Abstract This paper presents an overview of lexicographic projects related to Flemish Sign Language (Vlaamse Gebarentaal or VGT). VGT consists of five regional varieties. There was a time when this lexical variation was considered a disadvantage in deaf education and L2 teaching. That is why in the 1980s and early 1990s the very first lexicographic project related to Flemish signs concentrated on lexical unification in Signed Dutch. However, when the Flemish Deaf community officially rejected the use of Signed Dutch and started promoting VGT, this also led to a change in focus with respect to sign language lexicography. In 2004, the first electronic bilingual Flemish Sign Language/Dutch dictionary was published online. Since by then the Flemish Deaf community had decided not to opt for standardisation from above, it was decided to also include regional variation in the dictionary and to label the regional variants as such. By looking at past and current projects and initiatives planned for the near future, it has become clear that both societal and technological developments have been shaping lexicographic work related to Flemish Sign Language. We also look forward to future developments that will pave the way for exciting possibilities with respect to sign language lexicography. 1. Introduction 1.1. Introducing Flemish Sign Language Sign languages are the natural, visual-gestural languages of Deaf communities.1 They are not universal – they differ from region to region. Even where a common spoken language is used, the sign languages may be different (e.g. while Dutch is spoken in Flanders and the Netherlands, Flemish Sign Language and Sign Language of the Netherlands are two distinct sign languages (see Vermeerbergen et al. 2013). This illustrates how sign languages have evolved independently from spoken languages, although there is clear evidence of language contact. Sign language linguistics is a young field of study: modern sign linguistics was pioneered by Tervoort’s (1953) doctoral dissertation on the signing of deaf children in the Netherlands, and Stokoe’s (1960) linguistic analysis of American Sign Language (ASL). Today, while some sign languages have been studied in relative detail, many others remain under- or even un-described. Research has clearly revealed that signed and spoken languages share fundamental properties at all levels of linguistic structure, but it has also shown that modality matters. Modality-specific characteristics of sign languages include the use of space and of highly simultaneous linguistic structures, both of which can be seen at the level of the individual sign as well. Indeed, lexical signs are composed of four simultaneously produced sublexical units or ‘parameters’: handshape, movement, orientation (of both palm and fingers) and place of articulation (or location). The sociolinguistic context of sign languages also differs from that of spoken languages. For example, over 95% of deaf children have non-deaf, non-signing parents so that for the majority of sign language users the sign language acquisition path is atypical. In the past, Flemish deaf children in hearing families usually started to acquire VGT when entering a (pre-)school for the deaf, mostly on the playground in contact with other signing peers. Today, this is no longer the case. The majority of deaf children are mainstreamed in a ‘hearing school’, and as a result there are more substantial differences in terms of the age at which deaf children first come into contact with Flemish Sign Language. Flemish Sign Language is the sign language used in Flanders, the northern part of Belgium. The Flemish Deaf community is estimated to include approximately 5000 to 6000 sign language users (Loots et al. 2003). VGT is clearly related to the Langue des Signes de Belgique Francophone (LSFB), the sign language used in Wallonia, the southern part of Belgium. Flemish Sign Language consists of five regional varieties, which developed in and around the different Flemish deaf schools located in Bruges (West Flanders), Ghent (East Flanders), Antwerp (Antwerp), Brussels (Brussels Capital Region and Flemish Brabant) and Hasselt (Limburg). Next to the differences between the areas, there is intra-regional variation, of which gender-related variation is one example. In 1997, the Flemish Deaf community rejected an imposed standardisation from above and decided to promote the ongoing standardisation process of Flemish Sign Language. This decision and other standardisation issues in Flemish Sign Language were discussed in Van Herreweghe and Vermeerbergen (2009) (See also Section 2.2.3.). In April 2006, after lengthy negotiations, Flemish Sign Language was officially recognised by the Flemish Parliament. The decree on the recognition also includes the recognition and funding by the Flemish Government of a centre of expertise with respect to VGT (Van Herreweghe et al. 2016). As a result, in 2008, the Flemish Sign Language Centre (Vlaams GebarentaalCentrum, VGTC), a non-profit organisation founded in 1997 (www.vgtc.be), was recognised as such a centre and started to receive some modest financial support from the Flemish government. In the first phase, from 2008 to 2011, the work concentrated on documenting grammatical aspects of VGT. However, since 2012, the centre has been focusing on lexicographic projects. 1.2. The established lexicon and the productive lexicon An important distinction related to the lexicon of sign languages is that between the established and the productive lexicons. Established signs, also called (fully) lexical signs, are highly conventionalised signs in both form and meaning. These signs can easily be listed in a dictionary. Brennan (1992: 45-46) describes them as ‘ready-made, off the shelf lexical items. They are already in existence: the signer simply has to pluck them from her/his mental lexicon and place them in the appropriate lexical contexts’. In contrast, productive signs are the result of a process which ‘involves selecting the component parts and putting them together in appropriate ways to create particular kinds of effects’ (Brennan 1990: 163). Productive signs are also called ‘partly-lexical signs’, but are normally not taken up in dictionaries. According to Johnston (2016: 15): Signs which are partly-lexical have one or both of these two important characteristics: (i) they have little or no conventionalised or language-specific meaning value in addition to that carried by their formational components (e.g. handshape, location, orientation etc.); (ii) they have a meaning that is incomplete in some way—one needs to refer to the context of utterance (the unfolding text and/or the actual utterance space) in a non-trivial way to ‘complete’ the meaning of the sign. They cannot be listed in a dictionary in any straightforward way, (…). An example from Vermeerbergen (1996: 40) will clarify the difference between the two. In VGT, there is an established sign for TO-DRIVE2 as in Figure 1 (which is also a lemma in the VGT dictionary). Figure 1. View largeDownload slide The established sign TO-DRIVE. Figure 1. View largeDownload slide The established sign TO-DRIVE. It is, however, also possible to express the type of driving in a more productive, visual way (but this can only be done in context) as in Figure 2 in which “driving-uphill-through-hairpin-bends” is expressed in one partly-lexical sign. This type of sign has not been taken up in the VGT dictionary. Figure 2. View largeDownload slide The productive sign with the meaning “driving-uphill-through-hairpin-bends”. Figure 2. View largeDownload slide The productive sign with the meaning “driving-uphill-through-hairpin-bends”. It should be noted here that the distinction between established and productive signs is not always clear-cut. As Ferrara (2012: 128) points out: (…) both entrenchment and conventionality are notions of degree. Sometimes, there is difficulty determining whether a particular form is entrenched and conventionalized enough to be considered a fully lexical (linguistic) sign. This complex situation is especially unclear with respect to the generally understudied signed languages of the world, where signers accept a high degree of variation in regards to “acceptable signing”. Different researchers have proposed different solutions to be able to make the distinction between the two. Overall, they are of three kinds: Based on use, involving notions such as “stability”, “conventionalisation”, “standardisation” and “entrenchment” (e.g. Johnston 2016, Ferrara 2012); Based on certain formal characteristics, e.g. complexity of form, eye gaze behaviour, presence/absence of mouthing, syntactic behaviour, etc. (e.g. Engberg-Pedersen 1993, Vermeerbergen 1996, Vermeerbergen et al. 2007b); Based on the signer’s intention, do signers want to ‘say’ something, or ‘show’ something (Cuxac 2000). For Flemish Sign Language, deciding on the boundary between the established and productive lexicons is work in progress (Vermeerbergen 2016). A number of researchers suggest a continuum, rather than a sharp boundary. We fully agree with this suggestion; however it doesn’t make it easier when it comes to e.g. consistency in transcribing large sets of data and/or to decisions on which signs to include in a dictionary. 2. The past 2.1. The origin: Signed Dutch dictionaries As was the case in many other sign language communities, the first attempts at lexicography related to signs in the Flemish Deaf community focused on a signed system rather than a fully-fledged sign language. Signed systems are artificial mixtures of the lexical system borrowed from the local sign language and the morpho-syntactic system borrowed from the local spoken language. They were mainly developed for use in deaf education, starting from the assumption that the surrounding spoken language could be visualised and hence would be more accessible for deaf children so that they would be able to acquire the surrounding spoken language more easily (Van Herreweghe 1996: 68). Already in the 1960s and 1970s, a number of such signed systems had been developed in the United States: SEE 1 (Seeing Essential English) (Anthony 1966, 1971), SEE 2 (Signing Exact English) (Gustason et al. 1972) and Signed English (Bornstein 1974, Bornstein et al. 1980). What they all had in common was that they were based on the morpho-syntax of English, so the English word order was adopted in these signed systems, and that (some of) the lexemes were borrowed from American Sign Language. The differences between these three systems are related to the borrowing of compounds, derivations and inflections, the artificial invention of signs (for certain bound morphemes or for free morphemes as well), etc. To give one example: in SEE1 for every free or bound morpheme in English a sign is used so that the English word butterflies is treated as a compound noun consisting of three separate morphemes and is consequently rendered with three separate signs, i.e. BUTTER, FLY and S. However, as Signed English mostly started from word-sign correspondences and only 14 affixes (including the plural morpheme) were explicitly visualised, here butterflies was represented by means of two signs BUTTERFLY and S. In Flanders, the signed system Signed Dutch (or Nederlands met Gebaren) was developed in the 1980s and 1990s (Buyens 1987), and closely resembles Signed English, except that the signs were borrowed from Flemish Sign Language and the morpho-syntax from Dutch. In general, there was a mapping of one-word-to-one-sign and there were artificially invented signs for some morphological markers in Dutch (e.g. markers for the comparative and superlative forms of adjectives, a diminutive marker, a plural marker on nouns, a past tense marker, etc.). The idea was that Signed Dutch would be used in deaf education in order to facilitate the acquisition of Dutch, but also to arrive at a type of lexical unification or ‘uniformisation’ (as it was called at the time) of the lexicon of the five regional varieties of Flemish Sign Language. The lexical unification was in the hands of a ‘sign committee’ with deaf signers from the different regional VGT varieties. Originally, in 1980, the committee consisted of about 40 members, but this quickly proved counterproductive and was reduced to 18 members. The committee consisted of deaf people from the Flemish National Deaf Association, Fevlado, and therefore worked within the Association. From 1980 up to and including 1994, the Flemish ‘sign committee’ met one Saturday a month to select a standard ‘unified’ sign for each Dutch word taken from a frequency list of about 9200 Dutch words (Geysels et al. 1989). Each meeting, a number of Dutch words were shown (in print) to the committee members and they all had to show the corresponding sign in their own variety of Flemish Sign Language. Then the committee members would decide on one sign corresponding to the Dutch word. A drawing was made of these signs and the drawings were published in a series of 25 booklets called Woord en Gebaar (‘Word and Sign’) from 1981 onwards. In 1983 the drawings found in the first booklets were gathered in a book, also called Woord en Gebaar, containing a glossary of 1200 drawings of the chosen signs. For a description of the methodology used to select the ‘unified’ signs we would like to refer the reader to Van Herreweghe and Vermeerbergen (2009: 312-313). At the end of the 15 years, a Signed Dutch dictionary (Buyens 1995) was published, which contained on the one hand a number of regional signs actually used in the Deaf community – and thus part of the lexicon of Flemish Sign Language – and on the other hand a number of signs invented by the sign committee. Information on the origin of the signs could not be given since no record had ever been kept during the meetings. As a consequence, for some of the signs signers themselves don’t know any more whether the origin of the sign is to be found in Signed Dutch or in VGT (Van Herreweghe and Vermeerbergen 2004). However, by 1997, some people in the Flemish Deaf community – mostly deaf people who were in some way involved in the Signed Dutch courses – started thinking differently about Signed Dutch. This was caused by changes within Fevlado (e.g. a new Board of Directors and the foundation of ‘leadership courses’ for deaf members, cf. De Clerck 2007: 9), international contacts (cf. De Clerck 2007: 9, Van Herreweghe and Vermeerbergen 2009: 316), and the first results of grammatical research into VGT (Van Herreweghe 1995, Vermeerbergen 1996, 1997). In a General Assembly of Fevlado in March 1997, the use of Signed Dutch and the Signed Dutch dictionary was rejected in favour of Flemish Sign Language. It was decided that the ‘communication courses’ that they offered would be changed from Signed Dutch courses to Flemish Sign Language courses (Van Herreweghe and Vermeerbergen 2009: 311). At the same time, researchers started to look for funding for lexicographic research into VGT, and they were successful towards the end of the 20th century, when a project was set up to look at lexical variation in VGT, culminating in the first dictionary containing VGT lexemes. 2.2. An online translating dictionary Dutch-Flemish Sign Language/Flemish Sign Language-Dutch From various angles, a pressing need for a bilingual dictionary had been expressed. Teachers in the Flemish deaf schools could not consult a dictionary when they did not know how to translate a certain Dutch word into Flemish Sign Language. Nor could the teachers who worked in the sign language interpreter-training programmes. The dictionary was also strongly desired by various organisations such as Fevlado and several Centres of Adult Education, all of them organising Flemish Sign Language courses throughout Flanders. From 1999 onwards, a number of (quite modest) lexicographic projects led to the launch in 2004 of the first, freely accessible, electronic bilingual dictionary Dutch/Flemish Sign Language - Flemish Sign Language/Dutch (http://gebaren.ugent.be). In 2008, the online dictionary was published as a DVD by Standaard Uitgeverij, and had exactly the same format as the online issue, except for the fact that the user didn’t need to be online. 2.2.1. Methodology Zwitserlood (2010:445-6) points out that whereas all lexicographers have to cope with challenges, sign language lexicographers are faced with a number of additional ones: First, there is no orthography for signed languages. Although several sign notation systems have been developed, none of them is generally accepted, and most are considered too difficult for sign representation in dictionaries. Second, in almost all countries there are only a few signed sources from which the meaning(s) and grammatical characteristics of signs can be deduced and frequencies can be calculated. Third, signed languages are quite different from many spoken languages, particularly from Indo-European languages. Because the sign language may have a complex structure where the spoken language has a simple word and vice versa, it is often rather difficult to give a sign-for-word or word-for-sign translation. Finally, sign language lexicographers have to overcome the legacy of a tradition of (inadequate) compilation of sign language dictionaries set by non-professionals. All of this holds true for Flemish Sign Language lexicography as well. When the research towards a Flemish Sign Language dictionary started in 1999, there were hardly any sources that could easily be used as the basis for a sign language dictionary, and since sign languages are face-to-face languages, it was also not possible to rely on written sources. Moreover, trying to start from a Dutch word list would yield problems with respect to assumed, but unrealistic, word-for-sign or sign-for-word translations. Finally, there was the legacy of the above-mentioned sign committee working on the Signed Dutch dictionary selecting ‘unified’ signs, and thus disregarding variation in the language. When the Signed Dutch ‘unified’ signs were promoted, many sign language users reacted quite negatively to at least some of them. Some people refused to use some of the invented signs (especially those signs that violated VGT phonological rules, such as the proposed signs for the colours, Van Herreweghe & Vermeerbergen 2009: 313), so that in the 1990s, many signers would say that they did not like the ‘new’ signs and preferred the ‘old’ ones. Moreover, the fact that when there were different regional variants of one sign, one variant was ‘elected’ and labelled as the ‘unified’ sign, was cause for concern. Some deaf signing parents heard from their children who attended a deaf school where Signed Dutch was used that the (regional) signs the parents used were incorrect, and that the Signed Dutch (‘unified’) signs used at school were correct. This caused a lot of insecurity within the Deaf community at the time, especially by implying that their own variety of VGT was inferior (Van Herreweghe & Vermeerbergen 2004: 116-117). It was therefore decided early on in the research towards the Flemish Sign Language dictionary that lexical variation would be taken into account, so as not to give the impression that one variety of VGT has a higher status than another. Priority lists From the beginning of the lexicographic project, it was decided to work with elicited video materials, but to try and make sure that the elicited language data were as natural as possible. Since the researchers had to start from zero, lists of priority concepts (not Dutch words) were developed, so that signs could be elicited in different thematic rounds. Because of the fact that in the Signed Dutch dictionaries (see Section 2.1.) artificial signs had been invented (especially in certain semantic fields), and that some of those signs were strongly rejected in the Flemish Deaf community, it was decided to first deal with those signs, i.e. signs for 0. the colours, the days of the week, the months of the year, the seasons, a number of temporal signs and question signs. The principle underlying the prioritisation of the rest of the concepts to be elicited was that the researchers first started with concepts related to (1) a person (and his/her close relations), then progressed to (2) where a person lives and what s/he does at home, (3) what a person encounters or does when leaving the house, and finally moving out to (4) the world at large. Inspiration for this was found in Flemish (spoken language) dialect lexicography (Van Keymeulen 1993, 2003), since for dialects it is also not possible to rely on a written corpus of language data. This yielded the following thematic lists: Family and friends, the body, health, hygiene, clothing; Types of houses and buildings, furniture, cleaning material, kitchen utensils, parties and ceremonies, food; Occupations and professions, at work, sports, holidays and free time, school; The world, countries and cities, fauna, flora, traffic. Finally, after having elicited signs in all of the above themes, it struck the researchers that hardly any abstract concepts had been elicited, so that a sixth list was added containing Abstract concepts. When developing the dictionary, the intention was to render the ‘basic’ Flemish Sign Language lexicon as accurately and completely as possible. The goal was to include all the existing sign language variants for a certain concept as they were used in Flanders at the time. Therefore, all regional variants that were recorded during the collection of the data were taken into account. However, due to the relatively small number of informants (see below), the researchers realised that it is simply impossible to record all existing signs. One cannot expect the informants to know all of the signs used in their region or to have produced/remembered all of them during the meetings. This does not mean that non-recorded signs are rejected by the researchers, or that the researchers only wanted to enter standardised signs, as was the case in the Sign Language of the Netherlands STABOL project (Schermer 2004, Zwitserlood 2010, Van Herreweghe and Vermeerbergen 2009). It simply means that they did not occur in the collected data. The risk, however, is that users will regard the dictionary as a prescriptive tool: ‘If it’s not in the dictionary, it’s wrong’. This is a hazard the researchers are well aware of, but it seems to be quite difficult to convince the users that that was not the intention. Van Herreweghe and Vermeerbergen (2004),3 includes a small-scale study that clearly showed the rapid effects of the codification in the form of lexical influence of Signed Dutch onto VGT. It was then maintained that the rapid impact of codification is due to (1) doubts in the Deaf community about their own linguistic competence in VGT, obviously related to the lower status of VGT compared to Dutch; (2) the exceptional language acquisition situation of deaf children where the overwhelming majority have hearing, non-signing parents so that most signers are (relatively) late learners of VGT; and (3) the fact that interpreters frequently function as linguistic models, especially in the case of educational interpreting (Heyerick and Vermeerbergen 2012). Informants Since there are five regional sign language varieties in Flanders (see Section 1.1.), five regional teams were assembled with deaf volunteers. Each regional team consisted of about half a dozen deaf informants, who all met the following description: between 20 and 50 years of age, having a thorough command of the studied sign language variety and using it as their first language, being an active member of the Flemish Deaf community and having received education in a deaf school. Both men and women were involved in the teams. In each group, one native signer was appointed as moderator. Only the moderators were allowed to look at the lists of priority concepts and they were in charge of eliciting the signs by means of the provided eliciting materials, and of filming the conversations (one video camera per group on a tripod). The informants themselves did not get to see any Dutch words. There were never any hearing people present. The latter was considered very important to avoid interference from Dutch as much as possible. Procedure The regional teams met six times in the course of 1999 and 2000 for each of the thematic rounds as described above. Prior to the meetings, the moderators were invited to Ghent University to get some basic training with respect to the priority lists, the elicitation materials, and the video equipment. During the team meetings, which were in informal settings, usually in the home of one of the informants, who took care of coffee and biscuits, the moderators either showed elicitation materials or started up a conversation by asking questions. As an example, the moderator would ask all the informants when their birthdays were, or what they did last week, etc., and continued asking questions until all the months of the year and the days of the week had been used in the conversations. The moderator could also use pictures or drawings to elicit conversations about the kitchen, for instance. The general idea was that the signs would be used in spontaneous conversations. Obviously, the informants frequently remarked on each other’s use of certain signs. The informants were seated in an elongated circle, usually around the dinner table, and all of the conversations were recorded by means of only one Hi8-video camera filming all the informants (except for the moderator) so as to avoid excessive intrusion. The research eventually resulted in about 90 hours of recorded language data (an average of three hours per session, per region). Analysis The videodata were all watched on a TV-set and a Hi8-player by a deaf researcher so as to be able to extract the relevant signs from it. It was necessary to be able to compare signs intra- and interregionally, but doing this with a linear video recording alone was hardly possible. One option to extract individual signs would be by describing (in written Dutch) each sign on the basis of the different parameters, but this would have been a very lengthy process. Therefore, the individual signs were transcribed in SignWriting, an American transcription system developed by Valerie Sutton and the Deaf Action Committee (http://www.signwriting.org). It allows for a visual rendering of the signs, using symbols to represent the handshapes, movements, facial expressions, and body shifting. The signs were analysed linguistically and catalogued in FileMaker Pro and then they were filmed again with a neutral background, so as to be able to include them in the online dictionary (see Figure 9 below). The analysis encountered a number of stumbling blocks related to decisions with respect to established or productive lexical elements, phonemic or allophonic variation, signs for occupations and signs for male, female and baby animals. It was sometimes hard to decide whether a sign (and its variants) should be included in the dictionary, or whether we were dealing with productive lexical elements (see Section 1.2.). A good example is to blow. In all of the teams a sign was produced with the dominant hand starting at the mouth as in Figure 3 (which is a sign for instance used for blowing out a candle). Figure 3. View largeDownload slide TO-BLOW, sign 1. Figure 3. View largeDownload slide TO-BLOW, sign 1. However, a second sign was also produced with a blowing mouth, but the dominant hand started in front of the body (to the ipsilateral side, See Figure 4). This sign is for instance used for the blowing of a leaf blower (cf. http://gebaren.ugent.be/alfabet.php?id=17294). Figure 4. View largeDownload slide TO-BLOW, sign 2. Figure 4. View largeDownload slide TO-BLOW, sign 2. It was very difficult to decide on whether the second sign is to be considered an established variant of the sign that needs to be included in the dictionary, or whether this can be considered a productive variant that would not need inclusion. This was discussed in the group at length, and in the end it was decided to rely on native signers’ intuitions and to ask whether the team members would consider this an established or a productive sign. The conclusion was that both variants should be included since nobody would use the sign in Figure 3 for the blowing of a leaf blower. A second pitfall is related to the decision as to whether two variants contain phonemic or rather allophonic differences. In the first instance, inclusion into the dictionary would be expected, in the second that would not be the case. Again, an example will clarify the issue. In the Antwerp variety of VGT the sign for August can be produced in slightly different ways. Two of them are shown in Figures 5 and 6 (cf. http://gebaren.ugent.be/alfabet.php?id=17184). Figure 5. View largeDownload slide AUGUST, sign 1. Figure 5. View largeDownload slide AUGUST, sign 1. Figure 6. View largeDownload slide AUGUST, sign 2. Figure 6. View largeDownload slide AUGUST, sign 2. In the first AUGUST sign (Figure 5), the sign is produced with the index finger and the middle finger touching the ipsilateral temple. In the second sign (Figure 6), the sign is produced with the index finger and the thumb touching the ipsilateral temple. Here it was decided to include both variants, again on the basis of the intuitions of the native signers in the Antwerp team who considered it phonemic rather than allophonic variation. For both stumbling blocks we tried to find some objective rules to disentangle the problems, but in both cases we ended up relying on native signers’ intuitions. Obviously, this has its risks, but it seemed to be the only solution. A third difficulty had to do with signs for occupations. In Dutch, words for occupations (e.g. teacher, baker) are often agentive deverbal nouns formed by means of the suffix –er following the related verb stem. In the elicited data for such occupations, the informants frequently produced the accompanying verb sign followed or – to a lesser extent – preceded by the sign PERSON, thus creating a lexeme which can be considered a compound sign. In line with Zwitserlood (2010: 468) who discussed compounds in NGT (Sign Language of the Netherlands), we are very cautious to label such signs compounds in sign languages (see also Section 3.4.): For instance, the use of the term ‘compound’ for particular sign combinations suggests a particular morphological structure, whereas it is unclear so far whether the sequential sign combinations termed as such are indeed complex signs or rather phrases, or else perhaps literal translations from Dutch compounds rather than NGT compounds. It may be the case that the formation of VGT signs with PERSON as a second element has been caused by interference from the equivalent Dutch morphological structure (-er). Indeed, in day-to-day conversation the signs with PERSON are hardly ever used, and the verb sign is also used for the agentive noun. Moreover, other signs for occupations such as banker (BANK∧PERSON) or cashier (TILL∧PERSON) were found in the elicited material which are again hardly ever used in day-to-day conversation. Instead, Flemish signers produce something like ‘WORKS IN BANK’ for BANKER or ‘WORKS IN SHOP’ for CASHIER. This type of construction will not work for every occupation, but it seems to occur quite frequently. Even though we were very careful about this (no elicitation materials using any Dutch, no hearing people present), the influence of Dutch on our informants’ sign language use may have been caused by the situation (presence of a camera, use of elicitation material), but this is very difficult to avoid in such a setting. One last problem (which caused some discussion in the teams) was related to the signs for animals. Dutch – like many other languages – has specific terms to denote the female, male and baby of a certain animal species (e.g. (1) schaap ‘sheep’: ram ‘ram’, ooi ‘ewe’, lam ‘lamb’; (2) rund ‘cattle’: stier ‘bull, koe ‘cow’, kalf ‘calf’). The informants did produce specific signs for these animals and those signs were included in the dictionary, but the teams also stated that these signs are hardly ever used. Usually the reference is made as follows: SHEEP∧MAN, SHEEP∧WOMAN, SHEEP∧SMALL. 2.2.2. Form and content The signs are rendered in the shape of both video clips and SignWriting. The innovative aspect of this dictionary (at the time of its first publication in 2004) was that it operates in two directions, i.e. from Dutch into VGT and from VGT into Dutch (cf. also Zwitserlood 2010: 451). The makers thought this to be very important so as to give equal status to both languages. When going from Dutch to VGT, the user can click on a Dutch word in an alphabetically organised word list (Figure 7) or type in a (partial) word in a search box (Figure 8). Figure 7. View largeDownload slide Alphabetically ordered word list in the Dutch-VGT dictionary. Figure 7. View largeDownload slide Alphabetically ordered word list in the Dutch-VGT dictionary. Figure 8. View largeDownload slide Search box (from Dutch to VGT). Figure 8. View largeDownload slide Search box (from Dutch to VGT). The user then gets the equivalent sign(s) for the word, including regional variants, which are marked on the webpage in the top right corner. An example is given in Figure 9, where the first two signs for potato are shown, the first one being used in Antwerp, Limburg, East Flanders and Flemish Brabant, and the second in Limburg and East Flanders (but the user can scroll down for more regional variants). Figure 9. View largeDownload slide Screenshot of first two signs for potato in VGT. Figure 9. View largeDownload slide Screenshot of first two signs for potato in VGT. However, for a majority of the signs there is at least one sign that is used all over Flanders, which is then marked as ‘Flanders’ (or ‘Vlaanderen’) in the top right corner, as in Figure 10. Figure 10. View largeDownload slide Screenshot of STRAWBERRY. Figure 10. View largeDownload slide Screenshot of STRAWBERRY. As just mentioned, the innovative aspect of the dictionary (at least at the time) was the option to search in VGT. This can be done by giving some of the parameters of the sign, i.e. handshape (of both hands), place of articulation and type of contact. For handshape, the user would click on one (or in the case of two-handed signs two) of the 107 SignWriting symbols for handshape. For place of articulation, the user would click on the head, torso, arms, hands, legs or on ‘no contact’ for a sign in neutral signing space. Finally, the user would select one of six different types of contact (touching, grabbing, putting in between, hitting, sweeping, rubbing). The result is a list of SignWriting symbols for the different signs with the selected characteristics. Figure 11 shows the two results (HISTORY and AUGUST) that can be found when clicking on one handshape with extended thumb and index finger touching the head. Figure 11. View largeDownload slide Screenshot of two results of the search action. Figure 11. View largeDownload slide Screenshot of two results of the search action. The user can then click on the SignWriting symbols or the words to see the video clips, and decide which sign they were looking for. The advantage of this system is that it is indeed possible to search for a sign via its parameters, but the detour via SignWriting does make it less user-friendly. Since 2004, other electronic dictionaries involving a sign language and search options in two directions have been published that are far more user-friendly, such as the electronic dictionaries containing Swedish Sign Language (SSL; Institutionen för Lingvistik 2009), Danish Sign Language (DTS; Center for Tegnsprog 2008), and Finnish Sign Language (SVK; Kuurojen Liittory 2003 – originally only in one direction, but since recently in two directions). 2.2.3. A Standard Flemish Sign Language? In 2004, 1401 concepts had been elicited yielding 9134 collected signs (but the dictionary has been elaborated since then - see Section 3.2.). This means that for a majority of the concepts more than one sign was collected (see Figure 12). Figure 12. View largeDownload slide One concept – one/more than one sign per regional variety. Figure 12. View largeDownload slide One concept – one/more than one sign per regional variety. De Weerdt et al. (2003) and Vanhecke and De Weerdt (2004) showed that, when looking at regional variation, 72.3% of the collected signs were similar or related. For 540 concepts there was a common sign in the five regions; for 33 concepts there was no sign in any of the regions; and for 828 concepts there were different variants spread over the regions. Of those 828 concepts: 43% had identical or similar signs; 12% had related signs; 45% had completely different signs. The general conclusion was (and is) that to date there is no Standard VGT, but a standardisation process is well under way (Van Herreweghe and Vermeerbergen 2009) and that the term VGT can be considered an umbrella term for the different regional varieties of VGT. This nicely corroborates the decision made by Fevlado in 1997 to reject an imposed standardisation from above (or a renewed ‘unification’), but to promote the ongoing spontaneous standardisation process. There were three reasons for the rejection of standardisation from above. First, there had been one previous attempt at standardisation (or ‘unification’ as it was called then; see section 2.1) from above, which had met with a lot of antagonism in the Deaf community. Second, none of the possible methods for an imposed standardisation were considered acceptable.4 Third, at the time there had not been enough in-depth research into the causes of variation: is the variation purely regionally defined or is it possible that there are also other, linguistically internal and external (social), factors that need to be taken into account. These can be the chosen register, the context in which a sign is used or its grammatical function (cf. Lucas & Bayley, 2005 for ASL), as well as the age or gender of the signer. This means that choosing a preferred lexeme over others could lead to awkward results. In the case of register variation, for instance, it is quite possible that a non-appropriate sign could be labelled as ‘preferred’ since we do not know yet which signs are normally used in which register. This can be compared with selecting an informal verb such as ‘to puke’ as the preferred lexeme instead of, for example, the more formal ‘vomit’, ‘regurgitate’, or more neutral ‘be sick’, or ‘throw up’ (Van Herreweghe and Vermeerbergen 2009: 318). Instead, it was decided to actively support the spontaneous standardisation process by activities such as: lobbying for VGT on national television; promoting the use of VGT on websites; supporting contacts between sign language users from different regions; supporting the use of VGT by deaf signing teachers in the classroom (in deaf and/or bilingual education); and actively engaging in lexical modernisation. 2.3. Lexical modernisation Since Flemish Sign Language has not been used as medium of communication in certain (scientific and academic) domains, there are certain lexical gaps in the language that can be filled by means of lexical modernisation. Reagan (2001: 150) defines this as the efforts made to increase the lexicon of a language in order to allow it to deal with new technological, political, economic, educational and social developments and concepts. Lexical modernization therefore constitutes a clear instance of corpus planning. All languages experience from time to time what can be termed lexical gaps; lexical modernization hence specifically refers to controlled and directed attempts to expand a given language’s lexicon in a systematic manner. In the latter half of the 1990s, it became clear that many educators and educational interpreters were confronted with lexical gaps in their daily educational practices (or perceived lexical gaps if they did not know signs for certain concepts). As a result, many of them either avoided the subject, did not use the sign but mouthed the words instead, or started to invent ad hoc signs for certain concepts, without being aware of the possibility to create productive lexical items in VGT (see Section 1.2.). Consequently, for certain educational concepts, educators and interpreters all over Flanders used different signs which they, as hearing foreign language learners, had invented. This raised concerns in the Deaf community, who asked for a solution to the problem. In 2001, members of the board of Fevlado, a group of Flemish sign language linguists, and a number of professionals in deaf education together set up a lexical modernisation project. Funds were secured for two project collaborators: one deaf native signer, and one hearing therapist/sign language interpreter with experience in deaf education. A decision was made to start with mathematical vocabulary, followed by wereldoriëntatie ‘world orientation’ (i.e. geographical and historical) lexicon the next year. First, drawing on handbooks, curricula, etc. for primary and first level of secondary education, as well as a questionnaire survey of teachers and sign language interpreters, a list of about 500 concepts per domain was compiled of mathematical/geographical/historical words in Dutch that were used in primary education, and in the first two years of secondary education (i.e. the first grade of secondary education in Flanders). As a second step, sign language users from the different regional varieties were asked whether they knew established or productive lexical elements for those concepts (without showing them the Dutch words, i.e. by using drawings, asking them to calculate something, and so on), or whether they would explain those concepts in any other way (e.g. by means of a paraphrase). As a result, 192 (out of the 500 prioritised concepts) lexical gaps for mathematics and 209 (out of 500) lexical gaps for world orientation were detected. The third step constituted of verifying with a number of deaf ‘experts’ (i.e. people who had studied mathematics/geography/history at a more advanced level) how they would express those concepts. Indeed, in nearly all cases, they could not explain them in VGT, but had to fall back on (written) Dutch. In the fourth step, this same group of experts suggested solutions to fill in the detected lexical gaps. The lexicon was expanded in one of the following ways: borrowing from other sign languages (e.g. formula); using a productive (depicting) sign (e.g. radius, prefab); compounding (e.g. right angle, pilgrimage); using fingerspelling (e.g. least common multiple, Celsius); paraphrasing (e.g. isosceles triangle); referring to a mathematical symbol (e.g. =, cm2); adopting a sign from the Signed Dutch Dictionary (when it was felt that the sign used there had already been conventionalised). In step five, a linguistic check was made and two of the proposed signs were rejected and replaced. The ‘new’ signs were presented to the Deaf community in a workshop and then made accessible via the online Flemish Sign Language dictionary, where the signs were flagged with ‘hiatenproject’ (i.e. lexical modernisation project), so that they could be evaluated in the following years. However, when for three years in a row there was only positive feedback and no negative feedback, it was decided to regard those signs as having been accepted by the Flemish Deaf community (De Weerdt and Rogiest 2003a, De Weerdt and Rogiest 2003b, Van Herreweghe and Vermeerbergen 2003). It is important to note that in all steps of the project, participation of the Deaf community was an absolute requirement. This leads to the issue as to who can decide whether signs are acceptable or not. For many spoken languages, there is a ‘prestigious, distinguished community’, i.e. a cultural and intellectual educated elite ‘who have sufficiently refined and developed linguistic intuitions’ (Heidbuchel 1963, Van Hoof and Jaspers 2012: 102, Van Herreweghe & Vandemeulebroucke 2016: 210, our translation), usually professors, teachers, linguists, writers, journalists, etc. In Deaf communities, it is more difficult to define that prestigious, distinguished community: ‘[People] traditionally considered “educated” would be those who have achieved reasonable literacy levels in what is essentially their second language’ (Eichmann 2008: 36), not their first (sign) language, and at least in the Flemish Deaf community there has not been any academic sign language elite. Native signers, i.e. deaf sign language people who have/had deaf signing parents might come closest to the concept of a sign language elite, but again, in Flanders they have not identified themselves as such. 3. Current activities 3.1. A Flemish Sign Language Corpus In 2012, a project started with the general objective of collecting and publishing a corpus of Flemish Sign Language data that would be a core linguistic source for any research effort, including lexicographic work, aimed at the analysis of VGT, or at cross-linguistic research including VGT. Leech (2000: 57) maintains that [t]here are two different ways of designing a spoken corpus in order to achieve ‘representativeness'. One is to select recordings of speech to represent the various activity types, contexts, and genres into which spoken discourse can be classified. This may be called genre-based sampling. A second method is to sample across the population of the speech community one wishes to represent, in terms of sampling across variables such as region, gender, age, and socio-economic group, so as to represent a balanced cross-section of the population of the relevant speech community. This may be called a demographic sampling. The Flemish Sign Language corpus has targeted a mixture of both genre-based and demographic-based sampling. Language data were collected of 119 deaf men and women from the five regions in Flanders, aged between 12 and 90. Signers were invited to an elicitation session of about 7 hours, and were filmed in pairs. In order to prevent interference from Dutch, the recording sessions were led by a deaf research assistant using VGT. Informants were requested to perform several language tasks so that the corpus consists of elicited grammatical and lexical data, elicited and spontaneous narratives as well as conversational data. Such a corpus is only accessible and searchable, if it is enriched with annotations. Annotation and transcription are done by means of the computer programme ELAN, developed by the Max Planck Institute for Psycholinguistics (MPI) in Nijmegen, the Netherlands, which is freely available from their website (http://tla.mpi.nl/tools/tla-tools/elan/, Crasborn and Sloetjes 2008). The software allows users to create, edit, visualise and search annotations for video (and audio) data, including precise time-alignment of annotations with the corresponding video (and audio) sources. Annotation guidelines were developed, based on the guidelines that were available for other corpora such as Australian Sign Language or Sign Language of the Netherlands. Lexical signs are annotated by means of glosses. It should be stressed here that a gloss is merely a label used to identify and represent a sign. The choice of a specific Dutch word as a gloss for a particular sign does not, for example, entail any grammatical characteristics of the sign (such as word class or grammatical function). This is especially important since in Flemish Sign Language, as in other sign languages, an important number of signs are polysemous and display a broad range of meanings. Fenlon et al. (2015) mention the example of BALL in British Sign Language. This lexeme has several associated keywords in the BSL SignBank demonstrating a broad range in meaning: BALL includes the keywords: ‘ball’, ‘football’, ‘globe’, ‘global’, ‘round’, ‘sphere’, ‘spherical’, ‘world’. In each case, there is no difference in the citation form of the sign used for each meaning and there is an obvious shared meaning between these keywords – e.g. the keywords BALL all refer to something ball-shaped. (p.195) It is essential that annotators gloss consistently, thus always using the same gloss label for the same sign. This is called ID-glossing (Johnston 2008, 2010). Without such a system, searching the corpus for all tokens of a given lexeme is impossible. A lexical database was developed that contains all ID-glosses with a brief description of the meaning and form of the sign. This database is saved as a Google sheet on a shared Google drive, which allows the corpus team members to access and edit the database from different computers. An xml script links the Google sheet to the ELAN files comprising the annotations. The advantage of the Google sheet is that the database thus functions as an external controlled vocabulary (ECV) for the annotations and in this way guarantees consistent glossing. Unfortunately, there are also important disadvantages. Google sheets have a static interface: only text can be entered, no visuals such as pictures or movies can be incorporated. Working with a Google sheet also makes one dependent on Google and the team has already experienced how changes to the Google software can cause problems with the script linking the sheet to the ELAN files. A second disadvantage is that the inventory contains limited information on the signs. It was designed to support the annotation work and to ensure consistency in ID-glossing, but it is not a full-blown lexical database. However, it is possible to extract information from it for lexicographic purposes. In October 2015, the corpus was published on a website (http://www.corpusvgt.be) as an open source, but annotation is still ongoing. Annotation in itself is a time-consuming job. The colleagues from the German Sign Language corpus (DGS-Korpus) estimate that basic annotation of sign language data is done at a ratio of 1:130, meaning that you need 130 minutes to annotate one minute of signing. Moreover, it is crucial that the glossing is done consistently. Experiences of colleagues working on other sign language corpora have taught us that different annotators do not always annotate in a consistent manner, which compromises the automatic search of the corpus. Therefore, team meetings and control mechanisms (by letting annotators check each other’s work) have become part of the whole process. At the time of writing, the VGT corpus consists of approximately 120 hours of video data, of which about 6 hours have been annotated. 3.2. Elaboration of the dictionary Since 2012, the Flemish Sign Language Centre (VGTC) has been conducting additional lexicographic research in order to elaborate the online dictionary. The methodology is largely identical to the original approach described in Section 2.2.1: Data were collected from five regional teams, each consisting of five deaf informants. In each group, one native signer was appointed as moderator. The VGTC provided technical support. Signs related to the following semantic fields have been elicited: Employment and unemployment Geographical places (towns and provinces, countries, etc.) Emergency services Technology (mainly related to communication) and social media Culture: theatre, film, music, dance, exposition, paintings, etc. Medical terms: the human body, (health) care, illness Hygiene Brands: cars, sports, shops/supermarkets Emotions Religion Animal species Politics, elections Deaf-community-related terminology Food and drink Buying and building a house School/education Finances Time Pregnancy In order to elicit these signs, 20 elicitation tasks were designed. For a number of semantic fields, e.g. animals, food and drinks, brands, finances, this involved showing pictures and other visual materials, and asking informants to produce all the different signs that may be used to name what they saw. For the other domains, informants were asked to talk about their own experiences, or to engage in a group discussion. For example, in order to elicit signs related to employment and unemployment, informants were invited to talk about their jobs and/or experiences with unemployment. Some elicitation tasks were designed to elicit signs related to several of the domains. To elicit signs related to emergency services, informants were shown video-footage related to the 9/11 attacks and invited to talk about this. The aim of the task was to elicit signs related to current events, politics, and religion but also signs for different emotions and name signs for the places and people involved. For other semantic fields, e.g. geographical places, buying and building a house, and time, video-recorded signs used in Flemish Sign Language classes were shown to the informants for evaluation. They were asked whether they knew the signs that were shown, whether they themselves used them, and in which contexts the signs would be used. All data collection tasks were video-recorded. Targeted signs were identified in these recordings and re-filmed individually. In the beginning, all signs were analysed in terms of their parameters to make a comparison of phonological variants possible. However, because this phonological analysis proved to be too time consuming, and it was felt there was no need for a phonological analysis of all the elicited signs, later on signs were only labelled by means of spoken language glosses, i.e. Dutch words that have a semantic link with the signs they represent (see also note 2). The sign for car, for instance, is glossed using the Dutch word auto: AUTO. These glosses were included in an online Excel-file and thematically organised. At the time of writing, all additional recordings have been analysed, and the team has just started to mine the Flemish Sign Language corpus to extract additional signs to be included in the dictionary. At the same time, they are working towards the inclusion of the signs in the online dictionary. For some of the collected signs, advice will be sought from an expert committee in order to determine whether the signs should be included in the dictionary (see also Section 4.1.). 3.3. VGT Drop In 2012, the Flemish public broadcasting company VRT started to provide sign language interpreting of the daily news and of Karrewiet, a daily news magazine for children. The interpreters working for the news are hearing, the interpreters working for Karrewiet are deaf. All interpreters, both hearing and deaf, frequently experience ‘lexical gaps’ in their knowledge, i.e. they come across Dutch terms, notions and/or names for which they themselves do not know a sign in Flemish Sign Language, either because no such sign exists, or because they cannot think of it. The VRT therefore asked the Flemish Sign Language Centre (VGTC) for support. In a first response, the VGTC started an online forum using Facebook, the Gebarenpagina (‘Sign page’). The initiative grouped 30 Flemish Sign Language users with an interest in Flemish Sign Language. The VGTC merely acted as the intermediary between the VRT and this group. When the VRT asked for a specific sign, the VGTC posted the request on the group’s Facebook page and members of the group were invited to react. Suggested signs were passed on to the VRT. However, this procedure proved not to be as easy as had been hoped. Requests from the VRT were always very urgent, and most often a notion, term, or name for which a proper translation was asked during the day (usually in the afternoon), was needed by the interpreter that very evening. This timing proved to be too strict for the members of the Sign page. In addition, the group became less and less active. An additional problem is that this approach is too much based on a one word-one sign translation strategy, while it may sometimes be better not to search for a specific sign, but rather provide a paraphrase or use the productive lexicon (see Section 1.2.). This is why it was decided to put the Gebarenpagina on hold. As an alternative, the VGTC has recently started the ‘VGT Drop’ initiative: www.vgtdrop.be. VGT Drop is also an online forum, but it does not use Facebook; rather it is a wiki. The aim of this crowdsourcing tool is to create a forum to publish specialised terminology in Flemish Sign Language, e.g. terminology related to sports, religion, current affairs, proper names for people and places, etc. Everyone can post signs on VGT Drop. The plan is to install an expert group forming a ‘distinguished community’ (see Section 4.1.) to evaluate the contributions to decide whether the proposed sign is acceptable in VGT. At the time of writing, this expert committee has not yet been installed. 3.4. The case of compounds One of the issues that the VGTC collaborators involved in current lexicographic work on VGT would like the expert group to advise on is the inclusion of compounds in the dictionary. Compounding refers to the process of creating new signs from two or more free morphemes. An example of a well-established compound sign in Flemish Sign Language (listed in the dictionary as used in 3 out of the 5 regions) is the sign PARENTS (FATHER∧MOTHER, See Figure 13). Figure 13. View largeDownload slide PARENTS (FATHER∧MOTHER). Figure 13. View largeDownload slide PARENTS (FATHER∧MOTHER). Here the two parts in the compound sign have undergone phonological changes. Whereas in the individual signs for MOTHER and FATHER, there is double contact between the thumb and the middle finger, this is reduced to one contact in the compound sign, resulting in a total of two contacts (rather than four). We should note, however, that signs in a compound may also keep their original form. In general, in dictionaries and lexical databases, compounds are considered unique lexemes and are entered as such. As pointed out by Fenlon et al. (2015: 189), one challenge for the sign language lexicographer is to determine when two signs represent a compound as opposed to when they represent a collocation (i.e. two signs that appear together frequently but do not represent a lexeme). The authors suggest making this decision based on meaning: whilst collocational pairs may look like compounds, their meaning remains predictable. That is, the meaning that is derived from the combination of the signs MAKE and TRUE is not novel and can easily be predicted. Only when such combinations take on broader, unpredictable meanings can we consider them as unique lexemes. (Fenlon et al. 2015: 190) Many compounds in sign languages are loan translations from the surrounding spoken language, or at least, they look like loan translations. The sign BAD∧KAMER (BATH∧ROOM), for example, consists of the signs BAD and KAMER, and as such closely resembles the Dutch compound badkamer (bathroom). However, in many other spoken (and most likely also sign) languages, a similar compound exists, so that the VGT compound sign may not be a loan translation after all but may have simply emerged in VGT itself.5 Nevertheless, it does seem to be the case that many sign languages include compounds that are based on the equivalent compound word in the surrounding spoken language. In fact, in Flemish Sign Language, creating a novel loan translation is used as a coping strategy when encountering lexical gaps. In a recent study investigating native signers and sign language interpreters’ translations for legal terminology (Van Themsche 2017), several informants created a novel loan translation to render the term bewijslast ‘onus of proof’. The compound was formed using the sign meaning bewijs (‘proof’, i.e. the noun) followed by the sign LASTIG meaning ‘troublesome, difficult’ as in ‘a troublesome, difficult child’. It seems that creating a loan translation is the preferred strategy when translating a newly coined Dutch compound. An example from some time ago is when the Dutch compound frietchinees ‘chips+Chinese’ began to be used (to talk about the fact that a growing number of Chinese people were opening chip shops in Flanders). In Flemish Sign Language this soon became a FRIET∧CHINEES. It is sometimes assumed that creating direct loan translations is done more often by hearing L2 users of the language than deaf L1 users, but preliminary research suggests that this is not the case (Gorrebeeck 2012). However, more research is needed before we can say anything conclusive. There are some interesting examples resulting from what may be called ‘Dutch-based compounding’ in VGT (see also the quote from Zwitserlood (2010) in Section 2.2.1). A sign which is no longer used, is the compound sign REGENEN∧PERSOON (RAIN∧PERSON) for the Dutch word regent, which is an older word for a type of secondary school teacher. Although pronounced differently, the written form of this word is identical to the written form of the 2nd and 3rd person singular of the verb regenen ‘to rain’, hence the compound sign REGEN∧PERSOON. A final example is the compound consisting of the sign MOEDER (MOTHER) and the sign KOEK (BISCUIT) meaning placenta. A synonym for placenta in Dutch is the compound moederkoek. In these examples, the iconic depiction present in one of the signs of the compound doesn’t ‘fit’ the overall meaning, i.e. it involves the iconic depiction of holding a round object (the biscuit) and putting it in one’s mouth. Fenlon et al. (2015: 196) discuss a somewhat related example of a loan translation: [B]oth the BSL signs HANGING and HANGOVER depict someone being hanged by the neck. With HANGOVER, this is not an iconic depiction, because the form has no bearing on the meaning of being ill due to excessive drinking. Instead, what has happened here is that semantic extension has occurred with HANGING based on the English word ‘hang’ which comprises part of the word ‘hangover’. Some members of the Flemish Deaf community argue against including such compounds in the dictionary because of the clear influence from Dutch, especially in the case of relatively recent coinages. This brings us back to the ‘who decides’ issue discussed towards the end of Section 2.3. 4. Plans for the immediate future In this section we briefly present some lexicographic initiatives that have already been decided on, and will be launched or continued in the near future. 4.1. VGT Drop – evaluating the contributions The VGT Drop system (see 3.3) allows anyone to post signs. It is therefore important to evaluate the contributions, and it was decided to ask the same expert committee that would be established within the framework of expanding the dictionary for this task. As mentioned in Section 3.3, at the time of writing, this committee has not yet been installed. However, the VGTC has already put forward some suggestions with regard to the committee’s composition and its activities related to VGT Drop. The panel will consist of a maximum of 15 members, from the five different regions. For each region, a maximum of three members will participate in the group. All participants will be deaf and native signers, preferably with deaf family members. Every region will have a contact person, who will help to select the other two members from the region. Meetings of the expert committee will be chaired by a staff member of the VGTC, and at least one sign language linguist will be present during those meetings. The most important question for the expert group is ‘Is this sign used in your region?’ The aim is to check in each of the regions whether a specific sign is considered a ‘proper sign’ in a given region in order to get an idea of the regional dispersion of the sign. When a specific sign is discussed, experts from a given region are expected to provide one answer for their region (rather than individual responses). At the moment, it is not yet clear how the outcomes of these discussions will be dealt with. For example, let us suppose that a specific sign is judged to be acceptable by the representatives of two regions, not known by the representatives of the other two regions and unacceptable by the fifth region. Will this information be added to the sign on the VGT Drop page? Will the sign be included in the dictionary as a sign used in two of the five Flemish provinces? 4.2. Lexical database As explained in Section 3.1., the Flemish corpus team developed the good practice of collecting all the ID-glosses in an online lexical Excel sheet that includes 8,000 glosses. Within the framework of the lexicographic research carried out by the VGTC aiming at the expansion of the online dictionary, the annotation and storage of the lexicon was done slightly differently. Every VGTC-researcher working on the project had their own Excel files with separate sheets for each theme. The intention was to combine and compare these files, but this turned out to be a very time-consuming task. It was therefore decided to create an online platform that can serve as a lexical database for researchers, on the one hand as a resource to assist in the annotation process of sign language data, and on the other as the basis for the expansion of the existing bilingual dictionary and a yet-to-be-developed monolingual Flemish Sign Language dictionary. The lexical database will include video clips showing the signs, the ID glosses, and different types of additional information for each of the lexemes included, e.g. the different meanings, phonological variants, links with/relation to other lexemes, regional origin, etc. Currently, the VGTC is exploring the possibility of making use of the Signbank software to develop this lexical database. The software has already been used for a number of sign languages including AUSLAN (Johnston 2001), BSL (Fenlon et al. 2014) and NGT (Crasborn et al. 2016). 4.3. Monolingual dictionary In September 2016, a meeting was organised by the VGTC to discuss the idea of a general purpose monolingual dictionary for Flemish Sign Language. There was agreement that a monolingual dictionary including information on the different meanings and usages of lexemes would be most useful, especially since the lexicon of Flemish Sign Language is expanding as a result of the emancipation processes of the Deaf community, and the growing societal role of Flemish Sign Language (Vermeerbergen and Van Herreweghe 2008, Van Herreweghe et al. 2016). However, such a project requires substantial financial resources, and these are not readily available. It was also decided to first concentrate on the creation of a proper lexical database, and on the expansion of the bilingual dictionary using data from the Corpus VGT, as well as on securing additional funding to continue annotation work on the Corpus. 4.4. Automatic sign language recognition A very exciting new project in VGT lexicography is related to automatic Sign Language Recognition or SLR (Pigou et al. 2016, 2018). SLR systems may have many different uses, two of which are (semi)automatic corpus annotation and recognition of signs in a sign language dictionary. With respect to the latter, the objective is that users of a sign language dictionary produce a sign (or a series of signs) in front of a camera on their device, and the online dictionary will attempt to detect what sign(s) have been produced, so that it can provide the user with the meaning(s) (in a monolingual dictionary), or translation(s) (in a bilingual dictionary). Pigou et al. (2016, 2018) are tackling the problem by means of deep learning models based on convolutional neural networks (CNNs) (Lecun et al. 1998). CNNs are among the most successful techniques in deep learning, a domain in machine learning that has proven to be very effective at recognizing patterns in high dimensional data such as images, videos, and audio (Pigou et al 2016, 2018). However, we have to admit that at the time of writing, unconstrained SLR remains a big challenge. Sign languages use multiple communication channels (Vermeerbergen et al. 2007a) combined with a high degree of both intra- and inter-sign variability. In addition, publicly available annotated corpora are scarce and not designed for SLR in the first place. That is why successful SLR is only expected in the medium term. 5. Concluding section The first attempts at lexicography related to signs used in the Flemish Deaf community date back to more than 30 years ago with the first larger-scale lexicographic research project on Flemish Sign Language starting almost 20 years ago. Looking back, what has become clear is that both societal and technological developments have played an important role in shaping VGT lexicography. For example, whereas the first sign dictionary (Buyens 1995) covered the unified Signed Dutch lexicon, the changed status of Flemish Sign Language resulted in the Flemish government funding research to determine the extent of regional variation and standardisation in Flemish Sign Language, which led to the first Flemish Sign Language dictionary. The use of Flemish Sign Language in education brought with it a need for lexical modernisation, as did the start of sign language interpreting of the daily news, which was also one of the most important reasons for the VGT Drop initiative. The impact of technological developments can be seen in the different forms of the dictionaries: whereas the Signed Dutch Gebarenwoordenboek (Buyens 1995) is a printed dictionary, with signs illustrated by means of drawings, the Flemish Sign Language/Dutch – Dutch/Flemish Sign Language dictionary is an electronic publication including videoclips for the signs, which allows for more user-friendly bi-directional searches. Moreover, whereas initially signs were collected from a limited number of signers performing different elicitation tasks, today inclusion of signs in dictionaries is done by exploiting the Flemish Sign Language Corpus and organising crowd-sourcing initiatives making use of the internet. Automatic sign language recognition opens up exciting new avenues in sign-language lexicography. Possibly somewhat less obvious from the overview of past and current lexicographic activities related to Flemish Sign Language is the fact that societal changes and technological advances – and the consequent changes related to sign language uses and practices – also have an impact on the language itself. Flemish signers are more actively involved in hearing society, using their sign language to communicate in that society in many different semantic domains, and this results in an exponential growth of the lexicon itself. The diminishing importance of deaf schools, together with an increased contact between deaf people from different regions, has most certainly had a major impact on the spontaneous standardisation process of Flemish Sign Language (Van Herreweghe and Vermeerbergen 2009). The fact that sign languages are increasingly used for remote communication (which involves signing in front of a camera, which in turn impacts on signing space) may in time have an impact on the structure of the language itself (Meurant et al. 2013). Obviously, societal and technological developments will continue to shape both the language itself and the form of lexicographic projects related to the language. We expect the current spontaneous standardisation process of Flemish Sign Language to gain even more momentum in future, as Flemish Sign Language becomes even more frequently used on television, the internet, and social media. The increase in educational interpreting in mainstream education, the resulting presence of deaf students in tertiary education, the emergence of the first deaf academics, and deaf high-profile professionals (e.g. politicians) using Flemish Sign Language as their first language will continue to contribute to the expansion of the lexicon. At the same time, technological advances in the form of cochlear implantation have led to a discouragement in the use of Flemish Sign Language by young deaf children and the encouragement of their transition to mainstream education (Van Herreweghe et al. 2016, Mouvet 2013). Whether this will have an impact on the form and structure of Flemish Sign Language and its lexicon remains to be seen. All natural languages evolve, which implies that a constant revision of lexicographic research and outcomes is necessary. It seems that, like many other sign languages, Flemish Sign Language is experiencing particularly interesting times, with important societal changes related to its uses and practices resulting in rapid changes. This, together with technological advances, may give rise to inspiring new lexicographic initiatives. We are convinced that there is much to look forward to when it comes to the future of sign language lexicography. Footnotes 1 It is a widely recognised convention to use upper case Deaf for reference to the linguistic community of sign language users. 2 A sign is usually transcribed in the form of a gloss: a word (or several words) from a spoken language used to ‘label’ the sign. The meaning of the spoken language word more or less resembles the meaning of the sign. Glosses are written in capital letters. When one sign is labeled by a gloss consisting of more than one word, the words are connected with a hyphen. A∧ between two words means it is a compound sign. 3 The study dates from before the publication of the online dictionary. 4 I.e., choosing one regional variety as the standard one, meaning that the others would be non-standard and would thus receive a lower status; choosing a set of ‘preferred’ signs, again implicitly rejecting the non-preferred signs, etc. 5 We may note here that the component parts of the compound BAD∧KAMER have not undergone any changes and that its meaning is predictable. Yet, it clearly is a compound sign and not a collocational pair. References Center for Tegnsprog . 2008 . Ordbog over Dansk Tegnsprog. http://www.tegnsprog.dk/. [The Danish Sign Language Dictionary.] De Weerdt K. , Van Herreweghe M. , Van Mulders K. , Vermeerbergen M. . 2004. Woordenboek Nederlands/Vlaamse Gebarentaal - Vlaamse Gebarentaal/Nederlands. http://gebaren.ugent.be/ [Dutch/Flemish Sign Language - Flemish Sign Language/Dutch Dictionary] Institutionen för Lingvistik . 2009 . Svenskt teckensprakslexikon. http://teckensprakslexikon.su.se [Swedish Sign Language Lexicon] Kuurojen Liittory . 2003 . Suvi Suomalaisen Viittomakielen Verkosanakirja. http://www.viittomat.net/ [Finnish Sign Language Dictionary] Anthony D. A. 1966 . Seeing Essential English. MA Thesis, Eastern Michigan University, Ypsilanti. Anthony D. A. 1971 . Seeing Essential English . California : Educational Services Division, Anaheim Union High School District . Bornstein H. 1974 . ‘ Signed English: A Manual Approach to English Language Development.’ Journal of Speech and Hearing Disorders 39 : 330 – 343 . Google Scholar CrossRef Search ADS PubMed Bornstein H. , Saulnier K. , Hamilton L. . 1980 . ‘ Signed English: A First Evaluation.’ American Annals of the Deaf 125 . 4 : 467 – 481 . Google Scholar CrossRef Search ADS PubMed Brennan M. 1990 . Word formation in British Sign Language. Ph.D. Thesis, University of Stockholm. Brennan M. 1992 . ‘The Visual World of British Sign Language: An Introduction.’ In Brien D. (ed), Dictionary of British Sign Language/English . London : Faber and Faber , 1 – 133 . Buyens M. 1987 . Nederlands met gebaren. Taalkundige, Sociale en Opvoedkundige Aspecten. Gent: Federatie van Vlaamse Dovenverenigingen [Signed Dutch. Linguistic, Social and Educational Aspects]. Buyens M. 1995 . Gebarenwoordenboek. Leuven/Apeldoorn: Garant [Sign Dictionary]. Cormier K. , Quinto-Pozos D. , Sevcikova Z. , Schembri A. . 2012 . ‘Lexicalisation and De-lexicalisation Processes in Sign Languages: Comparing Depicting Constructions and Viewpoint Gestures.’ Language and Communication 32 . 4 : 329 – 348 . Google Scholar CrossRef Search ADS PubMed Crasborn O. , Bank R. , Zwitserlood I. , van der Kooij E. , Ormel E. , Ros J. , Schüller A. , de Meijer A. , van Zuilen M. , Nauta Y.E. , van Winsum F. , Vonk M. . 2016 . NGT Signbank. Nijmegen: Radboud University, Centre for Language Studies. Crasborn O. , Sloetjes H. . 2008 . ‘Enhanced ELAN Functionality for Sign Language Corpora’ In Crasborn O. , T Hanke , Efthimiou E. , Zwitserlood I. , Thoutenhoofd E. (eds), Construction and Exploitation of Sign Language Corpora. 3rd Workshop on the Representation and Processing of Sign Languages. Parijs : ELRA , 39 – 43 Cuxac C. 2000 . La Langue des Signes Française. Les voies de l’iconicité. (Faits de langues, no 15-16.) Paris: Ophrys [French Sign Language. The Directions of Iconicity]. De Clerck G. 2007 . ‘ Meeting global Deaf peers, visiting ideal Deaf places: Deaf ways of education leading to empowerment, an exploratory case study.’ American Annals of the Deaf 152 . 1 : 5 – 19 . Google Scholar CrossRef Search ADS PubMed De Weerdt K. , Rogiest M. . 2003a . Invulling van Hiaten in het Lexicon van de Vlaamse Gebarentaal: Wiskunde. Gent: Cultuur voor Doven [Filling in Gaps in the Flemish Sign Language Lexicon: Mathematics]. De Weerdt K. , Rogiest M. . 2003b . Invulling van Hiaten in het Lexicon van de Vlaamse Gebarentaal: Aardrijkskunde en Geschiedenis. Gent: Cultuur voor Doven Doven [Filling in Gaps in the Flemish Sign Language Lexicon: Geography and History]. De Weerdt K. , Vanhecke E. , Van Herreweghe M. , Vermeerbergen M. . 2003 . Op (Onder)zoek naar de Vlaamse Gebaren-Schat. Gent: Cultuur voor Doven [In (Re)Search of Flemish Signs]. Eichmann H. 2008 . ‘Hands off our language!’, Deaf sign language teachers' perspectives on sign language standardisation. Ph.D. Thesis, University of Central Lancashire. Engberg-Pedersen E. 1993 . Space in Danish Sign Language . Hamburg : Signum-Verlag . Fenlon J. , Cormier K. , Schembri A. . 2015 . ‘Building BSL SignBank: The Lemma Dilemma Revisited ’. International Journal of Lexicography 28 . 2 : 169 – 206 Google Scholar CrossRef Search ADS Fenlon J. , Cormier K. , Rentelis R. , Schembri A. , Rowley K. , Adam R. , Woll B. . 2014 . BSL Signbank: A Lexical Database of British Sign Language ( First Edition ). London : Deafness, Cognition and Language Research Centre, University College London . http://bslsignbank.ucl.ac.uk. Ferrara L. 2012 . A Grammar of Depiction: Exploring Gesture and Language in Australian Sign Language (Auslan). Ph.D. Thesis, Macquarie University, Sydney. Geysels G. , Ostyn C. , Snoeys I. , Vlaeminck K. . 1989 . Basiswoordenschatlijst. Bron voor Verdere Ontwikkelingen binnen de Tolkenopleiding. Thesis, Glorieux, Gentbrugge [Basic Word List. Resource for Further Developments in the Interpreter Training Programme]. Gorrebeeck W. 2013 . Van Vogelhuisfilmpje tot Elandlasagne. Hoe Gaan Doven en Tolken om met Hiaten in het/hun Lexicon? MA Thesis. KU Leuven, Antwerpen [From Bird House Film to Moose Lasagne. How do Deaf People and Interpreters Deal with Gaps in the/their Lexicon?]. Gustason G. , Pfetzing D. , Zawolkow E. . 1972 . Signing Exact English . Rossmoor, L.A .: Modern Signs Press . Heidbuchel H. 1963 . ‘A.B.N.: Waarom en hoe?’ Bouw: Zesdaagsenummer VIII: 6-7 [Why and How?]. Heyerick I. , Vermeerbergen M. . 2012 . ‘Sign Language Interpreting in Educational Settings in Flanders, Belgium’ In Leeson L. , Vermeerbergen M. (eds), Working with the Deaf Community: Education, Mental Health and Interpreting . Dublin : Interesource Group (Ireland) Limited , 119 – 134 . Johnston T. 2001 . ‘The Lexical Database of Auslan (Australian Sign Language)’ . Sign Language & Linguistics 4 . 1/2 : 145 – 169 . Johnston T. 2008 . ‘Corpus Linguistics and Signed Languages: No lemmata, no corpus’ In Crasborn O. , Hanke T. , Efthimiou E. , Zwitserlood I. , Thoutenhoofd E. (eds), Construction and Exploitation of Sign Language Corpora. 3rd Workshop on the Representation and Processing of Sign Languages. Parijs : ELRA , 82 – 97 Johnston T. 2010 . ‘ From Archive to Corpus: Transcription and Annotation in the Creation of Signed Language Corpora’ . International Journal of Corpus Linguistics 15 . 1 : 104 – 129 . Google Scholar CrossRef Search ADS Johnston T. 2016 . Auslan Corpus Annotation Guidelines. Unpublished manuscript. Macquarie University. Lecun Y. , Bottou L. , Bengio Y. , Haffner P. . 1998 . ‘ Gradient-based Learning Applied to Document Recognition.’ Proceedings of the IEEE 86 . 11 : 2278 – 324 . Google Scholar CrossRef Search ADS Leech G. 2000 . ‘Same Grammar or Different Grammar? Contrasting Approaches to the Grammar of Spoken English Discourse’ In Sarangi S. , Coulthard M. (eds.), Discourse and Social Life . Harlow : Pearson Education , 48 – 65 . Loots G. , Devisé I. , Lichtert G. , Hoebrechts N. , Van De Ginste C. , De Bruyne I. . 2003 . De Gemeenschap van Doven en Slechthorenden in Vlaanderen. Communicatie, Taal en Verwachtingen omtrent Maatschappelijke Toegankelijkheid. Gent: Cultuur voor Doven [The Community of Deaf and Partially Deaf People in Flanders. Communication, Language and Expectations about Societal Accessibility]. Meurant L. , Sinte A. , Van Herreweghe M. , Vermeerbergen M. , 2013 . ‘Sign Language Research, Uses and Practices: A Belgian Perspective’ In Meurant L. , Sinte A. , Van Herreweghe M. , Vermeerbergen M. (eds), Sign Language Research, Uses and Practices: Crossing Views on Theoretical and Applied Sign Language Linguistics. (Sign Language and Deaf Communities 1), Boston/Berlin : De Gruyter Mouton & Nijmegen: Ishara Press , 1 – 14 . Google Scholar CrossRef Search ADS Mouvet K. 2013 . What about Sign Language? A Longitudinal Study of the Language Development of Young Deaf Children in Flanders in Times of Cochlear Implantation. Ph.D. Thesis, Ghent University & Vrije Universiteit Brussel. Pigou L. , Van Herreweghe M. , Dambre J. , 2016 . ‘Sign Classification in Sign Language Corpora with Deep Neural Networks.’ In Calzolari N. , Choukri K. , Declerck T. , Goggi S. , Grobelnik M. , Maegaard B. , Mariani J. , Mazo H. , Moreno A. , Odijk J. , Piperidis S. (eds). Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC) . Parijs : ELRA , 175 – 178 . Pigou L. , van den Oord A. , Dieleman S. , Van Herreweghe M. , Dambre J. . 2018 . ‘ Beyond Temporal Pooling: Recurrence and Temporal Convolutions for Gesture Recognition in Video .’ International Journal of Computer Vision 126 : 430 (doi:10.1007/s11263-016-0957-7). Google Scholar CrossRef Search ADS Reagan T. 2001 . ‘Language Planning and Policy’ In Lucas C. (ed.), The Sociolinguistics of Sign Languages . Cambridge : Cambridge University Press , 145 – 180 . Google Scholar CrossRef Search ADS Schermer T. 2004 . ‘Lexical Variation in Sign Language of the Netherlands.’ In Van Herreweghe M. , Vermeerbergen M. (eds), To the Lexicon and Beyond. Sociolinguistics in European Deaf communities . Washington DC : Gallaudet University Press , 91 – 110 . Stokoe W. C. 1960 . Sign Language Structure. An outline of the visual communication system of the American Deaf . Silver Spring MD : Linstok Press . Tervoort B. 1953 . Structurele Analyse van Visueel Taalgebruik binnen een Groep Dove Kinderen . Amsterdam : Noord-Hollandse Uitgevers Maatschappij [Structural Analysis of the Visual Language Use in a Group of Deaf Children ]. Vanhecke E. , De Weerdt K. . 2004 . ‘Regional Variation in Flemish Sign Language.’ In Van Herreweghe M. , Vermeerbergen M. (eds), To the Lexicon and Beyond. Sociolinguistics in European Deaf Communities . Washington DC : Gallaudet University Press , p. 27 – 38 . Van Herreweghe M. 1996 . Prelinguaal Dove Jongeren en Nederlands: een Syntactisch Onderzoek. Ph.D. Thesis. Universiteit Gent [Young Prelingually Deaf People and Dutch: A Syntactic Study]. Van Herreweghe M. 1995 . De Vlaams-Belgische Gebarentaal: Een eerste verkenning. Gent: Academia Press [Flemish-Belgian Sign Language: A First Exploration]. Van Herreweghe M. , De Meulder M. , Vermeerbergen M. . 2016 . ‘From Erasure to Recognition (and Back Again?). The Case of Flemish Sign Language’ In Marschark M. , Spencer P. (eds), The Oxford Handbook of Deaf Studies in Language: Research, Policy, and Practice . Oxford : Oxford University Press , p. 45 – 61 . Van Herreweghe M. , Vandemeulebroucke E. . 2016 . ‘ Vlaamse Gebarentaligen en standaard-VGT: Verstoten of Omarmen?’ Taal En Tongval 68 . 2 : 201 – 236 [Flemish Sign Language People and Standard VGT: Embraced or Dismissed?]. Google Scholar CrossRef Search ADS Van Herreweghe M. , Vermeerbergen M. . 2003 . ‘Het Opsporen en Invullen van “Gaten” in het Lexicon van de Vlaamse Gebarentaal’ In Koole T. , Nortier J. , Tahitu B. (eds), Artikelen van de 4e Sociolinguïstische Conferentie . Delft : Eburon , 445 – 454 [Tracing and Filling in ‘Gaps’ in the Flemish Sign Language Lexicon]. Van Herreweghe M. , Vermeerbergen M. . 2004 . ‘Flemish Sign Language: Some Risks of Early Codification’ In Van Herreweghe M. , Vermeerbergen M. (eds), To the Lexicon and Beyond. Sociolinguistics in European Deaf communities . Washington : Gallaudet University Press , 111 – 137 . Van Herreweghe M. , Vermeerbergen M. . 2009 . ‘ Flemish Sign Language Standardisation.’ Current Issues in Language Planning 10 . 3 : 308 – 326 . Google Scholar CrossRef Search ADS Van Hoof S. , Jaspers J. . 2012 . ‘Hyperstandaardisering’ . Tijdschrift voor Nederlandse taal- en letterkunde 128 . 2 : 97 – 125 [Hyperstandardisation]. Van Keymeulen J. 1993 . ‘De Algemene Woordenschat in het WVD: een Nieuwe Aanpak.’ WVD-KONTAKT: 18-20 [The General Lexicon in the WVD: a New Approach]. Van Keymeulen J. 2003 . ‘ Compiling a Dictionary of an Unwritten Language. A Non Corpus-based Approach.’ Lexikos 13 : 183 – 205 Van Themsche E. 2017 . Juridisch lexicon in Vlaamse Gebarentaal. MA Thesis. KU Leuven, Antwerpen [Legal Lexicon in Flemish Sign Language]. Vermeerbergen M. 1996 . ROOD KOOL TIEN PERSOON IN. Morfo-syntactische aspecten van gebarentaal. Ph.D. Thesis, Vrije Universiteit Brussel [RED CABBAGE TEN PEOPLE IN. Morpho-syntactic Aspects of Sign Language]. Vermeerbergen M. 1997 . Grammaticale Aspecten van de Vlaams-Belgische Gebarentaal . Gentbrugge: Cultuur voor Doven [Grammatical Aspects of Flemish-Belgian Sign Language]. Vermeerbergen M. 2016 . ‘When There Is Little Difference Between (The Frozen Sign) TAP, (The Productive Classifier Sign) “open-flask” and CA: “open-jar-with-great-difficulty”. Defining and Describing the “Productive Lexicon” in Signed Languages.’ Invited talk, “Sign Pop-Up”, Radboud University, Nijmegen. Vermeerbergen M. , Nijen Twilhaar J. , Van Herreweghe M. . 2013 . ‘Variation between and within Sign Language of the Netherlands and Flemish Sign Language’ In Taeldeman J. , Hinskens F. (eds), Language and Space. An International Handbook of Linguistic Variation. Dutch. (Handbücher zur Sprach- und Kommunikationswissenschaft / Handbooks of Linguistics and Communication Science (HSK), 30/3). Boston/Berlin : de Gruyter Mouton , 680 – 699 . Vermeerbergen M. , Leeson L. , Crasborn O. (eds). 2007a . Simultaneity in Signed Languages: Form and Function . Amsterdam : John Benjamins . Google Scholar CrossRef Search ADS Vermeerbergen M. , Van Herreweghe M. , Akach P. , Matabane E. . 2007b . ‘ Constituent Order in Flemish Sign Language and South African Sign Language. A Cross-linguistic Study.’ Sign Language & Linguistics 10 . 1 : 25 – 54 . Google Scholar CrossRef Search ADS Vermeerbergen M. , Van Herreweghe M. . 2008 . ‘De Status van de Vlaamse Gebarentaal: Van Ondergronds Bestaan tot Culturele Erkenning’ In Vermeerbergen M. , Van Herreweghe M. (eds), Wat (Geweest/Gewenst) Is. Organisaties van en voor Doven in Vlaanderen Bevraagd over 10 Thema’s . Gent : Academia Press / Fevlado-Diversus, 1–25 [The Status of Flemish Sign Language. From an Underground Existence to Cultural Recognition]. Zwitserlood I. 2010 . ‘ Sign Language Lexicography in the Early 21st Century and a Recently Published Dictionary of Sign Language of the Netherlands’ . International Journal of Lexicography 23 . 4 : 443 – 476 . Google Scholar CrossRef Search ADS © 2018 Oxford University Press. All rights reserved. For permissions, please email: firstname.lastname@example.org This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
International Journal of Lexicography – Oxford University Press
Published: May 8, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera