A Grammar of Sign Language of the Netherlands (NGT)

Much of Chapter 1 is based on the PhD thesis of van der Kooij (2002). The data she used came from the SignPhon database (Blees et al. 1996 in: van der Kooij 2002) in which citation forms of signs are stored. At the time of her research, this database contained at least 3,000 signs with a phonetic description. Additionally, van der Kooij consulted the signers of the SignPhon database for well-formedness judgements. The signers were all female native signers from Voorburg/Zoetermeer, Rotterdam and Amsterdam (van der Kooij 2002: 17). There is no specified methodology for the research of van der Kooij & Crasborn (2008), but the authors do mention the use of narratives (p. 1308) and the intuitions of two native signers (p. 1321).

The first descriptions of handshapes in NGT were made in the course of the KOMVA project. The data in this project consisted of a corpus of more than 15,000 signs from 100 signs, who all came from different regions (Groningen, Voorburg, Eindhoven, Rotterdam, and Amsterdam) (Harder & Schermer 1986; Trude Schermer, personal communication July 2020). The handshape drawings used in the tables in this chapter have been developed by the Dutch Sign Center and are used in the paper dictionary (Schermer & Koolhof (eds.) 2009), while photos of these handshapes are used in the online dictionary. The categorization of these handshapes into 31 combinations of phonological features is done following van der Kooij (2002). Most examples that were used to illustrate minimal pairs and phonological features were selected by myself. We used the online dictionary of the Dutch Sign Centre (Schermer et al. 2013) to investigate possible combinations of selected fingers (PHONOLOGY 1.1.1.1), and to deduce the handshapes that form a selected set of alphabetic and numeral handshapes, as described in PHONOLOGY 1.1.3, since the online dictionary makes it possible to look for specific handshapes.

In April 2020, the dictionary included 16,760 glosses, which could refer to about 20,000 signs (including variants) (Trude Schermer, personal communication April 2020). Since we consulted the dictionary in light of a phonological description of NGT, it must be noted that this online dictionary is not only a descriptive collection of vocabulary, but additionally has an educational and informative function. Signs that originated within the deaf community co-exist with signs that found their way into the dictionary in another way, e.g. upon request of signers who need signs for certain concepts. This is particularly relevant for the implementation of locally used signs for countries. As a consequence of adopting these local signs, some handshapes that are attested in the NGT dictionary seem to only occur in loan signs. This is relevant for PHONOLOGY 1.1.1.1 where we not only looked at possible combinations of selected fingers, but also checked whether the handshapes which were not attested by van der Kooij yielded results in the NGT online dictionary. Some handshapes yielded more results than others, and some turned out to be very infrequent, or turned out to be merely used for name signs and country signs – which may include extraordinary phonological features. The influence of these loan signs on the phonology on NGT is yet to be investigated. Still, to determine whether the phonological combinations of van der Kooij are relevant for NGT, we investigated whether native signs were included in the results, too. This always turned out to be the case for at least one (KOMVA) handshape per phonological combination.

The NGT dataset in the Global Signbank database (Crasborn et al. 2020) was used for more recent and more representative distributions of handshape features, location features, and handedness. The main purpose of this database is to store signs that are found in the data from the Corpus NGT, including phonological information of these signs. WeI received administrator rights to be able to conduct these analyses, and downloaded the frequencies of combinations of phonological handshape features (“handshapes”) articulated by the strong hand. For the analysis of handshapes, we relied on the analysis as shown on the NGT Signbank website, but did not include datapoints for which no information on handshape was available. After exclusion of these datapoints, 3,798 signs remained, on which the distributions in Table 2.7 are based. 

As for location, Klomp conducted her own analysis, and first downloaded a file with all available signs. As of July 2020, 4,162 signs were stored, of which 4,082 were datapoints extracted from the Corpus NGT, and the other 80 signs came from projects carried out at the Radboud University Nijmegen (which also hosts the database). Sheonly took signs from the Corpus NGT into account, and deleted signs which had an occurrence of zero. She then ranked the remaining signs based on their location specification. The datasets from van der Kooij and the NGT Signbank both included specifications on subareas, but distinguished these subareas slightly different. For the sake of comparison, she took the frequencies of the main areas from van der Kooij, and merged the frequencies of the subareas from the NGT Signbank to gain one frequency number per main area. Locations which were specified with a location change were included in the category of the start location. To be more precise, the main areas in the first column in Table 2.13 included the subareas in the second column. 

Table 2.13. The categorization of main locations from the NGT Signbank

(Crasborn et al. 2020).

The full dataset was additionally used to look for minimal pairs related to one- or two-handed articulation, and no examples were found. In some cases, the phonological specifications in the dataset implied a minimal pair, e.g. with skinny and obedient, which, indeed, are very similar, but the videos on the NGT Signbank website systematically showed two-handed articulations, thus weakening the difference. As for types of two-handed signs, the NGT Signbank distinguishes three types, and one of these types is called ‘2n’ and includes the group ‘symmetrical but not mirrored’. Only 31 signs in their dataset was specified for ‘2n’, and not all of them are ‘symmetrical but not mirrored’; we therefore conclude that only this latter group of two-handed signs is very small.

In order to verify the nature of manner features as described in PHONOLOGY 1.3 and PHONOLOGY 1.3.1, Klomp consulted a female fluent signer of 58 years old, who has lived in the South of the Netherlands and in the Amsterdam area. They discussed the examples given by van der Kooij and confirmed that the features are indeed phonologically distinctive in some of the cases proposed by van der Kooij. However, although originally proposed to be applicable to all movement types, they found that the manner features ‘tense’ and ‘directionality’ only apply to path movements. We therefore described these features in PHONOLOGY 1.3.1 and not in the introduction of PHONOLOGY 1.3.

The information on mouth actions is mainly based on the PhD dissertations of Schermer (1990) and Bank (2014). The former elicited data from six informants from Groningen and Amsterdam. The informants retold a written Dutch story, signed a story based on a picture-book, and/or engaged in a spontaneous conversation. Bank (2014) extracted data from the Corpus NGT (see Introduction to this thesis). For this particular study, 40 videos were analyzed. For somewhat more information, and for the information for Example 1 from Klomp (2019a), see Information on Data and Consultants at the end of Syntax, Chapter 3.

Concerning the examples shown in the figures and video clips, most were selected by Klomp to illustrate the phenomenon at stake. Whenever we took an example from another source, we cite the reference directly preceding or following the figure/clip.