The Word Rain visualisation technique is a development of the classic word cloud, providing more possibilities for analysis of the texts visualised. We here briefly describe the work carried out so far, the reasoning behind it, and what could lie ahead.

Modern språkdokumentation har gått från att fältlingvisten behandlar talarna av ett språk som ganska passiva informanter till att vara mycket aktiva språkkonsulter. I detta ingår att språkkonsulterna ska kunna hantera teknisk utrustning som inspelningsapparater och att kunna göra transkriptioner på egen hand.

Många verktyg för transkriptioner är dock komplicerade både att arbeta med och att exportera data från. Här presenteras ett språkoberoende verktyg för transkription av ljudfiler som är direkt anpassat för annotering enligt Leipzig glossing rules.

Verktyget presenterar ljudet på en webbsida som scrollas på vanligt sätt: uppåt och neråt. På detta sätt undviks den oöversiktliga vy som är vanlig i många verktyg där ljudet presenteras som en enda lång horisontell tidslinje och där användaren behöver zooma in och ut i materialet för att växla mellan översikt och detalj.

Ljudfilen delas upp i ungefär 10 sekunder långa delar, som sedan läggs under varandra. Verktyget försöker dela upp ljudfilen vid pauser, men om användaren hittar lämpligare punkter i ljudfilen kan punkterna enkelt justeras. Varje del av ljudfilen representeras av ett spektrogram som inte är tillräckligt högupplöst för ingående fonetiska analyser, men hjälper användaren att navigera och avgöra var yttranden börjar och slutar.

Användaren får sedan tillgång till tre transkriptionslager: fonetiskt/fonematiskt, glossning och fri översättning. Det fonetisk/fonematiska lagret ger ett stöd till glossningen. Glossningslagret används för morfosyntaktisk transkription och morfosyntaktisk förklaring enligt Leipzig glossing rules. Slutligen löper den fria översättningen över hela yttrandet.

Verktyget är tänkt att användas för ett lager i taget, där användaren markerar en region i spektrogrammet med musen, lyssnar på regionen med TAB-tangenten och sedan enkelt skriver in transkriptionen. Transkriptioner i det fonetisk/fonematiska lagret sprids ut över hela den markera- de regionen för att foner/fonem ska komma någorlunda nära respektive plats i spektrogrammet.

Verktyget kan generera direkt publicerbar LATEX-kod med formatering enligt Leipzig glossing rules och automatiska tidsstämplar för varje transkriberat yttrande i ljudfilen.

Verktyget utvecklas av Språkbanken Sam inom Nationella språkbanken.

Word Rain is a development of the classic word cloud. It addresses some of the limitations of word clouds, in particular the lack of a semantically motivated positioning of the words, and the use of font size as a sole indicator of word prominence. Word Rain uses the semantic information encoded in a distributional semantics-based language model – reduced into one dimension – to position the words along the x-axis. Thereby, the horizontal positioning of the words reflects semantic similarity. Font size is still used to signal word prominence, but this signal is supplemented with a bar chart, as well as with the position of the words on the y-axis. We exemplify the use of Word Rain by three concrete visualization tasks, applied on different real-world texts and document collections on climate change. In these case studies, word2vec models, reduced to one dimension with t-SNE, are used to encode semantic similarity, and TF-IDF is used for measuring word prominence. We evaluate the technique further by carrying out domain expert reviews.

Technology has historically had both beneficial and detrimental effects on languages. In our lifetimes, computers and computer based communication have gone from obscure to mainstream, even indispensable. Having technology thwart you when you write your language can be devastating and make you not want to use it. Being instead nudged by helpful keyboards, good spell checkers, and predictive language models can increase your enjoyment and make you write it more.

I will give a brief tour of the history of language technology in a broad sense, from typewriters to modern smartphones, and describe the challenges facing many languages today that lack the proper tools, focusing on the Swedish national minority languages. I will also describe what we do at the Language Council of Sweden to both raise awareness and develop concrete tools for these languages.

The current Lexin website is quite old and has not been optimised for today’s use of smartphones. Using a smartphone is very different from using a traditional desktop computer, not only because of the size of the screen but also because of the different accommodations needed to support low-precision input methods, such as using fingers as opposed to a mouse and keyboard. Our users also come from very diverse language backgrounds, and even if we have translated the dictionary into around 20 languages, there might not be a translation available in their first, or preferred, language, or the translation/version of the dictionary might not be comprehensive. To work around this, the user might perform multiple searches in different language versions of the dictionary, i.e., in different dictionaries, which is even harder in a smartphone environment. We have therefore made it possible to look up a word in multiple languages at once. We have also implemented an offline search where users can download one or more dictionaries and continue to use the website even when they have no internet connection. In short, adapting the dictionary to small screens has made us rethink and rebuild the whole user experience of our online dictionary.

The Language Council of Sweden at the Institute for Language and Folklore has a responsibility to support the development of language technology for the languages of Sweden. Together with the Department of National Minorities and Swedish Sign Language at the same institute, we are developing tools to collect and analyse texts in the national minority languages of Sweden. These texts will be available for use in research, for example in lexicographic work. We here present the specific challenges associated with building these tools for the Swedish national minority language Meänkieli.

The classic word cloud remains a popular visualisation technique, also to use for more advanced text exploration and comparison tasks. However, since the standard word cloud does not provide any support for these kinds of analytical tasks, we have created the Word Rain visualisation technique, which is a development of the classic word cloud. The Word Rain technique positions paradigmatically similar words close to each other on the x-axis, which makes it easier to identify semantic word clusters and to carry out comparison tasks. We have previously applied the technique on several different tasks, and we here show how the Word Rain visualisation can support a topical analysis of the text collection content. We first apply the topic modelling tool Topics2Themes to a collection of texts on the subject of climate change, and then use the Word Rain technique to visualise the automatically extracted topics. The Word Rain visualisation applied on the entire text collection provides an overview of its content, sorted according to paradigmatic similarity. When also creating focused word rain visualisations for the extracted topics, a visual semantic profile for each one of the topics is created, which supports the tasks of understanding and comparing topics. We have, thereby, here provided yet an example of how the Word Rain technique can be practically used for visualising and exploring texts.

Word Rain is a novel approach to the classic word cloud that uses word embeddings to make it useful for exploring the word content of a text or corpus. Downloading and running the code can, however, be prohibitively difficult or cumbersome for non-technical users and casual evaluation. Since Word Rain also requires a word embeddings model, the inexperienced or casual user would benefit greatly from a streamlined interface. We have therefore collected everything that is needed in a web based service and are making it available as a SWELANG K-centre resource.

The Swedish Language Act states that the public sector is responsible for protecting the Swedish national minority languages. Since these languages typically have few speakers, commercial actors do not create the language technology necessary in an increasingly digitalised society. One way for the public sector to facilitate creation of such technology is by making lexicographic resources available to the public. In this article, we present a survey of existing resources and guidelines for new resources, such as the early inclusion of a computational linguistic perspective.

Yiddish is one of the national minority languages of Sweden, and one of the languages for which the Swedish Institute for Language and Folklore is responsible for developing useful language resources. We here describe the web-based version of a Swedish–Yiddish/Yiddish–Swedish dictionary. The single search field of the web-based dictionary is used for incrementally searching all three components of the dictionary entries (the word in Swedish, the word in Yiddish with Hebrew characters and the transliteration in Latin script). When the user accesses the dictionary in an online mode, the dictionary is saved in the web browser, which makes it possible to also use the dictionary offline.