ArticlePDF Available

LL(O)D and NLP perspectives on semantic change for humanities research

Authors:

Abstract and Figures

This paper presents an overview of the LL(O)D and NLP methods, tools and data for detecting and representing semantic change, with its main application in humanities research. The paper’s aim is to provide the starting point for the construction of a workflow and set of multilingual diachronic ontologies within the humanities use case of the COST Action Nexus Linguarum, European network for Web-centred linguistic data science, CA18209. The survey focuses on the essential aspects needed to understand the current trends and to build applications in this area of study.
Content may be subject to copyright.
Semantic Web 0 (0) 1 1
IOS Press
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
LL(O)D and NLP Perspectives on Semantic
Change for Humanities Research
Florentina Armaselu a,*, Elena-Simona Apostol b, Anas Fahad Khanc, Chaya Liebeskind d,
Barbara McGillivray e,f, Ciprian-Octavian Truic˘
ag, Andrius Utka h, Giedr˙
e Val¯
unait˙
e Oleškeviˇ
cien˙
ei
Marieke van Erpj
aLuxembourg Centre for Contemporary and Digital History (C2DH), University or Luxembourg, Luxembourg
E-mail: florentina.armaselu@uni.lu
bComputer Science and Engineering Department, Faculty of Automatic Control and Computers, University
Politehnica of Bucharest, Romania
E-mail: elena.apostol@upb.ro
cIstituto di Linguistica Computazionale «A. Zampolli», Consiglio Nazionale delle Ricerche, Italy
E-mail: fahad.khan@ilc.cnr.it
dDepartment of Computer Science, Jerusalem College of Technology, Jerusalem, Israel
E-mail: liebchaya@gmail.com
eDepartment of Digital Humanities, King’s College London, United Kingdom
E-mail: barbara.mcgillivray@kcl.ac.uk
fThe Alan Turing Institute, United Kingdom
E-mail: bmcgillivray@turing.ac.uk
gComputer Science and Engineering Department, Faculty of Automatic Control and Computers, University
Politehnica of Bucharest, Romania
E-mail: ciprian.truica@upb.ro
hCentre of Computational Linguistics, Vytautas Magnus University, Kaunas, Lithuania
E-mail: andrius.utka@vdu.lt
iInstitute of Humanities, Mykolas Romeris University, Vilnius, Lithuania
E-mail: gvalunaite@mruni.eu
jDHLab, KNAW Humanities Cluster, Amsterdam, Netherlands
E-mail: marieke.van.erp@dh.huc.knaw.nl
Author contributions: F.A., sections 1, 2, 3, 5, 6, 8; E.S.A., section 5; A.F.K., section 4; C.L., section 5;
B.M., section 5; C.O.T., section 5; A.U., section 5; G.V.O., section 3; M.V.E., section 7. All the authors
critically revised and approved the final version of the manuscript submitted to the Journal.
Editor: Philipp Cimiano, Bielefeld University, Germany
Solicited reviews: Enrico Daga, The Open University, United Kingdom; Julia Bosque-Gil, University of Zaragoza, Spain; Thierry Declerck,
German Research Center for Artificial Intelligence, Germany
Abstract. This paper presents an overview of the LL(O)D and NLP methods, tools and data for detecting and representing
semantic change, with its main application in humanities research. The paper’s aim is to provide the starting point for the
construction of a workflow and set of multilingual diachronic ontologies within the humanities use case of the COST Action
Nexus Linguarum, European network for Web-centred linguistic data science, CA18209. The survey focuses on the essential
aspects needed to understand the current trends and to build applications in this area of study.
Keywords: linguistic linked open data, natural language processing, semantic change, ontologies, humanities
1570-0844/0-1900/$35.00 © 0 – IOS Press and the authors. All rights reserved
2F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
1. Introduction
Detecting semantic change in diachronic corpora
and representing the change of concepts over time
as linked data is a core challenge at the intersec-
tion between digital humanities (DH) and Seman-
tic Web (SW). Semantic Web technologies have al-
ready been used successfully in humanistic initiatives
such as the Mapping the Manuscripts project [1] and
in Pelagios [2]. They facilitate the creation, publi-
cation and interlinking of FAIR (Findable, Accessi-
ble, Interoperable and Reusable) datasets [3]. In par-
ticular, using a common data model, common for-
malisms and common vocabularies in linked data helps
to render datasets more interoperable; the use of read-
ily available technologies such as the query language
SPARQL also makes such data more (re-)usable. Se-
mantic change data can be highly heterogeneous and
potentially include linguistic, historical, bibliographic
and geographical information. The linked data model
is well suited to handling this. For instance, the lex-
ical aspect of semantic change data is already served
by the existing OntoLex-Lemon vocabulary and its ex-
tensions, and there are also numerous vocabularies and
datasets dealing with bibliographic metadata, histori-
cal time periods and geographic locations. In addition,
the Web Ontology Language (OWL) and associated
reasoning tools allow for basic ontological reasoning
to be carried out on such data, which is useful for deal-
ing with different classes of entities referred to by word
senses.
Although significant advances in the development
of natural language processing (NLP) methods and
tools for extracting historical entities and modelling di-
achronic linked data, as well as in the field of Linguis-
tic Linked (Open) Data (LL(O)D), 1have been made
so far [4–6], there is a need for a systematic overview
of this growing area of investigation. Some literature
surveys and overview papers on the state of the art in
lexical semantic change detection in NLP exist (e.g.
[7–10]), but none addresses the intersection of this line
of research with LL(O)D research. In particular, previ-
ous work has generally tended to focus on how to de-
tect semantic change (in both corpora, e.g., [11], and
linked data ontologies, e.g., [12]) but has generally not
*Corresponding author. E-mail: florentina.armaselu@uni.lu.
1We have added parentheses around the word ‘open’ because al-
though the focus is often on linked data, and in our case linguistic
linked data, that has been published with an open license, this is not
always the case and linked data may have other types of license.
provided an in-depth look at how to model and publish
semantic change datasets in Linked Open Data (LOD)
that result, at least in part, from these detection meth-
ods. 2
The contribution of this paper is a literature sur-
vey intended to consider these areas together. We posit
that to better contextualise and target the combina-
tion of NLP and LL(O)D techniques for detecting and
representing semantic change, the main workflow im-
plied in the process should be taken into account. The
term semantic change is used as generally referring
to a change in meaning, either of a lexical unit (word
or expression) or of a concept (a complex knowledge
structure that can encompass one or more lexical units
as well as relations among them and with other con-
cepts). Semantic change and other related terms, such
as semantic shift,semantic drift,concept drift,concept
shift,concept split, are also introduced and explained.
The current study is developed as part of the use
case in the humanities (UC4.2.1) carried out within the
COST Action European network for Web-centred lin-
guistic data science (Nexus Linguarum), CA18209. 3
The goal of the use case is to create a workflow for
the detection of semantic change in multilingual di-
achronic corpora from the humanities domain, and the
representation of the evolution of parallel concepts, de-
rived from these corpora as LLOD. The intended out-
come of UC4.2.1 is a set of diachronic ontologies in
several languages and methodological guidelines for
generating and publishing this type of knowledge us-
ing NLP and Semantic Web technologies.
The paper is organised in eight sections describing
the survey methodology and the state-of-the art in data,
tools, and methods for NLP and LL(O)D resources
that we deem important to a workflow designed for the
diachronic analysis and ontological representation of
concept evolution. Our main focus is concept change
for humanities research, which involves investigations
and data that include a time dimension, but the con-
cepts may also apply to other domains. The various
sections will focus on the essential aspects needed to
understand the current trends and to build applications
for detecting and representing semantic change. The
remainder of this paper is organised as follows. Sec-
tion 2 presents the methodology applied to build the
survey. Section 3 discusses existing theoretical frame-
works for tracing different types of semantic change.
2One exception is [13].
3https://nexuslinguarum.eu/.
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 3
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
Section 4 presents current LL(O)D formalisms (e.g.
RDF, OntoLex-Lemon, OWL-Time) and models for
representing diachronic relations. Section 5 is dedi-
cated to existing methods and NLP tools for the explo-
ration and detection of semantic change in large sets
of data, e.g. diachronic word embeddings, named en-
tity recognition (NER) and topic modelling. Section 6
presents an overview of methods and NLP tools for
(semi-) automatic generation of (diachronic) ontolog-
ical structures from text corpora. Section 7 provides
an overview of the main diachronic LL(O)D reposito-
ries from the humanities domain, with particular atten-
tion to collections in various languages, and emerging
trends in publishing ontologies representing semantic
change as LL(O)D data. The paper is concluded by
Section 8 where we discuss our findings and future di-
rections.
2. Survey methodology
The motivation of combining DH approaches with
semantic technologies is mainly related to the target
audiences of the survey. That is, researchers, students,
teachers interested in detecting how concepts in a cer-
tain domain evolve and how this evolution can be
represented via semantic Web and linked data tech-
nologies that support the production and dissemina-
tion of FAIR data on the Web. Therefore, the paper
addresses the study of semantic change and creation
of diachronic ontologies in connection with areas in
the humanities such as the history of concepts and
history of ideas on the one side, and linguistics on
the other. This topic may be of potential interest to
other researchers interested in semantic change detec-
tion within a particular domain and its modelling as
linked data. Scholars in Semantic Web technologies
may be interested in such areas of application and fur-
ther development of the linked data paradigm and the
possibilities of integrating diachronic representation of
data from the humanities into the LL(O)D cloud in the
future.
The scope of the paper covers diachronic corpora
that may span more distant or more recent periods in
time. Therefore, the article focuses on studies deal-
ing with diachronic variation, that is change over time,
but not with synchronic variation, which can refer, for
instance, to variation across genre (or register), class,
gender or other social category [14], within a given,
more limited period of time. The survey also targets the
construction of diachronic ontologies that, unlike syn-
Fig. 1. Generic workflow and related sections
chronic ontologies ignoring the historical perspective,
allow us to capture the temporal dimension of concepts
and investigate gradual semantic changes and concept
evolution through time [15].
As mentioned above, the survey follows a work-
flow for detecting and representing semantic change
as LL(O)D ontologies, based on diachronic corpora.
Figure 1 illustrates the main building blocks of such a
workflow and the possible interconnections among the
various areas of research considered relevant for the
study. Each block can be mapped onto one of the sub-
sequent sections (referred to as S3 - S7, in Fig. 1). It
should be noted that not all relationships displayed in
the figure are explicitly expressed in the surveyed liter-
ature. Some of them represent work in progress or pro-
jections of possible developments implied by the in-
tended workflow. For instance, we consider that theo-
retical modelling of semantic change in diachronic cor-
pora can play an important role in designing the fol-
lowing steps in the workflow, such as LL(O)D mod-
elling, detection of lexical semantic change and ontol-
ogy generation, and thus, a survey of this area is worth
investigating together with the other blocks. More-
over, approaches from the domain of lexical seman-
tic change detection may inform and potentially bring
about new perspectives on learning or generating (di-
achronic) ontologies from unstructured texts, which in
turn, connects with existing or future means of pub-
lishing such ontologies in the LL(O)D cloud.
Our methodology consisted of three phases: (1)
selecting or searching for (recent) surveys or refer-
ence works in areas related to the five blocks depicted
in Fig. 1; (2) expanding the set by considering rel-
evant references cited in the works collected during
the previous phase; (3) refining the structure of the
covered areas and corresponding sections and sub-
sections, as shown in table 1. The first phase started
4F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
with works already known to the authors, as related
to their field of research, or resulting from search-
ing by keywords such as “semantic change/shift/drift”,
“history of concepts/ideas”, “historical linguistics/se-
mantics”, “diachronic/synchronic variation/ontology”,
“ontology generation/acquisition/extraction/learning”,
“semantic change” + “word embeddings”. Keyword
search mainly involved the use of Google and se-
lection of journal articles, conference papers, book
sections usually made available via ResearchGate,
arXiv.org, ACL Anthology, IEEE Xplore, Semantic
Scholar, Google Scholar, Academia.edu, open source
journals, such as Journal of Web Semantics and Se-
mantic Web, and institutional libraries. The filtering
process included criteria such as relevance to the topic
discussed in a certain section, subsection and the work-
flow as a whole, and timeframe with reference, when
available, to recent research (in particular, last decade).
Publication year and citation number provided by var-
ious platforms, e.g. Google Scholar, ACL Anthology,
were also taken into account as pointing to newer and
influential research. Finally, the co-authors reached a
consensus on the works to be analysed and cited. Table
1 summarises the structure and size of the referenced
material presented in the survey.
3. Theoretical frameworks
Different disciplines (within or applied in the hu-
manities) make use of different interpretations, theo-
retical notions and approaches in the study of seman-
tic change. In this section, we survey various theoret-
ical frameworks that rest in the domain of either lin-
guistics or knowledge representation and that can serve
the theoretical modelling purposes of block 1 in the
generic workflow (Fig. 1). These theoretical frame-
works come from two distinct lines of enquiry, arising
from two traditions: one coming from philosophy, his-
tory of concepts and history of ideas, the other from
linguistics. Although there are no strict demarcations
between the two threads and some overlap exists, the
first is more closely associated with Semantic Web
technologies (and the corresponding representation of
knowledge, including ontologies), and the second with
corpus-based analysis.
3.1. Knowledge-oriented approaches
Scholars in domains such as history of ideas, his-
tory of concepts and philosophy focus on concepts as
Table 1. Structure and size of the surveyed material
Sec-
tion
Related research areas Cited
works
S1,
S2
Contextualisation of the topic, survey
methodology
15
S3 History of ideas, history of concepts,
philosophy, knowledge organisation
10
Lexical semantics, cognitive linguistics,
diachronic lexicology, terminology,
pragmatics, discourse analysis
20
S4
The OntoLex-Lemon model 3
Etymologies as LL(O)D 9
SW-based modelling of diachronic relations 7
SW resources for temporal information 4
S5
Overview 20
NLP Challenges 32
NER and NEL 24
Word embeddings 14
Transformer-based language models 5
Topic modelling 14
S6
Ontology learning 10
Diachronic constructs 11
Generating linked data 8
S7 Diachronic datasets in the LL(O)D cloud,
publishing diachronic ontologies as LL(O)D
9
Total (215 - 20 repeated citations) 195
units of analysis. In his comparative reading of Ger-
man and English conceptual history, Richter [16] ac-
counts for the distinction between words and concepts
in charting the history of political and social concepts,
where a concept is understood as a “forming part of a
larger structure of meaning, a semantic field, a network
of concepts, or as an ideology, or a discourse” (p. 10).
Basing his study on three major reference works by
20th-century German-speaking theorists, Richter notes
that outlining the history of a concept may sometimes
require tracking several words to identify continuities,
alterations or innovations, as well as a combination
of methodological tools from history, diachronic, and
synchronic analysis of language, semasiology, ono-
masiology, and semantic field theory. He also high-
lights the importance of sources (e.g. dictionaries, en-
cyclopaedias, political, social, and legal materials, pro-
fessional handbooks, pamphlets and visual, nonver-
bal forms of expression, journals, catechisms and al-
manacs) and procedures to deal with these sources,
employed in tracing the history of concepts in a cer-
tain domain, as demonstrated by the reference works
mentioned in his analysis.
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 5
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
Within the framework of intellectual history, Kuukka-
nen [17] proposes a vocabulary allowing for a more
formal description of conceptual change, in response
to critiques of Lovejoy’s long-debated notion of “unit-
ideas” or “unchangeable concepts”. Assuming that
a concept X is composed by two parts, the “core”
and the “margin”, underlain by context-unspecific and
context-specific features, Kuukkanen describes the
core as “something that all instantiations must satisfy
in order to be ‘the same concept”’, and the margin
as “all the rest of the beliefs that an instantiation of
X might have” (p. 367). This paradigm enables us to
record a full spectrum of possibilities, from conceptual
continuity, implying core stability and different de-
grees of margin variability, to conceptual replacement,
when the core itself is affected by change.
Another type of generic formalisation, combining
philosophical standpoints on semantic change, theory
of knowledge organisation and Semantic Web tech-
nologies, is proposed by Wang et al. [12] who con-
sider that the meaning of a concept can be defined in
terms of “intension, extension and labelling applicable
in the context of dynamics of semantics” (p. 1). Thus,
since reflecting a world in continuous transformation,
concepts may also change their meanings. This pro-
cess, called “concept drift”, 4occurs over time but
other kinds of factors, such as location or culture, may
be taken into account. The proposal is framed by two
“philosophical views” on the change of meaning of
a concept over time assuming that: (1) different vari-
ants of the same concept can have different mean-
ings (concept identity hypothesis); (2) concepts grad-
ually evolve into other concepts that can have almost
the same meaning at the next moment in time (con-
cept morphing hypothesis). In line with a tradition in
philosophy, logic and semiotics going back to Frege’s
“sense” and “reference” [19] and de Saussure’s “sig-
nifier” [20], Wang et al. formally describe the mean-
ing of a concept C as a combination of three “aspects”:
a “set of properties (the intension of C)”, a “subset
of the universe (the extension of C)”, and a “String”
(the label) [12, p. 6]. Based on these statements, they
develop a system of formal definitions that allows us
to detect different forms of conceptual drift, includ-
ing “concept shift” (where “part of the meaning of
a concept shifts to some other concept”) and “con-
cept split” (when the “meaning of a concept splits into
4The term “semantic drift” is also used, although the difference is
not explicitly defined. See also the discussion on [18].
several new concepts”) (pp. 2, 10). Various similarity
and distance measures (e.g. Jaccard and Levenshtein)
are computed for the three aspects to identify such
changes, according to the two philosophical perspec-
tives mentioned above. Within four case studies, the
authors apply this framework to different vocabular-
ies and ontologies in SKOS, RDFS, OWL and OBO 5
from the political, encyclopaedic, legal and biomedical
domains.
Drawing upon methodologies in history of philoso-
phy, computer science and cognitive psychology, and
elaborating on Kuukkanen’s and Wang et al.s formal-
isations, Betti and Van den Berg [21] devise a model-
based approach to the “history of ideas or concept drift
(conceptual change and replacement)” (p. 818). The
proposed method deems ideas or concepts (used inter-
changeably in the paper) as models or parts of mod-
els, i.e. complex conceptual frameworks. Moreover,
the authors consider that “concepts are (expressible in
language by) (categorematic) terms, and that they are
compositional; that is, if complex, they are composed
of subconcepts” (p. 813). Arguing that both the in-
tension and the extension of a concept should be in-
cluded in the study of concept drift, Betti and Van den
Berg identify the former with the core and margin, or
meaning, and the latter with the reference. To illustrate
their proposal, the authors use a model to represent
the concept of “proper science” as a relational struc-
ture of fixed conditions (core) containing sub-concepts
that can be instantiated differently within a certain cat-
egory, i.e. of expressions referring to something that
can be true, such as ‘propositions’, ‘judgements’ or
‘thoughts’ (margin) (pp. 822 - 824). According to [21],
such a model would support the study of the develop-
ment of ideas by enabling the representation of “con-
cept drift as change in a network of (shifting) relations
among subideas” and “fine-grained analyses of con-
ceptual (dis)continuities” (pp. 832 - 833).
Starting with an overview of concept change ap-
proaches in different disciplines, such as computer sci-
ence, sociology, historical linguistics, philosophy, Se-
mantic Web and cognitive science, Fokkens et al. [13]
propose an adaption of [17]’s and [12]’s interpre-
tations for modelling semantic change. Unlike [12],
[13] argue that only changes in the concept’s inten-
sion (definitions and associations), provided that the
core remains intact, are likely to be understood as con-
5SKOS (Simple Knowledge Organization System); RDFS (RDF
Schema), RDF (Resource Description Format); OWL (the W3C
Web Ontology Language); OBO (Open Biomedical Ontologies).
6F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
cept drift across domains; what belongs to the core
being decided by domain experts (oracles). Changes
to the core would determine conceptual replacement
(following [17]), while changes in the concept’s ex-
tension (reference) or label (words used to refer to it)
are considered related phenomena of semantic change
that may or may not be relevant and indicative of con-
cept drift. Fokkens et al. [13] apply these definitions
in an example using context-dependent properties and
an RDF representation in Lemon 6[22], the predeces-
sor of the OntoLex-Lemon model which is discussed
in Subsection 4.1. 7The authors also draw attention
to the fact that making the context of applicability of
certain definitions explicit can help in detecting con-
ceptual changes in an ontology and distinguish be-
tween changes in the world, which need to be formally
tracked, and changes due to corrections of inadequate
or inaccurate representations. However, obtaining the
required information for the former case is a challeng-
ing task. A possible path of investigation mentioned
in the paper refers to recent advances in distributional
semantics that can be effective in capturing semantic
change from texts.
A different interpretation is offered by Stavropou-
los et al. [18] through a background study intended to
describe the usage of terms such as semantic change,
semantic drift and concept drift in relation to ontol-
ogy change over time and according to different ap-
proaches in the field. Thus, from the perspective of
evolving semantics and Semantic Web, the authors
frame semantic change as a “phenomenon of change in
the meaning of concepts within knowledge represen-
tation models”. More precisely, semantic change de-
notes “extensive revisions of a single ontology or the
differences between two ontologies and can, therefore,
be associated with versioning” (p. 1). Within the same
framework, they define semantic drift as referring to
the gradual change either of the features of ontology
concepts, when their knowledge domain evolves, or
of their semantic value, as it is perceived by a rele-
vant user community. Further distinction are drawn be-
tween intrinsic and extrinsic semantic drift, depending
on the type of change in the concept’s semantic value.
That is, in respect to other concepts within the ontol-
ogy or to the corresponding real world object referred
by it. Originated from the field of incremental con-
6Lemon (the Lexicon Model for Ontologies).
7Note that although [13] cites the original Lemon model the ex-
ample featured in that article seems to be using the later OntoLex-
Lemon model.
cept learning [23] and adapted to the new challenges
of the Semantic Web dynamics [24], concept drift is
described in [18, p. 3] as a “change in the meaning of
a concept over time, possibly also across locations or
cultures, etc.”. Following [12], three types of concept
drifts are identified as operating at the level of label,in-
tension and extension. Stavropoulos et al. transfer this
type of formalisation to measure semantic drift in a
dataset from the Software-based Art domain ontology,
via different similarity measures for sets and strings,
by comparing each selected concept with all the con-
cepts of the next version of the ontology and iterat-
ing across a decade. The two terms, semantic drift and
concept drift, initially emerged from different fields
but according to [18] an increasing number of studies
show a tendency to apply notions and techniques from
a field to the other.
3.2. Language-oriented approaches
Scholars from computational semantics employ a
slightly different terminology from scholars from his-
tory of ideas, history of concepts and philosophy. Ku-
tuzov et al. [9], for example, describe the evolution of
word meaning over time in terms of “lexical semantic
shifts” or “semantic change”, and identify two classes
of semantic shifts: “linguistic drifts (slow and regu-
lar changes in core meaning of words) and cultural
shifts (culturally determined changes in associations of
a given word)” (p. 1385).
Disciplines from more traditional linguistics-related
areas provide other types of theoretical bases and ter-
minologies to research semantic change and concept
evolution. For instance, Kvastad [25] underlines the
distinction made in semantics between concepts and
ideas, on one side, and terms, words and expressions,
on the other side, where a “concept or idea is the
meaning which a term, word, statement, or act ex-
presses” (p. 158). Kvastad also proposes a set of meth-
ods bridging the field of semantics and the study of the
history of ideas. Such approaches include synonymity,
subsumption and occurrence analysis allowing histori-
ans of ideas to trace and interpret concepts on a sys-
tematic basis within different contexts, authors, works
and periods of time. Other semantic devices listed by
the author can be used to define and detect ambiguity
in communication between the author and the reader,
formalise precision in interpretation or track agree-
ment and disagreement in the process of communica-
tion and discussion ranging over centuries.
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 7
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
Along a historical timeline, spanning from the mid-
dle of the 19th century to 2009, Geeraerts [26] presents
the major traditions in the linguistics field of lexical
semantics, with a view on the theoretical and method-
ological relationships among five theoretical frame-
works: historical-philological semantics, structuralist
semantics, generativist semantics, neostructuralist se-
mantics and cognitive semantics. While focusing on
the description of these theoretical frameworks and
their interconnections in terms of affinity, elabora-
tion and mutual opposition, the book also provides
an overview on the mechanisms of semantic change
within these different areas of study. The main classi-
fications of semantic change resulted from historical-
philological semantics include on one hand, the se-
masiological mechanisms (meaning-related) that “in-
volve the creation of new readings within the range
of application of an existing lexical item”, with sema-
siological innovations endowing existing words with
new meanings. On the other hand, the onomasiologi-
cal (or “lexicogenetic”) mechanisms (naming-related)
“involve changes through which a concept, regardless
of whether or not it has previously been lexicalised,
comes to be expressed by a new or alternative lexi-
cal item”, with onomasiological innovations coupling
“concepts to words in a way that is not yet part of
the lexical inventory of the language” (p. 26). Further
distinctions within the first category refer to lexical-
semantic changes such as specialisation and gener-
alisation, or metonymy and metaphor. On the other
hand, the second category is related to the process
of word formation that implies devices such as mor-
phological rules for derivation and composition, trans-
formation through clipping or blending, borrowing
from other languages or onomatopoeia-based develop-
ment. Geeraerts also points out the general orientation
of historical-philological semantics as diachronic and
predominantly semasiological rather than onomasio-
logical, with a focus on the change of meaning under-
stood as a result of psychological processes, and an
“emphasis on shifts of conventional meaning” and thus
an empirical basis consisting “primarily of lexical uses
as may be found in dictionaries” (p. 43). In this sense,
historical-philological semantics links up with lexi-
cography, etymology and history of ideas (“meanings
are ideas”) (p. 9). Moreover, the author distinguishes
three main perspectives: structural that looks at the
“interrelation of [linguistic] signs” (sign-sign relation-
ship), pragmatic that considers the “relation between
the sign and the context of use, including the lan-
guage user” (sign–use(r) relationship), and referential
that delineates the “relation between the sign and the
world” (sign–object relationship). According to [26],
the evolution of lexical semantics (and implicitly of
the way meaning and semantic change are reflected
upon) can be characterised therefore by an oscillation
along these three dimensions. A historical-philological
stage dominated by the referential and pragmatic per-
spective, a structuralist phase centred on structural,
sign–sign relations, an intermediate position shaped
by generativist and neostructuralist approaches, and a
current cognitive stance that recontextualises seman-
tics within the referential and pragmatic standpoint and
displays a certain affinity with usage-based approaches
such as distributional analysis of corpus data (pp. 278 -
279, 285).
In cognitive linguistics and diachronic lexicology,
Grondelaers et al. [27] also identify that semantic
change could be approached by applying two differ-
ent perspectives – onomasiological and semasiologi-
cal. The onomasiological approach focuses on the ref-
erent and studies diachronically the representations
of the referent, whereas the semasiological approach
investigates the linguistic expression by researching
diachronically the variation of the objects identified
by the linguistic expressions under the investigation.
There is a tendency to apply the semasiological ap-
proach in computational semantic change research be-
cause it relies on words or phrases extracted from
the datasets; however, the extraction of concept rep-
resentations from linguistic data poses certain chal-
lenges and requires either semi-automatically or auto-
matically learning ontologies to trace concept drift or
change as it was discussed above.
In other fields, such as terminology, semasiological
and onomasiological approaches may encompass ei-
ther a concept- or a term-oriented perspective [28, 29].
Other standpoints, framed for instance in a sociocog-
nitive context, attempt to take into account both the
principles of stability, univocity of “one form for one
meaning” and synchronic term-concept relationship
from traditional terminology, and the need for under-
standing and interpreting the world and language in
their dynamics as they change over time, and for ap-
plying more flexible tools when analysing semantic
change in a specialised domain, such as prototype the-
ory [30, pp. 126, 130)].
Diachronic change at the level of pragmatics re-
quires a special endeavor as it is context specific.
While analysing diachronic change of discourse mark-
ers, first it should be stressed that the notion of dis-
course marker was introduced by Schiffrin [31] and
8F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
the author considered phrases such as ‘I think’ a dis-
course marker performing the function of discourse
management deictically “either point[ing] backward in
the text, forward, or in both directions”. Fraser [32]
provided a taxonomy of pragmatic markers drawn
from syntactic classes of conjunctions, adverbials and
prepositional phrases followed by Aijmer [33] sug-
gesting that ’I think’ is a “modal particle”. Over the last
few decades the research on discourse markers has de-
veloped into a considerable and independent field ac-
cepting the term of discourse markers [34–36]
Through the manual analysis of diachronic change
of discourse markers, e.g., Waltereit and Detges [37]
analysed the development of the Spanish discourse
marker bien derived from the Latin manner adverb
bene (‘well’) and showed that the functional differ-
ence between discourse markers and modal particles
can be related to different diachronic pathways. Cur-
rently, corpus-driven automatic analysis is acquiring
the impetus, e.g. Stvan [38] uses corpus analysis relat-
ing early 20th-century American texts with modern TV
shows to research diachronic change in the discourse
markers ‘why’ and ‘say’ in American English. How-
ever, there are still challenges analysing diachronic
change on the pragmatic layer as there is a need for a
move from queries based on individual words towards
larger linguistic units and pieces of text.
In addition to linguistic approaches focusing on
text linguistics and pragmatics, discourse analysis in a
broad sense studies naturally occurring language refer-
ring to socio-related textual characteristics in human-
ities and social sciences. According to Foucault, one
of the key theorists of the discourse analysis, the term
“discourse” refers to institutionalized patterns and dis-
ciplinary structures concerned with the connection of
knowledge and power [39]. Discourse analysis ap-
proaches language as a means of social interaction and
is related to the social contexts embedding the dis-
course. Within this framework, the discourse-historical
approach (DHA) is of particular interest, as part of the
broader field of critical discourse analysis (CDA) that
investigates “language use beyond the sentence level”
and other “forms of meaning-making such as visuals
and sounds” as elements in the “(re)production of so-
ciety via semiosis” [40]. Thus, based on the principle
of “triangulation”, DHA takes into account a variety of
datasets, methods, theories and background informa-
tion to analyse the historical dimension of discursive
events and the ways in which specific discourse gen-
res are subject to diachronic change. Recent studies on
linguistic change using diachronic corpora and a com-
bination of computational methods, such as word em-
beddings, and discourse-based approaches argue that
a discourse-historical angle can provide a better un-
derstanding of the interrelations between language and
social, cultural and historical factors, and their change
over time [41, 42].
4. LOD formalisms
Having given an overview of different theoretical
perspectives on semantic change across numerous dis-
ciplines in (digital) humanities-related areas, we will
look at how some of these perspectives can be mod-
elled as linked data in this section. In particular, we
survey possible modalities for formally representing
the evolution of word meanings and their related con-
cepts over time within a LL(O)D and Semantic Web
framework (also in connection to block 2, Fig. 1). In
Subsection 4.1, we will look at the OntoLex-Lemon
model for representing lexicon-ontologies as linked
data. This model is useful for representing the relation-
ship between a lexicon and a set of concepts, some-
thing that is relevant for both knowledge-oriented and
language-oriented approaches mentioned in Section 3.
Next, in Subsection 4.2, we look at the representa-
tion of etymologies or word histories in linked data
as these are particularly useful in language-oriented
approaches to semantic change. Afterwards, in Sub-
section 4.4 we look at how to explicitly represent di-
achronic relations in RDF; this is useful for any sit-
uation in which we have to model dynamic informa-
tion and is relevant to both of the general approaches
in Section 3 and is not limited only to linked data. Fi-
nally, we look at resources for representing temporal
information in linked data, in Subsection 4.4.
4.1. The OntoLex-Lemon model
OntoLex-Lemon [43] is the most widely used model
for publishing lexicons as linked data. For what re-
gards its modelling of the semantics of words, it repre-
sents the meaning of any given lexical entry “by point-
ing to the ontological concept that captures or rep-
resents its meaning”. 8In OntoLex-Lemon, the class
LexicalSense is defined as “[representing] the lexical
meaning of a lexical entry when interpreted as refer-
ring to the corresponding ontology element”, that is
8Lexicon Model for Ontologies: Community Report, 10 May
2016 (w3.org) https://www.w3.org/2016/05/ontolex/#semantics
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 9
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
Fig. 2. OntoLex-Lemon core model
“a reification of a pair of a uniquely determined lexi-
cal entry and a uniquely determined ontology entity it
refers to”. Moreover, the object property sense is de-
fined in the W3C Community Report as “[relating] a
lexical entry to one of its lexical senses” and the ob-
ject propertyreference as “[relating] a lexical sense to
an ontological predicate that represents the denotation
of the corresponding lexical entry”. See Figure 2 for a
schematic representation of the OntoLex-Lemon core.
Another property that is relevant to the modelling of
lexical meaning is denotes which is equivalent to the
property chain sense o reference.9In addition, the Us-
age class allows us to describe sense usages of indi-
viduals of LexicalSense.
OntoLex-Lemon also allows users the possibility of
modelling usage conditions on a lexical sense – condi-
tions that reflect pragmatic constraints on word mean-
ing such as those which concern register – via the (ap-
propriately named) object property usage.10 The use
of this property is intended to complement the lexical
sense rather than to replace it.
To summarise, OntoLex-Lemon offers users a model
for representing the relationship between a lexical
sense and an ontological entity in linked data. The
relationship between lexical and conceptual aspects,
or more broadly speaking, linguistic and conceptual
aspects of meaning 11 are important for many of the
approaches listed in Section 3. This holds for both
the knowledge-oriented approaches described in Sub-
9Here ostands for the relation composition operator, i.e., (a,b)
RoS⇔ ∃c.(a,c)R&(c,b)S
10https://www.w3.org/2016/05/ontolex/#usage
11Note that ontologies are usually described as conceptualisations
and of consisting of concepts [44] which makes them an ideal pre-
requisite for modelling conceptual shift.
section 3.1 such as those of Richter, as well as the
language-oriented approaches of Subsection 3.2. Note
that the work of [13] described above in Subsection 3.1
is already based on lemon, the immediate pre-cursor of
OntoLex-Lemon.
Another OntoLex-Lemon class for modelling mean-
ing is LexicalConcept. This is defined as “a mental ab-
straction, concept or unit of thought that can be lexical-
ized by a given collection of senses” in the OntoLex-
Lemon guidelines. 12 It is related to LexicalEntry via
the evokes class which relates a lexical entry to a
“mental concept that speakers of a language might as-
sociate when hearing [the entry]”. From this definition
a lexical entry for the word grape could be related via
evokes to the concept of ’wine’ or ’harvest’ or spe-
cific geographical regions such as Burgundy or Con-
cord. This can be useful in tracing the different associ-
ations and related concepts that a word picks up over
time, while sense and reference are used to look at the
core intensional and extensional meanings of the same
words.
Work on a Frequency, Attestation and Corpus In-
formation module (FrAC) for OntoLex-Lemon is un-
derway in the OntoLex W3C group [45]. This mod-
ule, once finished, will enable the addition of corpus-
related information to lexical senses, including infor-
mation pertaining to word embeddings.
4.2. Representing etymologies and sense shifts in
LL(O)D
One important source of information on semantic
shifts are etymologies. These are defined as word his-
tories and include descriptions of both the linguis-
tic drifts and cultural shifts described by Kutuzov et
al. and other (language-related) approaches discussed
in Subsection 3.2. They can be used in some of the
knowledge-oriented approaches mentioned in Subsec-
tion 3.1 such as that of Richter. As well as being a
source of semantic change information, etymologies
can also be used to encode and to make semantic
change information accessible in lexical resources in a
standardised way; we can do this by making use of and
extending existing linked data vocabularies as we will
see in this section.
Current work in modelling etymology in LL(O)D
was preceded and influenced by similar work in related
standards such as the Text Encoding Initiative (TEI)
12https://www.w3.org/2016/05/ontolex/#lexical-concept-class
10 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
and the Lexical Markup Framework (LMF). This in-
cludes notably Salmon-Alt’s LMF-based approach to
representing etymologies in lexicons [46], as well as
Bowers and Romary’s [47] work which builds on al-
ready existing TEI provisions for encoding etymolo-
gies in order to propose a deep encoding of etymolog-
ical information in TEI. In the latter case, the authors’
approach entailed enabling an enhanced structuring of
lexical data that would allow for the identification of,
for instance, etymons and cognates in a TEI entry, as
well as the specification of different varieties of etymo-
logical change. This also coincides with the current de-
velopment of an etymological extension of LMF by the
International Standards Organization working group
ISO/TC 37/SC 4/WG 4 [48], see also [49] for exam-
ples of LMF encoding from a Portuguese dictionary,
the Grande Dicionário Houaiss da Língua Portuguesa.
Work on the representation of etymologies in RDF
includes de Melo’s [50] work on Etymological Word-
Net, as well as Chiarcos et al’s [51] definition of a min-
imal extension of the lemon model with two new prop-
erties cognate and the transitive derivedFrom for
representing etymological relationships. Khan [52] de-
fines an extension of OntoLex-Lemon that, like [47]
attempts to facilitate a more detailed encoding of ety-
mological information. Notably, this extension reifies
the notion of etymology defining individuals of the Et-
ymology class as containers for an ordered series of Et-
ymologicalLink individuals. The latter class is a reifi-
cation, this time of the notion of an etymological link.
These etymological link objects connect together Ety-
mon individuals and (OntoLex) Lexical Entries or in-
deed any other kinds of lexical element that can have
an etymology. We can subtype etymological links to
represent sense shifts within the same lexical entry.
Other work specifically on the modelling of sense shift
in LL(O)D includes the modelling of semantic shift in
Old English emotion terms in [53] in which semantic
shifts are reified and linked to elements in a taxonomy
of metonymy and metaphor which describe the con-
ceptual structure of these shifts.
Etymological datasets in LL(O)D include the Latin-
based etymological lexicon published as part of the
LiLa project and described in [54].
4.3. Representing diachronic relations
We have thus far looked at ways of representing
information about lexicons and the concepts which
they lexicalise in RDF and which are salient for
both knowledge-oriented and language-oriented ap-
proaches. However, as argued by [55], to be able to
represent changes in the meaning of concepts, as well
as the concepts themselves within the framework of
the OntoLex-Lemon model, it would be useful to be
able to add temporal parameters to (at least) the proper-
ties sense or reference, as well as possibly the evokes
property. We refer to such properties or relations that
can change with time as fluents. Due to a well known
expressive limitation of the RDF framework, it is not
possible to add a temporal parameter to a binary prop-
erties. To remedy this, we can either extend RDF or
use a number of suggested ontology design patterns
in order to stay within the expressive constraints of
RDF. An example of the first strategy is described
in [56] where Rizzolo et al. present a formal “RDF-like
model” for concept evolution. This is based both on
the idea of temporal knowledge bases, in which tempo-
ral intervals or lifespans are associated with resources
as well as new relations for expressing parthood and
causality between concepts. These relations underpin
the authors’ representation of concept evolution via
specialised terms. Finally, they present a special exten-
sion of SPARQL based on their new framework and
which permits the querying of temporal databases for
questions relating to the evolution of a concept over
a time period. In [57], Gutierrez et al. propose an ex-
tension of RDF which permits temporal reasoning and
which describes so-called temporal RDF graphs. They
present a syntax, semantics as well as an inference sys-
tem for this new extension, 13 as well as a new tem-
poral query language. Another more recent solution
which is still under active development at the time of
the writing of this paper is RDF*. 14 In RDF*, triples
can be embedded in and therefore described by other
triples. This means for instance that a relationship such
as sense can be associated with temporal properties
which delimit its temporal validity.
In terms of the second solution, there are numer-
ous design patterns for adding temporal information
to RDF and permitting temporal reasoning over RDF
graphs without adding extra constructs to the language.
We will look very briefly at a few of the most promi-
nent of these. We refer the reader to [58] for a more
detailed survey.
The first pattern we will look at is to reify the re-
lation in question, that is turn it into an object, which
13They are able to show that their entailment for temporal RDF
graphs does not lead to an asymptotic increase in complexity.
14A draft of the specification can be found at this link: https://w3c.
github.io/rdf-star/cg-spec/editors_draft.html
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 11
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
was proposed by the W3C as a general strategy for
representing relations with an arity greater than 2. Ac-
cording to this pattern, we can turn OntoLex-Lemon
sense and reference relations into objects. This pattern
has the disadvantage of being too prolix and creating a
profusion of new objects, it also means that we cannot
use certain OWL constructs for reasoning (see [59] for
more details).
Other prominent patterns take the perdurantist ap-
proach by modelling entities as having temporal parts,
as well as (for physical objects) physical parts. Per-
haps the most influential of these is the Welty-Fikes
pattern introduced in [59] where fluents are repre-
sented as holding between temporal parts of entities
rather than the entities themselves. For instance, the
OntoLex-Lemon property sense would hold between
temporal parts of LexicalSense individuals rather than
the individuals themselves. The Welty-Fikes pattern is
much less verbose than the first pattern, and also al-
lows us to use the OWL constructs alluded to in the
last paragraph. However the fact that the Welty-Fikes
pattern constrains us into redefining fluent properties
as holding between temporal parts rather than between
the original entities (so sense, or the temporal version,
would no longer have the OntoLex-Lemon classes Lex-
icalEntry as a domain and LexicalSense as a range)
could be seen as a serious disadvantage. A simplifica-
tion to the Welty-Fikes pattern is proposed in [60] in
which “what has been an entity becomes a time slice”.
This implies that fluents hold between perdurants, that
is entities with a temporal extent, but these can be, in
our example, lexical entries and senses. This is the ap-
proach taken in [61] to model dynamic lexical infor-
mation, and where lexical entries and senses (among
other OntoLex-Lemon elements) were given temporal
extents.
4.4. OWL-Time ontology and other Semantic Web
resources for temporal information
The most well known linked data resource for en-
coding temporal information is the OWL-Time ontol-
ogy [62]. As of March 2020, it is a W3C Candidate
Recommendation. OWL-Time allows for the encoding
of temporal facts in RDF, both according to the Grego-
rian calendar as well as other temporal reference sys-
tems, including alternative historical and religious cal-
endars. It includes classes representing time instants
and time intervals as well as a provision for represent-
ing topological relationships among intervals and in-
stants and in particular those included in the Allen tem-
poral interval algebra [63]. This allows for reasoning
to be carried out over temporal data that uses the Allen
properties, in conjunction with an appropriate set of
OWL axioms and SWRL rules, such as those described
in [64].
Other useful resources that should be mentioned
here are PeriodO, 15 an RDF-based gazetteer of tem-
poral periods which are salient for work in archaeol-
ogy, history and art-history [65], and LODE, an ontol-
ogy for Linking Open Descriptions of Events.16 These
resources are useful both for approaches which deal
specifically with linguistic linked data as well as those
which deal with shifts in concepts over time more gen-
erally.
5. NLP for detecting lexical semantic change
Given the possibilities described above for mod-
elling semantic change via LL(O)D formalisms, we
will address the question of automatically capturing
such changes in word meaning (block 3, Fig. 1) by
analysing diachronic corpora available in electronic
format. This section provides an overview of existing
methods and NLP tools for the exploration and detec-
tion of lexical semantic change in large sets of data,
e.g. related to diachronic word embeddings, named en-
tity recognition (NER) and topic modelling.
5.1. Overview
The past decade has seen a growing interest in com-
putational methods for lexical semantic change detec-
tion. This has spanned across different communities,
including NLP and computational linguistics, informa-
tion retrieval, digital humanities and computational so-
cial sciences. A number of different approaches have
been proposed, ranging from topic-based models [66–
68], to graph-based models [69, 70], and word embed-
dings [11, 71–77]. [8], [7], and [9] provide compre-
hensive surveys of this research until 2018. Since then,
this field has advanced even further [78–81].
In spite of this rapid growth, it was only in 2020
that the first standard evaluation task and data were
created. [10] present the results of the first SemEval
shared task on unsupervised lexical semantic change
detection, which represents the current NLP state of
the art in this field. Thirty-three teams participated in
15https://perio.do/en/
16https://linkedevents.org/ontology/
12 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
the shared task, submitting 186 systems in total. These
systems use a representation of the semantics of words
from the input diachronic corpus, which is usually
split into subcorpora covering different time intervals.
The majority of the methods proposed rely on embed-
ding technologies, including type embeddings (i.e. em-
beddings representing a word type) and token embed-
dings (i.e. contextualised embeddings for each token).
Once the semantic representations have been built,
a method for aligning these representations over the
temporal sub-corpora is needed. The alignment tech-
niques used include orthogonal Procrustes [11], vector
initialisation [71] and temporal referencing [80]. Fi-
nally, to detect any significant shift which can be in-
terpreted as semantic change, the change between the
representations of the same word over time needs to
be measured. The change measures typically used in-
clude distances based on cosine and local neighbours,
Kullback-Leibler divergence, mean/standard deviation
of co-occurrence vectors, or cluster frequency. The
systems which participated in the shared task were
evaluated on manually-annotated gold standards for
four languages (English, German, Latin and Swedish)
and two sub-tasks, both aimed at detecting lexical se-
mantic change between two time periods. Given a list
of words, the binary classification sub-task aimed at
detecting which words lost or gained senses between
the two time periods, while the ranking sub-task con-
sisted in ranking the words according to their degree
of semantic change between the two time periods. The
best-performing systems all use type embedding mod-
els, although the quality of the results differs depend-
ing on the language. Averaging over all four languages,
the best result had an accuracy of 0.687 for sub-task
1 and a Spearman correlation coefficient of 0.527 for
sub-task 2.
5.2. NLP Challenges
Detecting lexical semantic change via NLP implies
a series of challenges, related to the digitisation, prepa-
ration and processing of data, as discussed below.
Applying NLP tools, such as POS taggers, syntac-
tic parsers, and named entity recognisers to historical
texts is difficult, because most existing NLP tools are
developed for modern languages [82, 83] and histor-
ical language use often differs significantly from its
modern counterpart. The two often have different lin-
guistic aspects, such as lexicon, morphology, syntax,
and semantics which make a naive use of these tools
problematic [84, 85]. One of the most prevalent dif-
ferences is spelling variation. The detection of spelling
variants is an essential preliminary step for identifying
lexical semantic change. A frequently suggested solu-
tion for the spelling variation issue is normalisation.
Normalisation is generally described as the mapping of
historical variant spellings into a single, contemporary
“normal form".
Recently, Bollmann [86] systematically reviewed
automatic historical text normalisation. Bollmann di-
vided the research data into six conceptual or method-
ical approaches. In the first approach, each historical
variant is checked in a compiled list that maps its ex-
pected normalisation. Although this method does not
generalise patterns for variants not included in the list,
it has proved highly successful as a component of sev-
eral other normalisation systems [87, 88]. The sec-
ond approach is rule-based. The rule-based approach
aims to encode regularities in the form of substitu-
tion rules in spelling variations, usually including con-
text information to distinguish between different char-
acter uses. This approach has been adopted to vari-
ous languages including German [89], Basque, Span-
ish [90], Slovene [91], and Polish [92]. The third ap-
proach is based on editing distance measures. Dis-
tance measures are used to compare historical vari-
ants to modern lexicon entries [88, 93, 94]. Normalisa-
tion systems often combine several of these three ap-
proaches [87, 94–96]. The fourth approach is statisti-
cal. The statistical approach models normalisation as
a probability optimisation task, maximising the prob-
ability that a certain modern word is the normalisa-
tion of a given historical word. The statistical approach
has been applied as a noisy channel model [91, 97],
but more commonly as character-based statistical ma-
chine translation (CSMT) [98–100], where the histor-
ical word is “translated” as a sequence of characters.
The fifth approach is based on neural network archi-
tectures, where the encoder–decoder model with recur-
rent layers is the most common [101–105]. The en-
coder–decoder model is the logical neural counterpart
of the CSMT model. Other works modelled the nor-
malisation task as a sequence labelling problem and
applied long short-term memory networks (LSTM)
neural networks [106, 107]. Convolutional networks
were also used for lemmatisation [108]. In the sixth
approach Bollmann [86] included models that use con-
text from the surrounding tokens to perform normal-
isation [109, 110]. Bollmann [86] also compares and
analyses the performance of three freely available tools
that cover all types of proposed normalisation ap-
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 13
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
proaches on eight languages. The datasets and scripts
are publicly available.
Other studies in detecting lexical semantic change
pointed out different types of challenges. For instance,
in their analysis of markers of semantic change and
leadership in semantic innovation using diachronic
word embeddings and two corpora containing scien-
tific articles and legal opinions from 20 and 18 cen-
tury to present, [111] reported difficulties posed by
names and abbreviations in identifying genuine candi-
dates of semantic innovations. They applied a series of
post-processing heuristics to alleviate these problems,
by training a feed-forward neural network and using
a pre-trained tagger to label names and proper nouns
or to detect abbreviations under a certain frequency
threshold, and discarding them from the list of candi-
dates.
[112] addressed the scalability and interpretability
issues observed in semantic change detection with
clustering of all word’s contextual embeddings for
large datasets, mainly related to high memory con-
sumption and computation time. The authors used a
pre-trained BERT model (see Subsection 5.5) to de-
tect word usage change in a set of multilingual corpora
(in German, English, Latin and Swedish) of COVID-
19 news from January to April 2020. To improve scal-
ability, they limited the number of contextual embed-
dings kept in memory for a given word and time slice
by merging highly similar vectors. The most changing
words were identified according to divergence and dis-
tance measures of usage computed between successive
time slices. The most discriminating items from the
clusters of usage corresponding to these words were
then used by the researchers and domain experts in the
interpretation of results.
Another type of challenge is that of assessing the
impact of OCR (Optical Character Recognition) qual-
ity on downstream NLP tasks, including the com-
bined effects of time, linguistic change and OCR qual-
ity when using tools trained on contemporary lan-
guages to analyse historical corpora. [113] performed
a large-scale analysis of the impact of OCR errors
on NLP applications, such as sentence segmentation,
named-entity recognition (NER), dependency parsing
and topic modelling. They used datasets drawn from
historical newspapers collections and based their tests
and evaluation on OCR’d and human-corrected ver-
sions of the same texts. Their results showed that the
performance of the examined NLP tasks was affected
to various degrees, with NER progressively degrading
and topic modelling diverging from the “ground truth”,
with the decrease of OCR quality. The study demon-
strated that the effects of OCR errors on this type of
applications are still not fully understood, and high-
lighted the importance of rigorous heuristics for mea-
suring OCR quality, especially when heritage docu-
ments and a temporal dimension are involved.
5.3. Named-entity recognition and named-entity
linking
Named-entity recognition (NER) and named-entity
linking (NEL) which allow organisations to enrich
their collections with semantic information have in-
creasingly been embraced by the digital humanities
(DH) community. For many NLP-based systems, iden-
tifying named-entity changes is crucial since fail-
ure to know various names referring to the same en-
tity greatly affects their efficiency. Temporal NER
has been mostly studied in the context of histori-
cal corpora. Various NER approaches have been ap-
plied to historical texts including early rule-based
approaches [114–116] through unsupervised statis-
tical approaches [117], conventional machine learn-
ing approaches [118–120] and to deep learning ap-
proaches [121–125]. Named-entity disambiguation
(NED) was also investigated and Agarwal et al. [126]
introduced the first time-aware method for NED of di-
achronic corpora.
Different eras, domains, and typologies have been
investigated, so comparing different systems or algo-
rithms is difficult. Thus, [127] recently introduced the
first edition of HIPE (Identifying Historical People,
Places and other Entities), a pioneering shared task
dedicated to the evaluation of named entity processing
on historical newspapers in French, German and En-
glish [128]. One of its subtasks is Named Entity Link-
ing (NEL). This subtask includes the linkage of the
named entity to a particular referent in the knowledge
base (KB) (Wikidata) or a NEL node if the entity is not
included in the base.
Traditionally, NEL has been addressed in two main
approaches: text similarity-based and graph-based.
Both of these approaches were adapted to historical
domains mostly as ‘of-the-shelf’ NEL systems. While
some of the previous works perform NEL using the
KB unique ids [128, 129], other works use LL(O)D
formalisms [130–133]. One of the aims of the HIPE
shared task was to encourage the application of neural-
based approaches for NER which has not yet been ap-
plied to historical texts. This aim was achieved suc-
cessfully. Teams have experimented with various en-
14 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
tity embeddings, including classical type-level word
embeddings and contextualised embeddings, such as
BERT (see Subsection 5.5). The manual annotation
guidelines of the HIPE corpus were derived from the
Quaero annotation guide [134] and thus, the HIPE
corpus mostly remains compatible with the NewsEye
project’s NE Finnish, French, German, and Swedish
datasets. 17 Pontes et al. [135] analysed the perfor-
mance of various NEL methods on these two multilin-
gual historical corpora and suggested multiple strate-
gies for alleviating the effect of historical data prob-
lems on NEL. In this respect, they pointed out the vari-
ations in orthographic and grammatical rules, and the
fact that names of persons, organisations, and places
could have significantly changed over time. [135] also
mentioned potential avenues for further research and
applications in this area. This may include the use of
entity linking in historical documents to improve the
coverage and relevance of historical entities within
knowledge bases, the adaptation of the entity linking
approaches to automatically generate ontologies for
historical data, and the use of diachronic embeddings
to deal with named entities whose name have changed
through the time.
Social media communication platforms such as
Twitter, with their informal, colloquial and non-standard
language, have led to major changes in the charac-
ter of written languages. Therefore, in recent years,
there has been research interest in NER for social
media diachronic corpora. Rijhwani and Preo¸tiuc-
Pietro [136] introduced a new dataset of 12,000 En-
glish tweets annotated with named entities. They ex-
amined and offered strategies for improving the utili-
sation of temporally-diverse training data, focused on
NER. They empirically illustrated how temporal drift
affects performance and how time information in doc-
uments can be leveraged to achieve better models.
5.4. Word embeddings
A common approach for lexical semantic change de-
tection is based on semantic vector spaces meaning
representations. Each term is represented as two vec-
tors representing its co-occurring statistics at various
eras. The semantic change is usually calculated by dis-
tance metric (e.g. cosine), or by differences in contex-
tual dispersion between the two vectors.
Previously, most of the methods for lexical semantic
change detection built co-occurrence matrices [137–
17https://www.newseye.eu/.
139]. While in some cases, high-dimensional sparse
matrices were used, in other cases, the dimensions of
the matrices were reduced mainly using singular value
decomposition (SVD) [140]. Yet, in the last decade,
with the development of neural networks, the word
embedding approach commonly replaced the mathe-
matical approaches for dimensional reduction.
Word embedding is a collective name for neural
network-based approaches in which words are em-
bedded into a low dimensional space. They are used
as a lexical representation for textual data, where
words with a similar meaning have similar represen-
tation [141–144]. Although these representations have
been used successfully for many natural language pre-
processing and understanding tasks, they cannot deal
with the semantic drift that appears with the change of
meaning over time if they are not specifically trained
for this task.
In [145], a new unsupervised model for learning
condition-specific embeddings is presented, which en-
capsulates the word’s meaning whilst taking into ac-
count temporal-spatial information. The model is eval-
uated using the degree of semantic change, the discov-
ery of semantic change, and the semantic equivalence
across conditions. The experimental results show that
the model captures the language evolution across both
time and location, thus making the embedding model
sensitive to temporal-spatial information.
Another word embedding approach for tracing the
dynamics of change of conceptual semantic relation-
ships in a large diachronic scientific corpus is pro-
posed in [146]. The authors focus on the increasing
domain-specific terminology emerging from scientific
fields. Thus, they propose to use hyperbolic embed-
dings [147] to map partial graphs into low dimen-
sional, continuous hierarchical spaces, making more
explicit the latent structure of the input. Using this ap-
proach, the authors built diachronic semantic hyper-
spaces for four scientific topics (i.e., chemistry, phys-
iology, botany, and astronomy) over a large historical
English corpus stretching for 200 years. The experi-
ments show that the resulting spaces present the char-
acters of a growing hierarchisation of concepts, both in
terms of inner structure and in terms of light compari-
son with contemporary semantic resources, i.e., Word-
Net.
To deal with the evolution of word representa-
tions through time, the authors in [148] propose three
LSTM-based sequence to sequence (Seq2Seq) mod-
els (i.e., a word representation autoencoder, a future
word representation decoder, and a hybrid approach
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 15
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
combining the autoencoder and decoder) that mea-
sure the level of semantic change of a word by track-
ing its evolution through time in a sequential man-
ner. Words are represented using the word2vec skip-
gram model [141]. The level of semantic change of a
word is evaluated using the average cosine similarity
between the actual and the predicted word representa-
tions through time. The experiments show that hybrid
approach yields the most stable results. The paper con-
cludes that the performance of the models increases
alongside the duration of the time period studied.
Word embeddings are also used to capture synthetic
distortions in textual corpora. In [149], the authors pro-
pose a new method to determine paradigmatic (i.e.,
a term can be replaced by a word) and syntagmatic
association (i.e., the co-occurrence of terms) shifts.
The study employs three real-world datasets, i.e., Red-
dit, Amazon, and Wikipedia, with texts, collected be-
tween 1996-2018 for the experiments. The analysis
concludes that local neighborhood [150], which de-
tects shifts via the knearest neighbors, is sensitive
to paradigmatic shifts while the global semantic dis-
placement [150], which detects shifts within word co-
occurrence using the cosine similarity of embeddings,
is sensitive to syntagmatic shifts in word embeddings.
Furthermore, the experimental results show that words
undergo paradigmatic and syntagmatic shifts both sep-
arately and simultaneously.
5.5. Transformer-based language models
The current state of the art in word representation
for multiple well known NLP tasks is established by
transformer-based pre-trained language models, such
as BERT (Bidirectional Encoder Representations from
Transformers) [151], ELMo [152] and XLNet [153].
Recently, transformers were also used in lexical se-
mantic change tasks. In [154], the authors present
one of the first unsupervised approaches to lexical-
semantic change that utilise a transformer model. Their
solution uses the BERT transformer model to obtain
contextualised word representations, compute usage
representations for each occurrence of these words,
and measure their semantic shifts along time. For eval-
uation, the authors utilise a large diachronic English
corpus that covers two centuries of language use. The
authors provide an in-depth analysis of the proposed
model, proving that it captures a range of synchronic,
e.g., syntactic functions, literal and metaphorical us-
age, and diachronic linguistic aspects. In [155], dif-
ferent clustering methods are used on contextualised
BERT word embeddings to quantify the level of se-
mantic shift for target words in four languages, i.e.,
English, Latin, German, Swedish. The proposed so-
lutions outperform the baselines based on normalised
frequency difference or cosine distance methods.
5.6. Topic modelling
Topic modelling is another category of methods
proposed for the study of semantic change. Topic
modelling often refers to latent Dirichlet allocation
(LDA) [156], a probabilistic technique for modelling a
corpus by representing each document as a mixture of
topics and each topic as a distribution over words. LDA
is referred to either as an element of comparison or as
a basis for further extensions that take into account the
temporal dimension of word meaning evolution. Fr-
ermann and Lapata [68] draw ideas from such an ex-
tension, the dynamic topic modelling approach [157],
to build a dynamic Bayesian model of Sense ChANge
(SCAN) that defines word meaning as a set of senses
tracked over a sequence of contiguous time intervals.
In this model, senses are expressed as a probability
distribution over words, and given a word, its senses
are inferred for each time interval. According to [68],
SCAN is able to capture the evolution of a word’s
meaning over time and detect the emergence of new
senses, sense prevalence variation or changes within
individual senses such as meaning extension, shift, or
modification. Frermann and Lapata validate their find-
ings against WordNet and evaluate the performance of
their system on the SemEval-2015 benchmark datasets
released as part of the diachronic text evaluation exer-
cise.
Pölitz et al. [158] compare the standard LDA [156]
with the continuous time topic model [159] (called
“topics over time LDA” in the paper), for the task
of word sense induction (WSI) intended to automati-
cally find possible meanings of words in large textual
datasets. The method uses lists of key words in con-
text (KWIC) as documents, and is applied to two cor-
pora: the dictionary of the German language (DWDS)
core corpus of the 20th century and the newspaper cor-
pus Die Zeit covering the issues of the German weekly
newspaper from 1946 to 2009. The paper concludes
that standard LDA can be used, to a certain degree,
to identify novel meanings, while topics over time
LDA can make clearer distinctions between senses but
sometimes may result in too strict representations of
the meaning evolution.
16 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
[66, 67] apply the hierarchical Dirichlet process
technique [160], a non-parametric variant of LDA, to
detect word senses that are not attested in a reference
corpus and to identify novel senses found in a cor-
pus but not captured in a word sense inventory. The
two studies include experiments with various datasets,
such as selections from the BNC corpus (British En-
glish from the late 20th-century), ukWaC Web corpus
(built from the .uk domain in 2007), SiBol/Port collec-
tion (texts from several British newspapers from 1993,
2005, and 2010) and domain-specific corpora such as
sports and finance. Another example is [161] that ap-
plies topic modelling to the corpus of Hartlib Papers,
a multilingual collection of correspondence and other
papers of Samuel Hartlib (c.1600-1662) spanning the
period from 1620 to 1662, to identify changes in the
topics discussed in the letters. They then experimented
with using topic modelling to detect semantic change,
following the method developed in [162].
Based on these overviews and state of the art, we
can say that automatic lexical semantic change de-
tection is not yet a solved task in NLP, but a good
amount of progress has been achieved and a great
variety of systems have been developed and tested,
paving the way for further research and improvements.
An important aspect to stress is that this research has
rarely reached outside the remit of NLP, and relatively
few applications have involved humanities research
(e.g., [41, 42, 163]). This is not particularly surprising,
as it usually takes time for foundational research to find
its way into application areas. However, as pointed out
before (cf. [164]), given the high relevance of seman-
tic change research for the analysis of concept evolu-
tion, this lack of disciplinary dialogue and exchange is
a limiting factor and we hope that it will be addressed
by future multidisciplinary research projects.
6. NLP for generating ontological structures
While automatic detection of lexical semantic change
has shown advances in recent years despite a still in-
sufficient interdisciplinary dialogue, the field of gen-
erating ontologies from diachronic corpora and rep-
resenting them as linked data on the Web needs also
further development of multidisciplinary approaches
and exchanges, given the inherent complexity of the
work involved. In this section, we discuss the main as-
pects pertaining to this type of task (block 4, Fig. 1),
by taking account of previous research in areas such
as ontology learning, construction of ontological di-
achronic structures from texts and automatic genera-
tion of linked data.
6.1. Ontology learning
Iyer et al. [165] survey the various approaches for
(semi-)automatic ontology extraction and enrichment
from unstructured text, including research papers from
1995 to 2018. They identify four broad categories of
algorithms (similarity-based clustering, set-theoretic
approach, Web corpus-based and deep learning) allow-
ing for different types of ontology creation and updat-
ing, from clustering concepts in a hierarchy to learn-
ing and generating ontological representations for con-
cepts, attributes and attribute restrictions. The authors
perform an in-depth analysis of four “seminal algo-
rithms” representative for each category (guided ag-
glomerative clustering, C-PANKOW, formal concept
analysis and word2vec) and compare them using on-
tology evaluation measures such as contextual rele-
vance, precision and algorithmic efficiency. They also
propose a deep learning method based on LSTMs, to
tackle the problem of filtering out irrelevant data from
corpora and improve relevance of retained concepts in
a scalable manner.
Asim et al. [166] base their survey on the so-called
“ontology learning layer cake” (introduced by Buite-
laar et al. [167]), which illustrates the step-wise pro-
cess of ontology acquisition starting with terms, and
then moving up to concepts, concept hierarchy, re-
lations, relation hierarchy, axioms schemata, and fi-
nally axioms. The paper categorises ontology learning
techniques into linguistic, statistical and logical tech-
niques, and presents detailed analysis and evaluation
thereof. For instance, good performance is reported
in the linguistic category for (lexico-)syntactic parsing
and dependency analysis applied in relation extraction
from texts in various domains and languages. C/NC-
value (see also 6.3) and hierarchical clustering from
the statistical group are featured for the tasks of ac-
quiring concepts and relations respectively, while in-
ductive logical programming from the logical group
is mentioned for both tasks. Among the tools making
use of such techniques considered by the authors as
most prominent and widely used for ontology learn-
ing from text are Text2Onto [168], ASIUM [169]
and CRCTOL [170], in the category hybrid (linguistic
and statistical), OntoGain [171] and OntoLearn [172],
solely based on statistical methods, and TextStorm/-
Clouds [173] and Syndikate [174], from the logi-
cal category. Domain-specific or more wide-ranging
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 17
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
datasets, such as Reuters-21578 18 and the British Na-
tional Corpus, 19 are also included in the description,
as commonly used for testing and evaluating different
ontology learning systems. Although published just
one year earlier than [165], the survey does not men-
tion any techniques based on neural networks. How-
ever, the authors state that ontology learning can ben-
efit from incorporating deep learning methods into the
field. Importantly, Asim et al. advocate for language
independent ontology learning and for the necessity of
human intervention in order to boost the overall quality
of the outcome.
6.2. Diachronic constructs
He et al. [15] use the ontology learning layer cake
framework and a diachronic corpus in Chinese (Peo-
ple’s Daily Corpus), spanning from 1947 to 1996,
to construct a set of diachronic ontologies by year
and period. Their ontology learning system deals only
with the first four bottom layers of the ‘cake’ (see
also [166] and [167] above), for term extraction, syn-
onymy recognition, concept discovery and hierarchical
concept clustering. The first layer is built by segment-
ing and part of speech (POS) tagging the raw text using
a hierarchical hidden Markov model (HHMM) for Chi-
nese lexical analysis [175] and retaining all the words,
except for stopwords and low frequency items. For
synonymy detection, He et al. apply a distributional se-
mantic model taking into account both lexical and syn-
tactic contexts to compute the similarity between two
terms, a method already utilised in diachronic corpus
analysis in [176]. Cosine similarity and Kleinberg’s
“hubs and authorities” methodology [177] are used to
group terms and synonyms into concepts and to select
the top two terms with highest authority as semantic
tags or labels for the concepts. An iterative K-means
algorithm [178] is adopted to create a hierarchy of con-
cepts with highly semantically associated clusters and
sub-clusters. He et al. employ this four-step approach
to build yearly/period diachronic XML ontologies for
the considered corpus and evaluate concept discovery
and clustering by comparing their results with a base-
line computed via a Google word2vec implementation.
The authors report that the proposed method outper-
formed the baseline in both concept discovery and hi-
erarchical clustering, and that their diachronic ontolo-
18https://archive.ics.uci.edu/ml/datasets/reuters-21578+text+
categorization+collection
19http://www.natcorp.ox.ac.uk/
gies were able to capture semantic changes of a term
through comparison of its neighbouring terms or clus-
ters at different points in time, and detect the apparition
of new topics in a specific era. [15] also provides ex-
amples of diachronic analysis based on the ontologies
derived from the studied corpus, such as shift in mean-
ing from a domain to another, semantic change leading
to polysemy or emergence of new similar terms as a
result of real-world phenomena occurring in the period
covered by the considered textual sources.
Other papers addressed the question of conceptual-
ising semantic change using NLP techniques and di-
achronic corpora [146, 179, 180] implying various de-
grees of ontological formalisation.
Focusing on the way conceptual structures and the
hierarchical relations among their components evolve
over time, Bizzoni et al. [146] explore the direction of
using hyperbolic embeddings for the construction of
corpus-induced diachronic ontologies (see also Sub-
section 5.4). Using as a dataset the Royal Society Cor-
pus, with a time span from 1665 to 1869, they show
that such a method can detect symptoms of hierarchi-
sation and specialisation in scientific language. More-
over, they argue that this type of technology may of-
fer a (semi-)automatic alternative to the hand-crafted
historical ontologies that require considerable amount
of human expertise and skills to build hierarchies of
concepts based on beliefs and knowledge of a different
time.
In their analysis of changing relationships in tem-
poral corpora, Rosin and Radinsky [179] propose sev-
eral methods for constructing timelines that support
the study of evolving languages. The authors intro-
duce the task of timeline generation that implies two
components, one for identifying “turning points”, i.e.
points in time when the target word underwent signif-
icant semantic changes, the other for identifying as-
sociated descriptors, i.e. words and events, that ex-
plain these changes in relation with real-world triggers.
Their methodology includes techniques such as “peak
detection” in time series and “projected embeddings”,
in order to define the timeline turning points and cre-
ate a joint vector space for words and events, repre-
senting a specific time period. Different approaches
are tested to compare vector representations of the
same word or select the most relevant events caus-
ing semantic change over time, such as orthogonal
Procrustes [11], similarity-based measures, and su-
pervised machine learning (random forest, SVM and
neural networks). After assessing these methods on
datasets from Wikipedia, the New York Times archive
18 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
and DBpedia, Rosin and Radinsky conclude that the
best results are yielded by a supervised approach lever-
aging the projected embeddings, and the main factors
affecting the quality of the created timelines are word
ambiguity and the available amount of data and events
related to the target word. Although [179] does not ex-
plicitly refer to ontology acquisition as a whole, au-
tomatic timeline generation provides insight into the
modalities of detecting and conceptualising seman-
tic change and word-event-time relationships that may
serve with the task of corpus-based diachronic ontol-
ogy generation.
Gulla et al. [180] use “concept signatures”, i.e.
representations constructed automatically from textual
descriptions of existing concepts, to capture seman-
tic changes of concepts over time. A concept signa-
ture is represented as a vector of weights. Each ele-
ment in the vector corresponds to a linguistic unit or
term (e.g. noun or noun phrase) extracted from the tex-
tual description of the concept, with its weight calcu-
lated as a tf-idf (term frequency - inverted document
frequency) score. The process of signature building
includes POS tagging, stopword removal, lemmatisa-
tion, noun/phrase selection and tf-idf computing for
the selected linguistic units. According to Gulla et al.,
this type of vector representation enables comparisons
via standard information retrieval measures, such as
cosine similarity and Euclidian distance, that can un-
cover semantic drift of concepts in the ontology, both
with respect to real-world phenomena (extrinsic drift)
and inter-concept (taxonomic and non-taxonomic) re-
lationships (intrinsic drift). The proposed methodol-
ogy is applied to an ontology based on the Det Norske
Veritas (DNV) company’s Web site, 20 each Web page
representing a concept. The text of the Web pages is
used as a source for understanding the concepts and
constructing the corresponding signatures at different
points in time. [180] illustrates this procedure for var-
ious types of vector-based concept and relation com-
parison in the DNV ontology, computed for 2004 and
2008. The authors note that the size of the textual de-
scriptions of concepts is determinant for the signature
quality (too short descriptions may result in poor qual-
ity) and mention as further direction of research the
use of deeper grammatical analysis of sentences and
of semantic lexica for signature generation. Moreover,
Gulla et al. point out that since the automatic con-
struction of signatures relies on textual descriptions of
20A company specialising in risk management and certification.
existing concepts, the approach is primarily intended
to updating existing structures rather than developing
new ontologies.
6.3. Generating linked data
The transformation of the extracted information into
formal descriptions that can be published as linked
data on the Web is an important aspect of the process
of ontology generation from textual sources. A num-
ber of tools have been devised to implement an inte-
grated workflow for extracting concepts and relations,
and converting the derived ontological structure into
Semantic Web formalisations. While the first and sec-
ond subsections above provided an overview of var-
ious approaches for corpus-based production of on-
tologies and ontological constructs including a tempo-
ral dimension, this subsection focuses on means for
making the generated output available on the Web in
a structured and re-usable format. Three categories of
tools dedicated to such tasks are discussed, for extract-
ing information and linking entities to available on-
tologies on the Web, learning ontologies and translat-
ing the resulting models into Semantic Web represen-
tations, and for performing shallow conversion to RDF.
An example from the first category is LODifier [181],
which combines different NLP techniques for named
entity recognition, word sense disambiguation and se-
mantic analysis to extract entities and relations from
text and produce RDF representations linked to the
LOD cloud using DBpedia and WordNet 3.0 vocab-
ularies. The tool was evaluated on an English bench-
mark dataset containing newspapers, radio and televi-
sion news from 1998.
From the second category, OntoGain [171] is a plat-
form for unsupervised ontology acquisition from un-
structured text. The concept identification module is
based on C/NC-value [182], a method that enables the
extraction of multi-word and nested terms from text.
For the detection of taxonomic and non-taxonomic re-
lations, [171] applies techniques such as agglomera-
tive hierarchical clustering and formal concept analy-
sis in the first task, and association rules and condi-
tional probabilities in the second. OntoGain allows for
the transformation of the resulted ontology into stan-
dard OWL statements. The authors report assessment
including experiments with corpora from the medical
and computer science domain, and comparisons with
hand-crafted ontologies and similar applications such
as Text2Onto.
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 19
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
Concept-Relation-Concept Tuple-based Ontology
Learning (CRCTOL) [170] is a system for automat-
ically mining ontologies from domain-specific docu-
ments. CRCTOL adopts various NLP methods such as
POS tagging, multi-word extraction and tf-idf-based
relevance measures for concept learning, a variant of
Lesk’s algorithm [183] for word sense disambigua-
tion, and WordNet hierarchy processing and full text
parsing for the construction of taxonomic and non-
taxonomic relations. The derived ontology is then
modelled as a graph, with the possibility of exporting
the corresponding representation in RDFS and OWL
format. [170] presents two case studies, for building a
terrorism domain ontology and a sport event domain
ontology, as well as results of quantitative and qualita-
tive evaluation of the tool through various comparisons
with other systems or assessment references such as
Text-To-Onto/Text2Onto, WordNet, expert rating and
human-edited benchmark ontologies.
One of the systems often cited as a reference in on-
tology learning from textual resources (see also above)
is Text2Onto (the successor of TextToOnto) [168].
Based on the GATE framework [184], it combines
linguistic pre-processing (e.g. tokenisation, sentence
splitting, POS tagging, lemmatisation) with the use
of a JAPE transducer and shallow parsing run on the
pre-processed corpus to identify concepts, instances
and different types of relations (subclass-of, part-of,
instance-of, etc.) to be included in a Probabilistic
Ontology Model (POM). The model, independent of
any knowledge representation formalism, can be then
translated into various ontology representation lan-
guages such as RDFS, OWL and F-Logic. The paper
also describes a strategy for data-driven change dis-
covery allowing for selective POM updating and trace-
ability of the ontology evolution, consistent with the
changes in the underlying corpus. Evaluation is re-
ported with respect to certain tasks and a collection of
tourism-related texts, the results being compared with
a reference taxonomy for the domain.
Recent work accounts for more specialised tools,
from the third category, such as converters, making,
for instance, linked data in RDF format out of CSV
files (CoW 21 and cattle 22 [5]) or directly converting
language resources into LL(O)D (LLODifier 23 [185]).
As already pointed out at the beginning of this section,
the field may benefit from further exchanges among
21https://pypi.org/project/cow-csvw/
22http://cattle.datalegend.net/
23https://github.com/acoli-repo/LLODifier
scholars in different areas of studies such as theoreti-
cal and cognitive linguistics, history and philosophy of
language, digital humanities, NLP and Semantic Web.
7. LL(O)D resources and publication
In this section (related to block 5, Fig. 1), we outline
the existing resources on the Web including diachronic
representation of data from the humanities, with a view
towards the possibilities of integrating more resources
of this kind into the LL(O)D cloud in the future.
The main nucleus for linguistic linked open data
is the LL(O)D cloud [186], 24 which started in 2011
with less than 30 datasets, and at the time of writ-
ing consists of over 200 different datasets. The re-
sources linked in the LL(O)D cloud include corpora,
lexicons and dictionaries, terminologies, thesauri and
knowledge bases, linguistic resources metadata, lin-
guistic data categories, and typological databases. The
LL(O)D diagram is generated automatically from the
subset of Linghub 25 that is published as linked open
data.
Not all diachronic datasets are registered through
Linghub/LL(O)D Cloud. Within the CLARIAH project 26
several datasets have been converted from CSV for-
mat to linked open data, and published through project
websites or GitHub. For example, in [187], differ-
ent diachronic lexicons are modelled according to the
Lemon model and interlinked, such that one can query
across time and dialect variations.
Also in the Netherlands, the Amsterdam Time Ma-
chine connects attestations of Amsterdam dialects and
sociolects, cinema and theatre locations and tax infor-
mation to base maps of Amsterdam at various points
in time [188]. A combined resource like this allows
scholars to investigate ‘higher’ and ‘lower’ sociolects
in conjunction with ‘elite density’ in a neighbourhood
(i.e. the proportion of wealthier people that lived in
an area). Lexicologists at the Dutch Language Institute
have been creating dictionaries of Dutch that cover the
period from 500 to 1976 which are now being mod-
elled through OntoLex-Lemon [189].
Searching for and modelling diachronic change re-
quires rethinking some contemporary (Semantic) Web
infrastructure. As [190] shows, standardised language
24https://linguistic-lod.org/
25http://linghub.org
26https://clariah.nl
20 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
tags cannot capture the differences between Old-,
Middle- and Modern French resources.
Digital editions, often modelled in TEI [191], are a
rich resource of diachronic language variation. Some
corpora, such as the 15th-19th-century Spanish poetry
corpus described in [192] contain additional annota-
tions such as psychological and affective labels, but it
seems the study was not focused particularly on how
these aspects may have changed over time.
For humanities scholars such as historians, who deal
with source materials dating back to for example the
early modern period, language change is a given, but
the knowledge they gain over time is not always for-
malised or published as linked data. For example, a
project that analyses the representation of emotions
plays from the 17th to the 19th century, a dataset and
lexicon were developed, but these were not explicitly
linked to the LL(O)D cloud [193, 194]. 27 In con-
trast to [192], here the labels are explicitly grounded
in time. There is a task here for the Semantic Web
community to make it easier to publish and maintain
LL(O)D datasets for non-Semantic Web experts.
It should be also noted that while there do not
currently exist guidelines for publishing lexicons and
ontologies representing semantic change as LL(O)D
data, there are moves towards producing such material
within the Nexus Linguarum COST Action, however,
with particular reference to the overlap between differ-
ent working groups and UC4.2.1.
8. Conclusions
This paper presents a literature survey, bringing to-
gether various fields of research that may be of interest
in the construction of a workflow for detecting and rep-
resenting semantic change (Fig. 1). The state of the art
described in the paper also represents the starting point
in designing a methodology, based on this workflow,
for the humanities use case UC4.2.1 as an application
within the COST Action Nexus Linguarum, European
network for Web-centred linguistic data science. The
survey touches upon the use of multilingual diachronic
corpora from the humanities, and different approaches
from linguistics-related disciplines, NLP and Semantic
Web. The organisation of the sections and the themes
included in the outline reflects the heterogeneity and
complexity of the task and the necessity of a frame-
27https://www.esciencecenter.nl/projects/
from-sentiment-mining-to-mining- embodied-emotions/
work enabling interdisciplinary dialogue and collabo-
ration.
At this stage, the reviewed literature and main sur-
veyed approaches and tools (see Appendix) suggest
that the theoretical frameworks (Section 3) and the
NLP techniques for detecting lexical semantic change
(Section 5) show good levels of development, although
certain conceptual and technical difficulties are yet
to overcome. The fields dealing with the generation
of diachronic ontologies from unstructured text and
their representation as LL(O)D formalisms on the Web
(Section 4, 6, 7) would require further harmonisation
with the previous points and research investment.
Despite recent advances in creating and publish-
ing linguistic resources on the LL(O)D cloud, and the
availability of potentially relevant resources, human-
ities researchers working on the detection and repre-
sentation of semantic change as linked data on the
Web are still confronted with a series of challenges.
These include limitations in representing temporal and
dynamic aspects given the work in progress status of
some of the applicable Semantic Web technologies,
absence of guidelines for producing diachronic ontolo-
gies, and lack of ways to ease publication and main-
tenance of data for non-Semantic Web experts. An-
other point requiring further attention is the need for
building connections between the various areas of re-
search involved in the type of task described in the pa-
per. As we tried to illustrate through the structure of
the generic workflow and the discussions within the
related sections, the research agenda for attaining this
goal should include interdisciplinary approaches and
exchanges among the identified fields of study. The
results of the survey seem to suggest that there are
not yet enough interrelations and explicit connections
between these fields, and the area under investigation
would benefit from further developments in this direc-
tion.
We assume that, given the current progress in deep
learning, digital humanities and the ongoing under-
takings in LL(O)D, the detection and representation
of semantic change as linked data combined with the
analysis of large datasets from the humanities will ac-
quire the level of attention and dialogue needed for
the advancement in this area of study. Detecting and
representing semantic change as LL(O)D is an impor-
tant topic for the future development of Semantic Web
technologies, since learning to deal with the knowl-
edge of the past and its evolution over time also implies
learning to deal with the knowledge of the future.
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 21
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
Acknowledgment
This article is based upon work from COST Action
Nexus Linguarum, European network for Web-centred
linguistic data science, supported by COST (European
Cooperation in Science and Technology). www.cost.
eu.
References
[1] T. Burrows, E. Hyvönen, L. Ransom and H. Wijsman,
Mapping Manuscript Migrations: Digging into Data for
the History and Provenance of Medieval and Renaissance
Manuscripts, Manuscript Studies: A Journal of the Schoen-
berg Institute for Manuscript Studies 3(1) (2018), 249–252.
[2] L. Isaksen, R. Simon, E.T. Barker and P. de Soto Cañamares,
Pelagios and the emerging graph of ancient world data, in:
Proceedings of the 2014 ACM conference on Web science,
2014, pp. 197–201.
[3] M.D. Wilkinson, M. Dumontier, I.J. Aalbersberg, G. Ap-
pleton, M. Axton, A. Baak, N. Blomberg, J.-W. Boiten,
L.B. da Silva Santos, P.E. Bourne, J. Bouwman, A.J. Brookes,
T. Clark, M. Crosas, I. Dillo, O. Dumon, S. Edmunds,
C.T. Evelo, R. Finkers, A. Gonzalez-Beltran, A.J.G. Gray,
P. Groth, C. Goble, J.S. Grethe, J. Heringa, P.A.C. ’t Hoen,
R. Hooft, T. Kuhn, R. Kok, J. Kok, S.J. Lusher, M.E. Mar-
tone, A. Mons, A.L. Packer, B. Persson, P. Rocca-Serra,
M. Roos, R. van Schaik, S.-A. Sansone, E. Schultes, T. Sen-
gstag, T. Slater, G. Strawn, M.A. Swertz, M. Thompson,
J. van der Lei, E. van Mulligen, J. Velterop, A. Waag-
meester, P. Wittenburg, K. Wolstencroft, J. Zhao and B. Mons,
The FAIR Guiding Principles for scientific data manage-
ment and stewardship, Scientific Data 3(1) (2016), 160018.
doi:10.1038/sdata.2016.18.
[4] A. Meroño-Peñuela, A. Ashkpour, M. van Erp, K. Man-
demakers, L. Breure, A. Scharnhorst, S. Schlobach and
F. van Harmelen, Semantic technologies for historical re-
search: A survey, Semantic Web 6(6) (2014), 539–564.
doi:10.3233/SW-140158.
[5] A. Meroño-Peñuela, V. de Boer, M. van Erp, W. Melder,
R. Mourits, R. Schalk and R. Zijdeman, Ontologies in CLAR-
IAH: Towards Interoperability in History, Language and Me-
dia, https://arxiv.org/abs/2004.02845v2 (2020), 26.
[6] C. Chiarcos and A. Pareja-Lora, Open Data—Linked
Data—Linked Open Data—Linguistic Linked Open Data
(LLOD): A General Introduction, in: Development of linguis-
tic linked open data resources for collaborative data-intensive
research in the language sciences, A. Pareja-Lora, M. Blume,
B.C. Lust and C. Chiarcos, eds, MIT Press, 2019, pp. 1–18.
ISBN 978-0-262-53625-7.
[7] N. Tahmasebi, L. Borin and A. Jatowt, Survey of Computa-
tional Approaches to Lexical Semantic Change, arXiv: Com-
putation and Language (2018).
[8] X. Tang, A state-of-the-art of semantic change computa-
tion, Natural Language Engineering 24(5) (2018), 649–676.
doi:10.1017/S1351324918000220.
[9] A. Kutuzov, L. Øvrelid, T. Szymanski and E. Velldal, Di-
achronic word embeddings and semantic shifts: A survey, in:
Proceedings of the 27th International Conference on Compu-
tational Linguistics, Association for Computational Linguis-
tics, Santa Fe, New Mexico, USA, 2018, pp. 1384–1397.
[10] D. Schlechtweg, B. McGillivray, S. Hengchen, H. Du-
bossarsky and N. Tahmasebi, SemEval-2020 Task 1: Unsu-
pervised Lexical Semantic Change Detection, in: Proceedings
of the 14th International Workshop on Semantic Evaluation,
Association for Computational Linguistics, Barcelona, Spain,
2020.
[11] W.L. Hamilton, J. Leskovec and D. Jurafsky, Diachronic
Word Embeddings Reveal Statistical Laws of Semantic
Change, in: Proceedings of the 54th Annual Meeting of
the Association for Computational Linguistics, Vol. 1, 2016,
pp. 1489–1501.
[12] S. Wang, S. Schlobach and M. Klein, Concept drift and how
to identify it, Journal of Web Semantics First Look (2011).
http://dx.doi.org/10.2139/ssrn.3199520.
[13] A. Fokkens, S. Ter Braake, I. Maks and D. Ceolin, On the
Semantics of Concept Drift: Towards Formal Definitions of
Semantic Change, Drift-a-LOD@EKAW (2016).
[14] T. McEnery and A. Hardie, Corpus-based studies of syn-
chronic and diachronic variation, in: Corpus Linguis-
tics: Method, Theory and Practice, Cambridge Univer-
sity Press, 2011, pp. 94–121. ISBN 978-0-511-98139-5.
doi:10.1017/CBO9780511981395. http://ebooks.cambridge.
org/ref/id/CBO9780511981395.
[15] S. He, X. Zou, L. Xiao and J. Hu, Construction of Diachronic
Ontologies from People’s Daily of Fifty Years, Proceedings
of the Ninth International Conference on Language Resources
and Evaluation (LREC’14) (2014).
[16] M. Richter, The History of Political and Social Concepts: A
Critical Introduction, Oxford University Press, 1995.
[17] J.-M. Kuukkanen, Making Sense of Conceptual Change
47(3) (2008), 351–372. doi:https://doi.org/10.1111/j.1468-
2303.2008.00459.x. https://onlinelibrary.wiley.com/doi/abs/
10.1111/j.1468-2303.2008.00459.x.
[18] T.G. Stavropoulos, S. Andreadis, M. Riga, E. Kontopoulos,
P. Mitzias and I. Kompatsiaris, A Framework for Measuring
Semantic Drift in Ontologies, 2016.
[19] M. Fitting, Intensional Logic, in: The Stanford Encyclopedia
of Philosophy, Spring 2020 edn, E.N. Zalta, ed., Metaphysics
Research Lab, Stanford University, 2020. https://plato.
stanford.edu/archives/spr2020/entries/logic-intensional/.
[20] F. de Saussure, Cours de linguistique générale (1916), Payot,
1971. https://fr.wikisource.org/wiki/Cours_de_linguistique_
g%C3%A9n%C3%A9rale.
[21] A. Betti and H. van den Berg, Modelling the History of Ideas,
British Journal for the History of Philosophy 22(4) (2014),
812–835. doi:10.1080/09608788.2014.949217.
[22] J. McCrae, D. Spohr and P. Cimiano, Linking lexical re-
sources and ontologies on the semantic web with lemon, in:
Extended Semantic Web Conference, Springer, 2011, pp. 245–
259.
[23] G. Widmer and M. Kubat, Learning in the presence of concept
drift and hidden contexts, Machine Learning, Kluwer Aca-
demic Publishers, Boston. Manufactured in The Netherlands
23(1) (1996), 69–101.
22 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
[24] G. Antoniou, M. d’Aquin and J.Z. Pan, Semantic Web dy-
namics, Journal of Web Semantics 9(3) (2011), 245–246.
doi:10.1016/j.websem.2011.06.008.
[25] N.B. Kvastad, Semantics in the Methodology of the History
of Ideas, Journal of the History of Ideas, University of Penn-
sylvania Press 38(1) (1977), 157–174.
[26] D. Geeraerts, Theories of lexical semantics, Oxford Univer-
sity Press, 2010. ISBN 978-0-19-870031-9.
[27] S. Grondelaers, D. Speelman and D. Geeraerts, Lexical vari-
ation and change, in: The Oxford handbook of cognitive lin-
guistics, 2007.
[28] C. Roche, Ontoterminology: How to unify terminology
and ontology into a single paradigm, in: LREC 2012 -
Eighth international conference on Language Resources and
Evaluation, 2012, pp. 2626–2630. http://christophe-roche.fr/
Bibliographie/2012/567_Paper_Header.pdf.
[29] D. Gromann, Terminology meets the multilingual Semantic
Web: A semiotic comparison of ontologies and terminolo-
gies, in: Languages for Special Purposes in a Multilingual,
Transcultural World, Proceedings of the 19th European Sym-
posium on Languages for Special Purposes, G. Budin and
V. Lušicky, eds, University of Vienna, 2013, pp. 418–428.
ISBN 978-3-200-03674-1.
[30] R. Temmerman, Towards New Ways of Terminology De-
scription: The sociocognitive approach, Terminology and
Lexicography Research and Practice, Vol. 3, John Ben-
jamins Publishing Company, 2000. ISBN 978-90-272-2326-
5. doi:10.1075/tlrp.3. http://www.jbe-platform.com/content/
books/9789027298638.
[31] D. Schiffrin, Discourse markers, Vol. 5, Cambridge Univer-
sity Press, 1987.
[32] B. Fraser, Pragmatic markers, Pragmatics 6(2) (1996), 167–
190.
[33] K. Aijmer, I think–an English modal particle, Modality in
Germanic languages: Historical and comparative perspec-
tives 1(1997), 47.
[34] B. Fraser, What are discourse markers?, Journal of pragmat-
ics 31(7) (1999), 931–952.
[35] D. Schiffrin, Discourse marker research and theory: revisiting
and, Approaches to discourse particles 1(2006), 315–338.
[36] P. Auer and Y. Maschler, NU/NÅ: A family of discourse mark-
ers across the languages of Europe and beyond, Vol. 58, Wal-
ter de Gruyter GmbH & Co KG, 2016.
[37] R. Waltereit and U. Detges, Different functions, different
histories. Modal particles and discourse markers from a di-
achronic point of view, Catalan journal of linguistics (2007),
61–80.
[38] L.S. Stvan, Diachronic change in the uses of the dis-
course markers why and say in American English, Linguistic
Insights-Studies in Language and Communication 25 (2006),
61–76.
[39] L. Downing, The Cambridge Introduction to Michel Foucault
(2008).
[40] R. Wodak, Critical Discourse Analysis, Discourse-Historical
Approach, in: The International Encyclopedia of Lan-
guage and Social Interaction, 1st edn, K. Tracy, T. Sandel
and C. Ilie, eds, Wiley, 2015. ISBN 978-1-118-61110-4.
doi:10.1002/9781118611463.
[41] L. Viola and J. Verheul, One Hundred Years of Mi-
gration Discourse in The Times: A Discourse-Historical
Word Vector Space Approach to the Construction of
Meaning, Frontiers in Artificial Intelligence 3(2020), 64.
doi:10.3389/frai.2020.00064.
[42] S. Soni, L. Klein and J. Eisenstein, Abolitionist Net-
works: Modeling Language Change in Nineteenth-Century
Activist Newspapers, arXiv:2103.07538 [cs] (2021), arXiv:
2103.07538. http://arxiv.org/abs/2103.07538.
[43] J.P. McCrae, J. Bosque-Gil, J. Gracia, P. Buitelaar and
P. Cimiano, The OntoLex-Lemon Model: Development and
Applications (2017), 587–597, Publisher: Lexical Comput-
ing CZ s.r.o. https://elex.link/elex2017/wp-content/uploads/
2017/09/paper36.pdf.
[44] N. Guarino, D. Oberle and S. Staab, What Is an On-
tology?, in: Handbook on Ontologies, S. Staab and
R. Studer, eds, International Handbooks on Information Sys-
tems, Springer, Berlin, Heidelberg, 2009, pp. 1–17. ISBN
9783540926733. doi:10.1007/978-3-540-92673-3_0. https://
doi.org/10.1007/978-3-540-92673-3_0.
[45] C. Chiarcos, M. Ionov, J. de Does, K. Depuydt, A.F. Khan,
S. Stolk, T. Declerck and J.P. McCrae, Modelling Frequency
and Attestations for OntoLex-Lemon, in: Proceedings of the
2020 Globalex Workshop on Linked Lexicography, European
Language Resources Association, Marseille, France, 2020,
pp. 1–9. ISBN 979-10-95546-46-7. https://www.aclweb.org/
anthology/2020.globalex-1.1.
[46] S. Salmon-Alt, Data structures for etymology: towards an et-
ymological lexical network., BULAG 31 (2006), 1–12.
[47] J. Bowers and L. Romary, Deep encoding of etymological
information in TEI, Journal of the Text Encoding Initiative
(2016).
[48] L. Romary, M. Khemakhem, F. Khan, J. Bowers, N. Calzo-
lari, M. George, M. Pet and P. Ba´
nski, LMF Reloaded, arXiv
preprint arXiv:1906.02136 (2019).
[49] F. Khan, L. Romary, A. Salgado, J. Bowers, M. Khemakhen
and T. Tasovac, Modelling Etymology in LMF/TEI, in: Pro-
ceedings of the 12th Conference on Language Resources and
Evaluation (LREC 2020), European Language Resources As-
sociation (ELRA), 2020.
[50] G. de Melo, Etymological Wordnet: Tracing The History of
Words., in: Proceedings of the 9th Conference on Language
Resources and Evaluation (LREC 2014), European Language
Resources Association (ELRA), 2014.
[51] C. Chiarcos, F. Abromeit, C. Fäth and M. Ionov, Etymology
Meets Linked Data. A Case Study In Turkic., in: Digital Hu-
manities 2016. Krakow, 2016.
[52] F. Khan, Towards the Representation of Etymological and
Diachronic Lexical Data on the Semantic Web, in: Proceed-
ings of the Eleventh International Conference on Language
Resources and Evaluation (LREC 2018), Miyazaki, Japan.
European Language Resources Association (ELRA). event-
place: Miyazaki, Japan, 2018.
[53] F. Khan, J. Díaz-Vera and M. Monachini, Representing mean-
ing change in computational lexical resources; the case of
shame and embarrassment in Old English, Formal Represen-
tation and the Digital Humanities (2018), 59.
[54] F. Mambrini and M. Passarotti, Representing Etymology in
the LiLa Knowledge Base of Linguistic Resources for Latin,
in: Proceedings of the 2020 Globalex Workshop on Linked
Lexicography, European Language Resources Association,
Marseille, France, 2020, pp. 20–28. ISBN 979-10-95546-46-
7. https://www.aclweb.org/anthology/2020.globalex-1.3.
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 23
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
[55] F. Khan, A. Bellandi and M. Monachini, Tools and Instru-
ments for Building and Querying Diachronic Computational
Lexica, in: Proceedings of the Workshop on Language Tech-
nology Resources and Tools for Digital Humanities (LT4DH),
The COLING 2016 Organizing Committee, pp. 164–171.
https://www.aclweb.org/anthology/W16-4022.
[56] F. Rizzolo, Y. Velegrakis, J. Mylopoulos and S. Bykau, Mod-
eling concept evolution: a historical perspective, in: Interna-
tional Conference on Conceptual Modeling, Springer, 2009,
pp. 331–345.
[57] C. Gutierrez, C. Hurtado and A. Vaisman, Temporal RDF, in:
The Semantic Web: Research and Applications, A. Gómez-
Pérez and J. Euzenat, eds, Springer Berlin Heidelberg, Berlin,
Heidelberg, 2005, pp. 93–107. ISBN 978-3-540-31547-6.
[58] P. Garbacz and R. Trypuz, Representation of Tensed Rela-
tions in OWL, in: Metadata and Semantic Research, Vol. 755,
E. Garoufallou, S. Virkus, R. Siatri and D. Koutsomiha,
eds, Springer International Publishing, 2017, pp. 62–73,
Series Title: Communications in Computer and Informa-
tion Science. ISBN 978-3-319-70862-1 978-3-319-70863-8.
doi:10.1007/978-3-319-70863-8_6. http://link.springer.com/
10.1007/978-3-319-70863-8_6.
[59] C. Welty, R. Fikes and S. Makarios, A reusable ontology for
fluents in OWL, in: FOIS, Vol. 150, 2006, pp. 226–236.
[60] H.-U. Krieger, A detailed comparison of seven approaches
for the annotation of time-dependent factual knowledge in
RDF and OWL, in: Proceedings 10th Joint ISO-ACL SIGSEM
Workshop on Interoperable Semantic Annotation, 2014, p. 1.
[61] F. Khan and J. Bowers, Towards a Lexical Standard for the
Representation of Etymological Data, in: Convegno annuale
dell’Associazione per l’Informatica Umanistica e la Cultura
Digitale, 2020.
[62] J.R. Hobbs and F. Pan, Time ontology in OWL, W3C working
draft 27 (2006), 133.
[63] J.F. Allen, Maintaining knowledge about temporal intervals,
Communications of the ACM 26(11) (1983), 832–843.
[64] S. Batsakis, E.G. Petrakis, I. Tachmazidis and G. Antoniou,
Temporal representation and reasoning in OWL 2, Semantic
Web 8(6) (2017), 981–1000.
[65] P. Golden and R. Shaw, Period assertion as nanopublication:
The PeriodO period gazetteer, in: Proceedings of the 24th In-
ternational Conference on World Wide Web, 2015, pp. 1013–
1018.
[66] P. Cook, J.H. Lau, D. McCarthy and T. Baldwin, Novel word-
sense identification, in: Proceedings of COLING 2014, the
25th International Conference on Computational Linguistics:
Technical Papers, 2014, pp. 1624–1635.
[67] J.H. Lau, P. Cook, D. McCarthy, S. Gella and T. Bald-
win, Learning word sense distributions, detecting unattested
senses and identifying novel senses using topic models, in:
Proceedings of the 52nd Annual Meeting of the Associa-
tion for Computational Linguistics (Volume 1: Long Papers),
Vol. 1, 2014, pp. 259–270.
[68] L. Frermann and M. Lapata, A Bayesian model of diachronic
meaning change, Transactions of the Association for Compu-
tational Linguistics 4(2016), 31–45.
[69] S. Mitra, R. Mitra, S.K. Maity, M. Riedl, C. Biemann,
P. Goyal and A. Mukherjee, An automatic approach to iden-
tify word sense changes in text media across timescales, Nat-
ural Language Engineering 21(5) (2015), 773–798.
[70] N. Tahmasebi and T. Risse, Finding Individual Word Sense
Changes and their Delay in Appearance, in: Proceedings of
the International Conference Recent Advances in Natural
Language Processing, RANLP 2017, 2017, pp. 741–749.
[71] Y. Kim, Y. Chiu, K. Hanaki, D. Hegde and S. Petrov, Tempo-
ral Analysis of Language through Neural Language Models,
in: LTCSS@ACL, Association for Computational Linguistics,
2014, pp. 61–65.
[72] P. Basile and B. McGillivray, Discovery Science, in Lec-
ture Notes in Computer Science, Vol. 11198, Springer-Verlag,
2018, Chapter Exploiting the Web for Semantic Change De-
tection.
[73] V. Kulkarni, R. Al-Rfou, B. Perozzi and S. Skiena, Statisti-
cally significant detection of linguistic change, in: Proceed-
ings of the 24th International Conference on World Wide Web,
International World Wide Web Conferences Steering Com-
mittee, 2015, pp. 625–635.
[74] H. Dubossarsky, D. Weinshall and E. Grossman, Outta Con-
trol: Laws of Semantic Change and Inherent Biases in Word
Representation Models, in: Proceedings of the 2017 Confer-
ence on Empirical Methods in Natural Language Processing,
2017, pp. 1136–1145.
[75] N. Tahmasebi, A Study on Word2Vec on a Historical Swedish
Newspaper Corpus, in: CEUR Workshop Proceedings. Vol.
2084. Proceedings of the Digital Humanities in the Nordic
Countries 3rd Conference, Helsinki Finland, March 7-9,
2018., University of Helsinki, Faculty of Arts, Helsinki,
2018.
[76] M. Rudolph and D. Blei, Dynamic Embeddings for Language
Evolution, in: Proceedings of the 2018 World Wide Web Con-
ference on World Wide Web, 2018, pp. 1003–1011.
[77] A. Jatowt, R. Campos, S.S. Bhowmick, N. Tahmasebi and
A. Doucet, Every Word has its History: Interactive Explo-
ration and Visualization of Word Sense Evolution, in: Pro-
ceedings of the 27th ACM International Conference on Infor-
mation and Knowledge Management, ACM, 2018, pp. 1899–
1902.
[78] A. Kutuzov, Distributional word embeddings in modeling di-
achronic semantic change, PhD thesis, University of Oslo,
2020.
[79] V. Perrone, M. Palma, S. Hengchen, A. Vatri, J.Q. Smith
and B. McGillivray, GASC: Genre-Aware Semantic Change
for Ancient Greek, in: Proceedings of the 1st International
Workshop on Computational Approaches to Historical Lan-
guage Change, Association for Computational Linguistics,
Florence, Italy, 2019, pp. 56–66. https://www.aclweb.org/
anthology/W19-4707.
[80] H. Dubossarsky, S. Hengchen, N. Tahmasebi and
D. Schlechtweg, Time-Out: Temporal Referencing for Robust
Modeling of Lexical Semantic Change, in: Proceedings of
the 57th Annual Meeting of the Association for Computa-
tional Linguistics (Volume 1: Long Papers), Association for
Computational Linguistics, Florence, Italy, 2019.
[81] P. Shoemark, F. Ferdousi Liza, D. Nguyen, S. Hale and
B. McGillivray, Room to Glo: A Systematic Comparison of
Semantic Change Detection Approaches with Word Embed-
dings, in: Proceedings of the 2019 Conference on Empirical
Methods in Natural Language Processing and the 9th Inter-
national Joint Conference on Natural Language Processing
(EMNLP-IJCNLP), Association for Computational Linguis-
tics, 2019, pp. 66–76.
24 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
[82] M. Piotrowski, Natural Language Processing for Historical
Texts, Morgan & Claypool, 2012.
[83] B. McGillivray, Methods in Latin Computational Linguistics,
Brill, Leiden, 2014.
[84] P. Rayson, D.E. Archer, A. Baron, J. Culpeper and N. Smith,
Tagging the Bard: Evaluating the accuracy of a modern POS
tagger on Early Modern English corpora, in: Proceedings of
the Corpus Linguistics conference: CL2007, 2007.
[85] S. Scheible, R.J. Whitt, M. Durrell and P. Bennett, Evaluating
an ‘off-the-shelf’POS-tagger on Early Modern German text,
in: Proceedings of the 5th ACL-HLT workshop on language
technology for cultural heritage, social sciences, and human-
ities, 2011, pp. 19–23.
[86] M. Bollmann, A large-scale comparison of historical text
normalization systems, arXiv preprint arXiv:1904.02036
(2019).
[87] A. Baron and P. Rayson, VARD2: A tool for dealing with
spelling variation in historical corpora, in: Postgraduate con-
ference in corpus linguistics, 2008.
[88] M. Bollmann, automatic normalization of historical texts us-
ing distance measures and the Norma tool, in: Proceedings of
the second workshop on annotation of corpora for research in
the humanities (ACRH-2), Lisbon, Portugal, 2012, pp. 3–14.
[89] M. Bollmann, F. Petran and S. Dipper, Rule-based normaliza-
tion of historical texts, in: Proceedings of the Workshop on
Language Technologies for Digital Humanities and Cultural
Heritage, 2011, pp. 34–42.
[90] J. Porta, J.-L. Sancho and J. Gómez, Edit transducers for
spelling variation in Old Spanish, in: Proceedings of the work-
shop on computational historical linguistics at NODALIDA
2013; May 22-24; 2013; Oslo; Norway. NEALT Proceed-
ings Series 18, Linköping University Electronic Press, 2013,
pp. 70–79.
[91] I. Etxeberria, I. Alegria, L. Uria and M. Hulden, Evaluating
the noisy channel model for the normalization of historical
texts: Basque, Spanish and Slovene, in: Proceedings of the
Tenth International Conference on Language Resources and
Evaluation (LREC’16), 2016, pp. 1064–1069.
[92] K. Jassem, F. Grali´
nski, T. Obr˛ebski and P. Wierzcho´
n, Au-
tomatic Diachronic Normalization of Polish Texts, Investiga-
tiones Linguisticae 37 (2017), 17–33.
[93] M. Kestemont, W. Daelemans and G. De Pauw, Weigh your
words—memory-based lemmatization for Middle Dutch, Lit-
erary and Linguistic Computing 25(3) (2010), 287–301.
[94] E. Pettersson, B. Megyesi and J. Nivre, Normalisation of his-
torical text using context-sensitive weighted Levenshtein dis-
tance and compound splitting, in: Proceedings of the 19th
Nordic conference of computational linguistics (Nodalida
2013), 2013, pp. 163–179.
[95] Y. Adesam, M. Ahlberg and G. Bouma, bokstaffua, bok-
staffwa, bokstafwa, bokstaua, bokstawa... Towards lexical
link-up for a corpus of Old Swedish., in: KONVENS, 2012,
pp. 365–369.
[96] H. van Halteren and M. Rem, Dealing with orthographic vari-
ation in a tagger-lemmatizer for fourteenth century Dutch
charters, Language resources and evaluation 47(4) (2013),
1233–1259.
[97] C. Oravecz, B. Sass and E. Simon, Semi-automatic normal-
ization of Old Hungarian codices, in: Proceedings of the
ECAI 2010 Workshop on Language Technology for Cultural
Heritage, Social Sciences, and Humanities (LaTeCH 2010),
2010, pp. 55–59.
[98] F. Sánchez-Martínez, I. Martínez-Sempere, X. Ivars-Ribes
and R.C. Carrasco, An open diachronic corpus of historical
Spanish: annotation criteria and automatic modernisation of
spelling, arXiv preprint arXiv:1306.3692 (2013).
[99] E. Pettersson, Spelling normalisation and linguistic analysis
of historical text for information extraction, PhD thesis, Acta
Universitatis Upsaliensis, 2016.
[100] M. Domingo and F. Casacuberta, Spelling normalization of
historical documents by using a machine translation approach
(2018).
[101] M. Bollmann, J. Bingel and A. Søgaard, Learning attention
for historical text normalization by learning to pronounce, in:
Proceedings of the 55th Annual Meeting of the Association for
Computational Linguistics (Volume 1: Long Papers), 2017,
pp. 332–344.
[102] N. Korchagina, Normalizing Medieval German Texts: from
rules to deep learning, in: Proceedings of the NoDaLiDa 2017
Workshop on Processing Historical Language, 2017, pp. 12–
17.
[103] A. Robertson and S. Goldwater, Evaluating Historical Text
Normalization Systems: How Well Do They Generalize?, in:
Proceedings of the 2018 Conference of the North American
Chapter of the Association for Computational Linguistics:
Human Language Technologies, Volume 2 (Short Papers),
2018, pp. 720–725.
[104] M. Hämäläinen, T. Säily, J. Rueter, J. Tiedemann and
E. Mäkelä, Normalizing early English letters to present-
day English spelling, in: Proceedings of the Second Joint
SIGHUM Workshop on Computational Linguistics for Cul-
tural Heritage, Social Sciences, Humanities and Literature,
2018, pp. 87–96.
[105] S. Flachs, M. Bollmann and A. Søgaard, Historical Text Nor-
malization with Delayed Rewards, in: Proceedings of the 57th
Annual Meeting of the Association for Computational Lin-
guistics, 2019, pp. 1614–1619.
[106] M.A. Azawi, M.Z. Afzal and T.M. Breuel, Normalizing
historical orthography for OCR historical documents using
LSTM, in: Proceedings of the 2nd International Workshop on
Historical Document Imaging and Processing, 2013, pp. 80–
85.
[107] M. Bollmann and A. Søgaard, Improving historical spelling
normalization with bi-directional LSTMs and multi-task
learning, arXiv preprint arXiv:1610.07844 (2016).
[108] M. Kestemont, G. De Pauw, R. van Nie and W. Daelemans,
Lemmatization for variation-rich languages using deep learn-
ing, Digital Scholarship in the Humanities 32(4) (2017), 797–
815.
[109] P. Mitankin, S. Gerdjikov and S. Mihov, An approach to unsu-
pervised historical text normalisation, in: Proceedings of the
First International Conference on Digital Access to Textual
Cultural Heritage, 2014, pp. 29–34.
[110] N. Ljubešic, K. Zupan, D. Fišer and T. Erjavec, Normalis-
ing Slovene data: historical texts vs. user-generated content,
in: Proceedings of the 13th Conference on Natural Language
Processing (KONVENS 2016), Vol. 16, 2016, pp. 146–155.
[111] S. Soni, K. Lerman and J. Eisenstein, Follow the Leader:
Documents on the Leading Edge of Semantic Change Get
More Citations, arXiv:1909.04189 [physics] (2020), arXiv:
1909.04189. http://arxiv.org/abs/1909.04189.
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 25
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
[112] S. Montariol, M. Martinc and L. Pivovarova, Scalable and
Interpretable Semantic Change Detection, in: Proceedings of
the 2021 Conference of the North American Chapter of the
Association for Computational Linguistics: Human Language
Technologies, Association for Computational Linguistics,
2021, pp. 4642–4652. doi:10.18653/v1/2021.naacl-main.369.
https://www.aclweb.org/anthology/2021.naacl-main.369.
[113] D. van Strien, K. Beelen, M. Ardanuy, K. Hosseini,
B. McGillivray and G. Colavizza, Assessing the Impact of
OCR Quality on Downstream NLP Tasks:, in: Proceedings
of the 12th International Conference on Agents and Artifi-
cial Intelligence, SCITEPRESS - Science and Technology
Publications, 2020, pp. 484–496. ISBN 978-989-758-395-7.
doi:10.5220/0009169004840496.
[114] L. Borin, D. Kokkinakis and L.-J. Olsson, Naming the
past: Named entity and animacy recognition in 19th cen-
tury Swedish literature, in: Proceedings of the Workshop on
Language Technology for Cultural Heritage Data (LaTeCH
2007)., 2007, pp. 1–8.
[115] C. Grover, S. Givon, R. Tobin and J. Ball, Named Entity
Recognition for Digitised Historical Texts, in: Proceedings of
the Sixth International Conference on Language Resources
and Evaluation (LREC’08), 2008.
[116] K. Kettunen and T. Ruokolainen, Names, right or wrong:
Named entities in an OCRed historical Finnish newspaper
collection, in: Proceedings of the 2nd International Confer-
ence on Digital Access to Textual Cultural Heritage, 2017,
pp. 181–186.
[117] N. Tahmasebi, G. Gossen, N. Kanhabua, H. Holzmann and
T. Risse, Neer: An unsupervised method for named entity
evolution recognition, in: Proceedings of the 24th Interna-
tional Conference on Computational Linguistics (COLING
2012), 2012, pp. 2553–2568.
[118] C. Neudecker, L. Wilms, W.J. Faber and T. van Veen, Large-
scale refinement of digital historic newspapers with named
entity recognition, in: Proc IFLA Newspapers/GENLOC Pre-
Conference Satellite Meeting, 2014.
[119] S. Mac Kim and S. Cassidy, Finding names in trove: named
entity recognition for Australian historical newspapers, in:
Proceedings of the Australasian Language Technology Asso-
ciation Workshop 2015, 2015, pp. 57–65.
[120] S.T. Aguilar, X. Tannier and P. Chastang, Named entity recog-
nition applied on a data base of Medieval Latin charters. The
case of chartae burgundiae, in: 3rd International Workshop on
Computational History (HistoInformatics 2016), 2016.
[121] R. Sprugnoli, Arretium or Arezzo? a neural approach to the
identification of place names in historical texts, in: Fifth Ital-
ian Conference on Computational Linguistics (CLiC-it 2018),
aAccademia University Press, 2018, pp. 360–365.
[122] M. Riedl and S. Padó, A named entity recognition shootout
for german, in: Proceedings of the 56th Annual Meeting of the
Association for Computational Linguistics (Volume 2: Short
Papers), 2018, pp. 120–125.
[123] H. Hubková, Named-entity recognition in Czech historical
texts: Using a CNN-BiLSTM neural network model, 2019.
[124] K. Labusch, P. Kulturbesitz, C. Neudecker and D. Zellhöfer,
BERT for Named Entity Recognition in Contemporary and
Historical German, in: Proceedings of the 15th Conference on
Natural Language Processing (KONVENS 2019), 2019.
[125] S. Schweter and J. Baiter, Towards robust named entity recog-
nition for historic german, arXiv preprint arXiv:1906.07592
(2019).
[126] P. Agarwal, J. Strötgen, L. Del Corro, J. Hoffart and
G. Weikum, Dianed: time-aware named entity disambigua-
tion for diachronic corpora, in: Proceedings of the 56th An-
nual Meeting of the Association for Computational Linguis-
tics (Volume 2: Short Papers), 2018, pp. 686–693.
[127] M. Ehrmann, M. Romanello, A. Flückiger and S. Clematide,
Extended overview of CLEF HIPE 2020: named entity pro-
cessing on historical newspapers, in: CLEF 2020 Work-
ing Notes. Conference and Labs of the Evaluation Forum,
Vol. 2696, CEUR, 2020.
[128] M. Ehrmann, M. Romanello, A. Flückiger and S. Clematide,
Overview of CLEF HIPE 2020: Named entity recognition and
linking on historical newspapers, in: International Confer-
ence of the Cross-Language Evaluation Forum for European
Languages, Springer, 2020, pp. 288–310.
[129] M. Rovera, F. Nanni, S.P. Ponzetto and A. Goy, Domain-
specific named entity disambiguation in historical memoirs,
in: CEUR Workshop Proceedings, Vol. 2006, RWTH, 2017,
p. Paper–20.
[130] F. Frontini, C. Brando and J.-G. Ganascia, Semantic web
based named entity linking for digital humanities and heritage
texts, 2015.
[131] S. Van Hooland, M. De Wilde, R. Verborgh, T. Steiner and
R. Van de Walle, Exploring entity recognition and disam-
biguation for cultural heritage collections, Digital Scholar-
ship in the Humanities 30(2) (2015), 262–279.
[132] M. De Wilde, S. Hengchen et al., Semantic enrichment of a
multilingual archive with linked open data, Digital Humani-
ties Quarterly (2017).
[133] C. Brando, F. Frontini and J.-G. Ganascia, REDEN: named
entity linking in digital literary editions using linked data
sets, Complex Systems Informatics and Modeling Quarterly
(2016), 60–80.
[134] S. Rosset, C. Grouin, K. Fort, O. Galibert, J. Kahn and
P. Zweigenbaum, Structured named entities in two distinct
press corpora: contemporary broadcast news and old newspa-
pers, in: Proceedings of the Sixth Linguistic Annotation Work-
shop, 2012, pp. 40–48.
[135] E.L. Pontes, L.A. Cabrera-Diego, J.G. Moreno, E. Boros,
A. Hamdi, N. Sidère, M. Coustaty and A. Doucet, En-
tity Linking for Historical Documents: Challenges and So-
lutions, in: International Conference on Asian Digital Li-
braries, Springer, 2020, pp. 215–231.
[136] S. Rijhwani and D. Preo¸tiuc-Pietro, Temporally-informed
analysis of named entity recognition, in: Proceedings of the
58th Annual Meeting of the Association for Computational
Linguistics, 2020, pp. 7605–7617.
[137] K. Gulordava and M. Baroni, A distributional similarity ap-
proach to the detection of semantic change in the Google
Books Ngram corpus., in: Proceedings of the GEMS 2011
workshop on geometrical models of natural language seman-
tics, 2011, pp. 67–71.
[138] C. Liebeskind, I. Dagan and J. Schler, Statistical thesaurus
construction for a morphologically rich language, in: * SEM
2012: The First Joint Conference on Lexical and Computa-
tional Semantics–Volume 1: Proceedings of the main confer-
ence and the shared task, and Volume 2: Proceedings of the
26 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
Sixth International Workshop on Semantic Evaluation (Se-
mEval 2012), 2012, pp. 59–64.
[139] A. Jatowt and K. Duh, A framework for analyzing semantic
change of words across time, in: IEEE/ACM Joint Conference
on Digital Libraries, IEEE, 2014, pp. 229–238.
[140] E. Sagi, S. Kaufmann and B. Clark, Tracing semantic change
with latent semantic analysis, Current methods in historical
semantics 73 (2011), 161–183.
[141] T. Mikolov, K. Chen, G. Corrado and J. Dean, Efficient esti-
mation of word representations in vector space, in: Interna-
tional Conference on Learning Representations, 2013, pp. 1–
12.
[142] J. Pennington, R. Socher and C. Manning, GloVe: Global
Vectors for Word Representation, in: Proceedings of the 2014
Conference on Empirical Methods in Natural Language Pro-
cessing (EMNLP), Association for Computational Linguis-
tics, 2014, pp. 1532–1543. doi:10.3115/v1/D14-1162.
[143] P. Bojanowski, E. Grave, A. Joulin and T. Mikolov, Enrich-
ing Word Vectors with Subword Information, Transactions of
the Association for Computational Linguistics 5(2017), 135–
146. doi:10.1162/tacl_a_00051.
[144] T. Mikolov, E. Grave, P. Bojanowski, C. Puhrsch and
A. Joulin, Advances in Pre-Training Distributed Word Rep-
resentations, in: International Conference on Language Re-
sources and Evaluation, 2018, pp. 52–55.
[145] H. Gong, S. Bhat and P. Viswanath, Enriching Word Embed-
dings with Temporal and Spatial Information, in: Proceed-
ings of the 24th Conference on Computational Natural Lan-
guage Learning, Association for Computational Linguistics,
Online, 2020, pp. 1–11. https://www.aclweb.org/anthology/
2020.conll-1.1.
[146] Y. Bizzoni, M. Mosbach, D. Klakow and S. Degaetano-
Ortlieb, Some steps towards the generation of diachronic
WordNets, in: Proceedings of the 22nd Nordic Conference
on Computational Linguistics, 2019, pp. 55–64. https://www.
aclweb.org/anthology/W19-6106.
[147] M. Nickel and D. Kiela, Poincaré Embeddings for Learn-
ing Hierarchical Representations, in: Proceedings of the 31st
International Conference on Neural Information Processing
Systems, 2017, pp. 6341–6350.
[148] A. Tsakalidis and M. Liakata, Sequential Modelling of the
Evolution of Word Representations for Semantic Change De-
tection, in: Proceedings of the 2020 Conference on Empirical
Methods in Natural Language Processing (EMNLP), Asso-
ciation for Computational Linguistics, 2020, pp. 8485–8497.
doi:10.18653/v1/2020.emnlp-main.682.
[149] A. Wegmann, F. Lemmerich and M. Strohmaier, Detecting
Different Forms of Semantic Shift in Word Embeddings via
Paradigmatic and Syntagmatic Association Changes, in: Lec-
ture Notes in Computer Science, Springer International Pub-
lishing, 2020, pp. 619–635. doi:10.1007/978-3-030-62419-
4_35.
[150] W.L. Hamilton, J. Leskovec and D. Jurafsky, Cultural Shift or
Linguistic Drift? Comparing Two Computational Measures of
Semantic Change, in: Proceedings of the 2016 Conference on
Empirical Methods in Natural Language Processing, Asso-
ciation for Computational Linguistics, 2016, pp. 2116–2121.
doi:10.18653/v1/D16-1229.
[151] J. Devlin, M.-W. Chang, K. Lee and K. Toutanova, BERT:
Pre-training of Deep Bidirectional Transformers for Lan-
guage Understanding, in: Conference of the North American
Chapter of the Association for Computational Linguistics,
Association for Computational Linguistics, 2019, pp. 4171–
4186. doi:10.18653/v1/N19-1423.
[152] M. Peters, M. Neumann, M. Iyyer, M. Gardner, C. Clark,
K. Lee and L. Zettlemoyer, Deep Contextualized Word
Representations, in: Proceedings of the 2018 Conference
of the North American Chapter of the Association for
Computational Linguistics: Human Language Technolo-
gies, Volume 1 (Long Papers), Association for Computa-
tional Linguistics, New Orleans, Louisiana, 2018, pp. 2227–
2237. doi:10.18653/v1/N18-1202. https://www.aclweb.org/
anthology/N18-1202.
[153] Z. Yang, Z. Dai, Y. Yang, J. Carbonell, R. Salakhutdinov and
Q.V. Le, XLNet: Generalized Autoregressive Pretraining for
Language Understanding, in: Advances in Neural Information
Processing Systems, 2019, pp. 5753–5763.
[154] M. Giulianelli, M.D. Tredici and R. Fernández, Analysing
Lexical Semantic Change with Contextualised Word Rep-
resentations, in: Proceedings of the 58th Annual Meeting
of the Association for Computational Linguistics, Associa-
tion for Computational Linguistics, 2020, pp. 3960–3973.
doi:10.18653/v1/2020.acl-main.365.
[155] V. Kanjirangat, S. Mitrovic, A. Antonucci and F. Rinaldi,
SST-BERT at SemEval-2020 Task 1: Semantic Shift Trac-
ing by Clustering in BERT-based Embedding Spaces, in: Pro-
ceedings of the Fourteenth Workshop on Semantic Evalua-
tion, SemEval@COLING 2020, Barcelona (online), Decem-
ber 12-13, 2020, A. Herbelot, X. Zhu, A. Palmer, N. Schnei-
der, J. May and E. Shutova, eds, International Committee for
Computational Linguistics, 2020, pp. 214–221. https://www.
aclweb.org/anthology/2020.semeval-1.26/.
[156] D.M. Blei, A.Y. Ng and M.I. Jordan, Latent Dirichlet Alloca-
tion, Journal of Machine Learning Research 3(2003), 993–
1022.
[157] D.M. Blei and J.D. Lafferty, Dynamic topic models, in: Pro-
ceedings of the 23rd international conference on Machine
learning - ICML ’06, ACM Press, 2006, pp. 113–120. ISBN
978-1-59593-383-6. doi:10.1145/1143844.1143859. http://
portal.acm.org/citation.cfm?doid=1143844.1143859.
[158] C. Pölitz, T. Bartz, K. Morik and A. Störrer, Investigation
of Word Senses over Time Using Linguistic Corpora, in:
Text, Speech, and Dialogue, P. Král and V. Matoušek, eds,
Lecture Notes in Computer Science, Vol. 9302, Springer
International Publishing, 2015, pp. 191–198. ISBN 978-
3-319-24032-9. doi:10.1007/978-3-319-24033-6_22. http://
link.springer.com/10.1007/978-3-319-24033-6_22.
[159] X. Wang and A. McCallum, Topics over time: a non-
Markov continuous-time model of topical trends, in: Pro-
ceedings of the 12th ACM SIGKDD international confer-
ence on Knowledge discovery and data mining - KDD
’06, ACM Press, 2006, p. 424. ISBN 978-1-59593-
339-3. doi:10.1145/1150402.1150450. http://portal.acm.org/
citation.cfm?doid=1150402.1150450.
[160] Y.W. Teh, M.I. Jordan, M.J. Beal and D.M. Blei, Hi-
erarchical Dirichlet Processes, Journal of the Ameri-
can Statistical Association 101(476) (2006), 1566–1581.
doi:10.1198/016214506000000302.
[161] B. McGillivray, R. Buning and S. Hengchen, Topic Mod-
elling: Hartlib’s Correspondence before and after 1650, in:
Reassembling the Republic of Letters in the Digital Age,
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 27
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
H. Hotson and T. Wallnig, eds, Göttingen University Press,
2019.
[162] S. Hengchen, When does it mean? Detecting semantic change
in historical texts, PhD thesis, Université libre de Bruxelles,
2017.
[163] B. McGillivray, S. Hengchen, V. Lähteenoja, M. Palma and
A. Vatri, A computational approach to lexical polysemy in
Ancient Greek, Digital Scholarship in the Humanities 34(4)
(2019), 893–907.
[164] B. McGillivray, Computational methods for semantic analy-
sis of historical texts, Routledge, 2020.
[165] V. Iyer, M. Mohan, Y.R.B. Reddy and M. Bhatia, A Survey
on Ontology Enrichment from Text (2019).
[166] M.N. Asim, M. Wasim, M.U.G. Khan, W. Mahmood
and H.M. Abbasi, A survey of ontology learning
techniques and applications, Database 2018 (2018).
doi:10.1093/database/bay101.
[167] P. Buitelaar, P. Cimiano and B. Magnini, Ontology Learning
from Text: An Overview, in: Ontology Learning from Text:
Methods, Evaluation and Applications, Vol. 123, IOS Press,
2005, pp. 3–12.
[168] P. Cimiano and J. Völker, Text2Onto. A Framework for
Ontology Learning and Data-driven Change Discovery,
in: Natural Language Processing and Information Sys-
tems, A. Montoyo, R. Mu´
noz and E. Métais, eds, Lec-
ture Notes in Computer Science, Vol. 3513, Springer Berlin
Heidelberg, 2005, pp. 227–238. ISBN 978-3-540-26031-8.
doi:10.1007/11428817_21.
[169] D. Faure and C. Nédellec, Asium: Learning subcategorization
frames and restrictions of selection (1998).
[170] X. Jiang and A.-H. Tan, CRCTOL: A semantic-based domain
ontology learning system, Journal of the American Society for
Information Science and Technology 61(1) (2010), 150–168.
doi:10.1002/asi.21231.
[171] E. Drymonas, K. Zervanou and E.G.M. Petrakis, Unsuper-
vised Ontology Acquisition from Plain Texts: The OntoGain
System, in: Natural Language Processing and Information
Systems, C.J. Hopfe, Y. Rezgui, E. Métais, A. Preece and
H. Li, eds, Lecture Notes in Computer Science, Vol. 6177,
Springer Berlin Heidelberg, 2010, pp. 277–287. ISBN 978-
3-642-13880-5. doi:10.1007/978-3-642-13881-2_29. http://
link.springer.com/10.1007/978-3-642-13881-2_29.
[172] R. Navigli and P. Velardi, Learning domain ontologies from
document warehouses and dedicated web sites, Computa-
tional Linguistics 30(2) (2004), 151–179.
[173] A. Oliveira, F.C. Pereira and A. Cardoso, Automatic Reading
and Learning from Text, in: Proceedings of the International
Symposium on Artificial Intelligence (ISAI), 2001.
[174] U. Hahn and K. Schnattinger, Towards text knowledge engi-
neering, Hypothesis 1(2) (1998).
[175] H.-P. Zhang, Q. Liu, X.-Q. Cheng, H. Zhang and H.-K. Yu,
Chinese lexical analysis using hierarchical hidden Markov
model, SIGHAN ’03: Proceedings of the second SIGHAN
workshop on Chinese language processing 17 (2003), 63–70.
[176] X. Zou, N. Sun, H. Zhang and J. Hu, Diachronic Corpus
Based Word Semantic Variation and Change Mining, in:
Language Processing and Intelligent Information Systems,
M.A. Kłopotek, J. Koronacki, M. Marciniak, A. Mykowiecka
and S.T. Wierzcho´
n, eds, Lecture Notes in Computer Science,
Vol. 7912, Springer Berlin Heidelberg, 2013, pp. 145–150.
ISBN 978-3-642-38633-6. doi:10.1007/978-3-642-38634-
3_16. http://link.springer.com/10.1007/978-3-642-38634-3_
16.
[177] J.M. Kleinberg, Authoritative Sources in a Hyperlinked Envi-
ronment, Journal of the ACM 46(5) (1999), 604–632.
[178] J.B. MacQueen, Some methods for classification and analy-
sis of multivariate observations, in: Proceedings of the Fifth
Berkeley Symposium on Mathematical Statistics and Prob-
ability, University of California Press, 1967, pp. 281–297.
https://projecteuclid.org/euclid.bsmsp/1200512992.
[179] G.D. Rosin and K. Radinsky, Generating Timelines by
Modeling Semantic Change, in: Proceedings of the 23rd
Conference on Computational Natural Language Learn-
ing (CoNLL), Association for Computational Linguistics,
2019, pp. 186–195. doi:10.18653/v1/K19-1018. https://www.
aclweb.org/anthology/K19-1018.
[180] J.A. Gulla, G. Solskinnsbakk, P. Myrseth, V. Haderlein and
O. Cerrato, Semantic Drift in Ontologies, in: WEBIST 2010,
Proceedings of the 6th International Conference on Web In-
formation Systems and Technologies, Vol. 2, 2010.
[181] I. Augenstein, S. Padó and S. Rudolph, LODifier: Gener-
ating Linked Data from Unstructured Text, in: The Seman-
tic Web: Research and Applications, E. Simperl, P. Cimi-
ano, A. Polleres, O. Corcho and V. Presutti, eds, Lecture
Notes in Computer Science, Vol. 7295, Springer Berlin
Heidelberg, 2012, pp. 210–224. ISBN 978-3-642-30283-
1. doi:10.1007/978-3-642-30284-8_21. http://link.springer.
com/10.1007/978-3-642-30284-8_21.
[182] K.T. Frantzi and S. Ananiadou, The C-value/NC-value
domain-independent method for multi-word term extraction,
Journal of Natural Language Processing 6(3) (1999), 145–
179. doi:10.5715/jnlp.6.3_145.
[183] M. Lesk, Automatic sense disambiguation using machine
readable dictionaries: How to tell a pine cone from an ice
cream cone, in: SIGDOC ’86: Proceedings of the 5th annual
international conference on Systems documentation, 1986,
pp. 24–26. https://dl.acm.org/doi/10.1145/318723.318728.
[184] H. Cunningham, D. Maynard, K. Bontcheva and V. Tablan,
GATE: A Framework and Graphical Development En-
vironment for Robust NLP Tools and Applications, in:
Proceedings of the 40th Annual Meeting of the Asso-
ciation for Computational Linguistics, 2002, pp. 168–
175. https://www.researchgate.net/publication/200044237_
GATE_A_Framework_and_Graphical_Development_
Environment_for_Robust_NLP_Tools_and_Applications.
[185] P. Cimiano, C. Chiarcos, J.P. McCrae and J. Gracia, Lin-
guistic Linked Data in Digital Humanities, in: Linguistic
Linked Data. Representation, Generation and Applications,
1st edn, Springer International Publishing, 2020. https://
www.springer.com/gp/book/9783030302245.
[186] P. Cimiano, C. Chiarcos, J.P. McCrae and J. Gracia, Linguistic
linked open data cloud, in: Linguistic Linked Data, Springer,
2020, pp. 29–41.
[187] I. Maks, M. van Erp, P. Vossen, R. Hoekstra and N. van der
Sijs, Integrating diachronous conceptual lexicons through
linked open data, DHBenelux, 2016.
[188] J. Noordegraaf, M. van Erp, R. Zijdeman, M. Raat, T. van
Oort, I. Zandhuis, T. Vermaut, H. Mol, N. van der Sijs,
K. Doreleijers, V. Baptist, C. Vrielink, B. Assendelft,
C. Rasterhoff and I. Kisjes, Semantic Deep Mapping in the
Amsterdam Time Machine: Viewing Late 19th- and Early
20th-Century Theatre and Cinema Culture Through the Lens
28 F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
of Language Use and Socio-Economic Status, 2021, Ac-
cepted for publication.
[189] K. Depuydt and J. De Does, The diachronic semantic lexicon
of dutch as linked open data, in: Proceedings of the Eleventh
International Conference on Language Resources and Evalu-
ation (LREC 2018). European Language Resources Associa-
tion (ELRA), Paris, France, 2018.
[190] S. Tittel and F. Gillis-Webber, Identification of Languages in
Linked Data: A Diachronic-Diatopic Case Study of French,
in: Electronic lexicography in the 21st century. Proceedings
of the eLex 2019 conference. 1-3 October 2019, Sintra, Por-
tugal, Lexical Computing, 2019, pp. 547–569.
[191] E. Vanhoutte, An Introduction to the TEI and the TEI Consor-
tium, Literary and linguistic computing 19(1) (2004), 9–16.
[192] A. Barbado, V. Fresno, Á.M. Riesco and S. Ros, DISCO PAL:
Diachronic Spanish Sonnet Corpus with Psychological and
Affective Labels, arXiv preprint arXiv:2007.04626 (2020).
[193] J.M. van der Zwaan, I. Maks, E. Kuijpers, I. Lee-
mans, K. Steenbergh and H. Roodenburg, Historic Em-
bodied Emotions Model (HEEM) dataset, Zenodo, 2016.
doi:10.5281/zenodo.47751.
[194] I. Leemans, E. Maks, J. van der Zwaan, H. Kuijpers and
K. Steenbergh, Mining Embodied Emotions: A Compara-
tive Analysis of Bodily Emotion Expressions in Dutch The-
atre Texts 1600-1800’, Digital Humanities Quarterly 11(4)
(2017).
[195] A.F. Khan, Towards the Representation of Etymological Data
on the Semantic Web 9(12) (2018), 304, Publisher: MDPI
AG. doi:10.3390/info9120304.
F. Armaselu et al. / LL(O)D and NLP Perspectives on Semantic Change for Humanities Research 29
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
30 30
31 31
32 32
33 33
34 34
35 35
36 36
37 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
Appendix
Table 2. Main theoretical approaches surveyed in S 3
Knowledge-oriented Language-oriented
Charting the history of
political and social
concepts [16]
Semasiological vs.
onomasiological mechanisms of
semantic change in lexical
semantics [26]
Formal description of
conceptual change
implying a “core” and a
“margin” [17]
Semasiological vs.
onomasiological mechanisms of
semantic change in cognitive
linguistics and diachronic
lexicology [27]
Defining the meaning of
a concept in terms of
“intension, extension and
labelling” [12]
Stability and univocity principles
vs. sociocognitive approaches to
understand world and language
change in terminology [30]
Model-based approach to
the “history of ideas or
concept drift” [21]
Diachronic change in the layer of
pragmatics [31]
Describing semantic
change, semantic drift,
concept drift in relation
to ontology change [18]
Discourse-historical approach
(DHA) and the principle of
"triangulation" [40]
Table 3. Main LL(O)D formalisms and resources surveyed in S 4
and S 7
Models OntoLex-Lemon [43]
Temporal RDF [57]; RDF-star
Approaches
Etymology modelling [50, 51, 195]
Perdurantist modelling [59]
OWL-based temporal reasoning [64]
Resources
General
LL(O)D cloud [186]
Linghub
For diachronic analysis
LiLa etymological lexicon [54]
OWL-Time ontology [62]; LODE ontology;
PeriodO gazetteer of periods
Diachronic semantic lexicon of Dutch [189]
Table 4. Main NLP methods for diachronic analysis surveyed in S 5
NER, NED,
NEL
NER: rule-based [114–116]; unsupervised,
statistical [117]; machine learning [118–120];
deep learning [121–125]
Time-aware NED, NER [126, 136]
LL(O)D-based NEL [130–133]
Word
embeddings
Unsupervised, with temporal-spatial
information [145]; hyperbolic [146, 147]
LSTM-based [148]; detecting paradigmatic
and syntagmatic shifts [149]
Transformer-
based
BERT [151]; ELMo [152]; XLNet [153]
Unsupervised, with contextualised word
representations [154]; clustering [155]
Topic
modelling
SCAN [68]; topics over time LDA [158]
Hierarchical Dirichlet [66, 67]
LDA-based [161]
Table 5. Main NLP applications for generating (diachronic) ontolog-
ical and linked data structures surveyed in S 6
Learning diachronic
constructs
Ontologies [15, 146]
Timelines [179]
Concept signatures [180]
Learning ontologies
and producing linked
data
OntoGain [171]
CRCTOL [170]
TextToOnto [168]
Extracting information
and linking entities LODifier [181]
Converting to linked
data formats
CoW, cattle [5]
LLODifier [185]
Article
Motifs in folktales and myths have been identified and articulated by scholars, and the computational identification and discovery of such motifs is an area of ongoing research. Achieving this goal means meeting scientific requirements (that methods be comparable and replicable) and requirements for collaboration (that multi-disciplinary teams can reliably access data). To support those requirements, access to consistent reference datasets is needed. Unfortunately, these datasets are not openly available in a format that supports their use in data science. Here we report work in progress toward this goal, having converted the Ashliman Folktexts collection into a public dataset of annotated tale texts. The data can be accessed at doi.org/10.5281/zenodo.6575263.