Package {wordbankr}

wordbankr: Accessing the 'Wordbank' Database

Description

Connecting to 'Wordbank' https://wordbank.stanford.edu, an open repository for developmental vocabulary data from the MacArthur-Bates Communicative Development Inventories (Frank et al. 2017 doi:10.1017/S0305000916000209). Data are read from a versioned dataset hosted on 'Redivis', so analyses can be pinned to a specific release.

Author(s)

Maintainer: Mika Braginsky mika.br@gmail.com [copyright holder]

Other contributors:

• Daniel Yurovsky yurovsky@stanford.edu [contributor]

• Michael Frank mcfrank@stanford.edu [contributor]

• Danielle Kellier dkellier@stanford.edu [contributor]

• Alvin Tan tanawm@stanford.edu [contributor]

See Also

Useful links:

• https://langcog.github.io/wordbankr/

• https://github.com/langcog/wordbankr/

• Report bugs at https://github.com/langcog/wordbankr/issues/

Deprecated database arguments

Description

As of wordbankr 2.0, data are retrieved from the versioned Wordbank dataset on Redivis rather than a MySQL database; 'db_args' and 'connect_to_wordbank()' are deprecated and ignored.

Usage

check_db_args(db_args)

connect_to_wordbank(db_args = NULL)

get_wordbank_args()

Arguments

Value

check_db_args(): no return value, called for its side effect (a warning if db_args is supplied). connect_to_wordbank(): the Wordbank dataset reference, as returned by wb_dataset(). get_wordbank_args(): a list with elements organization, dataset, and version identifying the Redivis dataset.

Fit age of acquisition estimates for Wordbank data

Description

For each item in the input data, estimate its age of acquisition as the earliest age (in months) at which the proportion of children who understand/produce the item is greater than some threshold. The proportions used can be empirical or first smoothed by a model.

Usage

fit_aoa(
  instrument_data,
  measure = "produces",
  method = "glm",
  proportion = 0.5,
  age_min = min(instrument_data$age, na.rm = TRUE),
  age_max = max(instrument_data$age, na.rm = TRUE)
)

Arguments

Value

A data frame where every row is an item, the item-level columns from the input data are preserved, and the aoa column contains the age of acquisition estimates.

Examples

eng_ws_data <- get_instrument_data(language = "English (American)",
                                   form = "WS",
                                   items = c("item_1", "item_42"),
                                   administration_info = TRUE)
if (!is.null(eng_ws_data)) eng_ws_aoa <- fit_aoa(eng_ws_data)

Fit quantiles to vocabulary sizes using quantile regression

Description

Fit quantiles to vocabulary sizes using quantile regression

Usage

fit_vocab_quantiles(vocab_data, measure, group, quantiles = "standard")

Arguments

Value

A data frame with the columns "language", "form", "age", group (if specified), "quantile", and measure, where measure is the fit vocabulary value for that quantile at that age.

Examples

eng_wg <- get_administration_data(language = "English (American)",
                                  form = "WG",
                                  include_demographic_info = TRUE)
if (!is.null(eng_wg)) {
  vocab_quantiles <- fit_vocab_quantiles(eng_wg, production)
  vocab_quantiles_sex <- fit_vocab_quantiles(eng_wg, production, sex)
  vocab_quartiles <- fit_vocab_quantiles(eng_wg, production, quantiles = "quartiles")
}

Get the Wordbank by-administration data

Description

Get the Wordbank by-administration data

Usage

get_administration_data(
  language = NULL,
  form = NULL,
  filter_age = TRUE,
  include_demographic_info = FALSE,
  include_birth_info = FALSE,
  include_health_conditions = FALSE,
  include_language_exposure = FALSE,
  include_study_internal_id = FALSE,
  version = "current"
)

Arguments

Value

A data frame where each row is a CDI administration and each column is a variable about the administration or the corresponding child, including which dataset_version it came from.

Examples

english_ws_admins <- get_administration_data("English (American)", "WS")

Get cached age-of-acquisition estimates

Description

Age-of-acquisition estimates for every word item on every instrument, precomputed with fit_aoa (glm method, 50 data release. aoa is NA for items that do not reach the threshold within the instrument's age range.

Usage

get_aoa(language = NULL, form = NULL, measure = NULL, version = "current")

Arguments

Value

A data frame with one row per instrument item and measure: language, form, item_id, item_definition, category, uni_lemma, measure, aoa, dataset_version.

Examples

danish_aoa <- get_aoa(language = "Danish", form = "WS")

Get item-by-age summary statistics for items across languages

Description

Get item-by-age summary statistics for items across languages

Usage

get_crossling_data(uni_lemmas, version = "current")

Arguments

Value

A dataframe with a row for each combination of language, item, and age, and columns for summary statistics for the group: number of children (n_children), means (comprehension, production), standard deviations (comprehension_sd, production_sd); and item-level variables (item_id, definition, uni_lemma, lexical_category, lexical_class, dataset_version).

Examples

## Not run: 
# downloads item-level data from every instrument, which takes minutes
crossling_data <- get_crossling_data(uni_lemmas = "dog")

## End(Not run)

Get the uni_lemmas available in Wordbank

Description

Get the uni_lemmas available in Wordbank

Usage

get_crossling_items(version = "current")

Arguments

Value

A data frame with the columns uni_lemma and dataset_version.

Examples

uni_lemmas <- get_crossling_items()

Get the Wordbank data sources

Description

Get the Wordbank data sources

Usage

get_datasets(
  language = NULL,
  form = NULL,
  admin_data = FALSE,
  version = "current"
)

Arguments

Value

A data frame where each row is a particular dataset and its characteristics, including which dataset_version it came from.

Examples

english_ws_datasets <- get_datasets("English (American)", "WS")

Get multilingual item embeddings

Description

Semantic embeddings for every unique word item definition, computed with Google's multilingual gemini-embedding-001 model (768 dimensions). All languages share one embedding space, so cosine similarities are meaningful both within and across languages.

Usage

get_embeddings(language = NULL, version = "current")

Arguments

Value

A data frame with one row per unique item definition: language, item_definition, embedding (a list-column of numeric vectors), and dataset_version.

Examples

danish_embeddings <- get_embeddings(language = "Danish")

Get the Wordbank administration-by-item data

Description

Get the Wordbank administration-by-item data

Usage

get_instrument_data(
  language,
  form,
  items = NULL,
  administration_info = FALSE,
  item_info = FALSE,
  version = "current",
  ...
)

Arguments

Value

A data frame where each row contains the values (value, produces, understands) of a given item (item_id) for a given administration (data_id), with additional columns of variables about the administration and item, as specified, and dataset_version.

Examples

eng_ws_data <- get_instrument_data(language = "English (American)",
                                   form = "WS",
                                   items = c("item_1", "item_42"))

Get the Wordbank instruments

Description

Get the Wordbank instruments

Usage

get_instruments(version = "current")

Arguments

Value

A data frame where each row is a CDI instrument and each column is a variable about the instrument (instrument_id, language, form, form_type, age_min, age_max, has_grammar, unilemma_coverage, dataset_version).

Examples

instruments <- get_instruments()

Get the Wordbank by-item data

Description

Get the Wordbank by-item data

Usage

get_item_data(language = NULL, form = NULL, version = "current")

Arguments

Value

A data frame where each row is a CDI item and each column is a variable about it: item_id, item_kind, item_definition, english_gloss, language, form, form_type, category, lexical_category, lexical_class, complexity_category, uni_lemma, dataset_version.

Examples

english_ws_items <- get_item_data("English (American)", "WS")

Get item-by-age summary statistics

Description

Get item-by-age summary statistics

Usage

summarise_items(item_data, version = "current")

Arguments

Value

A dataframe with a row for each combination of item and age, and columns for summary statistics for the group: number of children (n_children), means (comprehension, production), standard deviations (comprehension_sd, production_sd); also retains item-level variables from lang_items (item_id, item_definition, uni_lemma, lexical_category) and dataset_version.

Examples

italian_items <- get_item_data(language = "Italian", form = "WG")
if (!is.null(italian_items)) {
  italian_dog <- dplyr::filter(italian_items, uni_lemma == "dog")
  italian_dog_summary <- summarise_items(italian_dog)
}

The Wordbank dataset on Redivis

Description

Returns a reference to the Wordbank Redivis dataset (https://stanford.redivis.com/datasets/627v-9ewzpdvz0).

Usage

wb_dataset(version = "current")

Arguments

Value

A redivis dataset reference.