Douglas Biber’s research while affiliated with Northern Arizona University and other places

Two recent studies (Biber, Larsson, & Hancock 2024a,b) evaluate how well theory-based models account for the distributions of complexity features across spoken and written texts. Those studies provide strong evidence for two groupings of complexity features: phrases functioning syntactically as noun modifiers, and finite dependent clauses functioning as clause-level constituents. At the same time, those studies fail to identify systematic patterns of covariation for the other complexity features (e.g., phrases functioning as clause-level constituents). The present study picks up where those previous studies left off, exploring the possibility that complexity features pattern together in systematic ways at the register level, even though they have less strong patterns of covariation across individual texts. The results show that: 1) all 25 features can be grouped into one of two groupings, referred to as the 'oral' and 'literate' complexity dimensions; and 2) those two dimensions have a strong complementary relation to one another. These general patterns are described and interpreted relative to the particular features grouped into each dimension, the register distributions associated with each dimension, and the extent to which these register-level patterns are found at the text level.

Rank-ordered lists of word types are ubiquitous in corpus linguistics and applied linguistics. Word lists are commonly developed as aids for language teaching and learning, vocabulary testing, and language description. Yet, these lists are often produced and used without evaluation of their stability — or replicability — across corpus samples. Our primary objective in this paper is to describe the cumulative state of knowledge regarding the stability of corpus-based word type lists, focusing on three goals that motivate the creation and use of rank-ordered lists: identifying key lexical items for learning or teaching, assessing vocabulary size or knowledge, and identifying all items in a language domain. We show that word type lists are far less stable than researchers and practitioners often assume, although there is substantial variability in stability depending on the goals and methods behind list creation.

TV news broadcasts (TVNBs) represent an unusual mix of oral/literate situational characteristics. In addition, TVNBs have shifted historically to become increasingly focused on ‘soft’ human-interest stories. Thus, we have good reason to predict that present-day TVNBs will incorporate a hybrid blend of oral and literate lexico-grammatical characteristics. The present study explores this possibility through a corpus-based analysis of TVNBs compared to conversation and newspaper writing. In part, the results confirm these prior expectations. Surprisingly, though, a more detailed analysis shows that this mixture of situational characteristics has given rise to a unique grammatical style of discourse rather than a hybrid style. In particular, TVNBs can be analyzed to a large extent as main clauses where the finite verb has been deleted. This discourse style is interesting because it cannot be attributed to challenging production circumstances. Rather, the functional motivation seems related to creating a perception of urgency and excitement.

It is not always clear how a cumulative knowledge mindset can permeate the normal primary research studies that we do. That is, both secondary research and replication studies are focused on the goal of establishing what we know; specifying the state of the art for our current state of knowledge. But what about new primary studies, designed with the intent of moving the field forward? Is it possible to also base those on a foundation of cumulative knowledge? We imagine that most readers will react to this question by saying ‘yes, of course – this is what I always do!’. However, there is a crucial distinction that we hope to explore: the difference between motivating a new primary study by identifying a gap that has never been studied before, versus motivating a new primary study by first documenting what we currently know about a topic, and then using that to identify the mysteries that need to be resolved. Both approaches are based on a thorough survey of previous research. But the first focuses on what was done in previous studies, while the second focuses on what we have learned from previous studies. Our broader agenda in this special issue is to encourage the adoption of this second perspective – the cumulative knowledge approach – as normal practice for primary studies. Thus, we explore this possibility through a series of case studies. In this introductory article, we lay the groundwork for those case studies.

Larsson, T., Biber, D., Hancock, G. R. (2024). On the role of cumulative knowledge building and specific hypotheses: The case of grammatical complexity. Corpora, 19(3). Abstract As corpus linguistics matures as a field, there are an increasing number of research areas in which we have accrued sufficient knowledge that we can start building knowledge in a cumulative manner by (a) synthesizing findings and generalizations made by previous research and interpreting new findings in relations to those, and (b) formulating and testing increasingly specific predictions/hypotheses resulting from (a). The present paper outlines what a move toward cumulative knowledge building may look like for the field and offers a case study on grammatical complexity as illustration. In building knowledge in a more systematic way, we can engage more deeply with the claimed generalizable findings from previous research and help move the field's state-of-the-art forward.

Our response is focused on the analysis of grammatical complexity in learner language. We fully support the distinction between complexity and difficulty advocated by Bulté et al., and we agree with their characterization of 'complexity' (as opposed to 'difficulty') as referring 'to the structural characteristics of linguistic items/structures and texts.' We are surprised, though, by Bulté et al.'s recommendation that this construct of grammatical complexity can be adequately captured by omnibus measures that disregard and confound the different influences of multiple structural and syntactic factors. That is, even though Bulté et al. recommend using multiple omnibus measures that are intended to capture different 'subdimensions' of complexity, all of those measures are extremely general, disregarding analysis of particular grammatical structures and completely disregarding analysis of syntactic function. Thus, the methods recommended by Bulté et al. (and those generally practiced by SLA researchers) are based on the implicit assumption that grammatical complexity can be described without actually carrying out a careful grammatical/syntactic analysis. As opposed to that approach, we argue that any adequate description of grammatical complexity must be based on a principled linguistic analysis of grammatical structures and syntactic functions. To illustrate how anomalous the 'omnibus' approach is, imagine a team of biologists who want to describe and compare the complexity of forests. These researchers are aware of the incredible diversity in the composition of forests. For example, one forest is composed of deciduous and coniferous trees from many different species, at different stages of maturity, growing with different extents of density, with undergrowth representing many different plant species; another forest is composed entirely of pine trees all at a single stage of maturity with no undergrowth. However, the researchers decide to disregard what they know about the biology of trees and plants, and instead 1 It is surprising to us that researchers in SLA strongly prefer the label "syntactic" complexity' as opposed to "grammatical" complexity-even though almost all SLA studies of complexity completely disregard analysis of syntactic function, focusing instead solely on structural characteristics. We explain this distinction further in our response below.

The Lancaster-Northern Arizona Corpus of Spoken American English (LANA-CASE) is a collaborative project between Lancaster University and Northern Arizona University to create a publicly available, large-scale corpus of American English conversation. In this article, we describe the design of LANA-CASE in terms of the challenges that have arisen and how these have been addressed – including decisions related to operationalizing the domain, sampling the data, recruiting participants, and selecting instruments for data collection. In addressing these challenges, we were able to draw on and further develop strategies established in the creation of other spoken corpora (including the British English counterpart to LANA-CASE, the Spoken British National Corpus 2014) as well as to implement recent theoretical and technical innovations related to each step. We hope that this discussion can inform future projects focused on the design and construction of spoken corpora.

The present study tests the empirical adequacy of competing models of grammatical complexity in university student writing, based on analysis of disciplinary texts from L1-English and L2-English students. The results show that grammatical complexity in student writing must be treated as a multi-dimensional linguistic construct, distinguishing among both structural types and syntactic functions. We compare the results here to previous research (Biber et al., 2024a, b), showing a similar patterning of complexity features in student writing and the broader domain of general writing. Two of these groupings – dependent phrases functioning as noun modifiers, and finite dependent clauses functioning as clause-level constituents – are especially interesting. These two groupings represent the strongest co-occurrence patterns in general writing, but only the dependent clause grouping is represented in student writing. This discrepancy is interpreted relative to the development of advanced proficiency in the use of complexity features by university students.

During the past 20 years, corpus linguistic research on register variation has yielded important theoretical advances. The first part of this paper discusses these advances and the cumulative body of research that has produced them. In the second part of the paper, we focus on the goals of research on register variation. The traditional goal of the text-linguistic (TxtLx) approach to linguistic variation has been to describe registers and patterns of register variation: describing the linguistic and situational characteristics of registers. In this paper, we explore a related, but distinct, text-linguistic goal: to account for all linguistic variation among texts. Because the TxtLx framework assumes the importance of functional correspondence between linguistic characteristics and situational characteristics, it is reasonable to assume that in addition to register, we can use situational parameters coded continuously at the level of individual texts as additional predictors of text-linguistic variation. We describe the results of an empirical study to show that using both register categories and text-level situational parameters as predictors results in a more comprehensive and explanatory model of text-linguistic variation. In the conclusion we discuss the future of corpus-based register studies, focusing on unanswered questions related to theoretical claims about register.