From speech to sign – learning Swedish Sign Language as a second language
Most of us have experienced learning a second language (L2). However, what happens when the new subject of study is not spoken but is signed? Distinct from other languages, signed languages operate in a different modality, as they are expressed through the hands and face, and are perceived through the eyes. Swedish sign language (SSL) is the primary language of deaf people in Sweden and is used in communication between deaf people and with other proficient signers. However, most education in regard to SSL aims to teach the language as an L2 to hearing beginners. This education is often intended for students to later fill important roles in society, serving as sign language interpreters or teachers of the deaf.
Despite the long tradition of SSL being taught as an L2, there is a relative lack of research on its acquisition as a second language. Our research will address the question of what happens when one learns a sign language as an L2 as an adult. Of particular focus here will be the acquisition of two specific linguistic components that may be difficult for hearing learners to acquire: i) mouth actions, which together with the hands express the meaning of signs, and ii) modifying signs and classifier signs (i.e. signs that are modified according to meaning and reference). Results from this study will contribute significantly to our understanding of the mechanisms underlying the learning of a (signed) language and how to best promote the teaching of SSL as an L2.
Despite the long tradition of SSL being taught as an L2, there is a relative lack of research on its acquisition as a second language. Our research will address the question of what happens when one learns a sign language as an L2 as an adult. Of particular focus here will be the acquisition of two specific linguistic components that may be difficult for hearing learners to acquire: i) mouth actions, which together with the hands express the meaning of signs, and ii) modifying signs and classifier signs (i.e. signs that are modified according to meaning and reference). Results from this study will contribute significantly to our understanding of the mechanisms underlying the learning of a (signed) language and how to best promote the teaching of SSL as an L2.
Final report
When people learn a new language other than their own language, it usually takes place within the same language modality, i.e. spoken languages: a person with Spanish as a first language could learn Swedish for example. Based on the research field of second language acquisition, we know quite a lot about the prerequisites and what happens when a person learns a second language. But what happens when a person learns a language that is not a spoken language but operates in another modality, namely the gestural-visual modality? Here, research has been sparse, even though the field has started to emerge in the last decade. The overall aim of this project has been to fill the gap in this research area i.e. the second language acquisition of a sign language. The purpose of the project has been to investigate the learning and development of linguistic structures in a group of learners with Swedish sign language (STS) as a second language (L2).
More specifically, the project has had two sub-objectives for investigation:
1) the acquisition of the non-manual component mouth actions and
2) how morphology with respect to modified signs and productive signs is acquired and developed by learners
To investigate these objectives, we started from an existing learning corpus in Swedish sign language and expanded it. In addition, we also constructed a control corpus consisting of data from deaf native speakers of Swedish sign language.
PROJECT IMPLEMENTATION
A large part of the project has been the construction of a learner corpus consisting of data from second language signers of Swedish sign language, especially in the initial phase. Prior to the start of the project in 2017, there was a smaller existing learner corpus consisting of 18 L2 STS speakers, which we expanded to a total of 38 speakers during the project. In addition, we collected data from 9 L1 STS speakers for the purpose of a control corpus. The data collection was completed in December 2018. The construction of a corpus in sign language is usually time consuming due to the process of annotating data in order to make it searchable. In parallel with the data collection, the corpus data was annotated and by the end of the project, 100% of the data had been annotated in sign glosses and 76% had been translated into Swedish. In total, the L2 STS corpus consists of 25:06 hours of language data and approximately 40,000 sign tokens, and the control corpus consists of 1:22 hours and 6,500 sign tokens. In addition to the gloss annotations, there is also a considerable amount of data linked to the use of non-manual components, e.g. mouth actions (58% annotated).
The project has proceeded according to schedule and in accordance with the purpose, method, materials and implementation described in the application and according to project plan. Minor adjustments and additions have been made to data collection and stimuli.
THE PROJECT'S THREE MOST SIGNIFICANT RESULTS
The single most significant result of the project is the corpus itself. As the world’s probably first L2 corpus on sign language, it has a great potential to generate important results even long after the end of the project, thanks to its extensive size (Mesch & Schönström, 2018). This will be an important resource in future research on sign language as a second language (Schönström, 2021). The corpus is semi-accessible to other researchers via the Svensk nationell datatjänst (snd.gu.se).
The scientific results that have emerged through the two sub-studies in the project are largely related to the discussion about the role of first language in second language acquisition. Several of our analyses show that the learners' first language, Swedish, influences their Swedish sign language in a number of different ways. Among other things, the non-manual part of the sign language seems to be affected in the first place, i.e. the interplay between mouth movements and manual characters. We also see that second language learners overuse the mouth actions that are based on articulations of Swedish words, the so-called borrowed mouth actions. The results are interesting in view of the different modalities of languages and in relation to previous research on signed languages as second languages which claimed that L1 transfer has a limited role in L2 sign language due to the difference in modality. We show this in various publications (Mesch & Schönström, 2021; in press; Schönström, in press). We also see that learners acquire certain mouth actions later than others, such as adverbial mouth actions.
Furthermore, we have used the corpus to explore the interlanguage of L2 learners with regard to the use of the different lexical categories within Swedish sign language with division into 1) lexical signs, 2) fingerspelling and 3) depicting signs. Our analyses show that L2 learners initially fingerspell to a greater extent. They are initially influenced by spoken Swedish, and benefit from the structural similarities that exist between STS and Swedish. With regard to depicting signs, the results indicate that these seem to be acquired later in the L2 learners (Schönström & Mesch, submitted). This is discussed in terms of the importance of gestures for learning as well as the linguistic complexity of depicting signs.
NEW RESEARCH QUESTIONS
The analyses of complex constructions such as depicting signs, use of viewpoint markers for perspective shifting and several of the non-manual components raised questions about how L2 learners acquire syntactic structure. We have seen that L2 learners' development of lexical knowledge should be seen in the light of their development of syntax knowledge. For the future, there are good opportunities to identify development phases in hearing L2 learners of Swedish sign language through syntactic analysis.
Furthermore, the project has taken a step into the cognitive and functional aspects of language learning, more specifically the discussion of the iconic and gestural properties of sign languages and their significance for L2 acquisition. Previous studies have shown that the use of gestures plays a significant role in the L2 acquisition of a spoken language. In the case of sign language, there is not always a clear distinction between lexical units (signs) and gestures, but even here, second language learners seem to be scaffolded by their internal knowledge of gestures.
DISSEMINATION OF RESULTS AND COLLABORATION
Thanks to our collaboration with the European network for Sign Language Second Language Acquisition, led by Prof. Dr. Gary Morgan at City University, London and funded by Leverhulme, some new research questions were initiated during the project. The questions concerned the second language learners' acquisition of viewpoint markers for perspective shifting in sign languages. Here we have carried out a number of different analyses of our corpus data regarding the use of perspectives, some of which have resulted in an MA thesis.
Furthermore, researchers at Trinity College Dublin, Ireland, Humboldt University in Berlin, Germany and the Universidade Federal de Santa Catarina in Florianópolis, Brazil have initiated their own projects in which our corpus design has been replicated. Collaboration in this area enables possibilities of future cross-linguistic comparisons in different sign languages.
Our collaborations will continue even into 2021 with digital workshops together with colleagues from University College of London, led by Prof. Dr. Chloé Marshall. The goal of the workshops is to find new research issues to collaborate on.
In terms of disseminating the results, the project has generated a number of research articles, all open access, (see list of publications) and about twenty lectures at various research conferences and workshops, including a keynote lecture at the EuroSLA conference in Lund 2019. Finally, data from the project have been used in some students’ master’s theses. A major activity around Sweden is also the teaching of future sign language interpreters in Swedish Sign Language. Hopefully, the new insights from the project have generated new knowledge about the acquisition of sign languages as a second language and contributed to an impact on the teaching of sign languages.
More specifically, the project has had two sub-objectives for investigation:
1) the acquisition of the non-manual component mouth actions and
2) how morphology with respect to modified signs and productive signs is acquired and developed by learners
To investigate these objectives, we started from an existing learning corpus in Swedish sign language and expanded it. In addition, we also constructed a control corpus consisting of data from deaf native speakers of Swedish sign language.
PROJECT IMPLEMENTATION
A large part of the project has been the construction of a learner corpus consisting of data from second language signers of Swedish sign language, especially in the initial phase. Prior to the start of the project in 2017, there was a smaller existing learner corpus consisting of 18 L2 STS speakers, which we expanded to a total of 38 speakers during the project. In addition, we collected data from 9 L1 STS speakers for the purpose of a control corpus. The data collection was completed in December 2018. The construction of a corpus in sign language is usually time consuming due to the process of annotating data in order to make it searchable. In parallel with the data collection, the corpus data was annotated and by the end of the project, 100% of the data had been annotated in sign glosses and 76% had been translated into Swedish. In total, the L2 STS corpus consists of 25:06 hours of language data and approximately 40,000 sign tokens, and the control corpus consists of 1:22 hours and 6,500 sign tokens. In addition to the gloss annotations, there is also a considerable amount of data linked to the use of non-manual components, e.g. mouth actions (58% annotated).
The project has proceeded according to schedule and in accordance with the purpose, method, materials and implementation described in the application and according to project plan. Minor adjustments and additions have been made to data collection and stimuli.
THE PROJECT'S THREE MOST SIGNIFICANT RESULTS
The single most significant result of the project is the corpus itself. As the world’s probably first L2 corpus on sign language, it has a great potential to generate important results even long after the end of the project, thanks to its extensive size (Mesch & Schönström, 2018). This will be an important resource in future research on sign language as a second language (Schönström, 2021). The corpus is semi-accessible to other researchers via the Svensk nationell datatjänst (snd.gu.se).
The scientific results that have emerged through the two sub-studies in the project are largely related to the discussion about the role of first language in second language acquisition. Several of our analyses show that the learners' first language, Swedish, influences their Swedish sign language in a number of different ways. Among other things, the non-manual part of the sign language seems to be affected in the first place, i.e. the interplay between mouth movements and manual characters. We also see that second language learners overuse the mouth actions that are based on articulations of Swedish words, the so-called borrowed mouth actions. The results are interesting in view of the different modalities of languages and in relation to previous research on signed languages as second languages which claimed that L1 transfer has a limited role in L2 sign language due to the difference in modality. We show this in various publications (Mesch & Schönström, 2021; in press; Schönström, in press). We also see that learners acquire certain mouth actions later than others, such as adverbial mouth actions.
Furthermore, we have used the corpus to explore the interlanguage of L2 learners with regard to the use of the different lexical categories within Swedish sign language with division into 1) lexical signs, 2) fingerspelling and 3) depicting signs. Our analyses show that L2 learners initially fingerspell to a greater extent. They are initially influenced by spoken Swedish, and benefit from the structural similarities that exist between STS and Swedish. With regard to depicting signs, the results indicate that these seem to be acquired later in the L2 learners (Schönström & Mesch, submitted). This is discussed in terms of the importance of gestures for learning as well as the linguistic complexity of depicting signs.
NEW RESEARCH QUESTIONS
The analyses of complex constructions such as depicting signs, use of viewpoint markers for perspective shifting and several of the non-manual components raised questions about how L2 learners acquire syntactic structure. We have seen that L2 learners' development of lexical knowledge should be seen in the light of their development of syntax knowledge. For the future, there are good opportunities to identify development phases in hearing L2 learners of Swedish sign language through syntactic analysis.
Furthermore, the project has taken a step into the cognitive and functional aspects of language learning, more specifically the discussion of the iconic and gestural properties of sign languages and their significance for L2 acquisition. Previous studies have shown that the use of gestures plays a significant role in the L2 acquisition of a spoken language. In the case of sign language, there is not always a clear distinction between lexical units (signs) and gestures, but even here, second language learners seem to be scaffolded by their internal knowledge of gestures.
DISSEMINATION OF RESULTS AND COLLABORATION
Thanks to our collaboration with the European network for Sign Language Second Language Acquisition, led by Prof. Dr. Gary Morgan at City University, London and funded by Leverhulme, some new research questions were initiated during the project. The questions concerned the second language learners' acquisition of viewpoint markers for perspective shifting in sign languages. Here we have carried out a number of different analyses of our corpus data regarding the use of perspectives, some of which have resulted in an MA thesis.
Furthermore, researchers at Trinity College Dublin, Ireland, Humboldt University in Berlin, Germany and the Universidade Federal de Santa Catarina in Florianópolis, Brazil have initiated their own projects in which our corpus design has been replicated. Collaboration in this area enables possibilities of future cross-linguistic comparisons in different sign languages.
Our collaborations will continue even into 2021 with digital workshops together with colleagues from University College of London, led by Prof. Dr. Chloé Marshall. The goal of the workshops is to find new research issues to collaborate on.
In terms of disseminating the results, the project has generated a number of research articles, all open access, (see list of publications) and about twenty lectures at various research conferences and workshops, including a keynote lecture at the EuroSLA conference in Lund 2019. Finally, data from the project have been used in some students’ master’s theses. A major activity around Sweden is also the teaching of future sign language interpreters in Swedish Sign Language. Hopefully, the new insights from the project have generated new knowledge about the acquisition of sign languages as a second language and contributed to an impact on the teaching of sign languages.