question

Language Sample Analysis

answer

Collecting a Sample

question

Importance Language sampling is an important part of the assessment process because it:

answer

Allows the SLP to collect information not available via standardized testing; Provides a sample of authentic communication (if done correctly); Allows for multiple analyses to be completed on a single sample, each measuring a different aspect of communication.

question

Focus: We can analyze any and all of the following, using language sampling

answer

Syntax Vocabulary Morphology Semantics Pragmatics articulation

question

Purpose: Before we actually collect a language sample, we need to identify our REASON for collecting it.

answer

Are we trying to determine how accurate our standardized test results are? Are we using language sampling in place of standardized tests - perhaps because the tests are not available for our client's ethnic or racial group?

question

Purpose: Before we actually collect a language sample, we need to identify our REASON for collecting it.

answer

Are we trying to "hone in" on more detailed information related to some specific language skills? Or are we using the language sample for some combination of the above?

question

Once we have a clear idea of WHY we are collecting a language sample,

answer

we can determine HOW we will collect it

question

Example 1

answer

If our goal is to supplement results from the PLS-5 with peer to peer conversational skills, then we will want to construct a situation that allows us to collect a language sample of our client conversing with one or more children.

question

Example 2: Perhaps we are unable to obtain valid results on a standardized test because the child refused to engage in the tasks and was reluctant to interact with the evaluator.

answer

In that case, we may provide the parent with some toys, and instructions to engage in interactive play with his/her child while the evaluator remains in the background, taping the session.

question

Materials, Participants & Method

answer

Materials: we can use story books, wordless books, picture sequences, toys, or other manipulatives; Locations: child's home, a therapy room, classroom (though a classroom is typically NOT an ideal location);

question

Materials, Participants & Method

answer

Participants: can be peer to peer; parent to child; evaluator to child; small group of children; Methods: can include conversation, expository language, narration, or a combination of these.

question

Collecting the Language Sample Use an Audio or Video tape:

answer

videotaping is preferable because it captures more. If this isn't possible, the SLP can use an audiotape

question

The benefit of either of these tools is that an utterance can be replayed repeatedly to ensure accurate transcription.

answer

In rare cases, neither is available. This poses a problem for the SLP because it is very difficult to transcribe while engaged in conversation with a child, particularly if the child is very talkative. In such a case, a second transcriber could be used to minimize errors. However, the transcript must still be viewed with caution, and may yield little benefit.

question

Setting up the situation(s)

answer

Elicit a spontaneous sample vs. a set of responses During your interaction with the child, you MUST remember to allow for as much spontaneous conversation as possible.

question

Setting up the situation(s)

answer

DO NOT fall into the trap of asking question after question. Sometimes the best elicitation method is silence - sit back and allow the child to initiate the conversation.

question

Setting up the situation(s)

answer

If you have interesting materials and the child is at ease, he/she will talk readily. If you find that you must ask a probing question, make it open-ended rather than one that requires a yes/no or minimal response.

question

Eliciting a spontaneous sample

answer

Parallel talk is another way of stimulating conversation without asking a question - you "talk your way" through manipulation of materials.

question

For example, if you have a set of blocks and the child is shy about beginning to use them, you might start building a construction while describing your own actions:

answer

"I think I'll build a house"; "my house will be a little house so only little people can live in it"; "oh no, I don't have any more blocks"; "I don't know what to do"..... Often, the child will begin to problem-solve for you and start to converse more openly.

question

Selecting the corpus

answer

Once you have collected the sample, you need to determine how much of it will be used in your analysis

question

Determine the number of utterances & the variety of language uses

answer

...

question

Which utterances to include: A basic rule of thumb is to omit the first 10 utterances from the analysis.

answer

Often, a child's first few utterances may be less representative of his/her true skills. Once a child is comfortable, a truer sample of his/her language skills is presented. (If the child is familiar with the evaluator this may not be an issue).

question

How Many Utterances Do We Count?

answer

Depending upon the types of analyses the evaluator plans to use, the minimum number of requisite utterances varies.

question

For example, to obtain a more accurate indication of a child's lexical diversity (functional vocabulary), the TTR strongly suggests a minimum sample of 1000 words.

answer

For MLU analysis, a minimum of 100 utterances is preferred. For DSS analysis, the minimum is 50 subject-predicate utterances.

question

Intelligibility: The rule of thumb:

answer

if an utterance is unintelligible, it should be omitted.

question

If part of an utterance is unintelligible,

answer

the SLP must decide whether enough of it is intelligible to analyze, or if the accuracy of the analysis is impacted.

question

In my opinion, it is preferable to include fully intelligible utterances in your sample.

answer

However, you will want to make note of how many utterances are intelligible vs. those that are unintelligible - this gives you another piece of information about the functional quality of the child's communication skills.

question

Transcribing: When using language sample analysis, it is preferable that the person who collected the sample be the person to transcribe it.

answer

This is particularly true for samples that are not videotaped. Only the person who collected the sample is knowledgeable of the body language, context, facial and gestural interactions, etc.

question

Should you transcribe all the participants or just the client?

answer

If you are collecting a sample for ongoing analysis of one of your clients and you are the only person who will be using the sample, you may decide to transcribe just the child's portion.

question

Should you transcribe all the participants or just the client?

answer

If, however, you are sharing information with parents or colleagues, you may want to transcribe all participants. This is helpful when presenting information because you are able to provide the context in which the utterances occurred.

question

Should you transcribe all the participants or just the client?

answer

ALWAYS include a description of the task, setting, materials and methods used!

question

Orthography vs. phonetics

answer

Most people find it easier to transcribe a sample orthographically while others insist that phonetic transcription is essential to maintain accuracy.

question

Orthography vs. phonetics

answer

My own experience is that orthographic transcription is adequate with the exception of words that are mispronounced or unintelligible. In those instances, phonetic transcription allows for accurate recording of the utterance.

question

Orthography vs. phonetics

answer

It also provides a side-benefit of an articulation/ phonological inventory obtained in conversational context - one that can be used to supplement formal articulation testing or that can be used in place of it.

question

Transcription also requires accurate differentiation of utterances.

answer

This is easily done by attending to the supra-segmental characteristics such as pausing and intonation patterns.

question

Transcription also requires accurate differentiation of utterances.

answer

A pause of a second or more, indicates the end of one utterance and beginning of another. Likewise, a drop in intonation frequently indicates the end of an utterance.

question

Warning! Be aware of any specific rules attached to the analysis you will be using. For example, DSS analysis requires arbitrary division of compound sentences that use the same conjunction more than once.

answer

This statement, though uttered as one utterance, "Mom got me a cat and she got my brother a dog and she got my sister a fish", is scored with DSS as two utterances: 1) "Mom got me a cat and she got my brother a dog"; 2) "(and) she got my sister a fish".

question

Include a "key" to your Transcription

answer

A final consideration in transcription is the inclusion of a "key". Even if you are the only one who will be using the sample, you will want to designate what your abbreviations, participants, etc. are.

question

Include a "key" to your Transcription

answer

You will not remember this information several months from now (or even several weeks for some of us)!!

question

Some Common Language Sample Analyses

answer

Type-Token Ratio (TTR) Number of Different Words (NDW) Mean length of utterance (MLU) Developmental Sentence Scoring (DSS) Developmental Sentence Types (DST) T-Unit (terminal unit) C-unit (clause unit) Narrative Analysis Pragmatics Assessment All of these can be computed by hand.

question

Some Common Language Sample Analyses

answer

In addition to hand-scoring, there are computerized systems. However, these require the SLP to code utterances in accordance with individual program requirements. They also require the SLP to review the analyses carefully for errors - they are NOT foolproof. A few examples include:

question

SALT (Sentence Analysis for Language Transcription)

answer

is a computerized program that allows you to enter the sample and have a variety of analyses performed, including TTR, MLU, frequency of morpheme type, mazes, discourse.

question

The CHILDES database (Child Language Data Exchange System).

answer

Provides information and tools for language transcription and analysis - The CLAN program allows for a variety of analyses

question

Computerized Profiling

answer

Developed by Steven Long, provides a broad variety of analyses, including MLU and DSS

question

Practice with sample analysis

answer

In the next few slides, we'll practice using some of the techniques listed in the previous slides

question

These are by no means the only ways to analyze language samples but they will give you a good basis for understanding the process and its value, and they are some of the more common analyses used by SLPs

answer

We will score these by hand - it is important that you learn HOW to do this before relying on any computerized program.

question

TTR (Type-Token Ratio)& NDW (Number of Different Words)

answer

You will see from the resources available in this unit that TTR & NDW are variations of the same technique - for one you compute a ratio, and for the other you merely count.

question

On the next slide you will see a brief language "sample". This will serve as our practice sample.

answer

Using the TTR document & worksheet provided in the Unit resources, complete the analysis for the practice sample...

question

Mommy, I wanna go to the store with Daddy. Can I go, please? Why won't you answer me? Can I go? Thank-you, Mommy. Daddy, I'm gonna get my coat. Wait for me! What? What did you say? No, I want my blue coat. I don't want the other one.

answer

TTR & NDW results Be sure to refer to the TTR/NDW resources provided in the Unit. What did you find? Here's what I got: TTR = .64 NDW = 32 (different words) out of 50 (total)

question

According to the rules of TTR/NDW: "won't" & "don't" are counted as 1 word - "will + n't (not)" & "do + n't (not)". Contracted subject + verb (I'm) = 2 words

answer

I counted "wanna" and "gonna" each as a single word. These two forms are learned early, and as a "whole". Use of the infinitive "to" typically comes in later. I also counted "thank-you" as one word because it, too, is learned as a "whole".

question

Obviously - we don't have anywhere near a 1000 words so this likely isn't a true representation, but you understand how to complete this measure.

answer

Equally as important as any score, is my critical analysis of the sample contents in relation to the type of analysis I'm using. What else can you say about the vocabulary in this sample?

question

TTR & NDW

answer

When I look at my charted sample, I can see the following: The child used a few adjectives (other, blue) and articles (the) All utterances were simple statements - he/she didn't use any conjunctions All pronouns used were 1st & 2nd person AND were used correctly (I, you, my, me) Simple present (want, wanna, gonna, answer, get wait) & past (did) tense verbs were used, as well as some auxiliaries (can, will, contracted "am") Simple negation was used (no, contracted "not") A few prepositions (with, to, for) were present in the sample Overall, this child used a variety of vocabulary forms, with no observable errors

question

TTR & NDW: So, if we pretend that we had a 1000 word sample, we would have a TTR score (which would be at least average) and, just as importantly, we would be able to provide a descriptive analysis of this child's vocabulary.

answer

We can use this information to compare the TTR with vocabulary scores from std. tests - are both showing a similar result or are there inconsistencies? If we noted errors, we could also use this information to develop therapy goals, such as targeting specific types of vocabulary

question

While TTR/NDW gives us an idea of a child's lexical diversity, it doesn't take into consideration other aspects of the child's language

answer

But we can easily include these other aspects through application of some additional analyses Let's take a look.....

question

MLU (Mean Length of Utterance)

answer

MLU doesn't assess lexical diversity. It focuses on the number of morphemes the child uses in typical utterances From the information in your unit resources, take the same practice sample that we used for the TTR/NDW analysis and compute the MLU What did you get?

question

I got MLU = 4.63 (51 morphemes divided by 11 utterances).

answer

Let's say you had a sufficient size sample and the child was 4.3 yrs old. What does the MLU tell you? Could you also develop a descriptive summary as you did for the TTR/NDW?

question

Here are my answers to the previous 2 questions: The MLU score tells me that the child's expressive morphology is essentially within normal limits.

answer

I CAN develop a descriptive summary, i.e.: "This child demonstrates appropriate MLU for his/her age. His/her sample included appropriate use of contracted negatives (won't, don't), simple pronouns and simple verb forms. This child used both statement and question forms."

question

MLU But, let's see what happens if we change the sample slightly....

answer

How does it impact the MLU score? How does it impact the descriptive summary? Check out the altered sample on the next slide and compute the MLU....

question

Mommy, me wanna go to the store with Daddy. me can go, please? Why you not answer me? me can go? Thank-you, Mommy. Daddy, me gonna get me blue coat. Wait for me! What? What you did say? No, me want me blue coat. me not want the other one.

answer

The MLU = 4.54 (50 morphemes divided by 11 utterances) and is within normal limits, according to this measure. But are you satisfied with this? Do you see any problems with saying this child, who is over 4 yrs. old, has normal expressive language? This is where your descriptive analysis becomes very important....

question

The descriptive analysis may look something like this:

answer

"Though the child's MLU score suggests age-appropriate morphology, his/her sample included the following errors, which are atypical for his/her age: consistent substitution of me/I; inconsistent use of interrogative reversals (ie: "Can I go?", "Why you not answer me?"); inconsistent simplification of negatives (not/don't; not/won't)."

question

As you can see, MLU looks at expressive language in relation to the number of morphemes used but it doesn't take into consideration the errors - morphemes are counted whether they are used correctly or not

answer

The descriptive analysis is helpful in revealing the limitations of the numeric score but you, as an SLP, could find yourself in a difficult situation if your only analysis was the MLU and you are trying to justify the child's need for therapy.

question

DSS (Developmental Sentence Scoring) Let's take a look at the same samples, with the application of Lee's DSS

answer

Using the documents in this unit's resource section, compute the DSS for the first sample; and again for the altered sample What did you find?

question

For the first sample, the DSS = 11.77 (weighted scores total of 106 divided by 9 utterances - note that 2 utterances could not be included because they did not contain both a subject & predicate component)

answer

For the altered sample, the DSS = 5.22 (weighted scores total of 47 divided by 9 utterances) So, what does this mean according to Lee's weighted scores?

question

The mean DSS score for a child between 4.0 and 4.11 years is 8.04 with a standard deviation of 1.64 (so, a score of 6.40 and above is at least within normal limits for this age).

answer

If we are using percentiles, a score below the 10th% indicates a score below the average range, and a need for intervention.

question

For the first sample, a DSS score of 11.77 would be within the third standard deviation above the mean for a child of 4 yrs., and above the 90th % - well within the average range.

answer

For the second sample, the score is within the second standard deviation below the mean and below the 10th% rank, which is below the average range, and warrants therapy.

question

As with MLU, the descriptive analysis is very important. But because this particular measure includes an error weighting, the numeric score reflects the difference in syntax development between the two samples and provides a more objective measure for an IEP or IFSP Team in support of your recommendations.

answer

You can also see the importance of including both analyses in your evaluation.

question

In many cases, you will find an SLP using MLU analysis, or some variation of it, and no other analysis.

answer

While DSS is more time intensive until the SLP becomes familiar with it, its use with young children, and with older children who have significant language delays, is extremely beneficial.

question

In my opinion, both analyses should be used in order to provide a more comprehensive analysis of the child's language structure.

answer

...

question

Both MLU and DSS analyses are useful with children up to the age of 6 years or so, and with school-aged children who demonstrate significant delays in language structure.

answer

However, once a child is using the early syntactic forms correctly, and is over the age of 6 years, these analyses have limited usefulness in identifying expressive structural language difficulties. Two additional analyses the SLP may find helpful are T-Unit and C-Unit

question

T-Unit means "terminal unit" and was initially identified by Kellogg Hunt in the mid-1960's. He used it as a method to analyze written language skills of school-aged children. Simply stated, a t-unit equals a complete thought - a sentence including all its dependent clauses

answer

C-Unit stands for "communication unit" and is the equivalent measure for spoken language

question

You will see these measures used in research on language development and delay, particularly with older children, but they are also useful for assessment, in combination with other standardized tests. We will examine them in greater depth in an additional lecture.

answer

...

question

Some final thoughts on Language Sample Analysis

answer

You probably experienced some frustration and had many questions when attempting the analyses.

question

It takes lots and lots and lots of practice to become skilled in language sample analysis. As with standardized testing, you don't want to rely on only one measure.

answer

By adding language sample analysis to your assessment repertoire, you Increase validity & reliability of your assessment Decrease potential cultural, racial and ethnic bias Collect descriptive information that is critical to development of appropriate recommendations and, should intervention be warranted - development of appropriate goals & objectives

Language Sample Analysis – Flashcards

Unlock all answers in this set

Haven't found what you were looking for?

Search for samples, answers to your questions and flashcards