TBDBr

The goal of the TBDBr package is to provide easy access to TalkBankDB through an R API.

TalkBank (www.talkbank.org) is a vast collection of audio and video recordings across more than 50 languages linked to richly annotated transcriptions. The goal of TalkBank is to foster fundamental research in the study of human communication with an emphasis on spoken communication. The collection includes repositories in 14 research areas, including language development, adult conversation, multilingualism, and disorders which affect communication. These collections are freely available to support the research of human language and communication. TalkBank has been funded by the National Science Foundation and National Institute of Health since 1999.

TalkBankDB (www.talkbank.org/DB) is a database and set of tools for exploring TalkBank’s media and transcripts, specify data to be extracted, and pass these data on to statistical programs for further analysis. The TBDBr package (TalkBankDataBase - r) provides easy access to all information within TalkBankDB, including clinical collections. Clinical Banks are password protected. Visit www.talkbank.org to learn about gaining access to these collections.

Installation

You can install the released version of talkbank from GitHub with:

# install.packages("devtools")
# devtools::install_github("Talkbank/TBDBr")
library(TBDBr)

Functionality

TBDBr allows access to data from TalkBankDB through several functions. For example, to get a table of utterances from a particular transcript in the childes/Eng-NA/MacWhinney collection:

example_utterances <- getUtterances(corpusName = 'childes',
                                    corpora = c('childes', 'Eng-NA', 'MacWhinney', '010411a'));
#> [1] "Fetching data, please wait..."
#> [1] "Success!"
example_utterances[10:14,]
#>    filename                              path uid who         role postcodes
#> 10  010411a childes/Eng-NA/MacWhinney/010411a   9 CHI Target_Child      NULL
#> 11  010411a childes/Eng-NA/MacWhinney/010411a  10 FAT       Father      NULL
#> 12  010411a childes/Eng-NA/MacWhinney/010411a  11 CHI Target_Child      NULL
#> 13  010411a childes/Eng-NA/MacWhinney/010411a  12 FAT       Father      NULL
#> 14  010411a childes/Eng-NA/MacWhinney/010411a  13 CHI Target_Child      NULL
#>    gems                                         utterance startTime endTime
#> 10 NULL                                      what's that     33.398  33.714
#> 11 NULL                          taperecorder over there     33.714  34.884
#> 12 NULL                                               hm     34.884  35.999
#> 13 NULL do you have some nice little things to say to it     35.999  37.818
#> 14 NULL                                               hi     37.818  38.394

The available functions for accessing different data sets are below. The arguments for each function are documented, which can be accessed by typing ?functionName (e.g. ?getTranscriots()).

Because the options for each talkbank collection and function vary, the getLegalValues() function can help determine the possible options. getLegalValues() will interactively return options available at each level of the database.

For troubleshooting, an additional function, validPath(), will return whether a given path is valid.

validPath(c('childes', 'childes', 'Clinical'));
#> [1] "Invalid path at:  childes"
#> [1] FALSE

If the path is not valid, it will return which level of the query is incorrect

validPath(c('childes', 'childes', 'somethingThatDoesNotExist'))
#> [1] "Invalid path at:  childes"
#> [1] FALSE

To access clinical collections, use the argument auth = TRUE. A pop-up will ask you to select the clinical bank you are trying to access and to enter the correct username and password for that database

aphasia_transcrips <- getTranscripts(corpusName = 'aphasia',
                                     corpora = c('aphasia', 'English', 'Aphasia', 'Adler'),
                                     auth = TRUE);

More examples are located here