Futrell and Mahowald's assumptions about parallels between language models and human language are primarily based on LMs' performance in English. In our commentary, we discuss how this impacts some of their key takeaways and highlight LM shortcomings in languages other than English (specifically, Icelandic), which are only alluded to in the target article.

This paper evaluates current Large Language Model (LLM) benchmarking for Icelandic, identifies problems, and calls for improved evaluation methods in low/medium-resource languages in particular. We show that benchmarks that include synthetic or machine-translated data that have not been verified in any way, commonly contain severely flawed test examples that are likely to skew the results and undermine the tests' validity. We warn against the use of such methods without verification in low/medium-resource settings as the translation quality can, at best, only be as good as MT quality for a given language at any given time. Indeed, the results of our quantitative error analysis on existing benchmarks for Icelandic show clear differences between human-authored/-translated benchmarks vs. synthetic or machine-translated benchmarks.

INTRODUCTIONInterventions to treat speech-language difficulties in primary progressive aphasia (PPA) often use word accuracy as a highly comparable outcome. However, there are more constructs of importance to people with PPA that have received less attention.METHODSFollowing Core Outcome Set Standards for Development Recommendations (COSSTAD), this study comprised: Stage 1 - systematic review to identify measures; Stage 2 - consensus groups to identify important outcome constructs for people with PPA (n = 82) and care partners (n = 91); Stage 3 - e-Delphi consensus with 57 researchers.RESULTSThe systematic review identified 84 Outcome Measurement Instruments. Core outcome constructs identified included: (1) Participate in conversations with family and friends, (2) get words out, (3) be more fluent, (4) convey a message by any means, and (5) understand what others are saying. Researchers were unable to reach a consensus on measurement instruments.DISCUSSIONFurther work is required to develop appropriate measurement instruments that address all core outcome constructs important to key stakeholders.Highlights We introduce new symptom-led perspectives on primary progressive aphasia (PPA). The focus is on non-fluent/agrammatic (nfvPPA) and semantic (svPPA) variants. Foregrounding of early and non-verbal features of PPA and clinical trajectories is featured. We introduce a symptom-led staging scheme for PPA. We propose a prototype for a functional impairment scale, the PPA Progression Planning Aid.

Rising public concern regarding the linguistic effects of an increased, digitally mediated English presence in Iceland was one of the factors motivating the research project Modeling the Linguistic Consequences of Digital Language Contact (MoLiCoDiLaCo), in 2016-2019. In public discourse, a causal relationship between digital English input and reduced/incompletely acquired Icelandic had been (and still is) assumed, without scientific evidence. While the MoLiCoDiLaCo project sought to provide evidence for this assumption, the results did not show extensive English input effects on Icelandic linguistic variables. The main results of the project are discussed in the article with reference to a recent shift in the public narrative of perceived threats to the future of Icelandic. The authors show that the focus has in part moved away from digital language contact between English and Icelandic, and toward the rise of English as a lingua franca in the interactions of L1 and L2 speakers of Icelandic. This shift is discussed in terms of recent research on language acquisition, multilingualism, linguistic gatekeeping, and the preservation of globally small languages.

In this article, we report on a number of violations of (or exceptions to) the so-called V2 constraint in different variants of Icelandic. The main purpose is to investigate what these violations can tell us about the nature of the V2 constraint, its vulnerability, the limits of syntax, and about children's ability to sort out what is relevant and what is not in the input they hear during the acquisition period. Three main explanatory possibilites are taken under consideration: the use and acceptance of sentences with V2 violations in Icelandic (i) is due to English influence, (ii) indicates an expansion of patterns existing in the language for language-internal reasons, (iii) is due to a task effect. In brief, our results support (i) for heritage Icelandic but not for non-heritage Icelandic, while different subsets of our data are best accounted for in terms of either (ii) or (iii).

The use of English in Iceland has increased, in part due to globalisation and technological advancements. In the research project Modeling the Linguistic Consequences of Digital Language Contact (https://molicodilaco.hi.is), a nationwide profile was constructed of the amount of Icelandic and English input that Icelandic speakers of different ages receive, their language use and skills, and their attitudes towards the two languages. In this chapter, some of the results from our 3–15-year-old participants are presented, focusing on the children’s digital usage, their input and language use, video game input and language attitudes.

Niðurstöður fyrri rannsókna benda til að fjöltyngd börn á Íslandi séu lengur að tileinka sér íslensku sem annað mál en börn í stærri málsamfélögum. Markmið rannsóknarinnar sem hér er til umfjöllunar var tvíþætt. Í fyrsta lagi að meta íslenskukunnáttu tvítyngdra leikskólabarna með ítarlegum mælingum á málfærni þeirra og bera saman við meðalfærni eintyngdra jafnaldra. Í öðru lagi að bera saman niðurstöður af mælingum með málþroskaprófum við mælingar á sjálfsprottnu tali með málsýnum. Þátttakendur voru 25 tvítyngd leikskólabörn af Suðurnesjum á aldrinum fimm til sex ára. Börnin voru öll fædd á Íslandi og höfðu dvalið í íslenskum leikskóla í nokkur ár. Báðir foreldrar höfðu sama móðurmál sem jafnframt var talað heima. Heimamál barnanna var því ekki íslenska. Helstu niðurstöður voru að tvítyngdu börnin sýndu marktækt slakari færni í íslensku í samanburði við meðalgetu eintyngdra jafnaldra á öllum athugunum sem voru gerðar. Á staðlaða málþroskaprófinu MELB voru börnin að meðaltali tveimur til þremur staðalfrávikum frá meðalgetu jafnaldra og sýndu mjög slaka færni í þáttum sem reyndu á orðaforða og málfræði. Sterkust voru þau í þáttum sem reyndu á hljóðkerfið þar sem færnin var í lágu meðaltali. Á orðaforðaprófinu Ísl-PPVT-4 var meðaltal tvítyngdu barnanna meira en fjórum staðalfrávikum frá meðaltali eintyngdra jafnaldra. Málsýni af sjálfsprottnu tali sýndu að tvítyngdu börnin tjáðu sig að meðaltali í marktækt styttri segðum (setningum), notuðu marktækt færri og ekki eins fjölbreytt orð og gerðu marktækt fleiri villur í samanburði við eintyngda jafnaldra. Fylgni var milli hlutfallslegs fjölda villna í málsýnum og niðurstaðna málþroskaprófa. Niðurstöðurnar eru mjög alvarlegar og kalla á breytt viðhorf í málörvun tvítyngdra leikskólabarna. Nauðsynlegt er að gera gangskör að eflingu íslenskukunnáttu tví- og fjöltyngdra barna á Íslandi.

This article reviews prominent research on non-English-speaking children’s extramural acquisition of English through digital media, and examines the understudied scenario of possible effects of such second language (L2) English input on domestically dominant but globally small first languages (L1s), with Icelandic as the test case. We outline the main results of the children’s part of the Modeling the Linguistic Consequences of Digital Language Contact (MoLiCoDiLaCo) research project, which targeted 724 3–12-year-old Icelandic-speaking children. The focus is on English input and its relationship to the children’s Icelandic/English vocabulary and Icelandic grammar. Although a causal relationship between digital English and reduced/incompletely acquired Icelandic is often assumed in public discourse, our results do not show large-scale effects of L2 digital English on L1 Icelandic. English still seems to be a relatively small part of Icelandic children’s language environment, and although we find some indications of contact induced/reinforced language change, i.e. in the standard use of the subjunctive, as well as reduced MLU/NDW (mean length of utterance/number of different words) in the Icelandic language samples, we do not find pervasive effects of L2 English on L1 Icelandic. On the other hand, the results show contextual L2 learning of English by Icelandic-speaking children through mostly receptive digital input. Thus, the results imply that English digital language input contributes mainly to L2 English skills without adversely affecting L1 Icelandic.

Undanfarin ár hefur umræða um áhrif aukinnar enskunotkunar hér á landi á íslenskufærni barna verið áberandi. Í þeirri umræðu hefur jafnvel verið gert ráð fyrir orsakasamhengi, þar sem aukin enska í málumhverfi barna er talin valda slakari færni þeirra í íslensku, án þess að niðurstöður rannsókna liggi þar að baki. Þessi umræða um hugsanlegar afleiðingar nánara málsambýlis íslensku við ensku, meðal annars í kjölfar snjalltækjavæðingar og fleiri tækninýjunga, voru hvatinn að rannsóknarverkefninu „Greining á málfræðilegum afleiðingum stafræns málsambýlis“ (http://molicodilaco.hi.is), sem hlaut öndvegisstyrk Rannsóknasjóðs Íslands á árunum 2016–2019. Í öndvegisverkefninu var meðal annars leitast við að setja umræðuna í fræðilegt samhengi og afla rannsóknargagna til þess að svara þeirri spurningu hvort alþjóðamálið enska, sem er ríkjandi mál í stafrænum miðlum, geti haft áhrif á máltöku móðurmáls eða fyrsta máls, þrátt fyrir að hið síðarnefnda sé aðalmál samfélagsins. Slíkt málsambýli hefur lítið sem ekkert verið rannsakað út frá sjónarhóli máltöku fyrsta máls, en í fræðilegri umfjöllun í þessari grein er gerð grein fyrir niðurstöðum rannsókna á þremur ólíkum rannsóknarsviðum sem tengjast viðfangsefninu. Þá er fjallað um öndvegisverkefnið og gerð grein fyrir niðurstöðum barnahluta þess sem tók til 3–12 ára barna. Lögð er áhersla á ílag og málnotkun barnanna og samband þessara þátta við íslenska og enska málfærni þeirra, bæði orðaforða og málfræði. Þær niðurstöður sem fjallað er um í greininni byggja á svörum 724 barna og foreldra þeirra við viðamikilli netkönnun, sem send var til 1.500 barna (svarhlutfall 48%), og ítarlegri prófunum og viðtölum við 106 börn (og foreldra þeirra) sem tóku þátt í netkönnuninni. Niðurstöður tölfræðilegrar greiningar á gögnunum úr barnahluta öndvegisverkefnisins gefa ekki til kynna víðtæk áhrif ensks ílags og málnotkunar á íslenskufærni barnanna enn sem komið er, þótt marktæk tengsl komi fram á milli ensku í málumhverfinu og nokkurra mállegra breyta, til dæmis hefðbundinnar notkunar viðtengingarháttar í íslensku og meðallengdar segða í íslensku málsýnunum. Mælingar á magni hvors tungumáls í málumhverfi barnanna og málnotkunar þeirra leiða í ljós að mikil enskunotkun er ekki endilega alltaf á kostnað íslenskunnar, auk þess sem enskumagn og enskuáhugi barnanna tengist aukinni enskufærni þeirra. Almennt má segja að niðurstöður okkar bendi til að enskt ílag og málnotkun spái enn sem komið er að litlu leyti fyrir um íslenskufærni 3–12 ára barna en spái mun betur fyrir um enskufærni þeirra.

This chapter examines Icelandic Dative Substitution (DS) and argues that two of its diachronic side effects, referred to as Person-Specific Retention and an Elsewhere Condition Death Rattle, are explained by the way in which formal features constrain the trajectory of morphosyntactic change. The study is based on naturally occurring data and a recent language acquisition experiment and the analysis draws on weighted grammatical rules as well as Yang’s Tolerance Principle. It is argued that DS does not necessarily involve a change in the syntax, but should rather be viewed as a change in the interpretation of syntactic information at the interface with morphology, resulting in variability on the surface. The results are analyzed in the context of recent theories on specialization in linguistic change and the dynamics of variation in individuals.

Icelandic is one of the smallest independent languages in Europe. According to traditional measures of language vitality such as the EGIDS (Lewis & Simons 2010) and the UNESCO vitality scale (2013), Icelandic fulfills the highest criteria for a national language and thus appears to have a strong status. This categorization is based on the fact that Icelandic is the only official language of a nation-state (apart from Icelandic sign language), has an established literary tradition, and is almost the sole language in government, public administration, workplaces, education at lower levels, and most domains of society. Iceland additionally has a strong tradition for language planning, preservation, and prescriptivism (Ottósson 1990) and the country has long been largely monolingual, although in recent years immigration has increased and the tourism industry has grown exponentially (Hagstofan 2017). However, language vitality scales typically do not consider the technological and sociological changes which in recent years have brought Icelandic into digital language contact with English. With this term, we refer to the situation in which speakers come in contact with another language in the digital domain as opposed to eye to eye contact in more traditional domains. Recipients of English digital contact in Iceland now increasingly include young children and teenagers, who are still in the acquisition stage and hence are more receptive to English exposure, e.g. through playing interactive computer games. Today, the globally dominant English is often perceived as a threat to Icelandic in public discourse (Kristinsson 2019). A growing body of recent studies shows that children in traditionally nonEnglish-speaking language communities acquire some of their English skills before the language is introduced in formal education, and continue learning extramurally through contextual learning (Lindgren & Muñoz 2012; Sylvén &

Stafræna byltingin og tækniframfarir síðustu fimm til tíu árin hafa haft í för með sér miklar samfélagsbreytingar. Með óteljandi tækifærum sem fylgja auknu aðgengi að notendavænni og öflugri tækni hafa einnig vaknað spurningar um nytsemi hennar og áhrif. Þessar spurningar hafa verið áberandi þegar kemur að máltöku, málþroska og málanámi almennt – en þegar þessi viðfangsefni eru sett í víðara samhengi fela þau einnig í sér spurningar um varðveislu og lífvænleika tungumála. Þótt tæknin geti verið kjörið tækifæri til þess að færa út umdæmi smærri málsamfélaga er iðulega bent á að í stafrænu, alþjóðlegu samhengi sé erfitt að veita framboði enskunnar samkeppni. Þegar kafað er dýpra í svo margþætt viðfangsefni verður þó fljótt ljóst að erfitt er að alhæfa. Er aukin enska í málumhverfi íslenskra barna alltaf á kostnað íslenskunnar? Er öll stafræn notkun á tungumálinu jafngild? Er til að mynda sambærilegt að streyma myndböndum án texta og spila tölvuleiki með samskiptum við aðra spilara? Getur tæknin jafnvel stuðlað að bættum málþroska barna? Eða ýtt undir tileinkun nýrra mála? Hér verður rýnt í innlendar og erlendar rannsóknir sem gefa vísbendingar um svörin við þessum spurningum, auk þess sem tæpt verður á nokkrum af frumniðurstöðum rannsóknarverkefnisins Greining á málfræðilegum afleiðingum stafræns málsambýlis (verkefnisstjórar: Sigríður Sigurjónsdóttir og Eiríkur Rögnvaldsson, 2016-2019). Verkefnið er styrkt af Rannsóknasjóði Rannís en markmið þess er meðal annars að kortleggja dreifingu og eðli ensks og íslensks ílags (máláreitis) í íslensku málsamfélagi. Í þessari umfjöllun er lögð áhersla á það sem vitað er um tengsl tækni, málþroska og málanáms og fjallað um það hvernig aukið magn ensku í umhverfi íslenskra barna og unglinga gæti haft áhrif á mál þeirra og stöðu íslenskunnar almennt. Þegar fjallað er um mál og tækni er hér helst átt við svokallað stafrænt ílag (e. digital input/electronic media exposure), en það er skilgreint sem mállegt áreiti frá ýmsum tækjum, svo sem sjónvörpum, tölvum, snjalltækjum, bílum, heimilistækjum og fleiru sambærilegu.

The finite verb typically occurs in second position in main clauses in Germanic languages other than English. Hence they are often referred to as ʽverb-second languagesʼ or V2-languages for short. The difference between a V2-language and a non-V2 language is shown in (i)–(ii) with Icelandic examples and English glosses (the finite verb is highlighted): In example (i) the finite verb occurs in second position in Icelandic, immediately following the subject María in Icelandic but in the English gloss it occurs in third position, following the adverb never . In (ii) the finite verb immediately follows the fronted (topicalized) object Maríu in Icelandic but in the English gloss the finite verb again occurs in third position, this time following the subject. This article discusses the influence of intense language contact (English/Icelandic) on the two V2-order types in North American Icelandic (NAmIce), a heritage language spoken in former Icelandic conclaves in North America. We show that the subject-first V2-order is more robust in NAmIce than the topic-first V2-order and less vulnerable to English influence, although both types are affected to some extent. This is interesting for two reasons. First, it has been argued that word order is typically less prone to cross-linguistic influence than for instance morphology. Second, these results suggest that, contrary to common assumption, the two types of V2-orders discussed here may have different syntactic sources in Icelandic syntax.

Abstract We introduce “inverse attraction” (IA) (Bianchi 1999: 93) in non-standard Icelandic, a nearly unstudied phenomenon that provides special insight into relative clause formation, specificially regarding the copy deletion that occurs in the “matching analysis” derivation of relatives. Descriptively, IA seems to involve deletion of the higher, rather than the lower copy of the relative head. In order to study this phenomenon in more detail, we conducted a judgment study with 130 Icelandic-speaking participants. The results of our study indicate that the acceptability of IA depends on the matrix and embedded cases involved, on the position of the gap inside the relative clause, and on the position of the whole relativized DP.

Abstract Extensive syntactic surveys show that variation in subject case appears in both Insular Scandinavian languages. The results furthermore indicate that intra-speaker variation is widespread and that Person-Specific Retention (PSR) is one of the characteristics of the variation in both Icelandic and Faroese. Many questions on the nature of the variation remain unanswered and no formal account of the PSR has been provided. In this paper I shed light on the subject by presenting new data from Icelandic, Faroese and heritage North-American Icelandic. Experimental language acquisition data can be particularly decisive regarding the nature of the intra-speaker variation and the variation and PSR can be accounted for formally within Yang’s (2002, 2016) variational model of acquisition.

We investigate ‘by’-phrases in the Icelandic impersonal passive and argue that they are grammatical, contra previous reports in the literature. However, it is only acceptable to use them when there are agent-specific pressures to realize the agent on the linear right, i.e., when the agent expresses new information or when it is phonologically heavy. We develop a formal analysis in the spirit of the Voiceadjunction theory of ‘by’-phrases and consider facts from historical syntax.