Afri-Bigrams

This dataset contains a large-scale collection of word bigrams extracted from text across 154 African languages. It is meticulously designed to support general research and analysis of African languages, providing foundational data based on two-word sequences.

Background

This dataset provides foundational linguistic data, specifically word bigram frequencies, for African languages. This type of structured data is crucial for various computational linguistic tasks, allowing researchers to explore lexical co-occurrence patterns across 154 languages.

Approach Overview

The dataset exclusively includes word bigrams (sequences of two words).

The data provides the raw frequency counts of these word bigrams for each language. Word bigrams are valuable for analyzing lexical co-occurrence and capturing basic syntactic patterns within the corpus used for extraction.

Dataset Structure

Each row in the dataset represents one word bigram and its frequency for a specific language.

Column	Description
bigram	Extracted sequence (word pair).
count	Frequency of the bigram within that language.
lang	Two-letter language code.
lang_id	Numeric identifier for the language.

Use Cases

• Statistical analysis of word co-occurrence across African languages.
• Development of language models and embedding techniques.
• Linguistic pattern analysis focusing on common lexical sequences.
• Supporting general multilingual NLP research.

Languages

The dataset covers the following languages. Note that representation may vary across languages.

Acholi (ach), Dangme (ada), Afrikaans (afr), Akan (aka), Alur (alz), Amharic (amh), Algerian Arabic (arq), Moroccan Arabic (ary), Egyptian Arabic (arz), Bamanankan (bam), Basaa (bas), Baoulé (bci), Bemba (bem), Edo (bin), Bulu (bum), Bilen (byn), Chopi (cce), Chuwabu (chw), Chokwe (cjk), Coptic (cop), Seychelles French Creole (crs), Southwestern Dinka (dik), Dinka (din), Zarma (dje), Lukpa (dop), Duala (dua), Jula (dyu), Efik (efi), Éwé (ewe), Fon (fon), Pulaar (fuc), Fulah (ful), Nigerian Fulfulde (fuv), Ga (gaa), Gun (guw), Hausa (hau), Herero (her), Igbo (ibo), Esan (ish), Isoko (iso), Kabyle (kab), Kamba (kam), Kanuri (kau), Kabiyè (kbp), Kabuverdianu (kea), Gikuyu (kik), Kinyarwanda (kin), Kimbundu (kmb), Kongo (kon), Konzo (koo), Kaonde (kqn), Krio (kri), Kisi, Southern (kss), Oshiwambo (kua), Kwangali (kwn), Kikongo (kwy), Lamba (lam), Lingala (lin), Lozi (loz), Luba-Kasai (lua), Luba-Katanga (lub), Luvale (lue), Ganda (lug), Lunda (lun), Dholuo (luo), Mende (men), Morisyen (mfe), Mambwe-Lungu (mgr), Malagasy (mlg), Moore (mos), Mozambican Sign Language (mzy), Min Nan Chinese (nan), Nyemba (nba), Ndau (ndc), Ndonga (ndo), Lomwe (ngl), Northern Sotho (nso), Chichewa (nya), Nyaneka (nyk), Nyankore (nyn), Nyungwe (nyu), Nzema (nzi), Okpe (oke), Oromo (orm), Nigerian Pidgin (pcm), Merina Malagasy (plt), Tarifit (rif), Ruund (rnd), Rundi (run), Sango (sag), Sena (seh), South African Sign Language (sfs), Tachelhit (shi), Sidaama (sid), Shona (sna), Somali (som), Songe (sop), Swati (ssw), Swahili (swa), Congo Swahili (swc), Swahili (swh), Tigrigna (tir), Tiv (tiv), Tetela (tll), Tamashek (tmh), Tonga (tog), Tonga (toh), Tonga (toi), Tswa (tsc), Setswana (tsn), Tsonga (tso), Tooro (ttj), Tumbuka (tum), Umbundu (umb), Urhobo (urh), Venda (ven), Makhuwa (vmw), Wolaytta (wal), Cameroon Pidgin (wes), Wolof (wol), Xhosa (xho), Yao (yao), Yoruba (yor), Zimbabwe Sign Language (zib), Zande (zne), Zambian Sign Language (zsl), Zulu (zul)