We present the results of a study on whether writing-intensive learning techniques can assist beginner students in learning linguistic argumentation. The analysis is based on student submissions (eighty submissions from twenty students, 22,328 words) from a typical Introduction to Linguistics course, which were analyzed with the Coh-Metrix tool (McNamara et al. 2014), a suite of tests that measures cohesion of the linguistic formulation of the text and coherence of the mental representation. The essays show improvement in descriptive measures (lexical diversity, use of content words) and greater simplicity in terms of readability, suggesting a growth in the sophistication of the students' argumentation and disciplinary knowledge.

Old Icelandic relative clauses are frequently preceded by the pronoun sá, considered by most grammars to be a demonstrative. Using a large corpus of Old Icelandic prose, I show that when sá precedes relative clauses, it is often ambiguous between a cataphoric demonstrative (referring ahead to a relative clause) and relative pronoun (part of the relative clause). Syntactic and prosodic evidence indicates that, at least in some instances, sá is unambiguously a relative pronoun, used in tandem with the particle er; thus Old Icelandic relative clauses seem to have doubly filled COMP. A notable characteristic of relative sá is its pervasive attraction to the case of the matrix antecedent. I argue that case attraction represents an intermediate stage in the reanalysis of sá from a demonstrative to a true relative pronoun. Structurally, case-attracting relative pronouns and true relative pronouns occupy different functional positions within a split-CP system. Sá achieved the final stage of the development in the seventeenth century, but rapidly declined under competition with the complementizer sem, thus leaving the false impression that sá never developed beyond the case-attraction stage.

The structural focus of linguistics has led to a static and modular treatment of meaning. Viewing language as practice allows us to transcend the boundaries of subdisciplines that deal with meaning and to integrate the social indexicality of variation into this larger system. This article presents the expression of social meaning as a continuum of decreasing reference and increasing performativity, with sociolinguistic variation at the performative extreme. The meaning potential of sociolinguistic variables in turn is based in their form and their social source, constituting a cline of ‘interiority’ from variables that index public social facts about the speaker to more internal, personal affective states.

This article illustrates two-verb clusters in Viscri Saxon, a dialect of Transylvanian Saxon (TrSax) spoken in Viscri, Romania, along with Romanian and Standard German. The orders found in Viscri Saxon verb clusters are encountered in West Germanic varieties related to TrSax (e.g. Moselle Franconian, Luxembourgish), but the distributions differ from the ones discussed in other varieties (Dubenion-Smith 2010, Wurmbrand 2017). I argue that word-order variation in Viscri Saxon is the result of syntactic transfer from Standard German, and show that there is flexible distribution between possible word orders. Furthermore, speakers with different linguistic profiles use the available constructions to different degrees, thus illustrating the roles of German and Romanian in the progression of contact-induced changes in Viscri Saxon.

The birthdate of both generative linguistics and neural networks can be taken as 1957, the year of the publication of foundational work by both Noam Chomsky and Frank Rosenblatt. This article traces the development of these two approaches to cognitive science, from their largely autonomous early development in the first thirty years, through their collision in the 1980s around the past-tense debate (Rumelhart & McClelland 1986, Pinker & Prince 1988) and their integration in much subsequent work up to the present. Although this integration has produced a considerable body of results, the continued general gulf between these two lines of research is likely impeding progress in both: on learning in generative linguistics, and on the representation of language in neural modeling. The article concludes with a brief argument that generative linguistics is unlikely to fulfill its promise of accounting for language learning if it continues to maintain its distance from neural and statistical approaches to learning.

Court reporters are certified at either 95% or 98% accuracy, depending on their certifying organization; however, the measure of accuracy is not one that evaluates their ability to transcribe nonstandard dialects. Here, we demonstrate that Philadelphia court reporters consistently fail to meet this level of transcription accuracy when confronted with mundane examples of spoken African American English (AAE). Furthermore, we show that they often cannot demonstrate understanding of what is being said. We show that the different morphosyntax of AAE, the different phonological patterns of AAE, and the different accents in Philadelphia related to residential segregation all conspire to produce transcriptions that not only are inaccurate, but also change the official record of who performed what actions under which circumstances, with potentially dramatic legal repercussions for everyday speakers of AAE.

Language-contact studies have shown that the transfer of morphology from one language to another is relatively rare (Gardani 2008, Grant 2012, Matras 2015), and the copying of verbal inflectional morphology is particularly infrequent (Seifart 2017). Copied morphemes are frequently assumed to enter the recipient language via ‘indirect affix borrowing’, whereby complex lexemes are copied and subsequently analyzed into their component parts in the recipient language, thus enabling use of the copied affixes with native roots (Grant 2012, Seifart 2015, Evans 2016). Although ‘direct affix borrowing’, in which speakers of the recipient language identify the meaning of affixes in the model language and transfer them directly for use with native roots, is known to occur, it has until now been identified only for derivational morphemes (Seifart 2015). I here provide evidence that inflectional morphemes, namely four Sakha (Yakut) tense-aspect-mood markers plus associated subject agreement paradigms, were copied directly into the Lamunkhin dialect of Even by fully bilingual speakers. This argument is based on the absence of Sakha verbal roots found with these paradigms in a corpus of Lamunkhin Even recordings, as well as on patterns of cooccurrence of these morphemes in clauses with Even grammatical morphology.

Dissatisfied with traditional grading, we developed a grading system to directly assess whether students have mastered course material. We identified the set of skills students need to master in a course and provided multiple opportunities for students to demonstrate mastery of each skill. We describe in detail how we implemented the system for two undergraduate courses, Introductory Phonetics and Phonology I. Our goals were to decrease student stress, increase student learning and make students' study efforts more effective, increase students' metacognitive awareness, promote a growth mindset, encourage students to aim for mastery rather than partial credit, be fairer to students facing structural and institutional disadvantages, reduce our time spent on grading, and facilitate complying with new accreditation requirements. Our own reflections and student feedback indicate that many of these goals were met.

The rajaz meter of Hausa is based on syllable quantity. In its dimeter form, it deploys lines consisting of two metra, each usually containing six moras. A variety of metra occur, and the key analytic challenge is to single out the legal metra from the set of logically possible ones. We propose an analysis, framed in maximum entropy optimality theory, that does this and also accounts for the statistical distribution of metron types, varying from poem to poem, within the line and stanza. We demonstrate a law of comparative frequency for rajaz and show how it emerges naturally in the maxent framework when competing candidates are in a relationship of harmonic bounding.

Turning to how the verse is sung, we observe that rajaz verse rhythm is typically remapped onto a distinct sung rhythm. We consider grammatical architectures that can characterize this remapping. Lastly, we develop a maxent phonetic grammar to predict the durations of the sung syllables. Our constraints simultaneously invoke all levels of structure: the syllables and moras of the phonology, the grids used for poetic scansion, and the grids used for sung rhythm.

In May 2016, the Executive Committee of the Linguistic Society of America approved formation of an ad hoc AP Linguistics Committee (APLC) to study the creation of an Advanced Placement Linguistics course and examination for US high schools. In January 2017, the APLC convened and voted to proceed with the drafting of a formal AP Linguistics proposal to the College Board and to take whatever preparatory steps were required in that process. In this paper we sketch the AP Linguistics initiative, describing the potential benefits of linguistics for American high schools and their students, the attractions of high school linguistics for the field of Linguistics itself, the motivations for an AP Linguistics course in this context, the formal requirements of an AP Linguistics proposal to the College Board, and the steps being taken to meet those requirements.

We report a statistical test of a long-standing hypothesis in the literature: that phonological neutralization rules are more common at the ends of lexical domains than the beginnings (Houlihan 1975 et seq.). We collected descriptive grammars for an areally and genetically diverse set of fifty languages, identified all active phonological rules that target the edge of a lexical domain (root, stem, word, phrase, or utterance), and further coded each rule for whether it was phonemically neutralizing, that is, able to create surface homophony. We find that such neutralizing rules are strongly, significantly less common at the beginnings of lexical domains relative to ends, and that this pattern is strikingly consistent across all languages within the data set. We show that this pattern is not an artifact of a tendency for syllable codas to be a target for phonological neutralization, nor is it associated with a suffixing or prefixing preference. Consistent with previous accounts, we argue that this pattern may ultimately be based in the greater average information content of phonological categories early in the word, which itself is a consequence of incremental processing in lexical access.

Pater's (2019) expansive review is a significant contribution toward bridging the disconnect of generative linguistics with connectionism, and as such, it is an important service to the field. But Pater's efforts for inclusion and reconciliation obscure crucial substantive disagreements on foundational matters. Most connectionist models are antithetical to the algebraic hypothesis that has guided generative linguistics from its inception. They eschew the notions that mental representations have formal constituent structure and that mental operations are structure-sensitive. These representational commitments critically limit the scope of learning and productivity in connectionist models. Moving forward, we see only two options: either those connectionist models are right, and generative linguistics must be radically revised, or they must be replaced by alternatives that are compatible with the algebraic hypothesis. There can be no integration without structured representations.

This article argues that linguistics can be effectively incorporated into project-based learning (PBL) as an innovative approach to introducing young students to linguistics. It draws from eighteen months of a collaboration between a linguist and a middle-school humanities teacher, and builds on prior efforts by linguists to incorporate linguistics into the K–12 curriculum. The article reviews relevant literature, introduces PBL and linguistics in PBL as a model, and presents detailed descriptions of how the study of language and the linguist's ways of knowing were incorporated into two student projects. Evidence of student learning and the impact of linguistics is also presented. The article concludes by suggesting that this approach to incorporating linguistics into K–12 education can address the goals both of linguists—changing attitudes toward and understanding of language—and of educators—improving academic performance and addressing content standards.

Old English underwent diachronic change in its vowel inventory between its predecessor West Germanic and Middle English. We provide an analysis of the addition and loss of vowels in Old English from the perspective of modified contrastive specification (Dresher et al. 1994). Three main themes emerge from our analysis: (i) the phonological representation of contrast in the vowels in English has remained remarkably stable for over a thousand years, (ii) the proposed analysis improves upon and supersedes similar analyses proposed in Dresher 2015 and Purnell & Raimy 2015, and (iii) the adoption of privative features provides an improved representationally based understanding of phonological activity, feature geometry, and how phonology reflects general cognitive features of memory.

As is common in work on prosodic typology, the notions ‘tone’ and ‘stress’ play a key role in Wee's 2016 study of the tone system of Hong Kong English (HKE). Based on the absence of phonetic correlates of stress and the distribution of tone in polysyllabic words, Wee claims that HKE must have lexical (high) tone. In this reply, we argue that, even in the absence of phonetic correlates of stress, foot structure provides a more parsimonious account of the distribution of surface tones. Multiple high tones within words follow from predictable morphological structure and/or tonal spreading, rather than from lexical tone.

The target article (Pater 2019) proposes to use neural networks to model learning within existing grammatical frameworks. This is easier said than done. There is a fundamental gap to be bridged that does not receive attention in the article: how can we use neural networks to examine whether it is possible to learn some linguistic representation (a tree, for example) when, after learning is finished, we cannot even tell if this is the type of representation that has been learned (all we see is a sequence of numbers)? Drawing a correspondence between an abstract linguistic representational system and an opaque parameter vector that can (or perhaps cannot) be seen as an instance of such a representation is an implementational mapping problem. Rather than relying on existing frameworks that propose partial solutions to this problem, such as harmonic grammar, I suggest that fusional research of the kind proposed needs to directly address how to ‘find’ linguistic representations in neural network representations.

There are often practical barriers to doing fieldwork in a novel, remote location. I propose a model for linguistic research designed to overcome such barriers: a linguistic field station. It is a centralized facility that coordinates scientific research by providing (i) research infrastructure, (ii) access to specific social, biological, or ecological systems that are not immediately available otherwise, (iii) training for students at the graduate and undergraduate levels, and (iv) access to local communities with the goal of obtaining data from them as well as training local specialists. Field stations are particularly important for research on and documentation of Indigenous languages, including contexts where colonial languages are supplanting Indigenous ones. Although the field station model is not new in research outside of language sciences, it has not yet been utilized widely in language research. I describe how the proposed model has been implemented in Guatemala and compare the field station there with other linguistic field stations.

Joe Pater's (2019) target article calls for greater interaction between neural network research and linguistics. I expand on this call and show how such interaction can benefit both fields. Linguists can contribute to research on neural networks for language technologies by clearly delineating the linguistic capabilities that can be expected of such systems, and by constructing controlled experimental paradigms that can determine whether those desiderata have been met. In the other direction, neural networks can benefit the scientific study of language by providing infrastructure for modeling human sentence processing and for evaluating the necessity of particular innate constraints on language acquisition.