This study examines the use of “decontextualized language” in the early home environment of both typically-developing children and children with early brain injury.
What is decontextualized language?
Decontextualized (“dxt”) language is defined as language that is removed from the here-and-now. Examples of this kind of decontextualized language use are evident in early conversations between parents and children, particularly in uses of extended discourse such as explanations, narratives and pretend play.
We examined 5 types of decontextualized utterances:
We also examined a number of lexical measures for dxt utterances:
MSW (Mean Speakers per Word)
To make clear just what this means we first need to explain what we mean by a word’s “speaker count”.
For any given word, say “dog”, we can check our canonical corpus of transcripts to see how many many kids (or parents) said this word at least once in the course of a visit. So let’s say that the child subject A said “dog” at two visits. And let’s say that child subject B said “dog” at three visits. If these two kids happened to be the only subjects that used the word “dog” in the course of the first ten visits, the child speaker count for “dog” would be five (2 for A and 3 for B). That is, a “speaker count” tells you how many speakers used a particular word at least once in the course of a visit during the first ten visits.
So, when looking at a given speaker’s utterances for a particular transcript, we can get the “speaker counts” for each word and then take the average. This average we call the “mean speakers per word”.
The data files and other documents associated with this study can be found in Ece’s working directory on the LDP file server (Analord) in the Groups volume:
afp://analord.spc.uchicago.edu/Groups/Grads/Ece/Decontextualized
Decontextualized
├── FinalTranscripts
│ ├── 05 # transcripts for session 5
│ └── 08 # transcripts for session 8
└── TrainingManuals
├── DXTSubcategories.doc # coding manual
├── Examples
└── schedule.xls # coding schedule
No specific coding template was used for this study. Rather, annotaters simply addded two columns (mother_dxt and child_dxt) to copies of the original transcript.