Semantic Maps [Oxford Bibliographies Online].pdf

***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 Semantic Maps Thanasis Georgakopoulos LAST MODIFIED: 15 JANUARY 2019 DOI: 10.1093/OBO/9780199772810-0229 Introduction A semantic map is a method for visually representing cross-linguistic regularity or universality in semantic structure. This method has proved attractive to typologists because it provides a convenient graphical display of the interrelationships between meanings or functions across languages, while (at the same time) differentiating what is universal from what is language-specific. The semantic map model was initially conceived to describe patterns of polysemy (or, more generally, of co-expression) in grammatical categories. However, several studies have shown that it can be fruitfully extended to lexical items and even constructions, suggesting that any type of meaning can be integrated in a map. The main idea of the method is that the spatial arrangement of the various meanings reflects their degree of (dis)similarity: the more similar the meanings, the closer they are placed—in accordance with the so-called connectivity hypothesis. Within the semantic map tradition, closeness has taken different forms depending on the approach adopted. In classical semantic maps (alternative terms: “first generation,” “implicational,” “connectivity” maps), the relation between meanings is represented as a line. This is the graph-based approach. In proximity maps (alternative terms: “similarity,” “second generation,” “statistical,” “probabilistic” maps), the distance between two meanings in space— represented as points—indicates the degree of their similarity. In this scale- or distance-based approach, the maps are constructed using multivariate statistical techniques, including the family of methods known as multidimensional scaling (MDS). Both classical and proximity maps have been widely used, although the latter have recently gained interest and popularity under the assumption that they can cope with large data more efficiently than classical semantic maps. However, classical semantic maps continue to be useful for studies aiming to discover universal semantic structures. Most importantly, classical maps can integrate information about directionality of change by drawing an arrow on the line connecting two meanings or functions. Beyond the choice between the two types of maps, one of the issues that has sparked debate and critical reflection among researchers is the universal relevance of semantic maps. The main question that these researchers address is whether semantic maps reflect the global geography of the human mind. Another much discussed issue is the identification of the factors that increase the accuracy of semantic maps in a way that allows for valid cross‐linguistic generalizations. Such factors include the choice of a representative language sample, the quality of the collected cross‐linguistic material, and the establishment of valid cross-linguistic comparators. Acknowledgments: The author wishes to thank one anonymous reviewer for their useful comments. For discussion of the material in this article, the author is grateful to Stéphane Polis. 1 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 Early Studies and Theoretical Foundations The references in this section include the representative early treatments in the literature on semantic maps as well those studies that establish the theoretical foundations for the semantic map approach. Interest in semantic maps began with three articles in the early 1980s. Anderson 1982 is the first influential and detailed article that set the foundations for the research that followed. Among other things, it explains the representational conventions used for constructing a map, e.g., closeness of meanings reflecting similarity to each other and curved closed lines indicating language-specific category boundaries. Anderson 1986 establishes a map of evidential space. It illustrates the inductive method for constructing such a map and the logic behind the inference of similarity of meanings based on similarity of form. It also explains the challenges of such an approach for explaining meaning shifts. Lazard 1981 discusses the nature of semantic universals and devotes a small section to uncovering the universal semantic structure of “past” grammatical markers through the use of a map. Kemmer 1993 is a systematic study on the semantic relations holding among middle and other situation types. Stassen 1997 discusses the domain of intransitive predication and adds a wider typological perspective, since the sample employed in the study comprises 410 languages. Haspelmath 1997 is a good resource for finding details about the many distinct ways languages carve up the semantic space in the domain of indefiniteness. Van der Auwera and Plungian 1998 offers an integrated map of modality, suggests additional means to enrich maps, and discusses, in detail, the adjacency requirement imposed by the map (analyzed also in Haspelmath 1997). Croft 2001 highlights the multidimensional nature of the method and provides many examples to illustrate its scope. It also articulates the general principles governing the model, most importantly the semantic map connectivity hypothesis, which states that “any relevant language-specific and construction-specific category should map onto a CONNECTED REGION in conceptual space” (p. 96). Deviating from the terminology employed in most studies, Croft uses the term conceptual space to refer to language-universal conceptual structure and the term semantic map to describe language-specific semantic patterns. All studies of this first era use the graph-based approach, which explains why, at a later point, this approach was called “classical” or “first generation.” Several studies, such as Anderson 1982, Anderson 1986, Kemmer 1993, and van der Auwera and Plungian 1998, underscore the fact that the diachronic dimension should be treated as an inseparable part of the model.  Anderson, Lloyd B. 1982. “The “perfect” as a universal and as a language‐ particular category.” In Tense‐aspect: Between semantics & pragmatics, typological studies in language. Edited by Paul J. Hopper, 227–264. Amsterdam: John Benjamins. Anderson is the first to use the term map to refer to the geometrical representation of meanings in space. These meanings are arranged in a way reflecting their degree of (dis)similarity. Crucially, similarity is defined on the basis of cross-linguistic comparison. The paper introduces some important notational devices that were used in many later studies. It also stresses the fact that for a map to have universal validity, the diachronic dimension must also be considered. 2 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229  Anderson, Lloyd B. 1986. Evidentials, paths of change, and mental maps: Typologically regular asymmetries. In Evidentiality: The linguistic encoding of epistemology. Edited by Wallace Chafe and Johanna Nichols, 273–312. Norwood, NJ: Ablex. Anderson builds a map of evidential meanings, which gives information about which meanings are more closely related to each other synchronically as well as about certain patterns of historical development. The results of the study help to formulate historical hypotheses for evidentials in Tibeto-Burman and California Indian languages, and for moods and modalities in Indo-European languages.  Croft, William. 2001. Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford Univ. Press. Croft develops the notion of conceptual space (which is the term he uses to refer to semantic maps) in his Radical Construction Grammar. In this work, he introduces many of the guiding principles of the semantic maps model, most importantly the semantic map connectivity hypothesis. His case studies come from fundamental concepts in syntactic theory, mainly parts of speech and syntactic roles. Chapters 3 and 4 are the most relevant.  Haspelmath, Martin. 1997. Indefinite pronouns. Oxford: Oxford Univ. Press. One of the goals of Haspelmath’s book is to explain the variety of usage of indefinite pronouns in terms of a semantic map. Haspelmath builds such a map based on 40and 100-language samples. The article makes explicit several important methodological decisions (e.g., the use of the term function rather than the term meaning) and highlights the fact that semantic maps can be thought of as making hypotheses about implicational universals.  Kemmer, Susan. 1993. The middle voice. Amsterdam: Benjamins. Kemmer devotes a whole chapter of her book to the semantic relations between the reflexive and related situation types. Relying both on typological and on diachronic data, she places the various functions identified in the reflexive-middle domain on a semantic map and focuses on the predictions and hypotheses stemming from the semantic relations holding among these functions.  Lazard, Gilbert. 1981. La quête des universaux sémantiques en linguistique. Actes Sémiotiques—Bulletin 19:26–37. Lazard discusses the nature of semantic universals, which ultimately result from an experience of the world shared by all human beings. He distinguishes two approaches to semantic universals: one deductive and one inductive. The article argues in favor of the latter approach. To address the issue of cross-linguistic comparison, it suggests that the focus should be on the multidimensional semantic space occupied by a given signified (i.e., a linguistic form). Finally, the article proposes a map for “past” grammatical markers. Reprinted in Lazard’s Études de Linguistique Générale (Leuven, Belgium: Peeters), pp. 47–56. 3 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229  Stassen, Leon. 1997. Intransitive predication. Oxford: Oxford Univ. Press. This is a study of the predication of various semantic classes. Using a language sample of 410 languages, the author identifies four semantic predicate categories and visualizes them in the form of a semantic map that is language independent and determined by universal principles. The two main principles discussed are “relative time stability” and “locational specification.” The study also shows how the different languages carve up the semantic space.  van der Auwera, Johan, and Vladimir A. Plungian. 1998. Modality’s semantic map. Linguistic Typology 2.1: 79–124. The article focuses on the domain of modality and proposes a semantic map that represents the cross linguistically relevant connections between the various meanings in this domain. This map is supposed to have universal relevance both in synchrony and in diachrony. The authors also show how the method allows the integration of different types of relationships, i.e., metaphor, metonymy, specialization, and generalization. General Overviews of Semantic Maps This section includes work on semantic maps of a general nature. The general overviews are divided into general descriptions of the model, entries in encyclopedias and overview articles. All these studies may serve as an introduction to the subject. General Descriptions of the Model Haspelmath 2003 is particularly interesting because it gives a comprehensive overview of the model. It explains the basic mechanics of the method and analyzes, in detail, its advantages over other alternatives. For example, it stresses the fact that it is neutral with respect to the monosemy-polysemy-homonymy distinction. Although the article focuses on grammatical categories in synchrony, it suggests possible ways to expand the range of the model to the study of semantic change as well as to the study of the lexicon. Van der Auwera 2013 devotes a fair amount of attention to comparing classical semantic maps to maps generated through multivariate statistics. François 2008 provides a blueprint for plotting lexical semantic maps.  François, Alexandre. 2008. Semantic maps and the typology of colexification: Intertwining polysemous networks across languages. In From polysemy to semantic change: Towards a typology of lexical semantic associations. Edited by Martine Vanhove, 163–215. Amsterdam: John Benjamins. This paper has a methodological bearing in the semantic map model, as it extends its scope to lexical typology. François’s model for lexical typology sets out to combine language specific analyses with cross-linguistic comparison. François coins the term colexification, which corresponds to the notions of polysemy and multifunctionality. As a case study to illustrate the typological method, the paper focuses on the notion BREATHE in thirteen genetically unrelated languages. 4 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229  Haspelmath, Martin. 2003. The geometry of grammatical meaning: Semantic maps and cross‐linguistic comparison. In The new psychology of language. Vol. 2. Edited by Michael Tomasello, 211–243. Mahwah, NJ: Lawrence Erlbaum. This article provides a detailed description of semantic maps as a method for describing polysemy (or multifunctionality, as this is the term used in the article) patterns in grammatical typology. It explains in a step-by-step fashion the principles of the method (e.g., regarding the language sample), presents a set of examples from earlier studies (e.g., indefinite pronouns and instrumentals), and outlines its main advantages. The article concludes by highlighting the role the method can play in representing diachronic change.  van der Auwera, Johan. 2013. Semantic maps, for synchronic and diachronic typology. In Synchrony and diachrony: A dynamic interface. Edited by Anna Giacalone Ramat, Caterina Mauri, and Piera Molinelli, 153–176. Amsterdam: John Benjamins. This chapter is devoted to a detailed comparison between the classical semantic approach and the proximity approach. Drawing examples mainly from the domains of indefinite and impersonal pronouns, it presents the main arguments in favor of the use of classical semantic maps as a tool for portraying synchronic variation and semantic change. It also seeks to find a common denominator for both approaches to make their combination possible. Entries in Encyclopedias General introductions to semantic maps presented in the form of encyclopedic entries are van der Auwera and Temürcü 2006 and Levshina 2017.  Levshina, Natalia. 2017. Sémantická mapa. In CzechEncy: Nový encyklopedický slovník češtiny. Edited by Petr Karlík, Marek Nekula, and Jana Pleskalová. This article discusses the two main types of maps, namely classical maps, which are normally constructed manually, and proximity maps, which are built with the help of either multidimensional scaling (MDS) or correspondence analysis. It summarizes the results of some basic studies in the field.  van der Auwera, Johan, and Ceyhan Temürcü. 2006. Semantic maps. In Encyclopedia of language & linguistics. Vol. 11. Edited by Keith Brown, 131– 134. Amsterdam: Elsevier. The authors offer a summary of the semantic map method in grammatical typology considering both the synchronic and the diachronic dimension of the technique. Special emphasis is placed on the contiguity/adjacency requirement and its predictive power as well as the consequences in instances of noncontiguity. 5 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 Overview Articles These overview articles demonstrate the wide scope of the model. Cysouw, et al. 2010 introduces a special issue on semantic maps (see Special issue: Semantic Maps: Methods and Applications. Linguistic Discovery, cited under Special Issues in Journals) and includes an exhaustive list of the grammatical domains investigated. Koptjevskaja‐Tamm, et al. 2015 is a good resource for studies in the lexicon that use the semantic map tool. Georgakopoulos and Polis 2018a is an extended overview of the model, concentrating on its basic principles, its development over the years, the recent lexical turn in the field, and the paradigm shift toward automation of the plotting process and the pending issues for future research. Georgakopoulos and Polis 2018b is a learning guide that helps situate the readers in the key relevant works in the field as well as a teaching guide that aims to introduce the students to the semantic map approach.  Cysouw, Michael, Martin Haspelmath, and Andrej Malchukov. 2010. Introduction. In Special issue: Semantic maps: Methods and applications. Linguistic Discovery 8.1: 1–3. This succinct introduction contains a representative list of the literature on semantic maps, which to that point in time focused first and foremost on grammatical categories. The various studies are categorized according to the domains covered.  Georgakopoulos, Thanasis, and Stéphane Polis. 2018a. The semantic map model: State of the art and future avenues for linguistic research. Language and Linguistics Compass 8.1: 1–33. This article gives a comprehensive overview of the semantic map model. Both classical and proximity maps are presented, and a guide is provided for building such maps. The advantages and shortcomings of both types are also discussed, and the main theoretical debates are outlined. A separate section discusses the tools that allow for the automatic plotting of semantic maps based on cross‐linguistic polysemy data. Several pending issues for future research are highlighted.  Georgakopoulos, Thanasis, and Stéphane Polis. 2018b. Teaching & Learning Guide for: The semantic map model. Language and Linguistics Compass 12.8: 1–13. This teaching and learning guide provides key references for the semantic map approach, gives information about available online electronic resources, and develops a set of teaching material that can be used in order to introduce the model in the context of courses on linguistic typology, historical linguistics, and computational linguistics.  Koptjevskaja‐Tamm, Maria, Ekaterina Rakhilina, and Martine Vanhove. 2015. The semantics of lexical typology. In The Routledge handbook of semantics. Edited by Nick Riemer, 434–454. London: Routledge. Section 7 in this chapter is devoted to the use of semantic maps as a representational tool in lexical typology. The various studies outlined in the article are distinguished on the basis of the type of semantic maps employed, be they graph-based/implicational or scale-based/probabilistic. 6 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 Special Issues in Journals Two journals have devoted special issues to semantic maps. They are relatively recent and both reflect the great interest in the method at that time. Theoretical Linguistics 34.1 introduces an alternative visualisation method, the multidimensional scaling (MDS) procedure. Special issue: Semantic Maps: Methods and Applications. Linguistic Discovery covers more thematic areas and contains a large number of papers.  Special issue: Semantic maps: Methods and applications. Linguistic Discovery 8.1 (2010). This special issue of the journal collects papers focusing on many diverse topics, which include, among others: the theoretical and methodological status of semantic maps in linguistic theory, the different representational techniques for drawing semantic maps, limits and challenges of the approach, the universal relevance of semantic maps, and the use of methodologies coming from other fields to solve some of the pending issues within the semantic map approach.  Theoretical Linguistics 34.1 (2008). This issue contains a position paper by Croft and Poole, which introduces the MDS approach (see the description of Croft and Poole 2008, cited under Proximity Maps in Grammatical Typology), four commentaries on this position paper (see, e.g., the description of van der Auwera 2008, cited under Theoretical Debates), and a final response from Croft and Poole. Theoretical Debates With the growth in popularity of the semantic map methodology, a period ensued in which some criticisms were voiced against some of its fundamental underlying assumptions and theoretical premises. For example, Janda 2009 aims its criticism at the discrete nature of the semantic map model (see also Wälchli 2016, cited under Proximity Maps in Lexical Typology). It is not a coincidence that most of the criticisms were expressed in the context of the two journal issues devoted exclusively to semantic maps (see Special Issues in Journals), which aimed at providing a deeper understanding of the method. One of the thorniest issues that has received much attention is the question of the universal relevance of semantic maps. Some works, such as Cristofaro 2010, Janda 2009, and Sansò 2010, claim that semantic maps do not represent a mental semantic space or urge caution when making such a strong statement, while others, such as Croft 2010, argue the opposite. It should be underlined that in earlier studies the latter view was the dominant one. Another much debated issue is the choice between a graph-based and a scale-based approach in plotting a semantic map. Van der Auwera 2008, responding to Croft and Poole 2008 (cited under Proximity Maps in Grammatical Typology), defends the graphbased method by highlighting the importance of adding the diachronic dimension to the model, a dimension that normally cannot be captured by proximity maps (but 7 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 see Eckhoff 2011, cited under Proximity Maps in Grammatical Typology for a preliminary attempt to overcome this problem). Mauri 2010 proposes a hybrid map, which integrates many dimensions of variation, e.g., coding complexity and frequency of polysemy patterns. In the studies employing the semantic map model, another crucial distinction is highlighted beyond the graph-based versus scale-based one. This distinction pertains to whether a study uses top-down or bottom-up methods. Both methods have their advantages and disadvantages. Zwarts 2010 argues that the best way for describing cross-linguistic variation in semantic structure is a combined approach. A further challenge for the semantic map method is how to deal with cases in which some of its basic assumptions are violated. Malchukov 2010 elaborates on the factors that can lead to such violations, e.g., complex diachronic scenarios, and suggests solutions to some of the major problems.  Cristofaro, Sonia. 2010. Semantic maps and mental representation. Linguistic Discovery 8.1: 35–52. This article takes up the challenge of answering the question as to whether the similarity relationships between the meanings of linguistic forms, which are represented in semantic maps, also reflect similarity relationships in speakers’ mental representations and ultimately argues against this idea. It discusses two mechanisms that give rise to cross-linguistic polysemy patterns, namely metonymization and generalization processes—processes that do not involve similarity (the notion of similarity is fundamental for semantic maps).  Croft, William. 2010. What do semantic maps tell us? Comment on “Semantic Maps and Mental Representation” by Sonia Cristofaro. Linguistic Discovery 8.1: 53–60. Written as a reply to Cristofaro 2010, this paper defends the idea that semantic maps reflect the global geography of the human mind. It also makes the claim that the range of phenomena emerging due to similarity (if viewed holistically) is wider than outlined in Cristofaro 2010. It further acknowledges that there are certain phenomena to which the semantic maps model should not be applied.  Cysouw, Michael. 2010. Semantic maps as metrics on meanings. Linguistic Discovery 8.1: 70–95. This article reflects on the difficult task of measuring meaning. It operationalizes the notion of meaning as the sum of all appearances of the expression. Such an operationalization allows cross-linguistic comparisons between language-specific expressions from different languages in a non-introspective way. In Cysouw’s approach, a semantic map is the graphical representation of attested structure, which visualizes the (dis)similarities of particular sampled contexts.  Janda, Laura A. 2009. What is the role of semantic maps in cognitive linguistics? In Cognitive approaches to language and linguistic data: Studies in honor of Barbara Lewandowska‐Tomaszczyk. Edited by Piotr Stalmaszczyk and Wieslaw Oleksy, 105–124. Frankfurt: Peter Lang. This paper discusses the status and the scope of semantic maps in linguistic theory primarily by reference to notions of continuity and discreteness. It identifies some 8 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 instances in which it is unhelpful to use the model, e.g., when two languages use different strategies to encode a certain domain. Janda also challenges the view that there is such a thing as a universal conceptual space.  Malchukov, Andrej L. 2010. Analyzing semantic maps: A multifactorial approach. Linguistic Discovery 8.1: 176–198. This article raises some general questions related to the semantic maps approach. It focuses on the factors that may give rise to polysemy other than semantic similarity: markedness, economy, and distinguishability as well as structural and diachronic factors (e.g., reanalysis and gram replacement). These non-semantic factors can incur violations on the map, which are, in principle, problematic for the method. The paper suggests certain heuristics for circumventing this problem.  Mauri, Caterina. 2010. Towards a unified account of the coding degree, coding complexity, and coding distance of coordination relations. Linguistic Discovery 8.1: 210–232. This paper expands the scope of the model by also considering the presence of overt marking and its morphophonological complexity. The label coined for the new map is “coding” map. Ultimately, the article argues for a hybrid map, which integrates the coding map and the map generated through an MDS analysis. The case study analyzed is the cross-linguistic coding of coordination relations.  Sansò, Andrea. 2010. How conceptual are semantic maps? Linguistic Discovery 8.1: 288–309. This article addresses the question of whether the meanings on the map reflect the geography of the mind. It sets two requirements that have to be met for maps to be considered conceptual: (a) conceptual relevance and (b) high degree of regularity in the data material. The article uses the labels first-generation vs. secondgeneration semantic maps (replacing for classical and proximity maps, respectively).  van der Auwera, Johan. 2008. In defense of classical semantic maps. Theoretical Linguistics34.1: 39–46. Written as a reply to Croft and Poole 2008 (cited under Proximity Maps in Grammatical Typology), this paper presents the advantages of classical semantic maps and compares them to the multidimensional scaling method. One of the article’s main arguments in favor of classical maps is the option of integrating the diachronic information. The importance of the diachronic dimension is summarized as follows: “the best synchronic semantic map is a diachronic one” (p. 43). In conclusion, the article argues for a synergy between the two maps.  Zwarts, Joost. 2010. Semantic map geometry: Two approaches. Linguistic Discovery 8.1: 377–395. This article discusses two approaches to the geometry of semantic maps: the matrix driven and the space driven. In the former, the conceptual space is induced from a cross-linguistic lexical matrix. In the latter, the starting point is the conceptual space and from there the task is to fill in the lexical matrix and examine the way words are 9 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 mapped onto the space. The author illustrates the advantages and the limitations of both approaches and concludes that a combined model could help overcome some of the problems of each individual approach. Classical Semantic Maps The first era in the history of the semantic map model is defined primarily by the use of graph-based/ classical semantic maps. In these “first generation” maps (for the term, see Sansò 2010, cited under Theoretical Debates), the meanings are represented as nodes on the graph and the relation between meanings is represented as an edge (i.e., a connecting line). The decisive factor that determines which meanings will be connected and, subsequently, which meanings are more similar, is cross-linguistic comparison. This external criterion for meaning similarity has already been introduced in Anderson 1982 (cited under Early Studies and Theoretical Foundations). Note that Anderson 1982 often omits the edges, in which case similarity is signalled through spatial adjacency. Graph-based semantic maps can express implicational universals (see Haspelmath 1997, cited under Early Studies and Theoretical Foundations, which coins the term implicational map). This type of map has been intensively used in recent years for investigating a variety of cross-linguistic questions in a plethora of linguistic domains (for a full list of domains, see Overview Articles). It has been employed in typology of both grammatical and lexical semantics, but the former has received the greatest amount of attention. The two subsections provide references for both grammatical and lexical semantic maps. Classical Semantic Maps in Grammatical Typology The list of references in this subsection is not representative of the number of publications on the topic, since many grammatical studies using the graph-based approach are cited elsewhere (e.g., Early Studies and Theoretical Foundations, General Descriptions of the Model). Thus, this list should be seen as complementary to these studies. The references cited here cover the following semantic and syntactic domains (and other aspects of linguistic structure): the domain of imperative-hortatives (van der Auwera, et al. 2003); case marking, in particular the comitative-instrumental domain (Narrog and Ito 2007); clause linkage (Kortmann 1997, Malchukov 2004); negative existentials (Veselinova 2013); additive markers (Forker 2016); the ditransitive construction (Barðdal, et al. 2011); and secondary predication (van der Auwera and Malchukov 2005). This list also contains a contribution that makes use of certain typological hierarchies rather than just semantic maps. Typological hierarchies and semantic maps are not identical, as the former make predictions based not only on adjacency requirements but also on the directionality of the hierarchy. However, these two concepts are still to be conceived as similar leading to convergence with the semantic maps tradition. As a matter of fact, exploring typological hierarchies in the domain of valency classes and alternations, Malchukov 2015 employs techniques similar to those used to construct semantic maps. Notably, it presents a semantic map with directionalities not conceived primarily as diachronic in nature. 10 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229  Barðdal, Jóhanna, Kristian Emil Kristoffersen, and Andreas Sveen. 2011. West Scandinavian ditransitives as a family of constructions: With a special attention to the Norwegian “V-REFL-NP” construction. Linguistics 49.1: 53–104. The article depicts the semantics of the ditransitive construction in the form of a semantic map on the basis of data from West Scandinavian languages. The article suggests a semantic typology that includes eight higher-level semantic categories, such as actual transfer, intention, possession, etc. These categories are represented as nodes on the map, which occupy contiguous areas of the semantic space. The proposed map is also valid for Modern Icelandic and Old Norse.  Forker, Diana. 2016. Toward a typology for additive markers. Lingua 180:69–100. The author analyzes the various functions of additives based on data from forty-two languages. She presents the results of the study in the form of two different maps, which the author calls categorical (i.e., classical) and distance-based continuous (i.e., proximity) semantic maps. In categorical maps, the author manually manipulates the thickness of the line connecting two functions to represent their frequency of occurrence in the sample investigated.  Kortmann, Bernd. 1997. Adverbial subordination: A typology and history of adverbial subordinators based on European languages. Berlin: Mouton de Gruyter. This book investigates the morphology and the semantics of adverbial subordinators based on a sample of fifty-two languages. Chapter 7 is devoted to the semantic space of adverbial relations. In this chapter, Kortmann provides semantic maps of the most important semantic affinities within the semantic space of intercausal relations, which includes six sets of relations, namely temporal, modal, locative, causal, conditional, and concessive relations. Kortmann uses the term cognitive maps, which reflects his claim that these maps produce implicational universals of cognition.  Malchukov, Andrej. 2004. Towards a semantic typology of adversative and contrast marking. Journal of Semantics 21.2: 177–198. This paper aims to investigate the semantic relation between certain functions in the domain of adversativity and contrast. It proposes a semantic map for coordinating connectives, which contains nine functions, e.g., mirative, consecutive, contrastive, correction, etc. Among other things, the paper highlights the fact that a map can generate a set of implicational universals.  Malchukov, Andrej. 2015. Valency classes and alternations: Parameters of variation. In Valency classes in the world’s languages. Edited by Andrej Malchukov and Bernard Comrie, 73–149. Berlin: Mouton de Gruyter. Malchukov uses typological hierarchies and semantic maps to represent crosslinguistic generalizations and to uncover universal and language-particular properties of valency classes. He presents the points of convergence and divergence of the two concepts. Furthermore, the author analyzes the cross-linguistic variation in valency classes with reference to two general factors, namely iconicity and markedness. The author acknowledges the role of both competing mechanisms within the semantic map approach. 11 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229  Narrog, Heiko, and Shinya Ito. 2007. Re‐constructing semantic maps: The comitative‐instrumental area. STUF: Language Typology and Universals 60.4: 273–292. This paper addresses some of shortcomings of the semantic map model and offers solutions to many of them. The main problems identified are of a methodological nature, e.g., the issue of language-sampling, of replicability of findings, of the number of meanings included in the map. The authors investigate the comitative-instrumental area, relying on a sample of 200 languages, and they focus on the connections between the various functions in this domain. They suggest a formula, which measures the possibility that two or more functions are connected.  van der Auwera, Johan, Nina Dobrushina, and Valentin Goussev. 2003. A semantic map for imperative‐hortatives. In Contrastive analysis in language: Identifying linguistic units of comparison. Edited by Dominique Willems, Bart Defrancq, Timothy Colleman, and Dirk Noël, 44–66. Basingstoke, UK: Palgrave Macmillan. This paper investigates the structure of the imperative-hortative system with data from 376 languages. It proposes a semantic map that is built on three dimensions relating both to the nature and the number of the speech participants and to the nature of the imperative-hortative speech act.  van der Auwera, Johan, and Andrej Malchukov. 2005. A semantic map for depictive adjectivals. In Secondary predication and adverbial modification: The typology of depictive constructions. Edited by Nikolaus P. Himmelmann and Eva Schultze-Berndt, 393–423. Oxford: Oxford Univ. Press. Van der Auwera and Malchukov propose a map for depictive adjectival constructions with data from fifteen languages. They build the map, gradually providing justification at each step for the position of each node. The result is a syntactically based semantic map, which relies on shared syntactic rather than semantic features. They show that depictives are semantically close to four other semantic notions, namely predicatives, complementatives, attributives, and adverbials.  Veselinova, Ljuba. 2013. Negative existentials: A cross‐linguistic study. Rivista di Linguistica25.1: 107–145. This paper describes the semantics of negative existentials using the semantic map method. Veselinova builds her map from data collected from ninety-five genetically and areally diverse languages. One innovative methodological decision with respect to the spatial arrangement of the functions is that the position the functions occupy on the map reflects their frequency cross-linguistically: the most frequent ones are put in the center of the map, whereas the less frequent ones are put on the periphery (for a similar methodological choice, see Stassen 1997, cited under Early Studies and Theoretical Foundations). 12 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 Classical Semantic Maps in Lexical Typology In the semantic map literature, cross-linguistic studies on the polysemy of grammatical morphemes have dominated the field due to the general tendency of typologists to focus more on grammar rather than on the lexicon. Several reasons can be given for this imbalance. For example, the idea that the study of grammar produces more interesting results than the study of the lexicon has been very widespread. Furthermore, grammatical meanings have been considered more comparable across languages than the meanings of content words (for a discussion, see Haspelmath 2003 and François 2008, both cited under General Descriptions of the Model). Therefore, only a few cross-linguistic studies on the lexicon apply semantic maps. However, during the second half of the 2000s, the method has experienced a “lexical turn,” the starting point of which can be seen in François 2008(cited under General Descriptions of the Model; see also Majid, et al. 2007, cited under Proximity Maps in Lexical Typology). Perrin 2010 is another attempt to generate lexical semantic maps by investigating the semantic organization of polysemous qualitative concepts. Urban 2011 deals with formal and semantic patterns in the lexicon, placing particular emphasis on nominal expressions. Its empirical contribution lies in the fact that it makes generalizations concerning preferences for simple versus complex coding of certain referring expressions. Among other things, it discusses the links between meanings in certain semantic domains, the mechanisms accounting for these connections, and the issue of areal as well as universal lexico-semantic associations. It also makes a methodological contribution, moving from similarity maps to a version of classical maps. Rakhilina and Reznikova 2016 presents a frame-based methodology, in which semantic maps assist the visualization of the relations between the various meanings (which are called frames in this approach). Youn, et al. 2016 also focuses on the lexicon and, in particular, on a twenty-two-concept set of the Swadesh 200-word list of basic vocabulary. These twenty-two concepts belong to different domains, namely material entities, celestial objects, and natural and geographic features. It does not adopt the semantic map approach, but the method followed is very similar. Georgakopoulos, et al. 2016 investigates the polysemic patterns associated with the notion “soil/earth,” mostly in ancient languages.  Georgakopoulos, Thanasis, A. Daniel Werning, Jörg Hartlieb, et al. 2016. The meaning of ancient words for “earth”: An exercise in visualising colexification on a semantic map. eTopoi: Journal for Ancient Studies 6:1–36. This paper investigates the polysemic patterns of the concept “soil/earth” and produces a diagrammatic visualization of the semantic spaces of twenty lexemes in nine languages (mainly ancient). It also highlights some of the methodological problems associated with the dictionary-based method. In the notations used in the article, spatial contiguity is the most weighted factor given that the edges are omitted. The article comes with a very detailed appendix with glossed examples from all the languages studied.  Perrin, Loïc-Michel. 2010. Polysemous qualities and universal networks, invariance and diversity. Linguistic Discovery 8.1: 259–280. The author analyzes the semantic organization of quality expressions applying the semantic map method. He discusses networks with universal validity as well as polysemies that are language specific. In representational terms, the article adopts a 13 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 system with connecting lines, different colors, and bold versus standard font to capture certain generalizations (e.g., recurrent polysemies versus patterns specific to one language).  Rakhilina, Ekaterina, and Tatiana Reznikova. 2016. A frame‐based methodology for lexical typology. In The lexical typology of semantic shifts. Edited by Päivi Juvonen and Maria Koptjevskaja‐Tamm, 95–129. Berlin: Mouton de Gruyter. This chapter gives information on how to adapt semantic maps to the needs of lexical typology. It presents the points of convergence and divergence of grammatical and lexical semantics maps. What is important for the methodology used in the paper is that a map is built from frames. The case studies discussed are the domain of emptiness and that of aquamotion.  Urban, Matthias. 2011. Analyzability and semantic associations in referring expressions: A study in comparative lexicology. PhD diss., Univ. of Leiden. This dissertation pays attention to the nominal lexicon focusing on simple as well as morphologically complex expressions. It makes inferences about the patterns of associations in lexical items within certain semantic domains and visualizes these associations in the form of adjacency networks. These networks use several representational conventions, e.g., thickness of lines to represent the strength of the association in the languages of the world or arrows to indicate the directionality of the mapping as revealed by morphologically complex expressions.  Youn, Hyejin, Logan Sutton, Eric Smith, et al. 2016. On the universal structure of human lexical semantics. Proceedings of the National Academy of Sciences of the United States of America 113.7: 1766–1771. This paper provides a metric for measuring semantic similarity between concepts. The authors construct a highly structured network for twenty-two concepts from the Swadesh list, relying on data obtained from translations in a balanced sample of eighty-one languages. The lines connecting the concepts are given different widths depending on the frequency of attestation of the polysemy pattern cross-linguistically. The article includes an extended appendix with specifics, among others, on the methodology of data collection and the methods of network representation. Proximity Maps One of the main shortcomings of the graph-based/classical approach has been its inefficiency in handling large typological data sets (see, however, the improvement in Regier, et al. 2013, cited under Computational Approaches). The classical semantic map approach has also been criticized because it does not follow a well-established protocol, which linguistics can adopt and which will ensure findings are replicable (see also Theoretical Debates). As a reaction to these limitations, the second era of the model witnessed a shift in focus. More specifically, Croft and Poole 2008 (cited under Proximity Maps in Grammatical Typology) suggests an alternative method, based on a computationally tractable technique known as multidimensional scaling (MDS), which enables the inference of universals based on large data sets. MDS is 14 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 not the only multivariate statistical technique employed in this stream of research, and some studies resort to correspondence analysis as well. These multivariate statistical techniques produce proximity maps that plot linguistic or non-linguistic stimuli in a multidimensional space, with distances between points in the representation indicating the similarity or dissimilarity of meanings. Proximity maps have been used in grammatical as well as in lexical typology. Proximity Maps in Grammatical Typology As was the case with classical semantic maps, research on proximity maps prioritized the study of grammatical phenomena. Thus, proximity maps were initially used to analyze cross-linguistic variation and uncover grammatical universals. Croft and Poole 2008, which has been circulated among researchers since 2004, suggests the paradigm shift from graph-based maps to proximity maps. Clancy 2006 is one of the first studies applying the new method in a small set of related languages. Cysouw 2007 discusses the scope of semantic maps as a linguistic model, highlights the importance of adding to the map information about the frequency of polysemy patterns across languages, and underscores the need to operationalize meaning so that it can be measured (which will make the results replicable). The issue of comparability is also addressed in Hartmann, et al. 2014, which analyzes the semantic roles of individual verbs (microroles). These microroles are used as comparative concepts that help identify comparable phenomena across languages. In addition to its empirical contribution, the article makes a methodological contribution by coining the term coexpression, which, in the context of semantic maps, allows us to remain neutral between different interpretations (e.g., between vagueness and polysemy or even homonymy). The grammatical domains covered in the references cited in this subsection are: person marking (Cysouw 2007), case (Clancy 2006), possessive constructions (Eckhoff 2011), and analytic causative constructions (Levshina, et al. 2013).  Clancy, Steven J. 2006. The topology of Slavic case: Semantic maps and multidimensional scaling. Glossos 7:1–28. One of the first studies applying MDS analysis, which was pioneered in Croft and Poole 2008. The study investigates case semantics in three languages, namely Russian, Polish, and Czech.  Croft, William, and Keith T. Poole. 2008. Inferring universals from grammatical variation: Multidimensional scaling for typological analysis. Theoretical Linguistics 34.1: 1–37. This is a position paper published in a special issue of Theoretical Linguistics 34.1 (cited under Special Issues in Journals). The paper summarizes the advances of the semantic map method up to that point, but simultaneously points out a series of problems that challenge its theoretical value. It presents an alternative approach to the classical semantic map method, called the MDS model. The MDS model provides a powerful tool for inferring grammatical universals. 15 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229  Cysouw, Michael. 2007. Building semantic maps: The case of person marking. In New challenges in typology. Edited by Bernhard Wälchli and Matti Miestamo, 225–248. Berlin: Mouton de Gruyter. This article investigates the cross-linguistic diversity of categories in the domain of person marking. It discusses the methodological status of semantic maps in linguistic theory and theorizes about the prerequisites for a model of the linguistic variation in the form of a semantic map. The focus moves from the question of which mapping method is the best one to the question of which method is optimal for a particular purpose.  Eckhoff, Hanne M. 2011. Old Russian possessive constructions: A construction grammar approach. Berlin: Mouton de Gruyter. This is the first book in which the author uses semantic maps as the main theoretical tool. This synchronic and diachronic study investigates the semantics of five possessive constructions in Old Russian (11th–14th centuries) and Middle Russian (17th century). In the synchronic analysis, the semantic map is obtained through correspondence analysis, whereas, in the diachronic part, a preliminary attempt is made to combine proximity maps with classical maps (as suggested in van der Auwera 2008, cited under Theoretical Debates).  Hartmann, Iren, Martin Haspelmath, and Michael Cysouw. 2014. Identifying semantic role clusters and alignment types via microrole coexpression tendencies. Studies in Language38.3: 463–484. This paper presents a method for identifying clusters of semantic roles and provides a semantic map of the roles of individual verbs (microroles) across twenty-five languages drawn from the Valency Patterns Leipzig database. The authors analyze 181 microroles from eighty-seven different verb meanings. The authors coin the term coexpression to refer to the phenomenon in which a microrole A is coded like/aligned with a microrole B across diverse languages. Thus, the crucial parameter is that of the identity of coding between different microroles.  Levshina, Natalia, Dirk Geeraerts, and Dirk Speelman. 2013. Mapping constructional spaces: A contrastive analysis of English and Dutch analytic causatives. Linguistics 51.4: 825–854. This is a corpus-based, contrastive study of English and Dutch analytic causative constructions. The method used relies on the statistical technique of multiple correspondence analysis. Through this method a multidimensional conceptual space of the English and Dutch causative constructions is created, which allows the identification of both similarities and differences between the two languages. Proximity Maps in Lexical Typology The main practice when constructing proximity maps in lexical typology is either to rely on non-linguistic stimuli or to draw on data from parallel corpora. Majid, et al. 2007 uses pictures and videoclips (i.e., non-linguistic materials) to elicit descriptions 16 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 about ‘‘cutting and breaking” events and investigates how similar the semantic categories of these events are across languages. Wälchli and Cysouw 2012, on motion verbs, as well as Wälchli 2016, on perception verbs, build probabilistic maps (their term for proximity maps) from parallel texts. In the studies in which maps are plotted based on parallel corpora, comparability is ensured by comparing languageparticular form classes (instead of categories) from different language varieties (which need not be representative of a language) across a sample of contextually embedded situations (instead of functions).  Majid, Asifa, Melissa Bowerman, Miriam van Staden, and James S. Boster. 2007. The semantics of “cutting and breaking” events: A cross-linguistic perspective. Cognitive Linguistics 18.2: 133–152. The authors use correspondence analysis to produce semantic maps that visualize the similarity of semantic categories of ‘‘cutting and breaking” across twenty-eight typologically, genetically, and geographically diverse languages.  Wälchli, Bernhard. 2016. Non‐specific, specific and obscured perception verbs in Baltic languages. Baltic Linguistics 7:53–135. To overcome the discrete-feature bias (see also Janda 2009, cited under Theoretical Debates), inherent in the classical semantic approach, the author applies the method of proximity maps, which are built from parallel corpora, and suggests an upgraded role of psychological approaches to perception. The focus is on perception verbs in the Baltic languages.  Wälchli, Bernhard, and Michael Cysouw. 2012. Lexical typology through similarity semantics: Toward a semantic map of motion verbs. In Special issue: New directions in lexical typology, linguistics Edited by Maria Koptjevskaja‐ Tamm and Martine Vanhove. Linguistics 50.3: 671–710. Focusing on lexical semantics in motion verbs, the authors propose building semantic maps from exemplar data in parallel texts. They draw data from translations of the Gospel according to Mark across 101 doculects (a label that replaces the notion of language in the article). They stress the relevance of the proposed method for lexical typology and also include an extended appendix discussing the necessity of diachronic maps as well as the possibility of proximity maps to integrate diachrony. Semantic Maps in Diachrony Though research on semantic maps is mainly associated with synchrony, the role of diachrony in the model has been stressed since its early days (see Anderson 1982, Kemmer 1993, van der Auwera and Plungian 1998, all cited under Early Studies and Theoretical Foundations). The significance of the historical dimension for semantic maps is outlined in van der Auwera 2008 (cited under Theoretical Debates) and van der Auwera 2013 (cited under General Descriptions of the Model). Due to the exclusive synchronic orientation of proximity maps, it comes as no surprise that research aiming at reporting on the directionality of semantic extensions relied on the classical semantic approach. The common denominator in most diachronic studies 17 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 using the semantic map tool is their focus on the grammatical domain. Traugott 2016 is innovative in that it demonstrates the usefulness of semantic maps in visualizing the diachronic development of relationships among polysemous constructions and not just grammatical morphemes. The studies listed here reveal another striking fact: the range of linguistic phenomena covered is rather limited. In fact, the main bulk of research is concentrated on preposition and case functions, such as Luraghi 2003, Narrog 2010, Luraghi 2014, and Guardamagna 2016. Directionality of change is represented in the form of an arrow, which resembles the strategy employed in grammaticalization research (Narrog and van der Auwera 2011). It should be noted that, due to lack of diachronic data, directionality of semantic change is often established in the relevant literature on the basis of synchronic data (Narrog 2010). Regarding the representation of the diachronic information, Andrason 2016 takes a different perspective by modeling meaning in terms of a wave consisting of two axes, one corresponding to the grammaticalization cline and another correlating meanings with their frequency.  Andrason, Alexander. 2016. From vectors to waves and streams: An alternative approach to semantic maps. Stellenbosch Papers in Linguistics 45:1–29. This article suggests a method of representation of meaning relations that builds on classical semantic maps and extends them in two ways: first, by integrating into the diachronic map information relating to frequency of attestation of a certain meaning and, second, by considering the relation of a plotted map to other maps of markers with similar meanings. The new map takes the form of a wave.  Guardamagna, Caterina. 2016. A diachronic semantic map for the Latin preposition secundum. Journal of Latin Linguistics 15.2: 233–277. This article presents a diachronic semantic map of spatial-temporal, logical, and (inter)subjective meanings of the preposition secundum in four different diachronic stages of Latin. Since the focus of the paper is on one language, in some cases the judgments about what counts as meaning or not are backed up by cross-linguistic evidence. Note, however, that the four diachronic stages could be thought of as representing different languages.  Luraghi, Silvia. 2003. On the meaning of prepositions and cases: The expression of semantic roles in Ancient Greek. Amsterdam: John Benjamins. This study makes use of semantic maps in heuristic fashion to describe the meanings of prepositions and cases in Ancient Greek. It also reports patterns of semantic extension and considers these patterns as hints as to which meanings are contiguous in conceptual space. The book carefully articulates the conditions (and the restrictions) under which the semantic map model can be used to represent the different meanings of a specific form in just one language.  Luraghi, Silvia. 2014. Plotting diachronic semantic maps: The role of metaphor. In Perspectives on semantic roles. Edited by Silvia Luraghi and Heiko Narrog, 99–150. Amsterdam: John Benjamins. This paper discusses how metaphors can provide hints for constructing semantic maps that integrate diachronic information. 18 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229  Narrog, Heiko. 2010. A diachronic dimension in maps of case functions. Linguistic Discovery8.1: 233–254. This article shows why diachronic information is important for semantic maps focusing on case functions and, in particular, on three domains: companion-instrument, sourceagent, and goal-recipient. It presents evidence from uncontroversial directionalities, but also discusses cases from either controversial or overlooked connections.  Narrog, Heiko, and Johan van der Auwera. 2011. Grammaticalization and semantic maps. In The Oxford handbook of grammaticalization. Edited by Heiko Narrog and Bernd Heine, 318–327. Oxford: Oxford Univ. Press. This article demonstrates the efficiency of the semantic map tool in historical linguistics and, in particular, in grammaticalization studies.  Traugott, Elisabeth C. 2016. Do semantic modal maps have a role in a constructionalization approach to modals? Constructions and Frames 8.1: 98– 125. This paper illustrates how semantic maps can be useful for informing diachronic constructional analyses. It proposes a reconceptualization of semantic maps as a tool operating on two levels, a macro level and a micro level. The two levels correspond to two kinds of maps: schema‐construction maps and micro‐construction maps. The case study of the paper is the development of the modals BETTER, RATHER, SOONER. Semantic Maps in Language Contact Semantic maps have also been used to discuss contact-induced language change. Tenser 2016focuses on language-specific polysemic patterns, adjusts the connectivity hypothesis to the needs of maps in the context of contact-induced change, and shows that semantic maps can be borrowed from one language to another. Gast and van der Auwera 2012 elaborates on the process of semantic map assimilation, which can be bidirectional in that it may involve changes in both the source and the target language. Grossman and Polis 2017 deals with the semantics of adposition borrowing and addresses the question of what exactly can be borrowed. Gil 2017 discusses cases in which coexpression patterns are replicated through borrowing.  Gast, Volker, and Johan van der Auwera. 2012. What is “contact‐induced grammaticalization”? Examples from Mayan and Mixe‐Zoquean languages. In Grammatical replication and borrowability in language contact. Edited by Björn Wiemer, Bernhard Wälchli, and Björn Hansen, 381–426. Berlin: Mouton de Gruyter. This paper discusses the “semantic map assimilation” process, which refers to the phenomenon in which a marker in a language extends its meaning on the basis of the meaning of a similar marker of some contact language (cf. the term semantic map harmony in Tenser 2016). 19 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229  Gil, David. 2017. Roon ve, DO/GIVE Coexpression and language contact in Northwest New Guinea. NUSA 62:41–100. This article proposes a semantic map for the form ve in Roon, a language of the South Halmahera West New Guinea branch of Austronesian. It elaborates on each of the thirteen functions of ve, analyzes the relationships that hold between them, and discusses the motivations of twenty-three pairwise connections. The focus is mainly on the DO/GIVE coexpression (for the term coexpressionsee Hartmann, et al. 2014, cited under Proximity Maps in Grammatical Typology), which is cross linguistically uncommon, but occurs frequently in Northwest New Guinea.  Grossman, Eitan, and Stéphane Polis. 2017. Polysemy networks in language contact: The borrowing of the Greek‐origin preposition κατά/κατα in Coptic. In Greek influence on Egyptian‐Coptic: Contact‐induced change in an ancient African language. Edited by Eitan Grossman, Peter Dils, Sebastian Richter, and Wolfgang Schenkel, 335–367. Hamburg: Widmaier Verlag. Taking the Greek-origin Adposition κατα in Coptic as a case study, this paper shows that this Coptic adposition does not borrow the entire complex polysemy network of the Greek source lexical item. As a result of this selective copying (in fact the authors suggest the term adaptive copying because new functions are also added to the network), the language-specific map of the functions of κατα in Coptic violates the connectivity hypothesis (since one of the core spatial meanings is not borrowed in Coptic). The authors provide a diachronic explanation for this discontinuity.  Tenser, Anton. 2016. Semantic map borrowing—Case representation in northeastern Romani dialects. Journal of Language Contact 9.2: 211–245. The author uses semantic maps to describe and represent instances of contactinduced grammaticalization. By investigating a sample of Romani dialects, he shows that semantic maps are borrowed from the contact languages, i.e., the Slavonic languages. Semantic map borrowing seems to rely on what Tenser labels “semantic map harmony.” The article concludes that semantic maps are a useful tool that can advance the study of language contact. Computational Approaches This section contains a few studies that use innovative methods to overcome certain methodological problems associated with the semantic map model. Regier, et al. 2013 takes up the challenge of addressing one of the basic criticisms against the graph-based approach: the fact that manually building maps is practically impossible, especially when one needs to process large-scale cross-linguistic data sets. Even the potentiality of plotting such maps automatically has been questioned, since the semantic map inference problem has been considered to be computationally intractable (see Croft and Poole 2008, cited under Proximity Maps in Grammatical Typology). Regier, et al. acknowledge the intractability of the problem, but they present an algorithm that approximates the optimal solution and produces results of high quality. Ryzhova and Obiedkov 2017 suggests a new method for plotting semantic maps based on formal concept analysis. This method automatically produces 20 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 hierarchical graphs (thus, it is a graph-based method as well), which are more complex and more informative than classical semantic maps. The resulting map can be less reader-friendly, though, if large data sets are to be processed. van Trijp 2010 presents experiments on artificial language evolution and, based on the results of these experiments, takes issue with the assumption that semantic maps represent the global geography of the human mind.  Regier, Terry, Naveen Khetarpal, and Asifa Majid. 2013. Inferring semantic maps. Linguistic Typology 17.1: 89–105. The authors address the criticism against the classical semantic maps approach that semantic maps cannot be used with large and highly variable cross-linguistic data sets. They present an algorithm, originally used to address the social network inference problem, which produces automatically plotted maps. They test the algorithm, among other data sets, on the cross-linguistic data of Haspelmath 1997 (cited under Early Studies and Theoretical Foundations) on indefinite pronouns and they conclude that it achieves high quality results.  Ryzhova, Daria, and Sergei Obiedkov. 2017. Formal concept lattices as semantic maps. In Computational linguistics and language science: Proceedings of the Workshop on Computational Linguistics and Language Science, Moscow, Russia, April 2016. Edited by Ekaterina L. Chernyak, 78–87. Aachen, Germany: CEUR Workshop Proceedings. This paper proposes a new approach to automatically building semantic maps based on formal concept analysis, which seeks to overcome certain shortcomings of classical semantic maps. The key notion in this approach is that of “concept lattice,” which should be understood as having two dimensions, one that includes the complete list of words and one that contains the meanings of the lexemes in the data set. The authors claim that such an analysis more accurately captures typological regularities.  van Trijp, Remi. 2010. Grammaticalization and semantic maps: Evidence from artificial language evolution. Linguistic Discovery 8.1: 310–326. This paper presents experiments on artificial language evolution focusing on the domain of grammar and, in particular, on the formation of case markers for semantic roles. It discusses the results of the experiments with reference to the question of the existence of a universal conceptual space. Digital Resources The references in this section include tools that can assist the visualization of semantic maps as well as online resources that can be used by researchers to extract or to assist the study of cross-linguistic polysemy patterns. Gephi is designed for visualizing and exploring complex networks. It contains many features that enable the user to make sense of complex data. For example, the modularity feature helps detect clusters of meanings in large networks (by applying cluster analysis methods). The idea is that similar meanings will form a group and the modularity-based clustering will produce as many communities/clusters as the data dictate. Subsets of nodes (which 21 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 will be meanings in our case) and edges (i.e., links between the meanings) can also be hidden in the network. This manipulation could be useful, for example, in removing infrequent polysemy patterns from the network. CLICS2 is an online database that offers information on lexical associations in more than twelve hundred languages. The visualization component of the database consists of nodes, edges, and weights whose thickness reflects the number of a given colexification in the languages of the world. This component features several functionalities that allow the user to interact with the content of the platform (e.g., when moving the mouse over a link of the network all languages exemplifying the respective colexification pattern appear in a list). The rich material in the database, as well as the additional visualization component of CLICS2, can be exploited by researchers in lexico-typological studies (e.g., in detecting areal patterns). Concepticon is a meta-resource whose programmatic goal is to link the many different existing concept lists, which can be found in the linguistic literature. It links more than 30,000 concepts from 160 concept lists to approximately 2,500 concept sets (since the content in this resource is regularly updated, the number of concepts, concept lists, and concept sets increases). This resource gives access to a wealth of information, e.g., to studies on semantics extensions, cross-linguistic polysemies, and lexical associations, which can inform semantic map studies.  Bastian, Mathieu, Sebastien Heymann, and Mathieu Jacomy. 2009. Gephi: An open source software for exploring and manipulating networks. In ICWSM: Proceedings of the Third International AAAI Conference on Weblogs and Social Media, 17–20 May 2009, San Jose, California, USA. Edited by Association for the Advancement of Artificial Intelligence, 361–362. Menlo Park, CA: Association for the Advancement of Artificial Intelligence Press. A desktop, free and open source software, that was created for graph and network analysis. The webpage includes different tutorials, which guide the users to the visualization and layout settings of the program. Gephi is effective in visualizing large graphs with thousands of nodes. This visualization platform can aid the graphical representation of classical semantic maps (for examples, see Georgakopoulos, et al. 2016, cited under Classical Semantic Maps in Lexical Typology; and Georgakopoulos and Polis 2018b, cited under Overview Articles). Gephi is available online.  List, Johann-Mattis, Simon Greenhill, Cormac Anderson, Thomas Mayer, Tiago Tresoldi, and Robert Forkel, eds. 2018a. Database of Cross-Linguistic Colexifications. Jena, Germany: Max Planck Institute for the Science of Human History. The Database of Cross-Linguistic Colexifications (CLICS2) contains information about synchronic lexical associations in 1,220 language varieties of the world and includes 2,638 different colexification patterns. It is a useful tool for research in lexical typology as it facilitates the investigation of colexification patterns on both global and regional scales. The platform allows users to explore the data from various perspectives, e.g., through geographic maps.  List, Johann-Mattis, Michael Cysouw, Simon Greenhill, and Robert Forkel, eds. 2018b. Concepticon. Jena, Germany: Max Planck Institute for the Science of Human History. 22 ***This is a final post-review version of the article*** For the published version, please refer to Oxford Bibliographies in Linguistics DOI 10.1093/OBO/9780199772810-0229 Concepticon contains a large amount of different (published and popular) concept lists, which are used in the linguistic literature. This resource links concept labels, such as KNOW, to a certain concept set, which is given a unique definition (e.g., the definition provided for KNOW is “to be certain or sure about something”). It offers access to a rich amount of data, which can be used to study, among other things, cross-linguistic polysemy patterns and the semantic evolution of certain lexemes. It is published as Linked Open Data. 23