File size: 4,716 Bytes
5a99584
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
---
language:
  - luo
  - bxk
  - lri
  - rag
  - swa
license: cc-by-4.0
task_categories:
  - translation
tags:
  - kenyan-languages
  - dholuo
  - lubukusu
  - lumarachi
  - lulogooli
  - swahili
  - low-resource-languages
  - african-languages
pretty_name: KenTrans
size_categories:
  - 10K<n<100K
configs:
- config_name: dho
  data_files: "dho/*.parquet"
- config_name: lbk
  data_files: "lbk/*.parquet"
- config_name: lch
  data_files: "lch/*.parquet"
- config_name: llg
  data_files: "llg/*.parquet"
---

# KenTrans: Kenyan Languages to Swahili Translation Dataset

## Dataset Structure
**KenTrans** is a parallel corpus between **Swahili** and three Kenyan languages (with multiple Luhya dialects). The dataset contains **11,795** sentence pairs translated **into Swahili**:

- **Dholuo → Swahili:** 4,222 pairs
- **Luhya → Swahili (total):** 7,573 pairs across three dialects
  - **Lumarachi (lch):** 2,475 pairs
  - **Lulogooli (llg):** 3,692 pairs
  - **Lubukusu (lbk):** 1,406 pairs

The dataset is provided in **Parquet format**, which is compatible with Hugging Face `datasets` library version 4.0.0 and above.

Each example contains parallel text with the following fields:
- **source**: Original sentence in the source language
- **target**: Translation in Swahili

> Example
> ```python
> {
>   'source': 'OSIEPE MA KENDE',
>   'target': 'MARAFIKI WA DHATI'
> }
> ```

### Languages & Codes

| Language / Dialect | Code | Family / Notes                |
|--------------------|------|-------------------------------|
| Dholuo (Luo)       | dho  | Nilotic (western Kenya)       |
| Lubukusu (Bukusu)  | lbk  | Bantu, Luhya dialect          |
| Lumarachi (Marachi)| lch  | Bantu, Luhya dialect          |
| Lulogooli (Logooli)| llg  | Bantu, Luhya dialect          |
| Swahili            | swa  | Target language for all pairs |

---

## Usage

### Loading with 🤗 Datasets

**Compatible with datasets 4.0.0+** (No `trust_remote_code` needed!)

```python
from datasets import load_dataset

# Load Dholuo → Swahili
dho = load_dataset("Kencorpus/KenTrans", "dho")

# Load Lubukusu → Swahili
lbk = load_dataset("Kencorpus/KenTrans", "lbk")

# Load Lumarachi → Swahili
lch = load_dataset("Kencorpus/KenTrans", "lch")

# Load Lulogooli → Swahili
llg = load_dataset("Kencorpus/KenTrans", "llg")

# Access the data
print(dho['train'][0])
# Output: {'id': 'dho_dho_combined.txt_0', 'source': '6AM DALA FM NEWS...', 'target': 'VIDOKEZI VYA HABARI...', ...}
```

---

### Dataset Format

The dataset is stored in **Parquet format** with the following structure:
- Each language pair has its own Parquet file (e.g., `dho-train.parquet`)
- Each row represents a parallel sentence pair
- All metadata is included in the Parquet schema

## Dataset Format

The dataset is distributed as **Parquet files** for optimal performance and compatibility:

- **Format**: Apache Parquet (columnar storage)
- **Encoding**: UTF-8
- **File naming**: `{language}-train.parquet` (e.g., `dho-train.parquet`)
- **Compatibility**: Works with `datasets` 4.0.0+ without custom loading scripts

---

## Data Fields (Programmatic View)

When parsed into records, each example can be represented with:

- `id`: Unique identifier (e.g., `dho_0773_dho_tr.txt_0`)
- `source`: Source language sentence
- `target`: Target Swahili sentence
- `src_lang`: One of `dho`, `lbk`, `lch`, `llg`
- `tgt_lang`: Always `swa`
- `pair`: Language pair as `"{src}-{tgt}"` (e.g., `dho-swa`)
- `filename`: Source filename (e.g., `0773_dho_tr.txt`)

---


## Sources

## Translators (Acknowledgements)
**Dholuo → Swahili**
- Mercy Lavinca Oduoll (Coordinator)
- Bildad Okebe
- Immaculate Ochieng
- Mary Muma. 

**Luhya (Logooli) → Swahili**
- Phillip Lumwamu (Coordinator)
- Kints Mugoha Musungu
- Vivian Alivitsa
- Joseph Ambwere
- Joyline Ingasiani. 

**Luhya (Bukusu) → Swahili**
- Martin Barasa Mulwale (Coordinator)
- Samwel Ralph Nyongesa
- Tobias Shikuku
- Phelisters N Simiyu

**Luhya (Marachi) → Swahili**
- Judith Awinja (Coordinator)
- Evans Owino
- Belinda Oduor
- Frankline Mwaro




## Dataset Curator
- Indede, Florence (Maseno University)
- McOnyango, Owen (Maseno University)
- Wanzare, Lilian D.A. (Maseno University)
- Wanjawa, Barack (University of Nairobi)
- Ombui, Edward (Africa Nazarene University)
- Muchemi, Lawrence (University of Nairobi)

## Research Paper

- Wanjawa, B.W., Wanzare, L.D., Indede, F., McOnyango, O., Ombui, E., & Muchemi, L. (2022). Kencorpus: A Kenyan Language Corpus of Swahili, Dholuo and Luhya for Natural Language Processing Tasks. ArXiv, abs/2208.12081. https://arxiv.org/abs/2208.12081

## Links
Dataverse: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/KLCKL5