat.d. The cat was chased by the mouse.

Asked to paraphrase e-g in their own words, subjects ‘normalized’ the sentences 60% of the time:

e. John dressed and had a bath.f. John finished and wrote the article on the weekend.g. Don't print that or I won't sue you.

Asked if they saw any difference between g and their ‘incorrect’ paraphrase h, 53% still said no

h. If you print that, I'll sue you.

clearly, the ‘Reality Principle’ guides our comprehension of linguistic structures

8.1 Comprehension of sounds

How can we identify sounds and words when sounds vary?

How to wreck a nice beach = How to recognize speech

Notice positional variants

Consider necessity of top-down interpretation

Phoneme restoration effect

a. It was found that the -eel was on the axle. wheel

b. It was found that the -eel was on the shoe. heel

c. It was found that the -eel was on the orange. peel

d. It was found that the -eel was on the table. meal

We hear progressively different allophones of a single phoneme as the same:

spread p in peel versus puckered p in pool versus unaspirated p in speed or spool

but we hear separate phonemes as distinct although they also occupy points along a single continuum

pie and buy differ only in the initial consonant

we attend only to difference in Voicing Onset Time (VOT)

VOT for pie about 50 milliseconds later than for buy

even sounds halfway between p and b in VOT are heard as one or the other rather than as a combination of the two

this categorial perception of sounds is a distinctly human trait

sometimes cited as evidence of innate language ability

but differences between fricatives like fa tha sa sha are perceived continuously on basis of aperiodic noise

8.2 Comprehension of words

Parallel Distributed Processing (PDP):

separate, simultaneous and parallel processes work to identify words

by pronunciation: to recognize homophones leadN and ledV pst

by spelling: to recognize homographs windN and windV

by grammar: to recognize smell as noun or verbwhile hear can only function as verb

by semantics: synonyms like little and smallantonyms like little and bighyponyms like car versus vehicle

etc

PDP can link word meanings to perceptual and

functional paradigms (how a thing looks, sounds etc,

what it's used for)

consider Tip-of-the-tongue (TOT) phenomena

you're trying to recall the word for the belief that life's

events are preordained by a deity

you remember that the word begins with p, then that

word begins with pre-, and that it ends with -tion

Bathtub Effect: recall is best for beginnings and

ends of words, like the head and feet of a person

which are visible though the middle remains

submerged in the tub

you recall associated words like:

predilection pretension

Presbyterian preordained

you finally come up with: predestination

Spreading activation networks: as the search progresses, more words and concepts are accessed related in various ways,

including schematic knowledge e.g. the association of Presbyterian

and predestination via 'religion‘

Both comprehension and production of both speechand writing require accessing the mental lexicon. Garman (1990: 249) diagrams input-output relations as following: