LINGUIST List 25.2185

Sun May 18 2014

Review: Anthropological Linguistics; General Linguistics; Historical Linguistics; Sociolinguistics: Michaelis et al. (2013)

Editor for this issue: Rajiv Rao <>

Date: 21-Jan-2014
From: Heiko Narrog <>
Subject: The Atlas and Survey of Pidgin and Creole Languages
E-mail this message to a friend

Discuss this message

Book announced at

EDITOR: Susanne Michaelis
EDITOR: Philippe Maurer
EDITOR: Martin Haspelmath
EDITOR: Magnus Huber
TITLE: The Atlas and Survey of Pidgin and Creole Languages
SUBTITLE: Super Set: Four-volume Pack
PUBLISHER: Oxford University Press
YEAR: 2013

REVIEWER: Heiko Narrog, Tohoku University


This is a four-volume set aiming to provide a description of (mainly structural) features of pidgin and creole languages world-wide. The first volume, the Atlas (henceforth: APiCS) provides a survey of linguistic features across all languages regardless of area or lexifier language, while the three survey volumes (henceforth: SPiCL) provide descriptions of individual languages divided by lexifier. In what follows I will first describe the APiCS and then the SPiCL.

The APiCS has a general introduction to the set in which the design of the volumes is explained in some detail. The term “pidgin and creole languages” is used traditionally without an attempt at a new definition (xxxv). 76 pidgins and creoles were chosen, and all the descriptions provided throughout the four volumes are based on these 76 languages. Since the total number of pidgins and creoles world-wide is considerably higher, this is only a limited sample. Furthermore, as the editors frankly admit, and even emphasize, the sample is not necessarily balanced, and has a bias towards languages with English as the lexifier (27), followed by Portuguese (14) and French (9). Only a minority of languages (19) do not have a European lexifier. The idea for this volume was born at a conference in 2005, leading to the formation of the editorial team. According to the editors, the bias was unavoidable, as contributors for many languages that would have been interesting for the sake of more balance or for specific features were simply not available. Therefore, the results presented in the APiCS are not necessarily representative of pidgins and creoles in general. However, they reflect the current landscape of research, and are therefore close to the optimum of what is practically possible.

Readers familiar with the World Atlas of Language Structures (WALS; Haspelmath et al. 2005), will immediately identify the APiCS as a sister volume to this prominent predecessor. While the WALS presents descriptions and maps of 142 structural and lexical features in a sample of 200 languages world-wide, the APiCS has 130. For the sake of cross-linguistic comparability, the features are abstract ones, such as “order of adjective and noun” or “indefinite pronouns”, and are not bound to concrete forms. Thus, “adjectives” will include semantic-functional equivalents to adjectives, even if they morphosyntactically behave as verbs or nouns in a particular language. Many of the features are taken from the WALS — some of the WALS features were discarded, as they turned out to not be meaningfully applicable to the new sample of languages — and some features were taken from Holm & Patrick’s (2007) systematic survey of semantic and syntactic features in 18 creole languages. However, there are a number of decisive differences between WALS and APiCS. The most important one is that while in the WALS, the entries on each structural feature were written by experts on these features who extracted the required information from available language descriptions, in the APiCS, the expert on a particular language cooperating in the project contributed the data directly. Therefore, the “APiCS Consortium”, that is, the 88 experts cooperating in the project, are co-authors on each entry (the other authors are the editors in charge of a particular entry).

This approach has far-reaching consequences for the reliability of the data. In early reviews of the WALS (e.g. Bright 2007, Donohue 2006), the reviewers pointed out rather striking shortcomings in the analysis of data from languages of their own area of expertise. The reviewers estimated a margin of error of 10 to 20% in the data. Since the WALS has been continuously updated in its online version, many of these errors will have been remedied by now, but they remain in the book version. With the APiCS approach, such errors are practically ruled out, unless they are at a level of disagreement among experts on a particular language regarding how to analyze a specific phenomenon in that language. Inconsistencies, in terms of sample size and analysis across entries, which were another point of criticism against the WALS, are also practically ruled out in the APiCS. This apparently took an immense amount of work; the editors describe in the introduction that they repeatedly refined their questionnaire and had the authors review and revise their data for the survey.

Other differences include more flexibility in the classification of features (in the APiCS, a language is allowed to have two complementary values of a feature at the same time), and the lack of a CD-Rom. APiCS data are available online (, which eliminates the need for an additional data disk.

The 130 features are divided as follows: word order (12 features), nominal categories (24 features), nominal syntax (6 features), verbal categories (14 features), argument marking (16 features), clausal syntax (19 features), complex sentences (8 features), negation, questions and focusing (7 features), lexicon (11 features), and phonology (13 features). Each entry on a feature (“chapter”) is organized into two parts, a text and a map. The text consists of: (1) feature description, (2) the values, and (3) distribution, and usually has two pages. It is followed by a two-page world map in a Gall-Peter projection (which distinguishes itself from other world map projections by keeping the circumpolar regions small) on which colored dots, each representing a language, indicate the presence, absence, or nature of the particular feature in that language. The large majority of features are just what one would expect from such a map. Besides those already mentioned, we find entries such as “definite articles” (nominal categories), “marking of possessor noun phrases” (nominal syntax), “tense-aspect systems” (verbal categories), “expletive subject of existential verbs” (argument marking), “reflexive constructions” (clausal syntax), “subject relative clauses” (complex sentences), “polar questions” (negation, questions, and focusing), “hand and arm’” (lexicon), and “tone” (phonology). It would be futile to criticize the lack of features that were present in the WALS, since the editors ensure us that they went through a gradual process of narrowing down many more features than those that actually worked. Some features will strike the average reader, at least the non-creolist, as exotic. Thus, we find an “antidual of paired body-part terms” (ch. 27), or the lexicalization of Portuguese pequenino (ch. 109), and the Romance ‘know’ verbs saber (Portuguese, Spanish) and savoir (French) (ch. 110). The latter two are deviations from the volume design because they deal with concrete forms. However, the striking fact about them is that they can also be found in languages that have a non-Romance lexifier. Thus, these seemingly exotic chapters legitimately reflect what is found to be interesting or striking for creolists. The 130 entries are followed by a list of references, a subject index, an author index, and a language index.

The SPiCL is divided into three volumes on Germanic-based (i.e. English- and Dutch-based languages: Volume I; 28 Chapters), Romance-based (i.e. Portuguese-based, Spanish-based and French-based languages: Volume II; 27 Chapters), and contact languages based on languages from Africa, Asia, Australia, and the Americas (Volume III, 19 Chapters). Each chapter is written by the same expert or team of experts that contributed the respective information to the APiCS. With remarkable consequence, the SPiCL chapters also have a unified format with identical section titles, and almost identical length. Each chapters starts with (1) an introduction, which is accompanied by an areal map, followed by (2) sociohistorical background, (3) sociolinguistic situation, (4) phonology, (5) noun phrase, (6) verb phrase, (7) simple sentences, (8) complex sentences, a short glossed text, and the references, squeezed into 8 to 11 pages of relatively dense two-column print. A sound file for many of the glossed texts is freely available on the APiCS webpage. Despite the unified format, these chapters also provided the opportunity for authors to point out structural features that were not included in the APiCS survey. Thus, for example, we find some details on several types of adverbial clauses in section 8, and information on left dislocation and topicalization in section 9 of the Sranan chapter (Volume I; authored by Donald Winford and Ingo Plag). In contrast, the APiCS volume has no chapters dedicated to adverbial clauses and information structure. Generally, though, the information in the SPiCL chapters is fairly concise and almost bare-bones, given the obvious constraints on their length. The articles are followed by a general index. The closest predecessors to the SPiCL that I am aware of is Holm (1989) a single-authored work, which covers more than a 100 languages, but in considerably less detail, and the above-mentioned Holm & Patrick (2007), which has descriptions of 18 languages by experts on them. Furthermore, we have the Atlas of the Languages of Suriname, by Carlin & Arends (2002), which is very nicely edited with many maps and figures, but confined to the languages of that region.

The APiCS online ( provides the following: a map of all languages and a short description of each language with a link to each feature, the audio file of the glossed text (if available), a comprehensive list of sources on each language, a short description and map for each feature, a list of example sentences with glosses, a list of sources with more than 1,500 entries, and a list of authors with affiliations and homepage links. The list of sources, the list of languages, and a database of features are freely downloadable. The SPiCL is understandably not part of the online edition. In print, the combined APiCS and SPiCL are available both individually and as a four-volume set (at a discount). There is also the option to purchase a package of the three-volume SPiCL at a discount compared to the individual volumes.


From the description above, it is clear that the APiCS is groundbreaking in several respects. First, there has never been a systematic description of structural (and some lexical) features at the same scale before. The desirability of such a description is beyond doubt. Most linguists agree on the special status of pidgins and creoles in the study of language. The WALS 200-language sample only included a single creole language, and the WALS 100-language sample not even a single one, thus, this group of languages was not well served in the WALS. The unprecedented scale of systematic description is compounded by the flexibility with which the results can be used through the availability of data in electronic form. Even the accompanying SPiCL, which is not available in electronic form, is unprecedented in scale and systematicity.

Will the APiCS and SPiCL be useful to linguists and help to advance the field? There is little doubt about this. It is now possible to obtain and compare features of pidgins and creoles at a glance or a mouse-click. This is not only relevant for specialists in the field but also general linguists, typologists, students of syntax, of language acquisition, etc., who can get an immediate answer if they wonder about the presence or absence of specific linguistic features in pidgins and creoles. The availability and combinability of information on such features can also lead to new research questions, such as the correlation between specific features. The complementary SPiCL provides concise and reliable information on an exceptionally wide range of languages in one publication. We have, meanwhile, many encyclopedic volumes about language families and language areas through series such as the Routledge Language Family Series and the Cambridge Language Surveys, but not many can match the SPiCL with respect to number of languages covered in exactly the same depth and format.

Decisively for quality, the APiCS data and the SPiCL chapters were provided by experts on the respective languages. This guarantees a level of quality that was not available with the WALS. It means that quibbles over the language description and data interpretation are practically restricted to issues between experts on the languages. Since all data were thoroughly filtered by the editors, it also means that all data are fully comparable across languages. This was another complaint leveraged against the WALS that does not hold for the APiCS.

Finally, what is left to be desired? It would not be difficult to make up a list of languages and of linguistic features that might have been part of the APiCS (and SPiCL). I already mentioned adverbial clauses or information structure above. Furthermore, there are no German- or Russian-based languages included. Here, we have to trust the editors who tell us that they picked up as much as was practically possible. The 130 features across 76 languages are doubtlessly a great accomplishment. One possible objection is that linguistic features should have been included for which comparable information could have been extracted only in a limited number of languages. However, this would have led to an unevenness in description that might have provoked more complaints about the design of the APiCS than their omission.

One thing that I personally missed, and I think could have been provided, was a map of “all” pidgins and creoles worldwide, to the extent that they are known. Further distinguishing those included in the APiCS sample and those that were not may have helped to get a better picture of how representative or non-representative the 76-language sample is after all. The editors caution us that it is not necessarily representative but it would be good to know to what extent this is the case.

Overall, I think that the APiCS and the SPiCL are an outstanding achievement by the editors and authors, and a great service to the linguistic community.


Bright, William. 2007. Review of Haspelmath, Martin; Dryer, Matthew S.; Gil, David; Comrie, Bernard: The World Atlas of Language Structures. Oxford University Press 2005. International Journal of American Linguistics 73/2, 241-244.

Carlin, Eithne B. and Jacques Arends (eds.). 2002. Atlas of the Languages of Suriname. Leiden: KITLV Press.

Donohue, Mark. 2006. Review of Haspelmath, Martin et al. (eds) 2005. The World Atlas of Language Structures. Oxford University Press. LINGUIST List 17.1055.

Haspelmath, Martin, Matthew S. Dryer, David Gil & Bernard Comrie (eds). 2005. World Atlas of Language Structures. Oxford: Oxford University Press.

Holm, John. 1989. Pidgins and Creoles. Volume II. Reference Survey. Cambridge: Cambridge University Press.

Holm, John & Peter L. Patrick. 2007. Comparative Creole Syntax. Parallel Outlines of 18 Creole Grammars. Plymouth: Battlebridge.


Heiko Narrog is an associate professor at Tohoku University in Sendai, Japan. His research interests include language typology, historical linguistics, syntax and semantics, modality, and the Japanese language.

Page Updated: 18-May-2014