Data-Driven Syllabification for Middle Dutch

Wouter Haverals; F.B. Karsdorp; Mike Kestemont

doi:10.16995/dm.83

Data-Driven Syllabification for Middle Dutch

Wouter Haverals, F.B. Karsdorp, Mike Kestemont

Meertens Instituut - Nederlandse Etnologie

Onderzoeksoutput: Bijdrage aan wetenschappelijk tijdschrift/periodieke uitgave › Artikel › Wetenschappelijk › peer review

Samenvatting

The task of automatically separating Middle Dutch words into syllables is a challenging one. A first method was presented by Bouma and Hermans (2012), who combined a rule-based finite-state component with data-driven error correction. Achieving an average word accuracy of 96.5%, their system surely is a satisfactory one, although it leaves room for improvement. Generally speaking, rule-based methods are less attractive for dealing with a medieval language like Middle Dutch, where not only each dialect has its own spelling preferences, but where there is also much idiosyncratic variation among scribes. This paper presents a different method for the task of automatically syllabifying Middle Dutch words, which does not rely on a set of pre-defined linguistic information. Using a Recurrent Neural Network (RNN) with Long-Short-Term Memory cells (LSTM), we obtain a system which outperforms the rule-based method both in robustness and in effort.

Originele taal-2	Engels
Pagina's (van-tot)	1-23
Tijdschrift	Digital Medievalist
Volume	1
Nummer van het tijdschrift	2
DOI's	https://doi.org/10.16995/dm.83
Status	Gepubliceerd - 04 nov. 2019

Toegang tot document

10.16995/dm.83Licentie: CC BY

Citeer dit

@article{509b9c69f6bf4e5bb77e2058e4c6e7ef,

title = "Data-Driven Syllabification for Middle Dutch",

abstract = "The task of automatically separating Middle Dutch words into syllables is a challenging one. A first method was presented by Bouma and Hermans (2012), who combined a rule-based finite-state component with data-driven error correction. Achieving an average word accuracy of 96.5%, their system surely is a satisfactory one, although it leaves room for improvement. Generally speaking, rule-based methods are less attractive for dealing with a medieval language like Middle Dutch, where not only each dialect has its own spelling preferences, but where there is also much idiosyncratic variation among scribes. This paper presents a different method for the task of automatically syllabifying Middle Dutch words, which does not rely on a set of pre-defined linguistic information. Using a Recurrent Neural Network (RNN) with Long-Short-Term Memory cells (LSTM), we obtain a system which outperforms the rule-based method both in robustness and in effort.",

author = "Wouter Haverals and F.B. Karsdorp and Mike Kestemont",

year = "2019",

month = nov,

day = "4",

doi = "10.16995/dm.83",

language = "English",

volume = "1",

pages = "1--23",

journal = "Digital Medievalist",

issn = "1715-0736",

publisher = "Ubiquity Press Limited",

number = "2",

}

TY - JOUR

T1 - Data-Driven Syllabification for Middle Dutch

AU - Haverals, Wouter

AU - Karsdorp, F.B.

AU - Kestemont, Mike

PY - 2019/11/4

Y1 - 2019/11/4

N2 - The task of automatically separating Middle Dutch words into syllables is a challenging one. A first method was presented by Bouma and Hermans (2012), who combined a rule-based finite-state component with data-driven error correction. Achieving an average word accuracy of 96.5%, their system surely is a satisfactory one, although it leaves room for improvement. Generally speaking, rule-based methods are less attractive for dealing with a medieval language like Middle Dutch, where not only each dialect has its own spelling preferences, but where there is also much idiosyncratic variation among scribes. This paper presents a different method for the task of automatically syllabifying Middle Dutch words, which does not rely on a set of pre-defined linguistic information. Using a Recurrent Neural Network (RNN) with Long-Short-Term Memory cells (LSTM), we obtain a system which outperforms the rule-based method both in robustness and in effort.

AB - The task of automatically separating Middle Dutch words into syllables is a challenging one. A first method was presented by Bouma and Hermans (2012), who combined a rule-based finite-state component with data-driven error correction. Achieving an average word accuracy of 96.5%, their system surely is a satisfactory one, although it leaves room for improvement. Generally speaking, rule-based methods are less attractive for dealing with a medieval language like Middle Dutch, where not only each dialect has its own spelling preferences, but where there is also much idiosyncratic variation among scribes. This paper presents a different method for the task of automatically syllabifying Middle Dutch words, which does not rely on a set of pre-defined linguistic information. Using a Recurrent Neural Network (RNN) with Long-Short-Term Memory cells (LSTM), we obtain a system which outperforms the rule-based method both in robustness and in effort.

U2 - 10.16995/dm.83

DO - 10.16995/dm.83

M3 - Article

SN - 1715-0736

VL - 1

SP - 1

EP - 23

JO - Digital Medievalist

JF - Digital Medievalist

IS - 2

ER -

Data-Driven Syllabification for Middle Dutch

Samenvatting

Toegang tot document

Vingerafdruk

Citeer dit