Train & Constrain: Phonologically Informed Tongue-Twister Generation from Topics and Paraphrases
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914977200209920 |
|---|---|
| author | Loakman, Tyler Tang, Chen Lin, Chenghua |
| author_facet | Loakman, Tyler Tang, Chen Lin, Chenghua |
| contents | Previous work in phonologically and phonetically grounded language generation has mainly focused on domains such as puns and poetry. In this article, we present new work on the generation of English tongue twisters - a form of language that is required to be conditioned on a phoneme level to maximize sound overlap, while maintaining semantic consistency with an input topic or phrase and still being grammatically correct. We present TwisterLister, a pipeline for generating phonologically informed tongue twisters from large language models (LLMs) that we use to generate TwistList 2.0, the largest annotated dataset of tongue twisters to date, consisting of 17K+ examples from a combination of human and LLM authors. Our generation pipeline involves the use of a phonologically constrained vocabulary alongside LLM prompting to generate novel, non-derivative tongue twister examples. We additionally present the results of automatic and human evaluation of smaller models trained on our generated dataset to demonstrate the extent to which phonologically motivated language types can be generated without explicit injection of phonological knowledge. Additionally, we introduce a phoneme-aware constrained decoding module (PACD) that can be integrated into an autoregressive language model and demonstrate that this method generates good quality tongue twisters both with and without fine-tuning the underlying language model. We also design and implement a range of automatic metrics for the task of tongue twister generation that is phonologically motivated and captures the unique essence of tongue twisters, primarily based on phonemic edit distance (PED) |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_13901 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Train & Constrain: Phonologically Informed Tongue-Twister Generation from Topics and Paraphrases Loakman, Tyler Tang, Chen Lin, Chenghua Computation and Language Previous work in phonologically and phonetically grounded language generation has mainly focused on domains such as puns and poetry. In this article, we present new work on the generation of English tongue twisters - a form of language that is required to be conditioned on a phoneme level to maximize sound overlap, while maintaining semantic consistency with an input topic or phrase and still being grammatically correct. We present TwisterLister, a pipeline for generating phonologically informed tongue twisters from large language models (LLMs) that we use to generate TwistList 2.0, the largest annotated dataset of tongue twisters to date, consisting of 17K+ examples from a combination of human and LLM authors. Our generation pipeline involves the use of a phonologically constrained vocabulary alongside LLM prompting to generate novel, non-derivative tongue twister examples. We additionally present the results of automatic and human evaluation of smaller models trained on our generated dataset to demonstrate the extent to which phonologically motivated language types can be generated without explicit injection of phonological knowledge. Additionally, we introduce a phoneme-aware constrained decoding module (PACD) that can be integrated into an autoregressive language model and demonstrate that this method generates good quality tongue twisters both with and without fine-tuning the underlying language model. We also design and implement a range of automatic metrics for the task of tongue twister generation that is phonologically motivated and captures the unique essence of tongue twisters, primarily based on phonemic edit distance (PED) |
| title | Train & Constrain: Phonologically Informed Tongue-Twister Generation from Topics and Paraphrases |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2403.13901 |