Hugging Face's logo

Datasets:
WINGNUS
/
ACL-OCL

Tasks:
Token Classification
Languages:
English
Size:
10K<n<100K
Tags:
research papers
acl
License:
Dataset card Files Community
5
ACL-OCL / Base_JSON /prefixL /json /lrec /2020.lrec-1.117.json
Benjamin Aw
Add updated pkl file v3
6fa4bc9
raw
history blame contribute delete
116 kB
 {
"paper_id": "2020",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T11:45:46.161045Z"
},
"title": "A New Latin Treebank for Universal Dependencies: Charters between Ancient Latin and Romance Languages",
"authors": [
{
"first": "Flavio",
"middle": [],
"last": "Massimiliano",
"suffix": "",
"affiliation": {},
"email": ""
},
{
"first": "Timo",
"middle": [],
"last": "Korkiakangas",
"suffix": "",
"affiliation": {},
"email": "[email protected]"
},
{
"first": "Marco",
"middle": [],
"last": "Passarotti",
"suffix": "",
"affiliation": {},
"email": "[email protected]"
},
{
"first": "\u2022",
"middle": [
"\u2022"
],
"last": "Circse",
"suffix": "",
"affiliation": {},
"email": ""
},
{
"first": "Universit\u00e0",
"middle": [],
"last": "Cattolica",
"suffix": "",
"affiliation": {},
"email": ""
},
{
"first": "Del",
"middle": [],
"last": "Sacro",
"suffix": "",
"affiliation": {},
"email": ""
},
{
"first": "Helsingin",
"middle": [],
"last": "Yliopisto",
"suffix": "",
"affiliation": {},
"email": ""
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "The present work introduces a new Latin treebank that follows the Universal Dependencies (UD) annotation standard. The treebank is obtained from the automated conversion of the Late Latin Charter Treebank 2 (LLCT2), originally in the Prague Dependency Treebank (PDT) style. As this treebank consists of Early Medieval legal documents, its language variety differs considerably from both the Classical and Medieval learned varieties prevalent in the other currently available UD Latin treebanks. Consequently, besides significant phenomena from the perspective of diachronic linguistics, this treebank also poses several challenging technical issues for the current and future syntactic annotation of Latin in the UD framework. Some of the most relevant cases are discussed in depth, with comparisons between the original PDT and the resulting UD annotations. Additionally, an overview of the UD-style structure of the treebank is given, and some diachronic aspects of the transition from Latin to Romance languages are highlighted.",
"pdf_parse": {
"paper_id": "2020",
"_pdf_hash": "",
"abstract": [
{
"text": "The present work introduces a new Latin treebank that follows the Universal Dependencies (UD) annotation standard. The treebank is obtained from the automated conversion of the Late Latin Charter Treebank 2 (LLCT2), originally in the Prague Dependency Treebank (PDT) style. As this treebank consists of Early Medieval legal documents, its language variety differs considerably from both the Classical and Medieval learned varieties prevalent in the other currently available UD Latin treebanks. Consequently, besides significant phenomena from the perspective of diachronic linguistics, this treebank also poses several challenging technical issues for the current and future syntactic annotation of Latin in the UD framework. Some of the most relevant cases are discussed in depth, with comparisons between the original PDT and the resulting UD annotations. Additionally, an overview of the UD-style structure of the treebank is given, and some diachronic aspects of the transition from Latin to Romance languages are highlighted.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "1. Introduction",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "Universal Dependencies (UD) is an open-access and collaborative project that aims to provide \"a framework for consistent annotation of grammar (parts of speech, morphological features, and syntactic dependencies) across different human languages\" 1 (Nivre et al., 2016) . The current release (2.5 as of February, 2020) of UD includes 157 treebanks over 90 languages, among them several ancient languages (such as Akkadian and Gothic) and historical phases of modern ones (such as Old French or Classical Chinese).",
"cite_spans": [
{
"start": 249,
"end": 269,
"text": "(Nivre et al., 2016)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Universal Dependencies",
"sec_num": "1.1."
},
{
"text": "Currently, Latin features the most data (ca. 582 000 tokens) and the most treebanks of all the ancient languages of UD (three). The Latin treebanks are: PROIEL (Haug and J\u00f8hndal, 2008) , which includes the entire New Testament in Latin and texts from the Classical and Late era (ca. 200 000 tokens); the Latin Dependency Treebank (LDT) by the Perseus Digital Library (Bamman and Crane, 2006) , comprised of a small selection of texts by Classical authors (ca. 29 000 tokens); and the Index Thomisticus Treebank (IT-TB) (Passarotti, 2019), based on works written by Thomas Aquinas in the XIII century (353 000 tokens). This testifies to the huge diachronic, diatopic and diamesic variety of Latin: initially the language of Ancient Rome, it was for a long time the European lingua franca of arts, sciences and law, including official documents; it still is the official language of the Catholic Church (Cecchini et al., forthcoming) . Because of this variety, no single treebank alone can sufficiently represent the entire Latinity. 2 The present UD conversion the Late Latin Charter Treebank sion of the LLCT2 into the UD framework and is organised as follows: Section 2. describes the original treebank and sketches the process of conversion into UD; Section 3. goes into the detail of how some syntactic and linguistic phenomena are tackled during the conversion process, and Section 4. gives an overview of the major formal differences between the two versions of the LLCT2. Finally, Section 5. concludes the paper and points to future work.",
"cite_spans": [
{
"start": 160,
"end": 184,
"text": "(Haug and J\u00f8hndal, 2008)",
"ref_id": "BIBREF11"
},
{
"start": 367,
"end": 391,
"text": "(Bamman and Crane, 2006)",
"ref_id": "BIBREF0"
},
{
"start": 901,
"end": 931,
"text": "(Cecchini et al., forthcoming)",
"ref_id": null
},
{
"start": 1032,
"end": 1033,
"text": "2",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Universal Dependencies",
"sec_num": "1.1."
},
{
"text": "In this section we briefly present the annotation approach of the original LLCT2 treebank and discuss the technicalities of the UD conversion process.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Structure and Conversion Process of the LLCT2",
"sec_num": "2."
},
{
"text": "The original annotation of the LLCT2 mainly follows the LDT and IT-TB styles, which are based on (Bamman et al., 2007, henceforth Guidelines) . The latter rests on the annotation style for the analytical layer of the Prague Dependency Treebank (PDT) presented in (Haji\u010d et al., 2017) . The LLCT2 was first automatically annotated and then manually corrected. However, the Guidelines only apply to Latin that follows regular Classical Latin grammar. 4 Therefore, the non-Classical syntactic (and morphological) features had to be annotated following an additional set of rules described in (Korkiakangas and Passarotti, 2011) , some of which are more closely detailed in Section 3.",
"cite_spans": [
{
"start": 97,
"end": 141,
"text": "(Bamman et al., 2007, henceforth Guidelines)",
"ref_id": null
},
{
"start": 263,
"end": 283,
"text": "(Haji\u010d et al., 2017)",
"ref_id": "BIBREF8"
},
{
"start": 449,
"end": 450,
"text": "4",
"ref_id": null
},
{
"start": 589,
"end": 624,
"text": "(Korkiakangas and Passarotti, 2011)",
"ref_id": "BIBREF14"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "The Original Annotation of the LLCT2",
"sec_num": "2.1."
},
{
"text": "The morphological annotation in the LLCT2 is based on the categories present in Classical Latin grammar. Since the LLCT is meant to be used to study the evolution of Latin that had taken place between Classical and Early Medieval Latin, the non-classical word forms attested in LLCT are given the morphological analyses of their equivalents in Classical Latin. For example, in the LLCT phrase debeam reddere soledo uno 'I have to pay one gold coin', the form soledo 'gold coin' (the direct object of the verb reddere 'to pay') is analysed as a masculine singular accusative like its Classical ancestor solidum, although it is questionable whether the Latin of the VIIIth and IXth centuries still had a morphologically expressed o-stem case. The masculine singular ending -o of the Romance type (cf. Italian soldo 'coin; pay') is likely to have derived from the accusative case form, hence its analysis as such (Zamboni, 2007, \u00a73.1.6) . A somewhat more complex issue is gender change: offertas 'offerings, tithe' has a seemingly feminine plural ending -as, but the form is annotated as a neuter plural because the original Classical Latin-like form offerta was a neuter plural. It is known that Latin neuter plurals became feminine in Romance languages (V\u00e4\u00e4n\u00e4nen, 2012, \u00a7215) , but if offertas were annotated as a feminine, the possibility to track the underlying diachronic change would be lost.",
"cite_spans": [
{
"start": 910,
"end": 933,
"text": "(Zamboni, 2007, \u00a73.1.6)",
"ref_id": null
},
{
"start": 1252,
"end": 1274,
"text": "(V\u00e4\u00e4n\u00e4nen, 2012, \u00a7215)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "The Original Annotation of the LLCT2",
"sec_num": "2.1."
},
{
"text": "No consensus prevails about the grammatical categories of the spoken Latin of the time. Therefore, the evolutionary principle that exploits Classical Latin grammatical categories seems to be the only workable way to annotate non-Classical Latin. In sum, the annotation style of the LLCT2 incorporates a diachronic component and, consequently, the annotation of the LLCT2 does not claim to be a synchronic representation of any specific stage of Latin. This is at the same time an obstacle and an added value for typological comparisons, since it provides a collection of data belonging to a late and very specific variety of Latin, and yet is diachronically related to the Classical one. A prospective user of the UD version of the LLCT2 has to be aware of the peculiarities of its annotation.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Original Annotation of the LLCT2",
"sec_num": "2.1."
},
{
"text": "The scripts behind the conversion process are written in the Perl language 5 and function as modules of Treex's architecture. 6 The scripts were developed as part of the HamleDT (harmonised multi-language dependency treebank) project (Zeman et al., 2012; Zeman et al., 2014; Rosa et al., 2014) and can be broken down into two sections: the harmonisation phase and the UD conversion proper. Since its latest version (3.0), HamleDT has been using the UD syntactic annotation style, 7 and so the conversion process has been readapted for UD. 8 \"Harmonisation\" here means the adjustment of a treebank so as to more closely conform to the PDT morphosyntactic annotation style originally used by HamleDT (versions 1.0 and 2.0) that serves as the basis for the current UD conversion proper, which rewrites morphological features and intervenes on the structure of the syntactic trees in the second phase. Since the LLCT2 is very close to the PDT standard to start with (see Section 2.1.), in our case the harmonisation is mainly used to encode parts of speech (POS) and morphology by means of the Interset tagset (Zeman, 2008) 9 used by HamleDT and UD, and whose formal aspects and terminology are inspired by (Petrov et al., 2011) and described in (Sylak-Glassman, 2016; Zeman, 2018) . In this phase, we also make some preliminary structural and annotation modifications in view of the UD re-interpretation of some linguistic phenomena (a part of which is illustrated in Section 3.).",
"cite_spans": [
{
"start": 126,
"end": 127,
"text": "6",
"ref_id": null
},
{
"start": 234,
"end": 254,
"text": "(Zeman et al., 2012;",
"ref_id": "BIBREF29"
},
{
"start": 255,
"end": 274,
"text": "Zeman et al., 2014;",
"ref_id": "BIBREF30"
},
{
"start": 275,
"end": 293,
"text": "Rosa et al., 2014)",
"ref_id": "BIBREF23"
},
{
"start": 535,
"end": 540,
"text": "UD. 8",
"ref_id": null
},
{
"start": 1106,
"end": 1119,
"text": "(Zeman, 2008)",
"ref_id": "BIBREF31"
},
{
"start": 1203,
"end": 1224,
"text": "(Petrov et al., 2011)",
"ref_id": "BIBREF20"
},
{
"start": 1242,
"end": 1264,
"text": "(Sylak-Glassman, 2016;",
"ref_id": "BIBREF25"
},
{
"start": 1265,
"end": 1277,
"text": "Zeman, 2018)",
"ref_id": "BIBREF32"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Conversion into UD and the UD Style",
"sec_num": "2.2."
},
{
"text": "Finally, the conversion proper is carried out by the main script and the output in the UD style is returned according to the CoNLL-U format, i. e. as a plain-text, csv-like file that follows a revised version of the CoNLL-X format (Buchholz and Marsi, 2006) . 10 The harmonisation and conversion scripts can be downloaded from the Github pages of the Treex and HamleDT projects. The UD project also provides a validator script 11 in Python 12 to ensure that the final result complies with the general UD principles. The differences between the UD principles and the PDT style of the original LLCT2 form the basis of the discussion in Sections 3. and 4.",
"cite_spans": [
{
"start": 231,
"end": 257,
"text": "(Buchholz and Marsi, 2006)",
"ref_id": "BIBREF3"
},
{
"start": 260,
"end": 262,
"text": "10",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Conversion into UD and the UD Style",
"sec_num": "2.2."
},
{
"text": "In the following, we refer to labels used for syntactic relations as uppercase afuns (analytical functions) when dealing with the PDT style of the original LLCT2 treebank, 13 , and as lowercase deprels (dependency relations) with regard to the UD style. 14 When citing an LLCT2 sentence, we use \"s2019\" for the sentence number 2019, and \"a. o.\" (\"and others\") if the same formula also recurs in other sentences.",
"cite_spans": [
{
"start": 172,
"end": 174,
"text": "13",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Notation",
"sec_num": "2.3."
},
{
"text": "The differences between the annotation styles of LLCT2 and UD pose a challenge to any automated conversion process. The most marked difference is that in the LLCT2 conjunctions, prepositions and copulas govern phrases, while UD favours dependencies between content words so that function words tend to end up as leaves of the syntactic tree. 15 This has consequences on the resulting syntactic trees (see Section 4.), especially on the rendition of often recurring syntactic constructions like co-ordinations and appositions. Appositional constructions in particular entail many linguistically and technically complex aspects, and they are tackled in Section 3.1. Moreover, the Latin of the LLCT2 features a few syntactic constructions that are absent, or only rarely attested, in Classical Latin, and that have been varyingly, if at all, treated in other Latin treebanks. In the following, we describe some of these constructions, along with how we deal with them in the UD framework. Such interpretations have, as a broader consequence, the constructive revision and improvement of the current UD annotation schemes for Latin, which still do not have official guidelines.",
"cite_spans": [
{
"start": 342,
"end": 344,
"text": "15",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Challenging Syntactic Structures",
"sec_num": "3."
},
{
"text": "Due to the formal and legal nature of the LLCT2, the notaries regularly strove to specify all the contracting parties as precisely as possible. This involves the use of proper names with kinship terms, such as filius 'son' and germanus 'brother', or titles, such as imperator 'emperor' and archiepiscopus 'archbishop', but also other expressions of qualification, profession, social status and the like. Proper names of locations are also often specified in their nature by generic terms, such as locus 'place', rivus 'creek' or civitas 'city': e. g. civitate Papia '[in] the city of Pavia' (s7133), loco Cassio '[in] the place [called] Cassium' (s3910 a. o.).",
"cite_spans": [
{
"start": 628,
"end": 636,
"text": "[called]",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "On a syntactic level, such specifications give rise to appositions. Broadly speaking, the Latin apposition is defined as a noun phrase acting as the co-referential complement of another noun phrase by means of case agreement and simple juxtaposition, with very few exceptional cases of discontinuity (Longr\u00e9e, 1990; Spevak, 2014, Ch. 4) . 16 Of particular interest are the many appositions (15 414 occurrences out of a total of 16 760 appositional constructions) where at least one of the two participating elements is a proper name, like the above-mentioned loco Cassio or Gherardus episcopus 'Gherardus, the bishop'/'bishop Gherardus' (s5715 a. o.). In such cases, the position of the specifying element in relation to the proper noun is the criterion by which we can distinguish so-called close appositions from free ones (Longr\u00e9e, 1990; Spevak, 2014, Ch. 4) .",
"cite_spans": [
{
"start": 300,
"end": 315,
"text": "(Longr\u00e9e, 1990;",
"ref_id": "BIBREF16"
},
{
"start": 316,
"end": 336,
"text": "Spevak, 2014, Ch. 4)",
"ref_id": null
},
{
"start": 339,
"end": 341,
"text": "16",
"ref_id": null
},
{
"start": 825,
"end": 840,
"text": "(Longr\u00e9e, 1990;",
"ref_id": "BIBREF16"
},
{
"start": 841,
"end": 861,
"text": "Spevak, 2014, Ch. 4)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "A close apposition implies a stronger bond between its constituents than a free one, which on the contrary is more similar to a predicative construction than to an attributive one. While the sequence noun phrase + proper noun (as in loco Cassio) can undoubtedly be considered a close apposition where the first element acts as its (weak) head, more doubts arise whether the sequence proper noun + noun phrase (as in Gherardus episcopus) is to be held as a close or free apposition (Longr\u00e9e, 1990) . We argue that, in the LLCT2, the latter pattern is always a free, non-restrictive apposition with the proper noun as its head, and as such we assign it the UD deprel appos, as opposed to flat for close appositions (see Section 3.1.1.). 17 In the examples above:",
"cite_spans": [
{
"start": 481,
"end": 496,
"text": "(Longr\u00e9e, 1990)",
"ref_id": "BIBREF16"
},
{
"start": 735,
"end": 737,
"text": "17",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": ". . . in loco Cassio . . . . . . 16 We notice here that some constructions traditionally labeled as \"appositions\", where a term is specified later in the sentence by a phrase introduced by an adverbial element like scilicet or id est 'that is', actually belong to the realm of co-ordination, and their introductory elements are to be treated as (explicative) conjunctions. Uncertainties in their treatment in UD are discussed in (Cecchini et al., 2018, \u00a72.2.3 ).",
"cite_spans": [
{
"start": 33,
"end": 35,
"text": "16",
"ref_id": null
},
{
"start": 429,
"end": 459,
"text": "(Cecchini et al., 2018, \u00a72.2.3",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "17 Cf. the discussion at https://github.com/ UniversalDependencies/docs/issues/536# issuecomment-375030335.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "In the original LLCT2 annotation, appositional constructions of this kind are considered simple chains of noun phrases where the head noun is modified by another noun via the afun ATR. This does not, however, apply to non-classical constructions that indicate price or quantity, e. g. s6:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "recepi . . . praetium in praefinito, auri soledos numero quinque 'I received . . . the price as agreed, five gold coins', which are analyzed in the LLCT2 as appositive constructions and marked with APOS (praetium 'price' and soledos 'gold coins' depend on the intervening comma tagged as APOS).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "This use of APOS represents the standard way of treating appositions in the Guidelines (Bamman et al., 2007, p. 28) : the two noun phrases are made dependent on the comma as their \"connecting element\" and are labelled with the same syntactic function in the sentence, while the suffix _AP marks them as part of this construction: Here, the latter portionem 'portion, part' is even picked up by ut dicxi 'as I said'. This phenomenon is related to the fact that formulaic sentences are often long and that the legal validity of the document depends on its correct wording. In addition, the LLCT2 abounds with sentences that contain various combinations of appositions and reprisals.",
"cite_spans": [
{
"start": 87,
"end": 115,
"text": "(Bamman et al., 2007, p. 28)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "In our UD treatment of the LLCT2, we use the deprel dislocated, along with the already discussed apos, to represent the structure of such sentences. This is currently the best way to deal with reprisals of this kind, in which we can distinguish between a \"dislocated\" element, i. e. the element that is more peripheral with respect to the predicate and is usually also the longer and more complex one, and an \"anaphoric\" element, i. e. an abbreviated repetition of the former which is more \"core\" with respect to the predicate. Both elements are made dependent on the predicate, but while only the latter receives its core deprel, the former is tagged as dislocated with an appropriate subrelation. In our example: We identify 834 such cases (i. e. affecting the 9.24% of sentences in the corpus) in the LLCT2. Among them, we report 7 cases in which the UD relations required by the dislocated and by the anaphoric elements are asymmetric, such as in s357, where quidquid . . . remanserit 'whatever . . . is left' is clausal and gets the deprel ccomp, while meam portionem 'my part' (dislocated) would be assigned obj. Since in this and in all other such cases one of the two elements is a so-called free relative clause (Grosu, 2003) , the need for a coherent interpretation of this phenomenon in UD will require further analyses and, in the light of the appositional constructions at hand, might encourage a rethinking of part of the overall annotation strategy for Latin.",
"cite_spans": [
{
"start": 1221,
"end": 1234,
"text": "(Grosu, 2003)",
"ref_id": "BIBREF7"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "portionem",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": "Note that our UD conversion currently does not include 121 sentences of the original corpus wherein the use of the afun APOS is reputed to be problematic. This applies in particular to sentences where 1. the APOS relation is not binary, i. e. there are more than two elements marked by the _AP suffix, and where 2. the APOS appears as the root. Further investigation is needed to assess how these anomalous structures are to be analysed in the UD framework, and if their annotation in the LLCT2 should actually be modified. Table 1 presents figures concerning the above discussed appositions with proper nouns. For this analysis, only non co-ordinated appositions are examined. Our first observation is that noun phrases postposed to proper nouns almost exclusively concern terms related to kinship and social status, and, consequently, to human beings. On the contrary, all the toponymal specifications are preposed, except for a few postposed occurrences (civitas 'city' twice, and ecclesia 'church' and plebs 'parish' once each). Even then, there is at least one preposed occurrence, as civitate Papia, s7133, vs. Papia civitate, s1513). The only cases in which human proper nouns are regularly postposed are either when they follow a first-or second-person pronoun (ego, tu, nos 'we', vos 'you (pl.)'), or when they are part of a fixed expression (domnus). The only term showing unstable behaviour is the multifaceted and very generic term vir, which is always further specified by an epithet, a fact we will explain below. These patterns can be explained by means of a generic definiteness criterion. Although this does not apply particularly well to legal contracts where various specifications are often used to unambiguously identify the contracting parties, in general, a name is sufficient to unambiguously identify a person, while further qualifications added as postposed appositions only serve to better describe their background, without being necessary. Instead, personal pronouns can only denote one specific individual in the context in which they are uttered. Indeed, we only observe the pattern personal pronoun + proper noun and never the inverse: it is the proper noun here that acts as extra information. Consequently, we attest chains of appositions in this 'hierarchical' order, as in (s2406 a. o.) ego Petrus episcopus 'I, Petrus, the bishop' (as signature). Another sign that we are in the presence of free appositions, and also a motivation for their postposition, is their expandability with no functional changes, which signals a greater independence from its head, as in s7689: One reason for the prevalence of close appositions with place names might be the stronger equivalence between the nature of a place and the place itself (represented by its name). While a person called Petrus might be a gastaldus, a presbiter, a schabinus or even the apostolus, and change his social status or occupation as time goes by, the status of the fluvius Cassina is practically immutable. In fact, while the syntactic head might be detected in fluvius 'creek' on the basis of agreement or similar factors (Spevak, 2014, Ch. 4, \u00a73.4) , semantically it is the apposition Cassina that allows the unambiguous identification of this landmark. The expression might be substituted by is fluvius 'that creek', but not by fluvius alone (Longr\u00e9e, 1990) . This actual 'headlessness' is managed in UD by using the deprel flat. This also clarifies why domnus or vir beatissimus 'a most blessed man' are always preposed while imperator 'emperor' or vir excellentissimus 'a most excellent man' are postposed: the former ones are seen as innate qualities of dominance or sanctity of the person, domnus having become too generic with respect to other titles of nobility and vir beatissimus being very close to sanctus (which is also always preposed), while the latter ones pertain more to the described person's social position. In fact, the rank of a domnus is nearly always further specified by imperator or a similar title, whereas vir excellentissimus might also be translated as 'gentleman'. This mirrors the different placements of adjectives according to their \"qualifying\" or \"determining\" function (Iovino, 2012, Ch. 2) . A similar interpretation can be given to the constructions of the type nomine Ermilinda 'by the name of Ermilinda' (s140) or numero quatraginta 'in the number of forty' (s914), which we similarly analyse with flat. 18",
"cite_spans": [
{
"start": 3123,
"end": 3150,
"text": "(Spevak, 2014, Ch. 4, \u00a73.4)",
"ref_id": null
},
{
"start": 3345,
"end": 3360,
"text": "(Longr\u00e9e, 1990)",
"ref_id": "BIBREF16"
},
{
"start": 4208,
"end": 4229,
"text": "(Iovino, 2012, Ch. 2)",
"ref_id": null
}
],
"ref_spans": [
{
"start": 524,
"end": 531,
"text": "Table 1",
"ref_id": "TABREF4"
}
],
"eq_spans": [],
"section": "Apposition-like Constructions",
"sec_num": "3.1."
},
{
"text": ". . . tu",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Close versus Free Appositions",
"sec_num": "3.1.1."
},
{
"text": "In all other cases where we encounter a nominal phrase modifying another nominal phrase (where the former is usually in the genitive or introduced by a preposition), we deem it to have a simple (i. e. non co-referential) attributive function; hence the application of the deprel nmod.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Close versus Free Appositions",
"sec_num": "3.1.1."
},
{
"text": "In Classical Latin, the only verb considered an auxiliary is sum/esse 'to be'. 19 It is used exclusively for passive tenses bearing a perfect aspect, e. g. amatus sum 'I was loved' vs. amabar 'I was being loved' (Barbieri, 1995, \u00a7142-144, 147-148) , (Greenough and Allen, 2006, \u00a7184-188) , and passive periphrases, e. g. amandum est 'it is to be loved/everybody has to love' (Barbieri, 1995, \u00a7156, 167) , (Greenough and Allen, 2006, \u00a7193-196) . 20 Among other structural changes that eventually have given shape to modern Romance languages, Late Latin saw the emergence of new periphrastically expressed active tenses based again on sum/esse, but also on the auxiliary habeo/hab\u0113re 'to have', accompanied by a perfect participle. Their origin is probably to be sought in the resultative use of a subject or an object predicative, respectively (V\u00e4\u00e4n\u00e4nen, 2012, \u00a7298, 300) . For this reason, in these tense forms the perfect participle originally agreed with the subject or the object, respectively, while in modern Romance languages grammaticalization has been fully accomplished and agreement can be lost, especially in conjunction with the outcomes of the auxiliary habeo/hab\u0113re. Here, adimpletum (from adimpleo/adimpl\u0113re 'to fulfil') is a generic singular neuter accusative without an explicit object (abetis is a spelling variant of habetis 'you (pl.) have'), and pegiorate (a spelling variant of postclassical peioratae) is a feminine plural, like res 'things'. Moreover, here the voice of peiorate sunt may be active and not passive, as would have been expected in Classical Latin from such a construction; this interpretation of pegiorate sunt is also possible because the form occurs parallel to the active verb adimpletum habetis, and the Classical counterparts would be adimplevistis and peioraverunt, respectively. In both cases we use the deprel aux and not aux:pass, an unprecedented fact with respect to Classical Latin as well as to the learned Medieval Latin of the IT-TB (Cecchini et al., 2018) . E. g. , the UD tree structure for the second example is:",
"cite_spans": [
{
"start": 79,
"end": 81,
"text": "19",
"ref_id": null
},
{
"start": 212,
"end": 247,
"text": "(Barbieri, 1995, \u00a7142-144, 147-148)",
"ref_id": null
},
{
"start": 250,
"end": 287,
"text": "(Greenough and Allen, 2006, \u00a7184-188)",
"ref_id": null
},
{
"start": 375,
"end": 402,
"text": "(Barbieri, 1995, \u00a7156, 167)",
"ref_id": null
},
{
"start": 405,
"end": 442,
"text": "(Greenough and Allen, 2006, \u00a7193-196)",
"ref_id": null
},
{
"start": 445,
"end": 447,
"text": "20",
"ref_id": null
},
{
"start": 843,
"end": 870,
"text": "(V\u00e4\u00e4n\u00e4nen, 2012, \u00a7298, 300)",
"ref_id": null
},
{
"start": 1987,
"end": 2010,
"text": "(Cecchini et al., 2018)",
"ref_id": "BIBREF4"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Auxiliary Verbs of Active Tenses",
"sec_num": "3.2."
},
{
"text": ". . . res ipsas peiorate sunt . . .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Auxiliary Verbs of Active Tenses",
"sec_num": "3.2."
},
{
"text": "While we can easily see that the auxiliary habeo occurs 14 times in the LLCT2, the use of sum in active tenses is a priori indistinguishable from its use in passive ones, so that manual disambiguation will be necessary to assess the extent of this phenomenon.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "det nsubj aux",
"sec_num": null
},
{
"text": "Throughout its history, Latin has had a constant tendency to create new conjunctions from adverbialised forms of interrogative-relative pronouns or determiners, e. g. the conjunction cum 'when' from the accusative masculine singular form quom of qui 'who/which/that' (Palmer, 1988; Vineis, 2005) . We can observe that this process is still active in the Late Latin of the charters. An adverb with a relative aspect lies in a syntactically grey area, so to speak, in that it tends to function without referring to an (explicit) antecedent. If its antecedent can be interpreted to be a preceding clause, the relative adverb takes on the role of a clause connector and is gradually grammaticalized as a conjunction. This phenomenon is actually very common among European languages and beyond (Kortmann, 1997) .",
"cite_spans": [
{
"start": 267,
"end": 281,
"text": "(Palmer, 1988;",
"ref_id": "BIBREF19"
},
{
"start": 282,
"end": 295,
"text": "Vineis, 2005)",
"ref_id": "BIBREF27"
},
{
"start": 789,
"end": 805,
"text": "(Kortmann, 1997)",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "New and Old Conjunctions",
"sec_num": "3.3."
},
{
"text": "In the LLCT2, the most frequent of such adverbs is qualiter, 'in what way', formed from the interrogative-relative determiner qualis 'of what kind', but clearly used to introduce a conditional/comparative adverbial subordinate clause.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "New and Old Conjunctions",
"sec_num": "3.3."
},
{
"text": "Qualiter is always found at the beginning of a clause with no corresponding referent in the main clause, similarly to the conjunctional ut 'as, like'. It is frequently used in formulaic expressions, such as qualiter [supra/superius/. . . ] legitur 'as it can be read above' (179 out of 522 total occurrences of qualiter), or more freely, as in s34:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "New and Old Conjunctions",
"sec_num": "3.3."
},
{
"text": ". . . dispensare ipsa res mea secundum Deo qualiter melius potueritis pro anima mea remedium '. . . to spend my property in accordance with God, in the best way you (pl.) can, for the salvation of my soul'",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "New and Old Conjunctions",
"sec_num": "3.3."
},
{
"text": "The LLCT2 also has other words traditionally labelled as adverbs which behave like conjunctions (and are often coordinated among them, as in comodo et qualiter 'in what sort and manner', 12 occurrences): quando/quandoque 'when', quantum 'how much', quatenus 'how far', quit ( = quid) 'that', comodo/como/quomodo 'in what manner', ubicumque/ubicunque 'wherever', unde 'wherefore'. Some of them survive with an identical function in modern Romance languages, like Italian come 'how' (formerly como), ultimately from quomodo. We label them with the UD part of speech SCONJ 22 and not ADV, and they consequently receive the deprel mark.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "New and Old Conjunctions",
"sec_num": "3.3."
},
{
"text": "The LLCT2 charters are preserved as originals, which means that their language has not been emended during a long manuscript tradition, contrary to what has happened to literary Latin texts. Accordingly, the LLCT2 also features original misspellings, grammatical mistakes and gaps, unlike any existing Latin treebank. Any annotation style of such a language variety has to manage these mistakes and lacunae in some way.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Anacolutha and Gaps",
"sec_num": "3.4."
},
{
"text": "Contrary to what is suggested by the Guidelines, the afun ExD is not used to mark ellipsis in the LLCT2 (cf. Section 3.5.). Instead, ExD is used for three other purposes in the LLCT2:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Anacolutha and Gaps",
"sec_num": "3.4."
},
{
"text": "1. to mark the node closest to the root in sentences that do not contain a verbal predicate because they are mere nominal phrases, e. g. signum + manus Pertualdi clerici 'sign + of the hand of Pertualdus the cleric' (s12); 2. to mark phrases that remain disconnected from the overall syntactic structure of a sentence for various reasons; a typical example of which is id est 'that is' (s159 a. o. where only the first subordinating conjunction ut 'that' is functional, whereas the latter two ut were added by mistake, probably caused by the distance between the subordinating verb volo 'I want' and the subordinated verb sit '(that) it be';",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Anacolutha and Gaps",
"sec_num": "3.4."
},
{
"text": "3. to mark anacolutha, i. e. sentences that show an irreconcilable discontinuity in their syntactic structure, e. g. constat me Sanitulum . . . vendere et tradere praevideo 'I, Sanitulus, am determined . . . I arrange for the selling and trading' (s2 a. o.), with two unconnected predicate verbs (constat and praevideo) that are seemingly derived from two different formulae.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Anacolutha and Gaps",
"sec_num": "3.4."
},
{
"text": "For these reasons, the outcome of the afun ExD in the UD conversion is quite diversified, as will be shown in Section 4.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Anacolutha and Gaps",
"sec_num": "3.4."
},
{
"text": "Material gaps and fragmentary passages that occur in the text are also taken into account in the LLCT2. If the POS of a missing or fragmentary token can be deduced with a high degree of probability, but without certainty about the exact lexical item, it is marked in square brackets as an artificial placeholder token and lemmatised as missing\u02c6token, e. g. + Ego David filio [Propn] rogatus [--] '+ I, David, son of [Propn] , having been asked [--]' (s180), where a generic [Propn] stands for the proper noun expected in this context. If a gap of one or more words cannot be restored at all, like the last part of the above example, it is marked with an artificial token [--] , which is attached either to the node on which it is most likely dependent, or, in the case of uncertain dependency, to the sentence root. This kind of restoration is possible thanks due to the highly formulaic documentary language, where it is often obvious what type of word is missing (Korkiakangas and Lassila, 2013) . In our UD conversion, such missing\u02c6tokens maintain their forms and dependencies and bear their corresponding UD morphological and syntactic analysis.",
"cite_spans": [
{
"start": 416,
"end": 423,
"text": "[Propn]",
"ref_id": null
},
{
"start": 671,
"end": 675,
"text": "[--]",
"ref_id": null
},
{
"start": 965,
"end": 997,
"text": "(Korkiakangas and Lassila, 2013)",
"ref_id": "BIBREF13"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Anacolutha and Gaps",
"sec_num": "3.4."
},
{
"text": "Another phenomenon indicated by special placeholders is ellipsis. Ellipsis is usually disambiguated in the LLCT2 by adding an artificial token, similar to the one used for gaps, with the POS of the elided element in square brackets and with the lemma missing\u02c6token. This is typical, for example, of comparative clauses, like s485:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Ellipses",
"sec_num": "3.5."
},
{
"text": "set tibi obedire et servire debeat, sicut filius [verb] patri 'but he/she must obey and serve you, like the son [obeys and serves] the father'.",
"cite_spans": [
{
"start": 49,
"end": 55,
"text": "[verb]",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Ellipses",
"sec_num": "3.5."
},
{
"text": "While ellipsis is intentional by definition, the same kind of placeholder is also used to introduce unintentionally omitted tokens if their POS can be reliably reconstructed. For example, it is probable that the participle regnante 'reigning' is missing from the opening of s1524 a. o.: Here, the scribe seems to have forgotten the participial head required by the formula. As it is not clear whether the scribe meant to write exactly regnante 'reigning', the restoration only indicates the POS. Such constructions differ from anacolutha (see Section 3.4.) in that they do not seem to result from the overlap of two competing syntactic structures or formulae, but from a sheer lapse of concentration by the scribe. However, it is sometimes hard to draw the line between different cases, because one cannot be sure whether a construction in conflict with the standard grammar of Classical Latin is an unintentional anacoluthon, a lapsus, or an intentional, albeit non-standard syntactic choice. one should expect to find the subordinating conjunction quia 'because; that', which is normally placed between the predicates manifestu sum 'I have shown up' and offero 'I offer'. However, it is not obvious whether quia was omitted by negligence or because the scribe considered the latter predicate somehow subordinate to the previous one; in the LLCT2, predicates in such a relation are connected by the afun ExD.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Ellipses",
"sec_num": "3.5."
},
{
"text": "+",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Ellipses",
"sec_num": "3.5."
},
{
"text": "As with anacolutha and gaps, the corresponding tokens in the UD conversion maintain the forms and lemmas as in the LLCT2 and bear the corresponding morphological and syntactic UD tags.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Ellipses",
"sec_num": "3.5."
},
{
"text": "As mentioned in Section 3., applying the UD annotation principles to the LLCT2 has a significant impact on the structures of PDT-style syntactic trees. In particular, since function words (corresponding to afuns starting with Aux, cf. Table 3 ) no longer mediate between content words (e. g. between a predicate and one of its oblique arguments), trees are tendentiously shallower in UD, as shown in Table 2 , and their depth is more stable: even if the median is the same, variance is nearly halved. This makes parsing these trees for particular syntactic constructions easier, quicker and less convoluted, especially with regard to more complex structures for which the PDT style poses a problem. This is in line with the operational objectives of the UD project, and is also confirmed by the different frequency of so-called non-projective constructions or discontinuities (Haji\u010dov\u00e1 et al., 2004) in UD when confronted with the PDT style.",
"cite_spans": [
{
"start": 876,
"end": 899,
"text": "(Haji\u010dov\u00e1 et al., 2004)",
"ref_id": "BIBREF9"
}
],
"ref_spans": [
{
"start": 235,
"end": 242,
"text": "Table 3",
"ref_id": "TABREF10"
},
{
"start": 400,
"end": 407,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Annotation Statistics",
"sec_num": "4."
},
{
"text": "Non-projectivities \u03bc M V \u03bc M V LLCT2-PDT",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Depth of trees",
"sec_num": null
},
{
"text": "5.47 5.0 6.16 2.35 2.0 3.72 LLCT2-UD 4.74 5.0 3.27 2.16 1.0 3.52 Table 2 : Means (\u03bc), medians (M) and variances (V) of tree depths (i. e. maximum number of nodes in a path from the root to a leaf node) and non-projective (i. e. discontinuous) arrangements of nodes for the PDT and UD versions of the",
"cite_spans": [],
"ref_spans": [
{
"start": 65,
"end": 72,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Depth of trees",
"sec_num": null
},
{
"text": "The data for non-projective arrangements in Table 2 are computed only relative to the trees where at least one such construction appears, which amount to 2765 in the PDT version of the LLCT2, against 2582 in the UD conversion. We note here that our current UD conversion consists of 9023 sentences as compared to the 9246 of the original LLCT2, because some had to be excluded for further analysis (cf. Section 3.1.1.). In any case we observe a slight independent decrease in non-projective occurrences (a positive note for the training of syntactic parsers), which again is in line with UD objectives. Their persistence, however, can be interpreted as an inherent presence of syntactic discontinuities in Latin in general, and in the formulaic language of the charters in particular.",
"cite_spans": [],
"ref_spans": [
{
"start": 44,
"end": 51,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "LLCT2.",
"sec_num": null
},
{
"text": "As discussed in Sections 3.1. and 3.4., constructions that in the original LLCT2 are mostly dealt with only with the two afuns APOS and ExD are interpreted in the UD by means of many more syntactic relations, expressed by a variety of deprels, as seen in Table 3 (where appositions are represented by the suffix _AP instead of APOS). The deprel parataxis, used when two clauses are joined with no apparent or explicit syntactic connection, is not particularly frequent, and thus not in the Table, but it is significant for the treatment of anacolutha and ellipses (see Sections 3.4. and 3.5.). Altogether, the UD standard seems more capable of describing the heterogenous array of non-standard constructions of the LLCT2 than the PDT style. Indeed, the relative lack of representative means within the latter had the underspecification of some afuns as a logical consequence.",
"cite_spans": [],
"ref_spans": [
{
"start": 255,
"end": 262,
"text": "Table 3",
"ref_id": "TABREF10"
},
{
"start": 490,
"end": 496,
"text": "Table,",
"ref_id": null
}
],
"eq_spans": [],
"section": "LLCT2.",
"sec_num": null
},
{
"text": "Lastly, Table 4 further highlights the greater representativeness of UD with respect to parts of speech. Notably, this conversion of the LLCT2 systematically introduces the determinants (DET), which in the PDT style are mostly labelled as adjectives (a) or pronouns (p), a fact which can also be inferred from the presence of the deprel det as a Table 4 : Correspondences between POS-tags in the PROIEL and UD styles; only the most frequent ( > 5% of corresponding occurrences) UD POS-tags are shown, together with their respective relative occurrences in %.",
"cite_spans": [],
"ref_spans": [
{
"start": 8,
"end": 15,
"text": "Table 4",
"ref_id": null
},
{
"start": 346,
"end": 353,
"text": "Table 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "LLCT2.",
"sec_num": null
},
{
"text": "Along with introducing a new Latin treebank annotated according to the Universal Dependencies framework, the present work focuses on the technical and linguistic challenges that are posed by an automated conversion process into UD from a different annotation standard.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "5."
},
{
"text": "Besides significantly expanding the coverage of Latin in UD, the distinctive features of the Late Latin of the charters, especially with respect to \"standardised\" Classical Latin, have brought forth the reassessment and revision of many aspects of the annotation strategy for Latin in general, of which some of the most relevant instances have been presented in Section 3. The consequences of these reinterpretations will hopefully expand beyond the present corpus, as our plan is to compile a set of new comprehensive guidelines for Latin annotation in UD with the ulterior aim of enriching extant and future Latin treebanks and making them more consistent.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "5."
},
{
"text": "As the conversion of the LLCT2 is a particularly complex process, we still consider it to be a work in progress, albeit in a very advanced state. The current version will need some more refinements before being officially included in the UD collection, probably as soon as in version 2.6 (scheduled for May 2020).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "5."
},
{
"text": "From: http://universaldependencies.org/ 2 For instance,(Ponti and Passarotti, 2016) shows a dramatic decrease of accuracy rates of a dependency parsing pipeline trained on the IT-TB when applied to texts of the Classical era taken from the LDT.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "The Latin variety used by Thomas Aquinas in the IT-TB, albeit Medieval, is quite formal and tends to follow the morphology and syntax of Classical Latin.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "https://www.perl.org/ 6 Treex is a modular software system in Perl for Natural Language Processing. It is described in(Popel and \u017dabokrtsk\u00fd, 2010) and is available online at http://ufal.mff.cuni. cz/treex. 7 See https://ufal.mff.cuni.cz/hamledt/news. 8 The conversion process sketched here closely follows the one used for the UD conversion of the IT-TB, which is described in(Cecchini et al., 2018). 9 See https://ufal.mff.cuni.cz/interset. For details on the tagset, see also https://wiki.ufal.ms. mff.cuni.cz/user:zeman:interset:features.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "This fact is corroborated by the observation that the second element can always be formally and functionally labelled as an ablative, like nomine or numero, and hence respect the criterion of case agreement between the two elements that are in apposition.19 We note that eo/ire 'to go' in its present passive infinitive form iri is used, together with a verb's supine in -um (bearing a final valence), similarly to an auxiliary in the periphrastic construction of the future passive infinitive, e. g. ductum iri 'to [be going to] be conducted'(Barbieri, 1995, \u00a7163),(Greenough and Allen, 2006, \u00a7509). However, this use is marginal with respect to the sheer mass of formations using sum/esse. 20 Although it might be argued that sum/esse is used as a copula here.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "For example, in Italian there still is agreement in e. g. lui \u00e8 venuto / lei \u00e8 venuta 'he/she has come' (it. \u00e8 < lat. est 'he/she/it is'), but it is no longer present in e. g. ho mangiato la mela 'I ate the apple' (it. ho < lat. habeo 'I have'), where mela 'apple' is feminine, but -o is the masculine ending.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "Until the current, i. e. 2.5, version, UD does not allow a closer specification of conjunction types, such as \"relative\".",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "The design and use of a Latin dependency treebank",
"authors": [
{
"first": "D",
"middle": [],
"last": "Bamman",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Crane",
"suffix": ""
}
],
"year": 2006,
"venue": "Proceedings of the Fifth Workshop on Treebanks and Linguistic Theories (TLT 2006)",
"volume": "",
"issue": "",
"pages": "67--78",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bamman, D. and Crane, G. (2006). The design and use of a Latin dependency treebank. In Proceedings of the Fifth Workshop on Treebanks and Linguistic Theories (TLT 2006), pages 67-78, Prague, Czech Republic. Univerzita Karlova.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Guidelines for the Syntactic Annotation of Latin Treebanks (v. 1.3). Tufts Published Scholarship. Tufts University's Digital Collections and Archives",
"authors": [
{
"first": "D",
"middle": [],
"last": "Bamman",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Crane",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Passarotti",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Raynaud",
"suffix": ""
}
],
"year": 2007,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bamman, D., Crane, G., Passarotti, M., and Raynaud, S. (2007). Guidelines for the Syntactic Annotation of Latin Treebanks (v. 1.3). Tufts Published Scholarship. Tufts University's Digital Collections and Archives, Medford, MA, USA. Permanent URL: http://hdl.handle. net/10427/42683.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Nuovo corso di lingua latina",
"authors": [
{
"first": "G",
"middle": [],
"last": "Barbieri",
"suffix": ""
}
],
"year": 1995,
"venue": "Loescher Editore",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Barbieri, G. (1995). Nuovo corso di lingua latina. Loescher Editore, Turin, Italy.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "CoNLL-X shared task on multilingual dependency parsing",
"authors": [
{
"first": "S",
"middle": [],
"last": "Buchholz",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Marsi",
"suffix": ""
}
],
"year": 2006,
"venue": "Proceedings of the Tenth Conference on Computational Natural Language Learning (CoNLL-X)",
"volume": "",
"issue": "",
"pages": "149--164",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Buchholz, S. and Marsi, E. (2006). CoNLL-X shared task on multilingual dependency parsing. In Llu\u00eds M\u00e0rquez et al., editors, Proceedings of the Tenth Conference on Computational Natural Language Learning (CoNLL-X), pages 149-164, New York, NY, USA, June. Association for Computational Linguistics (ACL).",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Challenges in Converting the Index Thomisticus Treebank into Universal Dependencies",
"authors": [
{
"first": "F",
"middle": [
"M"
],
"last": "Cecchini",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Passarotti",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Marongiu",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2018,
"venue": "Special Interest Group on linguistic DATa and corpus-based approaches to NLP (SIGDAT), ACL",
"volume": "",
"issue": "",
"pages": "27--36",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Cecchini, F. M., Passarotti, M., Marongiu, P., and Zeman, D. (2018). Challenges in Converting the Index Thomisti- cus Treebank into Universal Dependencies. In Proceed- ings of the Second Workshop on Universal Dependencies (UDW 2018) at EMNLP 2018, pages 27-36, Bruxelles, Belgium. Special Interest Group on linguistic DATa and corpus-based approaches to NLP (SIGDAT), ACL.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Verba Bestiae: How Latin Conquered Heavy Metal",
"authors": [
{
"first": "F",
"middle": [
"M"
],
"last": "Cecchini",
"suffix": ""
},
{
"first": "G",
"middle": [
"H"
],
"last": "Franzini",
"suffix": ""
},
{
"first": "M",
"middle": [
"C"
],
"last": "Passarotti",
"suffix": ""
}
],
"year": null,
"venue": "Multilingual Metal: Sociocultural, Linguistic and Literary Perspectives on Heavy Metal Lyrics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Cecchini, F. M., Franzini, G. H., and Passarotti, M. C. (forthcoming). Verba Bestiae: How Latin Conquered Heavy Metal. In Riitta Valij\u00e4rvi, et al., editors, Mul- tilingual Metal: Sociocultural, Linguistic and Literary Perspectives on Heavy Metal Lyrics, Emerald Studies in Metal Music and Culture. Emerald group publishing, Bingley, UK.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Allen and Greenough's new Latin grammar",
"authors": [
{
"first": "J",
"middle": [
"B"
],
"last": "Greenough",
"suffix": ""
},
{
"first": "J",
"middle": [
"H"
],
"last": "Allen",
"suffix": ""
}
],
"year": 2006,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Greenough, J. B. and Allen, J. H. (2006). Allen and Gree- nough's new Latin grammar. Dover publications, Mine- ola, NY, USA.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "A Unified Theory of 'Standard' and 'Transparent' Free Relatives",
"authors": [
{
"first": "A",
"middle": [],
"last": "Grosu",
"suffix": ""
}
],
"year": 2003,
"venue": "Natural Language & Linguistic Theory",
"volume": "21",
"issue": "",
"pages": "247--331",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Grosu, A. (2003). A Unified Theory of 'Standard' and 'Transparent' Free Relatives. Natural Language & Lin- guistic Theory, 21(2):247-331, May.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Prague Dependency Treebank",
"authors": [
{
"first": "J",
"middle": [],
"last": "Haji\u010d",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Haji\u010dov\u00e1",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Mikulov\u00e1",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "M\u00edrovsk\u00fd",
"suffix": ""
}
],
"year": 2017,
"venue": "Handbook of Linguistic Annotation",
"volume": "",
"issue": "",
"pages": "555--594",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Haji\u010d, J., Haji\u010dov\u00e1, E., Mikulov\u00e1, M., and M\u00edrovsk\u00fd, J. (2017). Prague Dependency Treebank. In Nancy Ide et al., editors, Handbook of Linguistic Annotation, pages 555-594. Springer, Dordrecht, Netherlands.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Issues of Projectivity in the Prague Dependency Treebank",
"authors": [
{
"first": "E",
"middle": [],
"last": "Haji\u010dov\u00e1",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Havelka",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Sgall",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Vesel\u00e1",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2004,
"venue": "Prague Bulletin of Mathematical Linguistics",
"volume": "81",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Haji\u010dov\u00e1, E., Havelka, J., Sgall, P., Vesel\u00e1, K., and Ze- man, D. (2004). Issues of Projectivity in the Prague De- pendency Treebank. Prague Bulletin of Mathematical Linguistics, 81. Available at https://ufal.mff. cuni.cz/pbml/81/hajicova-et-al.pdf.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Prague Markup Language Framework",
"authors": [
{
"first": "J",
"middle": [],
"last": "Hana",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "\u0160t\u011bp\u00e1nek",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the Sixth Linguistic Annotation Workshop",
"volume": "",
"issue": "",
"pages": "12--21",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Hana, J. and \u0160t\u011bp\u00e1nek, J. (2012). Prague Markup Lan- guage Framework. In Nancy Ide et al., editors, Proceed- ings of the Sixth Linguistic Annotation Workshop, pages 12-21, Jeju, Republic of Korea, July. Association for Computational Linguistics (ACL).",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Creating a parallel treebank of the old Indo-European Bible translations",
"authors": [
{
"first": "D",
"middle": [
"T T"
],
"last": "Haug",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "J\u00f8hndal",
"suffix": ""
}
],
"year": 2008,
"venue": "Proceedings of the Second Workshop on Language Technology for Cultural Heritage Data",
"volume": "",
"issue": "",
"pages": "27--34",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Haug, D. T. T. and J\u00f8hndal, M. (2008). Creating a par- allel treebank of the old Indo-European Bible transla- tions. In Caroline Sporleder et al., editors, Proceed- ings of the Second Workshop on Language Technology for Cultural Heritage Data (LaTeCH 2008), pages 27- 34, Marrakesh, Morocco. European Language Resources Association (ELRA).",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "La sintassi delle espressioni nominali latine",
"authors": [
{
"first": "R",
"middle": [],
"last": "Iovino",
"suffix": ""
}
],
"year": 2012,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Iovino, R. (2012). La sintassi delle espressioni nomi- nali latine. Ph.D. thesis, Universit\u00e0 Ca' Foscari, Venice, Italy.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Abbreviations, fragmentary words, formulaic language: Treebanking medieval charter material",
"authors": [
{
"first": "T",
"middle": [],
"last": "Korkiakangas",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Lassila",
"suffix": ""
}
],
"year": 2013,
"venue": "Proceedings of The Third Workshop on Annotation of Corpora for Research in the Humanities (ACRH-3)",
"volume": "",
"issue": "",
"pages": "61--72",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Korkiakangas, T. and Lassila, M. (2013). Abbreviations, fragmentary words, formulaic language: Treebanking medieval charter material. In Francesco Mambrini, et al., editors, Proceedings of The Third Workshop on Annota- tion of Corpora for Research in the Humanities (ACRH- 3), pages 61-72, Sofia, Bulgaria, December. Bulgarian Academy of Sciences.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Challenges in Annotating Medieval Latin Charters",
"authors": [
{
"first": "T",
"middle": [],
"last": "Korkiakangas",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Passarotti",
"suffix": ""
}
],
"year": 2011,
"venue": "Annotation of Corpora for Research in the Humanities: Proceedings of the ACRH Workshop",
"volume": "26",
"issue": "",
"pages": "105--116",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Korkiakangas, T. and Passarotti, M. (2011). Challenges in Annotating Medieval Latin Charters. Journal for Language Technology and Computational Linguistics (JLCL), 26(2):105-116, november. In Francesco Mam- brini, et al., editors, Annotation of Corpora for Research in the Humanities: Proceedings of the ACRH Workshop, Heidelberg, 5. Jan. 2012.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Adverbial Subordination, volume 18 of Empirical Approaches to Language Typology. Mouton de Gruyter",
"authors": [
{
"first": "B",
"middle": [],
"last": "Kortmann",
"suffix": ""
}
],
"year": 1997,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kortmann, B. (1997). Adverbial Subordination, volume 18 of Empirical Approaches to Language Typology. Mou- ton de Gruyter, Berlin, Germany -New York, NY, USA.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "\u00c0 propos du concept d'\u00ab apposition \u00bb : les constructions rex Ancus et urbs Roma. L'Information Grammaticale",
"authors": [
{
"first": "D",
"middle": [],
"last": "Longr\u00e9e",
"suffix": ""
}
],
"year": 1990,
"venue": "",
"volume": "45",
"issue": "",
"pages": "8--13",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Longr\u00e9e, D. (1990). \u00c0 propos du concept d'\u00ab apposi- tion \u00bb : les constructions rex Ancus et urbs Roma. L'Information Grammaticale, 45:8-13.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "Universal Dependencies v1: A multilingual treebank collection",
"authors": [],
"year": null,
"venue": "Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016)",
"volume": "",
"issue": "",
"pages": "1659--1666",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Universal Dependencies v1: A multilingual treebank col- lection. In Proceedings of the Tenth International Con- ference on Language Resources and Evaluation (LREC 2016), pages 1659-1666, Portoro\u017e, Slovenia. European Language Resources Association (ELRA).",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "The Project of the Index Thomisticus Treebank",
"authors": [
{
"first": "L",
"middle": [
"R"
],
"last": "Palmer",
"suffix": ""
},
{
"first": "O",
"middle": [
"K"
],
"last": "Norman",
"suffix": ""
},
{
"first": "Usa",
"middle": [],
"last": "Reprint",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Passarotti",
"suffix": ""
}
],
"year": 1988,
"venue": "Age of Access? Grundfragen der Informationsgesellschaft",
"volume": "10",
"issue": "",
"pages": "299--320",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Palmer, L. R. (1988). The Latin language. Oklahoma Uni- versity Press, Norman, OK, USA. Reprint. Passarotti, M. (2019). The Project of the Index Thomisticus Treebank. In Monica Berti, editor, Dig- ital Classical Philology, volume 10 of Age of Ac- cess? Grundfragen der Informationsgesellschaft, pages 299-320. De Gruyter Saur, Munich, Germany, August. Open access at https://www.degruyter.com/ viewbooktoc/product/502894.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "A Universal Part-of-Speech Tagset",
"authors": [
{
"first": "S",
"middle": [],
"last": "Petrov",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Das",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Mcdonald",
"suffix": ""
}
],
"year": 2011,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {
"arXiv": [
"arXiv:1104.2086"
]
},
"num": null,
"urls": [],
"raw_text": "Petrov, S., Das, D., and McDonald, R. (2011). A Universal Part-of-Speech Tagset. ArXiv e-prints. arXiv:1104.2086 at https://arxiv.org/abs/1104.2086.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Differentia compositionem facit. A slower-paced and reliable parser for Latin",
"authors": [
{
"first": "E",
"middle": [
"M"
],
"last": "Ponti",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Passarotti",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016)",
"volume": "",
"issue": "",
"pages": "683--688",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ponti, E. M. and Passarotti, M. (2016). Differentia com- positionem facit. A slower-paced and reliable parser for Latin. In Proceedings of the Tenth International Con- ference on Language Resources and Evaluation (LREC 2016), pages 683-688, Portoro\u017e, Slovenia. European Language Resources Association (ELRA).",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "TectoMT: modular NLP framework",
"authors": [
{
"first": "M",
"middle": [],
"last": "Popel",
"suffix": ""
},
{
"first": "Z",
"middle": [],
"last": "\u017dabokrtsk\u00fd",
"suffix": ""
}
],
"year": 2010,
"venue": "Advances in Natural Language Processing",
"volume": "6233",
"issue": "",
"pages": "293--304",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Popel, M. and \u017dabokrtsk\u00fd, Z. (2010). TectoMT: modular NLP framework. In Hrafn Loftsson, et al., editors, Ad- vances in Natural Language Processing, volume 6233 of Lecture Notes in Computer Science, pages 293-304, Berlin -Heidelberg, Germany, August. Springer.",
"links": null
},
"BIBREF23": {
"ref_id": "b23",
"title": "HamleDT 2.0: Thirty Dependency Treebanks Stanfordized",
"authors": [
{
"first": "R",
"middle": [],
"last": "Rosa",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Ma\u0161ek",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Mare\u010dek",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Popel",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Zeman",
"suffix": ""
},
{
"first": "Z",
"middle": [],
"last": "\u017dabokrtsk\u00fd",
"suffix": ""
}
],
"year": 2014,
"venue": "Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC 2014)",
"volume": "",
"issue": "",
"pages": "2334--2341",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Rosa, R., Ma\u0161ek, J., Mare\u010dek, D., Popel, M., Zeman, D., and \u017dabokrtsk\u00fd, Z. (2014). HamleDT 2.0: Thirty De- pendency Treebanks Stanfordized. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC 2014), pages 2334-2341, Reyk- jav\u00edk, Iceland, may. European Language Resources As- sociation (ELRA).",
"links": null
},
"BIBREF24": {
"ref_id": "b24",
"title": "The Noun Phrase in Classical Latin Prose",
"authors": [
{
"first": "O",
"middle": [],
"last": "Spevak",
"suffix": ""
}
],
"year": 2014,
"venue": "Amsterdam Studies in Classical Philology",
"volume": "21",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Spevak, O. (2014). The Noun Phrase in Classical Latin Prose, volume 21 of Amsterdam Studies in Classical Philology. Brill, Leiden, Netherlands -Boston, MA, USA.",
"links": null
},
"BIBREF25": {
"ref_id": "b25",
"title": "The Composition and Use of the Universal Morphological Feature Schema (Uni-Morph Schema)",
"authors": [
{
"first": "J",
"middle": [],
"last": "Sylak-Glassman",
"suffix": ""
}
],
"year": 2016,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sylak-Glassman, J., (2016). The Composition and Use of the Universal Morphological Feature Schema (Uni- Morph Schema). Johns Hopkins University, June. Work- ing draft, v. 2.",
"links": null
},
"BIBREF26": {
"ref_id": "b26",
"title": "Introduction au latin vulgaire. Librairie Klincksieck -S\u00e9rie linguistique",
"authors": [
{
"first": "V",
"middle": [],
"last": "V\u00e4\u00e4n\u00e4nen",
"suffix": ""
}
],
"year": 2012,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "V\u00e4\u00e4n\u00e4nen, V. (2012). Introduction au latin vulgaire. Li- brairie Klincksieck -S\u00e9rie linguistique. Klincksieck, Paris, France. Reprint.",
"links": null
},
"BIBREF27": {
"ref_id": "b27",
"title": "Il latino. Introduzioni (Linguistica)",
"authors": [
{
"first": "E",
"middle": [],
"last": "Vineis",
"suffix": ""
}
],
"year": 1993,
"venue": "Il Mulino",
"volume": "",
"issue": "",
"pages": "289--348",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Vineis, E. (2005). Il latino. Introduzioni (Linguistica). Il Mulino, Bologna, Italy. Reprint from: Anna Giacalone Ramat et al., editors. (1993). Le lingue indoeuropee, pages 289-348. Il Mulino, Bologna, Italy.",
"links": null
},
"BIBREF28": {
"ref_id": "b28",
"title": "Alle origini dell'italiano",
"authors": [
{
"first": "A",
"middle": [],
"last": "Zamboni",
"suffix": ""
}
],
"year": 2007,
"venue": "Universit\u00e0 (Linguistica). Carocci",
"volume": "213",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zamboni, A. (2007). Alle origini dell'italiano, volume 213 of Universit\u00e0 (Linguistica). Carocci, Rome, Italy, sec- ond edition.",
"links": null
},
"BIBREF29": {
"ref_id": "b29",
"title": "Ham-leDT: To Parse or Not to Parse?",
"authors": [
{
"first": "D",
"middle": [],
"last": "Zeman",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Mare\u010dek",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Popel",
"suffix": ""
},
{
"first": "L",
"middle": [],
"last": "Ramasamy",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "\u0160t\u011bp\u00e1nek",
"suffix": ""
},
{
"first": "Z",
"middle": [],
"last": "\u017dabokrtsk\u00fd",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Haji\u010d",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC 2012)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zeman, D., Mare\u010dek, D., Popel, M., Ramasamy, L., , \u0160t\u011bp\u00e1nek, J., \u017dabokrtsk\u00fd, Z., and Haji\u010d, J. (2012). Ham- leDT: To Parse or Not to Parse? In Nicoletta Calzo- lari, et al., editors, Proceedings of the Eight Interna- tional Conference on Language Resources and Evalua- tion (LREC 2012), Istanbul, Turkey, May. European Lan- guage Resources Association (ELRA).",
"links": null
},
"BIBREF30": {
"ref_id": "b30",
"title": "HamleDT: Harmonized Multi-Language Dependency Treebank",
"authors": [
{
"first": "D",
"middle": [],
"last": "Zeman",
"suffix": ""
},
{
"first": "O",
"middle": [],
"last": "Du\u0161ek",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Mare\u010dek",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Popel",
"suffix": ""
},
{
"first": "L",
"middle": [
"R"
],
"last": "Ramasamy",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "\u0160t\u011bp\u00e1nek",
"suffix": ""
},
{
"first": "Z",
"middle": [],
"last": "\u017dabokrtsk\u00fd",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Haji\u010d",
"suffix": ""
}
],
"year": 2014,
"venue": "Language Resources and Evaluation",
"volume": "48",
"issue": "4",
"pages": "601--637",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zeman, D., Du\u0161ek, O., Mare\u010dek, D., Popel, M., Ra- masamy, L. R., \u0160t\u011bp\u00e1nek, J., \u017dabokrtsk\u00fd, Z., and Haji\u010d, J. (2014). HamleDT: Harmonized Multi-Language Depen- dency Treebank. Language Resources and Evaluation, 48(4):601-637.",
"links": null
},
"BIBREF31": {
"ref_id": "b31",
"title": "Reusable Tagset Conversion Using Tagset Drivers",
"authors": [
{
"first": "D",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2008,
"venue": "Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC 2008)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zeman, D. (2008). Reusable Tagset Conversion Us- ing Tagset Drivers. In Nicoletta Calzolari, et al., editors, Proceedings of the Sixth International Con- ference on Language Resources and Evaluation (LREC 2008), Marrakesh, Morocco, May. European Language Resources Association (ELRA). Pub- lished as CD. Available at http://ufal.mff. cuni.cz/~zeman/publikace/2008-02/ tagdrivers-marrakech-styl-lrec.pdf.",
"links": null
},
"BIBREF32": {
"ref_id": "b32",
"title": "The world of tokens, tags and trees",
"authors": [
{
"first": "D",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2018,
"venue": "\u00dastav form\u00e1ln\u00ed a aplikovan\u00e9 lingvistiky (\u00daFAL)",
"volume": "19",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zeman, D. (2018). The world of tokens, tags and trees, volume 19 of Studies in Computational and Theoreti- cal Linguistics. \u00dastav form\u00e1ln\u00ed a aplikovan\u00e9 lingvistiky (\u00daFAL), Prague, Czech Republic, first edition.",
"links": null
}
},
"ref_entries": {
"FIGREF1": {
"text": "For example, in s31, manifestu sum ego Aufuso presbitero filio quondam Gualfridi . . . offero Deo et tibi ecclesia sancti Fridiani . . . 'I, Aufusus priest, son of the late Gualfridus, have shown up here . . . I offer to God and you, the church of St. Fridianus, . . . ',",
"type_str": "figure",
"uris": null,
"num": null
},
"TABREF1": {
"text": "constructions of the LLCT2, the two or more juxtaposed items can be rather distant from each other: in s560 omnem rebus substantie mee . . .[23 intervening words] . . . tam coltum quam et desertum, the quality of rebus 'things, property' is specified by tam coltum quam et desertum 'both cultivated and deserted land' after lengthy subordinate clauses.Another peculiarity of the LLCT2 is that APOS is also used with reprisals of noun phrases. This phenomenon is sometimes caused by an interruption of the writer's train of thought, or by his wish to ensure that a specific noun phrase remains topical, i. e. under discussion, as is the case in s20 portionem de omne res eius, quot est sexta portionem de ex omnibus rebus meis . . . [53 intervening words] . . . de omnia, ut dicxi, sexta portionem . . . sit . . . '(a) part of all his property, which is the sixth part of all my property . . . as I said, the sixth part of everything . . . be . . . '.",
"content": "<table><tr><td/><td/><td/><td>APOS</td><td/><td>PRED</td><td/></tr><tr><td>OBJ_AP</td><td colspan=\"2\">OBJ_AP</td><td>ATR</td><td/><td/><td/></tr><tr><td>praetium</td><td>,</td><td>soledos</td><td>auri</td><td>. . .</td><td>recepi</td><td>. . .</td></tr><tr><td colspan=\"2\">In the appositive</td><td/><td/><td/><td/><td/></tr></table>",
"type_str": "table",
"html": null,
"num": null
},
"TABREF2": {
"text": "de omnes res . . . sexta portionem . . . sit",
"content": "<table><tr><td>dislocated:nsubj</td><td/><td/></tr><tr><td>nmod</td><td/><td/></tr><tr><td>case</td><td>nsubj</td><td>PRED</td></tr><tr><td>det</td><td>amod</td><td/></tr></table>",
"type_str": "table",
"html": null,
"num": null
},
"TABREF4": {
"text": "",
"content": "<table/>",
"type_str": "table",
"html": null,
"num": null
},
"TABREF7": {
"text": "). Another example is (s20) volo ut . . . filius meus Rachiprandu . . . , ut portionem de omne res eius . . . ut sit in potestatem Deusdedi presbiteri 'I want that . . . my son Rachiprandus . . . , that a part of all his possessions . . . that it be in the possession of Deusdedit priest',",
"content": "<table/>",
"type_str": "table",
"html": null,
"num": null
},
"TABREF9": {
"text": "PDT-style afunsUD deprels ADV obl, advmod, advcl, conj, nsubj APOS punct ATR nmod, det, appos, amod, acl, conj",
"content": "<table><tr><td>ATV</td><td>advcl, xcomp</td></tr><tr><td>AtvV</td><td>xcomp, conj</td></tr><tr><td>AuxC</td><td>mark</td></tr><tr><td>AuxP</td><td>case</td></tr><tr><td>AuxV</td><td>aux</td></tr><tr><td>AuxY</td><td>cc, case, fixed, mark</td></tr><tr><td>AuxZ</td><td>advmod</td></tr><tr><td>COORD</td><td>cc, punct</td></tr><tr><td>ExD</td><td>root, fixed, orphan, reparandum</td></tr><tr><td>OBJ</td><td>obl, obj, conj, xcomp, ccomp</td></tr><tr><td>OCOMP</td><td>xcomp, conj</td></tr><tr><td>PRED</td><td>root, conj, cop</td></tr><tr><td>Pnom</td><td>xcomp, acl, root, conj</td></tr><tr><td>ROOT</td><td>punct</td></tr><tr><td>SBJ</td><td>nsubj, conj</td></tr><tr><td>_AP</td><td>dislocated, obj, cc, conj</td></tr><tr><td>_CO</td><td>conj, obl</td></tr></table>",
"type_str": "table",
"html": null,
"num": null
},
"TABREF10": {
"text": "Correspondences between afuns and deprels: only the most frequent ( > 5% of corresponding occurrences) deprels are shown. Deprels are ordered for relative frequency, but percentages are not shown for visual clarity. possible interpretation of the universal afun ATR in Table 3. 50.85 , DET: 43.43 , NUM: 5.71 c CCONJ: 78.77 , SCONJ: 20.44 d ADV: 86.12 , PART: 6.64 , SCONJ: 5.83",
"content": "<table><tr><td>PROIEL</td><td>UD</td></tr><tr><td>POS-tags</td><td>POS-tags</td></tr><tr><td colspan=\"2\">a ADJ: e INTJ: 100</td></tr><tr><td>m</td><td>NUM: 100</td></tr><tr><td>n</td><td>NOUN: 99.93</td></tr><tr><td>p</td><td>PRON: 65.80 , DET: 33.70</td></tr><tr><td>Propn</td><td>PROPN: 100.0</td></tr><tr><td>Punc</td><td>PUNCT: 99.20</td></tr><tr><td>r</td><td>ADP: 100</td></tr><tr><td>t</td><td>VERB: 100</td></tr><tr><td>v</td><td>VERB: 87.56 , AUX: 12.44</td></tr></table>",
"type_str": "table",
"html": null,
"num": null
}
}
}
}