Publications

Abstract

In this work we perform multimodal detection and classification
of head movements from face to face video conversation data.
We have experimented with different models and feature sets
and provided some insight on the effect of independent features,
but also how their interaction can enhance a head movement
classifier. Used features include nose, neck and mid hip position
coordinates and their derivatives together with acoustic features,
namely, intensity and pitch of the speaker on focus. Results
show that when input features are sufficiently processed by in-
teracting with each other, a linear classifier can reach a similar
performance to a more complex non-linear neural model with
several hidden layers. Our best models achieve state-of-the-art
performance in the detection task, measured by macro-averaged
F1 score.

Abstract

How do languages express ideas of movement, and how do they package different components of moving, such as manner and path? This task supports detailed investigation of motion descriptions. The specific study goals are: (a) the coding of “via” grounds (i.e., ground objects which the figure moves along, over, around, through, past, etc.); (b) the coding of direction changes; (c) the spontaneous segmentation of complex motion scenarios; and (d) the gestural representation of motion paths. The stimulus set is 5 simple 3D animations (7-17 seconds long) that show a ball rolling through a landscape. The task is a director-matcher task for two participants. The director describes the path of the ball in each clip to the matcher, who is asked to trace the path with a pen in a 2D picture.

Abstract

Judgments we make about how similar or different events are to each other can reveal the features we find useful in classifying the world. This task is designed to investigate how speakers of different languages classify events, and to examine how linguistic and gestural encoding relates to non-linguistic classification. Specifically, the task investigates whether speakers judge two events to be similar on the basis of (a) the path versus manner of motion, (b) sub-events versus larger complex events, (c) participant identity versus event identity, and (d) different participant roles. In the task, participants are asked to make similarity judgments concerning sets of 2D animation clips.

Abstract

How do different languages treat a particular semantic domain? It has already been established that languages have widely varied words for talking about “cutting” and “breaking” things: for example, English has a very general verb break, but K’iche’ Maya has many different ‘break’ verbs that are used for different kinds of objects (e.g., brittle, flexible, long). The aim of this task is to map out cross-linguistic lexicalisation patterns in the cutting/breaking domain. The stimuli comprise 61 short video clips that show one or two actors breaking various objects (sticks, carrots, pieces of cloth or string, etc.) using various instruments (a knife, a hammer, an axe, their hands, etc.), or situations in which various kinds of objects break spontaneously. The clips are used to elicit descriptions of actors’ actions and the state changes that the objects undergo.

Abstract

Place-names (toponyms) are at the intersection of spatial language, culture, and cognition. This questionnaire prepares the researcher to answer three overarching questions: how to formally identify place-names in the research language (i.e. according to morphological and syntactic criteria); what places place-names are employed to refer to (e.g. human settlements, landscape sites); and how places are semantically construed for this purpose. The questionnaire can in principle be answered using an existing database. However, additional elicitation with language consultants is recommended.

Abstract

The spatial vocabulary of the Mayan language Tzeltal is dominated by an Absolute system of spatial reckoning, whereby an "uphill/downhill" coordinate abstracted from the lay of the land is used to reckon spatial relationships on the horizontal in both small-scale and long distance space. This system is used in lieu of a Front/Back/Left/Right system which does not exist in this language. The spatial vocabulary dedicated to this system (which I refer to in general as the UP/DOWN vocabulary) includes intransitive motion verbs (roughly translatable as "ascend"/"descend"), their transitivized counterparts ("make it ascend/descend"), directional adverbs ("uphillwards"/"downhillwards"), and possessed relational nouns ("uphill/downhill in relation to it"). This same vocabulary applies to spatial relations on the vertical axis. Two seemingly contradictory observations about children's early meanings for the spatial verbs dedicated to this system motivate the proposal put forward in this paper. On the one hand, Tzeltal children's UP/DOWN vocabulary shows very early sensitivity to the semantic structure of the language they are learning: the meanings for these verbs are from the first usages attached to the slope of the land, and to particular places; there is no evidence of an initial preference for the vertical meaning. On the other hand, children's meanings remain for a long time too specific, and errors of interpretation/production (using the verbs to mean 'local slope of land' rather than 'overall N/S slope of land direction) are evident in verbal productions of some children as late as age 7 or 8. The proposal is made that the highly specific nature of Tzeltal verbs at the basic level influences the children's hypotheses about what kinds of meanings verbs can have.

Abstract

This is a reprint of the Brown 1999 article.

Abstract

This is an encyclopedia entry surveying research and theoretical approaches to politeness phenomena in language usage.

Abstract

This chapter provides an introduction to the handbook. It succintly overviews the key questions in the field of L3/Ln acquisition and summarizes the scope of all the chapters included. The chapter ends by raising some outstanding questions that the field needs to address.

Abstract

In order to understand early word learning we need to better understand and quantify properties of the input that young children receive. We extended the human simulation paradigm (HSP) using egocentric videos taken from infant head-mounted cameras. The videos were further annotated with gaze information indicating in-the-moment visual attention from the infant. Our new HSP prompted participants for two types of responses, thus differentiating referential from semantic ambiguity in the learning input. Consistent with findings on visual attention in word learning, we find a strongly bimodal distribution over HSP accuracy. Even in this open-ended task, most videos only lead to a small handful of common responses. What's more, referential ambiguity was the key bottleneck to performance: participants can nearly always recover the exact word that was said if they identify the correct referent. Finally, analysis shows that adult learners relied on particular, multimodal behavioral cues to infer those target referents.

Abstract

Two groups of listeners, with Dutch and British English as their native language judged stimuli in Dutch and British English, respectively, on the scales CONFIDENT vs. NOT CONFIDENT and FRIENDLY vs. NOT FRIENDLY, two meanings derived from Ohala's universal Frequency Code. The stimuli, which were lexically equivalent, were varied in pitch contour and pitch range. In both languages, the perceived degree of confidence decreases and that of friendliness increases when the pitch range is raised, as predicted by the Frequency Code. However, at identical pitch ranges, British English is perceived as more confident and more friendly than Dutch. We argue that this difference in degree of the use of the Frequency Code is due to the difference in the standard pitch ranges of Dutch and British English.

Abstract

In this study, we analyze one French signing and one French speaking family’s interaction during dinner. The families composed of two parents and two children aged 3 to 11 were filmed with three cameras to capture all family members’ behaviors. The three videos per dinner were synchronized and coded on ELAN. We annotated all participants’ acting, and languaging.
Our quantitative analyses show how family members collaboratively manage multiple streams of activity through the embodied performances of dining and interacting. We uncover different profiles according to participants’ modality of expression and status (focusing on the mother and the younger child). The hearing participants’ co-activity management illustrates their monitoring of dining and conversing and how they progressively master the affordances of the visual and vocal channels to maintain the simultaneity of the two activities. The deaf mother skillfully manages to alternate smoothly between dining and interacting. The deaf younger child manifests how she is in the process of developing her skills to manage multi-activity. Our qualitative analyses focus on the ecology of visual-gestural and audio-vocal languaging in the context of co-activity according to language and participant. We open new perspectives on the management of gaze and body parts in multimodal languaging.

Abstract

During dialog, interlocutors take turns at speaking with little gap or overlap between their contributions. But language production in monolog is comparatively slow. Theories of dialog tend to agree that interlocutors manage these timing demands by planning a response early, before the current speaker reaches the end of their turn. In the first half of this chapter, I review experimental research supporting these theories. But this research also suggests that planning a response early, while simultaneously comprehending, is difficult. Does response planning need to be this difficult during dialog? In other words, is early-planning always necessary? In the second half of this chapter, I discuss research that suggests the answer to this question is no. In particular, corpora of natural conversation demonstrate that speakers do not directly respond to the immediately preceding utterance of their partner—instead, they continue an utterance they produced earlier. This parallel talk likely occurs because speakers are highly incremental and plan only part of their utterance before speaking, leading to pauses, hesitations, and disfluencies. As a result, speakers do not need to engage in extensive advance planning. Thus, laboratory studies do not provide a full picture of language production in dialog, and further research using naturalistic tasks is needed.

Abstract

Most psycholinguistic studies on morphological processing have examined the role of morphological structure in the visual modality. This chapter discusses morphological processing in the auditory modality, which is an area of research that has only recently received more attention. It first discusses why results in the visual modality cannot straightforwardly be applied to the processing of spoken words, stressing the importance of acknowledging potential modality effects. It then gives a brief overview of the existing research on the role of morphology in the auditory modality, for which an increasing number of studies report that listeners show sensitivity to morphological structure. Finally, the chapter highlights insights gained by looking at morphological processing not only in reading, but also in listening, and it discusses directions for future research

Abstract

Bisyllabic sequences which could be interpreted as one word or two were produced in sentence contexts by a trained speaker, and syllabic durations measured. Listeners judged whether the bisyllables, excised from context, were one word or two. The proportion of two-word choices correlated positively with measured duration, but only for bisyllables stressed on the second syllable. The results may suggest a limit for listener sensitivity to syllabic lengthening as a word boundary cue.

Abstract

Phoneme detection is a psycholinguistic task in which listeners' response time to detect the presence of a pre-specified phoneme target is measured. Typically, detection tasks have used consonant targets. This paper reports two experiments in which subjects responded to vowels as phoneme detection targets. In the first experiment, targets occurred in real words, in the second in nonsense words. Response times were long by comparison with consonantal targets. Targets in initial syllables were responded to much more slowly than targets in second syllables. Strong vowels were responded to faster than reduced vowels in real words but not in nonwords. These results suggest that the process of phoneme detection produces different results for vowels and for consonants. We discuss possible explanations for this difference, in particular the possibility of language-specificity.

Abstract

Many of the world’s languages feature an open lexical class of ideophones, words whose marked forms and sensory meanings invite iconic associations. Ideophones (also known as mimetics or expressives) are well-known from languages in Asia, Africa and the Americas, where they often form a class on the same order of magnitude as other major word classes and take up a considerable functional load as modifying expressions or predicates. Across languages, commonalities in the morphosyntactic behaviour of ideophones can be related to their nature and origin as vocal depictions. At the same time there is ample room for linguistic diversity, raising the need for fine-grained grammatical description of ideophone systems. As vocal depictions, ideophones often form a distinct lexical stratum seemingly conjured out of thin air; but as conventionalized words, they inevitably grow roots in local linguistic systems, showing relations to adverbs, adjectives, verbs and other linguistic resources devoted to modification and predication

Abstract

No class of words has better claims to universality than interjections. At the same time, no category has more variable content than this one, traditionally the catch-all basket for linguistic items that bear a complicated relation to sentential syntax. Interjections are a mirror reflecting methodological and theoretical assumptions more than a coherent linguistic category that affords unitary treatment. This chapter focuses on linguistic items that typically function as free-standing utterances, and on some of the conceptual, methodological, and theoretical questions generated by such items. A key move is to study these items in the setting of conversational sequences, rather than from the “flatland” of sequential syntax. This makes visible how some of the most frequent interjections streamline everyday language use and scaffold complex language. Approaching interjections in terms of their sequential positions and interactional functions has the potential to reveal and explain patterns of universality and diversity in interjections.

Abstract

How do neural oscillations support human audiovisual language and communication? Considering the rhythmic nature of audiovisual language, in which stimuli from different sensory modalities unfold over time, neural oscillations represent an ideal candidate to investigate how audiovisual language is processed in the brain. Modulations of oscillatory phase and power are thought to support audiovisual language and communication in multiple ways. Neural oscillations synchronize by tracking external rhythmic stimuli or by re-setting their phase to presentation of relevant stimuli, resulting in perceptual benefits. In particular, synchronized neural oscillations have been shown to subserve the processing and the integration of auditory speech, visual speech, and hand gestures. Furthermore, synchronized oscillatory modulations have been studied and reported between brains during social interaction, suggesting that their contribution to audiovisual communication goes beyond the processing of single stimuli and applies to natural, face-to-face communication.

There are still some outstanding questions that need to be answered to reach a better understanding of the neural processes supporting audiovisual language and communication. In particular, it is not entirely clear yet how the multitude of signals encountered during audiovisual communication are combined into a coherent percept and how this is affected during real-world dyadic interactions. In order to address these outstanding questions, it is fundamental to consider language as a multimodal phenomenon, involving the processing of multiple stimuli unfolding at different rhythms over time, and to study language in its natural context: social interaction. Other outstanding questions could be addressed by implementing novel techniques (such as rapid invisible frequency tagging, dual-electroencephalography, or multi-brain stimulation) and analysis methods (e.g., using temporal response functions) to better understand the relationship between oscillatory dynamics and efficient audiovisual communication.

Abstract

Acquiring a second language (L2) requires newly learned information to be integrated with existing knowledge. It has been proposed that several memory systems work together to enable this process of rapidly encoding new information and then slowly incorporating it with existing knowledge, such that it is consolidated and integrated into the language network without catastrophic interference. This chapter focuses on consolidation of L2 vocabulary. First, the complementary learning systems model is outlined, along with the model’s predictions regarding lexical consolidation. Next, word learning studies in first language (L1) that investigate the factors playing a role in consolidation, and the neural mechanisms underlying this, are reviewed. Using the L1 memory consolidation literature as background, the chapter then presents what is currently known about memory consolidation in L2 word learning. Finally, considering what is already known about L1 but not about L2, future research investigating memory consolidation in L2 neurocognition is proposed.

Abstract

“Recognitional” words and constructions enshrine our systematic reliance on shared knowledge in dedicated morphological forms and usage patterns. For example, English has a large range of terms for use when a speaker cannot locate the word or name for something or someone (e.g., whatsit, what’s-his-name), but thinks that the interlocutor knows, or can easily work out, what the speaker is talking about. This task aims to identify and investigate these kinds of expressions in the research language, including their grammaticalised status, meaning, distribution, and productivity. The task consists of a questionnaire with examples of relevant hypothetical scenarios that can be used in eliciting the relevant terms. The researcher is then encouraged to pursue further questions in regard to these items.

Abstract

This task investigates the extensional meaning of body part terms, in particular the terms for the upper and lower limbs. Two questions are addressed, namely (i) are the boundaries of these body parts universal, guided by proposed universals of object recognition? (ii) How can we compare the extensional meanings of body part terms within and across different systems of nomenclature? Consultants receive booklets with line drawings of a body and are asked to colour in specific parts of the body.

Abstract

Demonstratives are typically described as encoding degrees of physical distance between the object referred to, and the speaker or addressee. For example, this in English is used to talk about things that are physically near the speaker, and that for things that are not. But is this how speakers really choose between these words in actual talk? This task aims to generate spontaneous language data concerning deixis, gesture, and demonstratives, and to investigate the significance of different factors (e.g., physical distance, attention) in demonstrative selection. In the presence of one consultant (the “memoriser”), sixteen colour chips are hidden under objects in a specified array. Another consultant enters the area and asks the memoriser to recount the locations of the chips. The task is designed to create a situation where the speaker genuinely attempts to manipulate the addressee’s attention on objects in the immediate physical space.

Abstract

The study presented here focuses on two pragmatic gestures:
the hand flip (Ferré, 2011), a gesture of the Palm Up Open
Hand/PUOH family (Müller, 2004) and the closed hand which
can be considered as the opposite kind of movement to the open-
ing of the hands present in the PUOH gesture. Whereas one of
the functions of the hand flip has been described as presenting
a new point in speech (Cienki, 2021), the closed hand gesture
has not yet been described in the literature to the best of our
knowledge. It can however be conceived of as having the oppo-
site function of announcing the end of a point in discourse. The
object of the present study is therefore to determine, with the
study of prosodic features, if the two gestures are found in the
same type of speech units and what their respective scope is.
Drawing from a corpus of three TED Talks in French the
prosodic characteristics of the speech that accompanies the two
gestures will be examined. The hypothesis developed in the
present paper is that their scope should be reflected in the
prosody of accompanying speech, especially pitch key, tone,
and relative pitch range. The prediction is that hand flips and
closing hand gestures are expected to be located at the periph-
ery of Intonation Phrases (IPs), Inter-Pausal Units (IPUs) or
more conversational Turn Constructional Units (TCUs), and are
likely to be co-occurrent with pauses in speech. But because of
the natural slope of intonation in speech, the speech that accom-
pany early gestures in Intonation Phrases should reveal different
features from the speech at the end of intonational units. Tones
should be different as well, considering the prosodic structure
of spoken French.

Abstract

Despite the early emergence of pointing, children are generally not documented to produce iconic gestures until later in development. Although research has described this developmental trajectory and the types of iconic gestures that emerge first, there has been limited focus on iconic gestures within interactional contexts. This study identified the first 10 iconic gestures produced by five monolingual English-speaking children in a naturalistic longitudinal video corpus and analysed the interactional contexts. We found children produced their first iconic gesture between 12 and 20 months and that gestural types varied. Although 34% of gestures could have been imitated or derived from adult or child actions in the preceding context, the majority were produced independently of any observed model. In these cases, adults often led the interaction in a direction where iconic gesture was an appropriate response. Overall, we find infants can represent a referent symbolically and possess a greater capacity for innovation than previously assumed. In order to develop our understanding of how children learn to produce iconic gestures, it is important to consider the immediate interactional context. Conducting naturalistic corpus analyses could be a more ecologically valid approach to understanding how children learn to produce iconic gestures in real life contexts.

Abstract

What kinds of resources to languages have for describing location and position? For some languages, verbs have an important role to play in describing different kinds of situations (e.g., whether a bottle is standing or lying on the table). This task is designed to examine the use of positional verbs in locative constructions, with respect to the presence or absence of a human “positioner”. Participants are asked to describe video clips showing locative states that occur spontaneously, or because of active interference from a person. The task follows on from two earlier tools for the elicitation of static locative descriptions (BowPed and the Ameka picture book task). A number of additional variables (e.g. canonical v. non-canonical orientation of the figure) are also targeted in the stimuli set.

Abstract

This paper compares the focus marking systems and the focus size that can be expressed by the different focus markings in four Mande and three Atlantic languages and varieties, namely: Bambara, Dyula, Kakabe, Soninke (Mande), Wolof, Jóola Foñy and Jóola Karon (Atlantic). All of these languages are known to mark focus morphosyntactically, rather than prosodically, as the more well-studied Germanic languages do. However, the Mande languages under discussion use only morphology, in the form of a particle that follows the focus, while the Atlantic ones use a more complex morphosyntactic system in which focus is marked by morphology in the verbal complex and movement of the focused term. It is shown that while there are some syntactic restrictions to how many different focus sizes can be marked in a distinct way, there is also a certain degree of arbitrariness as to which focus sizes are marked in the same way as each other.

Abstract

Sign language is a rich form of communication, uniquely conveying meaning through a combination of gestures, facial expressions, and body movements. Existing research in sign language generation has predominantly focused on text-to-sign pose generation, while speech-to-sign pose generation remains relatively underexplored. Speech-to-sign language generation models can facilitate effective communication between the deaf and hearing communities. In this paper, we propose an architecture that utilises prosodic information from speech audio and semantic context from text to generate sign pose sequences. In our approach, we adopt a multi-tasking strategy that involves an additional task of predicting Facial Action Units (FAUs). FAUs capture the intricate facial muscle movements that play a crucial role in conveying specific facial expressions during sign language generation. We train our models on an existing Indian Sign language dataset that contains sign language videos with audio and text translations. To evaluate our models, we report Dynamic Time Warping (DTW) and Probability of Correct Keypoints (PCK) scores. We find that combining prosody and text as input, along with incorporating facial action unit prediction as an additional task, outperforms previous models in both DTW and PCK scores. We also discuss the challenges and limitations of speech-to-sign pose generation models to encourage future research in this domain. We release our models, results and code to foster reproducibility and encourage future research1.

Abstract

Gesture is an integral part of face-to-face communication, and provides a rich area for cross-cultural comparison. “Locally-anchored spatial gestures” are gestures that are roughly oriented to the actual geographical direction of referents. For example, such gestures may point to a location or a thing, trace the shape of a path, or indicate the direction of a particular area. The goal of this task is to elicit locally-anchored spatial gestures across different cultures. The task follows an interview format, where one participant prompts another to talk in detail about a specific area that the main speaker knows well. The data can be used for additional purposes such as the investigation of demonstratives.

Abstract

Interpreting the meaning of co-speech gesture often involves
identifying a gesture’s ‘lexical affiliate’, the word or phrase to
which it most closely relates (Schegloff 1984). Though there is
work within gesture studies that resists this simplex mapping of
meaning from speech to gesture (e.g. de Ruiter 2000; Kendon
2014; Parrill 2008), including an evolving body of literature on
recurrent gesture and gesture families (e.g. Fricke et al. 2014; Müller 2017), it is still the lexical affiliate model that is most ap-
parent in formal linguistic models of multimodal meaning(e.g.
Alahverdzhieva et al. 2017; Lascarides and Stone 2009; Puste-
jovsky and Krishnaswamy 2021; Schlenker 2020). In this work,
I argue that the lexical affiliate should be carefully reconsidered
in the further development of such models.
In place of the lexical affiliate, I suggest a further shift
toward a frame-based, action schematic approach to gestural
meaning in line with that proposed in, for example, Parrill and
Sweetser (2004) and Müller (2017). To demonstrate the utility
of this approach I present three types of compositional gesture
sequences which I call spatial contrast, spatial embedding, and
cooperative abstract deixis. All three rely on gestural context,
rather than gesture-speech alignment, to convey interactive (i.e.
pragmatic) meaning. The centrality of gestural context to ges-
ture meaning in these examples demonstrates the necessity of
developing a model of gestural meaning independent of its in-
tegration with speech.

Abstract

How do languages express ideas of movement, and how do they package different components of this domain, such as manner and path of motion? This task uses one large set of stimuli to gain knowledge of certain key aspects of motion verb meanings in the target language, and expands the investigation beyond simple verbs (e.g., go) to include the semantics of motion predications complete with adjuncts (e.g., go across something). Consultants are asked to view and briefly describe 96 animations of a few seconds each. The task is designed to get linguistic elicitations of motion predications under contrastive comparison with other animations in the same set. Unlike earlier tasks, the stimuli focus on inanimate moving items or “figures” (in this case, a ball).

Abstract

Demonstratives (e.g., words such as this and that in English) pivot on relationships between the item being talked about, and features of the speech act situation (e.g., where the speaker and addressee are standing or looking). However, they are only rarely investigated multi-modally, in natural language contexts. This task is designed to build a video corpus of cross-linguistically comparable discourse data for the study of “deixis in action”, while simultaneously supporting the investigation of joint attention as a factor in speaker selection of demonstratives. In the task, two or more speakers are asked to discuss and evaluate a group of similar items (e.g., examples of local handicrafts, tools, produce) that are placed within a relatively defined space (e.g., on a table). The task can additionally provide material for comparison of pointing gesture practices.

Abstract

This entry contains: 1. An invitation to think about to what extent the grammar of space and time share lexical and morphosyntactic resources − the suggestions here are only prompts, since it would take a long questionnaire to fully explore this; 2. A suggestion about how to collect gestural data that might show us to what extent the spatial and temporal domains, have a psychological continuity. This is really the goal − but you need to do the linguistic work first or in addition. The goal of this task is to explore the extent to which time is conceptualised on a spatial basis.

Abstract

The ECHO proposal aims to bring to life the cultural heritage of Europe, through internet technology that encourages collaboration across the Humanities disciplines which interpret it – at the same time making all this scholarship accessible to the citizens of Europe. An essential part of the cultural heritage of Europe is the diverse set of languages used on the continent, in their historical, literary and spoken forms. Amongst these are the ‘hidden languages’ used by minorities but of wide interest to the general public. We take the 18 Sign Languages of the EEC – the natural languages of the deaf - as an example. Little comparative information about these is available, despite their special scientific importance, the widespread public interest and the policy implications. We propose a research project on these languages based on placing fully annotated digitized moving images of each of these languages on the internet. This requires significant development of multi-media technology which would allow distributed annotation of a central corpus, together with the development of special search techniques. The technology would have widespread application to all cultural performances recorded as sound plus moving images. Such a project captures in microcosm the essence of the ECHO proposal: cultural heritage is nothing without the humanities research which contextualizes and gives it comparative assessment; by marrying information technology to humanities research, we can bring these materials to a wider public while simultaneously boosting Europe as a research area.

Abstract

As for 'Locally-anchored spatial gestures task, version 2', a major goal of this task is to elicit locally-anchored spatial gestures across different cultures. “Locally-anchored spatial gestures” are gestures that are roughly oriented to the actual geographical direction of referents. Rather than set up an interview situation, this task involves recording informal, animated narrative delivered to a native-speaker interlocutor. Locally-anchored gestures produced in such narrative are roughly comparable to those collected in the interview task. The data collected can also be used to investigate a wide range of other topics.

Abstract

Wearing the hat of a cognitive anthropologist rather than an historian, I will try to amplify the ideas of Renn’s cited above. I argue that a particular subclass of material objects, namely “cognitive artifacts,” involves a close coupling of mind and artifact that acts like a brain prosthesis. Simple cognitive artifacts are external objects that act as aids to internal
computation, and not all cultures have extended inventories of these. Cognitive artifacts in this sense (e.g., calculating or measuring devices) have clearly played a central role in the history of science. But the notion can be widened to take in less material externalizations of cognition, like writing and language itself. A critical question here is how and why this close coupling of internal computation and external device actually works, a rather neglected question to which I’ll suggest some answers.

Abstract

Large language models that exhibit instruction-following behaviour represent one of the biggest recent upheavals in conversational interfaces, a trend in large part fuelled by the release of OpenAI's ChatGPT, a proprietary large language model for text generation fine-tuned through reinforcement learning from human feedback (LLM+RLHF). We review the risks of relying on proprietary software and survey the first crop of open-source projects of comparable architecture and functionality. The main contribution of this paper is to show that openness is differentiated, and to offer scientific documentation of degrees of openness in this fast-moving field. We evaluate projects in terms of openness of code, training data, model weights, RLHF data, licensing, scientific documentation, and access methods. We find that while there is a fast-growing list of projects billing themselves as 'open source', many inherit undocumented data of dubious legality, few share the all-important instruction-tuning (a key site where human labour is involved), and careful scientific documentation is exceedingly rare. Degrees of openness are relevant to fairness and accountability at all points, from data collection and curation to model architecture, and from training and fine-tuning to release and deployment.

Abstract

Speech recognition systems are a key intermediary in voice-driven human-computer interaction. Although speech recognition works well for pristine monologic audio, real-life use cases in open-ended interactive settings still present many challenges. We argue that timing is mission-critical for dialogue systems, and evaluate 5 major commercial ASR systems for their conversational and multilingual support. We find that word error rates for natural conversational data in 6 languages remain abysmal, and that overlap remains a key challenge (study 1). This impacts especially the recognition of conversational words (study 2), and in turn has dire consequences for downstream intent recognition (study 3). Our findings help to evaluate the current state of conversational ASR, contribute towards multidimensional error analysis and evaluation, and identify phenomena that need most attention on the way to build robust interactive speech technologies.