Mean Length of Utterance (MLU) A measure of language ability A Measure of Language Ability.

Mean Length of Utterance (MLU)

A measure of language ability

A Measure

of Language

Ability

Mean Length of Utterance (MLU)

MLU is a way to score a child’s language ability

When scoring for MLU, researchers count the number of morphemes in the child’s utterances

It has been found that as age increases, so does the average number of morphemes used by the child

Morphemes

Smallest element of meaning in speech

E.g.:“Walk the dog” contains 3 morphemes

“The dog walked” contains 4 morphemes-The suffix “-ed” adds meaning

Remember: Morphemes and words are not the same thing

Morphemes

Free Morphemes:• These are words on their own:

• E.g. “Dog”, “Walk”, “Sit”, “House”

Bound Morphemes:• Prefixes and Suffixes• These are not words on their own:

• E.g. “re-”, “un-”, “pre-”, “-ed”, “-s”, “-ing”, “-ly”

Using MLU to Assess Language Ability

Shouldn’t be used as the only measure, but does correlate with other language measures

Rice, Redmond, & Hoffman (2006)• Showed MLU correlated positively with:

• Developmental Sentence Scoring (DSS)• Index of Productive Syntax (IPSyn)• MLU in Words (rather than Morphemes)

MLU and Age

Miller & Chapman (1981)• Found a positive correlation of r=.88 between age

and MLU

Lots of research provides us with the average MLU for children at each age• E.g.

- At 30 months, you can expect a MLU of 2.54 - At 60 months, expect MLU of 5.36

Can be a diagnostic tool for Language Impairment, but researchers caution it shouldn’t be the only one

Children’s Understanding of Morphemes

Berko, 1958 e.g.

• Used nonsense words and pictures

• Found that children aged 4-7 correctly knew how to pluralize by adding an s or z sound.

• Correctly understood the use of the d sound for past tense

• Understood the use of -ing

Scoring a Child for MLU

Ideally:

• Observe and record interaction for 30-60 mins where dialogue is likely, e.g. Playing with dolls with mom

• Pick out 100 utterances made by the child that are completely intelligible

• Transcribe the interaction (write out what was said)

->

Scoring a Child for MLU

What to count as a morpheme:

• Free Morphemes: (“truck” = 1)• -s (used as plural- “girls” = 2)• -ed (“jumped” = 2)• -ing (“dancing” = 2)• -s (used as possessive –”mom’s car” = 3)• Contractions (“she’s” = 2)

Scoring a Child for MLU

What not to count as a morpheme:

• Unintentional repetitions (“He he is there”=3)• Compound words (“doghouse” =1)• Reduplications (“Choo, choo” =1)• Proper Names (“Mickey Mouse” =1)• Irregular plurals (“pants” =1)• Catenatives (“gonna” =1)• Fillers (“umm” =0)

Example Utterances

Some Examples:

Mommy’s going downstairs” • Mommy =1• -’s =1• go =1• -ing= 1• downstairs =1

• Total Morphemes= 5

Example Utterances

“The truck, umm, the truck went vroom, vroom”

• The = 1• truck = 1• umm = 0 (filler)• the = 0 (repetition)• truck = 0 (repetition)• went = 1• vroom = 1• vroom = 0 (reduplication)

• Total Morphemes = 4

Scoring a Child for MLU

Once the morphemes have been counted for each utterance:

• Add up all the morphemes

• Divide by the number of utterances• Ideally 100

• Now have the child’s Mean Length of Utterance score

Example MLU Calculation

• 1) “Mommy’s going downstairs” = 5 Morphemes

• 2) “The truck, umm, the truck went vroom vroom” = 4 Morphemes

Total = 9 Morphemes

• 9 Morphemes divided by 2 utterances = 4.5

• MLU (in this very short transcript)= 4.5

Part I: Data Collection

• As we watch the videos, try to write down everything the child says

• Video 1 : Girl at 2.5 years

• Video 2: Same girl at 3 years

• You will then be given transcripts of the videos and will score MLU at both ages

Part II: Data Analysis

Open the linked SPSS Data File Save to Desktop, Open SPSS, Open File

Run a Dependent t-test in SPSS to see if MLU scores changed significantly in participants from Time 1 (2.5 years old) to Time 2 (3 years old)

Did the children’s language complexity increase in the space of 6 months? Each participant has been tested twice.

Note about t-tests:

Independent t-test: Compares two groups of different people E.g. Comparing the marks from one lab section to another

Dependent t-test: Compares people to themselves at different times. E.g. Comparing each student’s 260 midterm mark to their exam mark.

The t-test determines if two sets of scores are different from one another. When the Significance Level is less than 0.05, this tells us that there is only a 5% or less probability that the difference you found was not real. There is a 95% or more probability that this difference is real and would be found again and again.

Reporting results: t(df)= insert t-value, p___0.05. The ‘p’ stands for “probability”. If it is less than 0.05, insert the “<“ symbol, “>” if greater.