• Digital Humanities and Cultural Heritage
• 2018-2022close
• Open Access close
• Hal-Diderot close
• Hyper Article en Ligne - Sciences d... close
• Hyper Article en Ligne close

Out-of-vocabulary (OOV) words can pose serious challenges for machine translation (MT) tasks, and in particular, for low-resource language (LRL) pairs, i.e., language pairs for which few or no parallel corpora exist. Our work adapts variants of seq2seq models to perform transduction of such words from Hindi to Bhojpuri (an LRL instance), learning from a set of cognate pairs built from a bilingual dictionary of Hindi--Bhojpuri words. We demonstrate that our models can be effectively used for language pairs that have limited parallel corpora; our models work at the character level to grasp phonetic and orthographic similarities across multiple types of word adaptations, whether synchronic or diachronic, loan words or cognates. We describe the training aspects of several character level NMT systems that we adapted to this task and characterize their typical errors. Our method improves BLEU score by 6.3 on the Hindi-to-Bhojpuri translation task. Further, we show that such transductions can generalize well to other languages by applying it successfully to Hindi -- Bangla cognate pairs. Our work can be seen as an important step in the process of: (i) resolving the OOV words problem arising in MT tasks, (ii) creating effective parallel corpora for resource-constrained languages, and (iii) leveraging the enhanced semantic knowledge captured by word-level embeddings to perform character-level tasks. 47 pages, 4 figures, 21 tables (including Appendices)

ABSTRACTLeporids (Lagomorpha, Mammalia), especially the mountain hare (Lepus timidus) and the European rabbit (Oryctolagus cuniculus), have been among the most abundant small game in western Europe since the Pleistocene. These animals are preferred prey for more than 40 predators, which increases the chances of finding their skeletal remains at archaeological or natural sites, and in particular karstic formations alternately occupied by Palaeolithic human societies and other predators. Moreover, specific eco‐ethological characteristics for both species also offer the possibility to produce a substantial quantity of their bones accumulated by natural mortality without predation, making it more difficult to identify their origin in the fossil record. Despite this fact, the taphonomic signature of accidental‐natural accumulations of leporids in karsts, such as pitfalls, has never been properly characterised. In order to address these issues, we carried out a detailed taphonomic study of leporid remains in a pitfall without any evidence of human activity, namely at Coulet des Roches (Monieux, Vaucluse, south‐eastern France). At this site, leporids are the most abundant species throughout the Last Glacial Maximum sedimentary sequence. The bone accumulation was analysed in order to determine the most relevant criteria to identify a natural accumulation that results from the accidental fall of individuals in the pitfall. The relative proportions of each species are consistent with their current ethology, and their mortality profiles are compatible with the structure of a living natural population. The completeness rate of the different skeletal portions for both species is relatively high, bones are less fragmented and, unlike what might be expected, very few anatomical connections were preserved. The observed anatomical representation, breakage, and bone surface modification patterns are discussed and compared with available data concerning accumulations with predation. Taken together, the identified criteria contribute valuable help in the recognition of accidental mortality in the fossil record, as well as to evaluate the proportion of leporid bones accumulated naturally and those accumulated by predation, human or not, in a deposit.

Abstract Lacustrine sediments are particularly sensitive to modifications within the lake catchment. In a volcanic area, sedimentation rates are directly affected by the history of the volcano and its eruptions. Here, we investigate the impact of Mt. Fuji Volcano (Japan) on Lake Motosu and its watershed. The lacustrine infill is studied by combining seismic reflection profiles and sediment cores. We show evidence of changes in sedimentation patterns during the depositional history of Lake Motosu. The frequency of large mass-transport deposits recorded within the lake decreases over the Holocene. Before ~ 8000 cal yr BP, large sublacustrine landslides and turbidites were filling the lacustrine depression. After 8000 cal yr BP, only one large sublacustrine landslide was recorded. The change in sedimentation pattern coincides with a change in sediment accumulation rate. Over the last 8000 cal yr BP, the sediment accumulation rate was not sufficient enough to produce large sublacustrine slope failures. Consequently, the frequency of large mass-transport deposits decreased and only turbidites resulting from surficial slope reworking occurred. These constitute the main sedimentary infill of the deep basin. We link the change in sediment accumulation rate with (i) climate and vegetation changes; and (ii) the Mt. Fuji eruptions which affected the Lake Motosu watershed by reducing its size and strongly modified its topography. Moreover, this study highlights that the deposition of turbidites in the deep basin of Lake Motosu is mainly controlled by the paleobathymetry of the lakefloor. Two large mass-transport deposits, occurring around ~ 8000 cal yr BP and ~ 2000 cal yr BP respectively, modified the paleobathymetry of the lakefloor and therefore changed the turbidite depositional pattern of Lake Motosu.

International audience; When it comes to Early Middle Ages, literature tends to associate canons with common life, and even to define those referred to as canonici in the sources as clerics living a community life, and therefore distinct from the rest of the clergy. This association derives from the great Carolingian normative texts, first of all the institutio canonicorum from 816, which required that those “canonical clerics” lived a common life. But what was the actual situation on the ground some two centuries after the carolingian redefinition of the vita canonica? Did those communities identified as canonical practice the common life on the eve of the Gregorian reform? And if so, in what form? This issue is more often than not hard to answer for that period, because of data scarcity and imprecision. Italian archives, however, given their richness, allow a more precise grasp of the concrete modalities of the common life in the main urban canonicae (cathedral chapter and of collegiate churches). Those cases, presumably reflecting similar phenomena more difficult to observe in other European regions, show that the way of life followed by canons in the Italian kingdom’s was very heterogeneous in the first half of the 11th c. ; and thus despite the attempts of several bishops of the 980’s-1040’s to reinforce communal life in their churches. A range and gradation of practices appear according to location, revealing different views of the common life and of its importance or lack thereof for the clergy’s elevation and the service of the community of the faithful.; La littérature scientifique relative au haut Moyen Âge a tendance à associer chanoines et vie commune, voire à définir ceux que les sources désignent par le terme de canonici comme des clercs vivant en communauté, par opposition au reste du clergé. Cette association dérive des grands textes normatifs carolingiens, à commencer par l’institutio canonicorum de 816, qui voulaient que ces clercs « canoniques » mènent la vie commune, sur le modèle monastique. Mais qu’en était-il dans les faits quelque deux siècles après la redéfinition carolingienne de la vita canonica ? Pratiquait-on la vie commune dans les communautés qui se reconnaissaient comme canoniales à l’aube de la réforme grégorienne ? Et si oui, sous quelle forme ? Les sources diplomatiques italiennes, par leur richesse, permettent d’approcher d’un peu plus près les modalités concrètes de la vie commune dans les principales canonicae urbaines (chapitres cathédraux et de grandes collégiales). Ces exemples, susceptibles d’éclairer en partie des phénomènes qui se laissent moins bien apprécier dans d’autres régions d’Europe, témoignent avant tout de l’hétérogénéité du mode de vie canonial dans le royaume d’Italie à la veille du moment grégorien, en dépit des efforts d’un certain nombre d’évêques pour renforcer les pratiques communautaires des canonici de leurs églises dans les années 980-1040. Un éventail typologique de formes de vie se dessine selon les cités, reflets de positionnements différents quant à la vie commune et à son importance pour l’élévation des clercs et le service à la communauté des fidèles.

As a response to Hurricane Mitch and the resulting widespread loss of life and destruction of Honduran infrastructure in 1998, the United States Geological Survey (USGS) conducted the first wide-area airborne lidar topographic mapping project in Central America. The survey was executed by the Bureau of Economic Geology at the University of Texas at Austin (BEG) in 2000, and it was intended to cover 240 square kilometers distributed among 15 flood-prone communities throughout Honduras. The original data processing produced basic digital elevation models at 1.5-meter grid spacing which were used as inputs for hydrological modeling. The USGS published the results in a series of technical reports in 2002. The authors became interested in this dataset in 2013 while searching for geospatial data that would provide additional context and comparative references for an archaeological lidar project conducted in 2012 in the Honduran Mosquitia. After multiple requests to representatives from the USGS and BEG, we found various types of processed data in personal and institutional archives, culminating in the identification of 8-mm magnetic tapes that contained the original point clouds. Point clouds for the 15 communities plus a test area centered on the Maya site of Copán were recovered from the tapes (16 areas totaling 700 km2). These point clouds have been reprocessed by the authors using contemporary software and methods into higher resolution and fidelity products. Within these new products, we have identified and mapped multiple archaeological sites in proximity to modern cities, many of which are not part of the official Honduran site registry. Besides improving our understanding of ancient Honduras, our experiences dealing with issues of data management and access, ethics, and international collaboration have been informative. This paper summarizes our experiences in the hope that they will contribute to the discussion and development of best practices for handling geospatial datasets of archaeological value.

International audience; The machine, over the course of the 20th century, progressively integrated itself into all fields of human activity, including artistic creation; and indeed, with the first decades of that century having established a surprisingly vital and wide-ranging series of perspectives on the relationship between art and the machine, certain artists in the wake of the Second World War no longer felt compelled to treat the machine as a mere theme or source of inspiration: the machine itself becomes art-unless it is art which seeks to become mechanical? The artist mutates into "artist-engineer"; and this transition, resonating within a specific historical context, leads not only to a questioning of the nature of the work itself, but also to a broader questioning which places us within the realm of anthropology: what is this art telling us about the actual conditions of contemporary human society, and what is it telling us about the future to which we aspire? It is the goal of this special issue of Arts to stimulate an historically conscious, protean, and global (re)thinking of the cultural relationship between man and machine; and to this end, we welcome contributions falling anywhere within the nearly infinite spectrum represented by the prismatic period during the middle of the last century in which the machine became a legitimate artistic medium.

Introduction Vers la fin de sa vie, au cours du discours prononcé à l’occasion de son Jubilé cinquantenaire d’économiste à l’Université de Lausanne, intitulé « Ruchonnet et le socialisme scientifique », Léon Walras remarquait : Il est certain, d’abord, que la détermination du salaire au moyen de grèves et de lock-outs, c’est-à-dire, en définitive, par échange de coups de pierre, contre coups de fusil, est un procédé sauvage. Mais il l’est également que, pour substituer à ce procédé celui de l...