lingtypology: Typological databases API

George Moroz

2019-11-22

lingtypology provides an ability to download data from these typological databases

All database function names have identical structure: database_name.feature. All functions have as first argument feature. All functions create dataframe with column language that can be used in map.feature() function. It should be noted that all functions cut out the data that can’t be maped, so if you want to prevent functions from this behaviour set argument na.rm to FALSE.

1. WALS

The names of the WALS features can be typed in a lower case. This function preserves coordinates from WALS, so you can map coordinates from the WALS or use coordinates from lingtypology.

df <- wals.feature(c("1a", "20a"))
head(df)
map.feature(df$language,
            features = df$`1a`,
            latitude = df$latitude,
            longitude = df$longitude,
            label = df$language,
            title = "Consonant Inventories")

2. AUTOTYP

The AUTOTYP features are listed on the GitHub page. You can use more human way with spaces.

df <- autotyp.feature(c('Gender', 'Numeral classifiers'))
head(df)
map.feature(df$language,
            features = df$NumClass.Presence,
            label = df$language,
            title = "Presence of Numeral Classifiers")

3. PHOIBLE

I used only four features from PHOIBLE: the number of phonemes, the number of consonants, the number of tones and the number of vowels. If you need only a set of them, just specify it in the features argument. Since there is a lot of doubling information in the PHOIBLE database, there is an argument source.

df <- phoible.feature(source = "UPSID")
head(df)
map.feature(df$language,
            features = df$phonemes,
            label = df$language,
            title = "Number of Phonemes")

4. AfBo

The AfBo database has a lot of features that distinguish affix functions, but again you can use a bare function without any arguments to download the whole database. There will be no difference in time, since this function downloads the whole database to your PC. The main destinction is that this database provides recipient and donor languages, so other column names should be used.

df <- afbo.feature(c("adjectivizer", "adverbializer"))
head(df)
map.feature(df$Recipient.name,
            features = df$adjectivizer,
            label = df$Recipient.name,
            title = "Borrowed adjectivizer affixes")

5. SAILS

The SAILS database provide a lot of features, so the function work with their ids:

df <- sails.feature(features = "ics10")
head(df)
map.feature(df$language,
            features = df$ics10_description,
            longitude = df$longitude,
            latitude = df$latitude,
            label = df$language,
            title = "Are there numeral classifiers?")

6. ABVD

The ABVD database is a lexical database, so it is different from clld databases. First of all, ABVD has its own language classification ids. The information about the same language from different sources can be received from these database different ids. So I select several languages and map them coloring by word with the meaning ‘hand’.

df <- abvd.feature(50:55)
head(df)
new_df <- df[df$word == "hand",]
map.feature(new_df$language,
            features = new_df$item,
            label = new_df$language)

7. UraLex

uralex.feature downloads data from UraLex basic vocabulary dataset. Original language names are stored in the language variable. Converted language names for map.feature are stored in the language2 variable.

df <- uralex.feature()
df <- df[df$uralex_mng == "crush",]

map.feature(df$language2,
            label = df$item,
            title = "crush")