Avatars in Real Life
Placename homonymy in Iran

Vlad Atanasiu, 2010

If somebody would like to hide something precious in Iran, a good choice would be the town of Hoseynabad. There are indeed over 380 Hoseynabad, which would confuse for a long time any search party. (Hoseyn is a common personal name and reference to an important martyr in the Shi'a religion; -abad / abode, dwelling, city is the usual suffix for constructing a placename in Iran.)

Intelligence and military applications are not the only ones benefiting from knowledge on placename homonymy. This universal georeferencing issue provides also excellent material to teach cultural geography, its study bundling together Humanities disciplines like linguistics, history and sociology, with the Science areas of geographical information systems, visualization and computational knowledge modeling. Some of these topics will be presented in the following preliminary sketch of an ongoing research.

Density

Method - To study the density of homonymous populated places (HPPL) in Iran relative to the overall density of populated places (PPL), placenames and their coordinates where obtained from the Geonames website, two subsets constructed, one for the PPL (gray markers, Figure 1), the other for the HPPL (red markers). The datasets were processed in ArcGIS^? with Kernel Density, Spatial Analyst toolbox, to generate spatially continuous data (Figure 2-top-left & 2-top-right give the PPL and HPPL, respectively), the values in each point of the resulting maps being then divided, to get the ratio of HPPL in PPL (Figure 2-bottom). This ensures that HPPL density can be analyzed with much less dependency on PPL density, subduing the impact of natural features such as deserts and mountains and human factors like suburban dense settlements. For reference the provicial capitals are displayed in Figure 2.

Figure 1 - Populated places of Iran.

Figure 2 - Density of homonymous settlements in Iran.

Interpretation - By looking at the HPPL areas with low density (dark blue) we observe that they correlate with linguistic minorities: Turkmene, Gilani and Azeri on the Caspian shores, Kurdish and, farther South, Arabic at the border with Iraq and Baluchi on the Pakistani frontier. High density HPPL (dark brown) areas concentrate on the Iranian highland around urban centers (black markers), where Persian is the majority language and culture.

Persian representing also the high culture and language since Antiquity in this region of the world, it had an impact on the toponymy of other cultures of Iran, by lexical imposition and hybridization. Imposition (e.g. placenames derived from names of rulers, like the Banadr-e Pahlavi in Gilani territory named after the Pahlavi dynasty - and changed to Bandar-e Anazali after the 1979 Revolution) reduce the number of placenames available for naming in a minority language, while hybridization creates unique names, lowering the chances to be reproduced elsewhere. The HPPL density map reveals itself as a special visualization of the power centers and backwaters of Iran.

The hypothesis of hybridized cultures can be tested by expanding the study of HPPL to other countries.

World phenomenon

The HPPL density of France exhibits low values in minority languages and dialects areas: Alsace under German influence, Dunkirk region of Flemish heritage, Brittany, the Basque country and Corsica, and a bit of Romance "Langues d'Oc" neighbouring Spanish Catalonia (Figure 3). Soultzbach-les-Bains is a nice example of a hybrid toponym, of mixed French-German linguistic origins.

The phenomenon repeats itself in Austria, where most of the boundary regions have low densities, that is those in contact with Slave, Hungarian and Italian cultures. Paznaun is another nice Romance derivative and also an excellent ski area in Tirol.

The multi-ethnic heritage of Transylvania and Banat - Romanian, German, Hungarian, Székely, Serbian... - translates into a *Romania*n HPPL landscape still largely following the borders between these former Austro-Hungarian territories and Wallachia and Moldavia.

In Turkey the Sanjak of Alexandrette, from 1921 to 1939 an autonomous territory and later republic, and the Eastern Kurdish regions, stand out as low-percentage HPPL areas.

The case of Great Britain is interesting insofar the low densities in the Celtic ares (Scotland, Northern Ireland, Wales and Cornwall) are not so much a consequence of English dominance, than one due to the structure of the languages and the way in which toponyms are constructed in those cultures. Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch is visibly too long for the same combination of letters to occur somewhere else on this planet (it means, in Welsh, "St Mary's Church in a hollow of white hazel near the swirling whirlpool of the church of St Tysilio with a red cave").

Finally, in Japan, like in the rest of the world, the lowest HPPL densities are still those of hybridised cultures: the non-Japanese Ainu realm of the Northern Hokkaido island.

Figure 3 - Settlements homonymy in various countries.

Politics - It is noteworthy to consider where the low densities appear. Mostly they are at the periphery of the countries examined, a direct result of the radial spatial decay in dominance around power centers, as well as the progressive fashion in which dominance was established in time. Japan and Great Britain are outstanding examples in this matter (Hokkaido was integrated into Imperial Japan as late as the 19th century).

The "dot power", where dominance is concentrated around one region, is however not the only possible type, as shown by Iran. Here, the "hop power" is at work: there are multiple centers distributed around the country from where power is exerted, a likely result of power fragmentation in former times, frequent change of capital cities and the importance of urban settlements. Thus, HPPL density maps are also mirrors revealing the functioning modes of power structures in various countries.

Open questions - The first set of questions to ask when interpreting the these maps is methodological. Are there any "natural reasons" to some characteristics (like the conspicious knolling)? How is the computational method affect the outcome? Then there are questions about the HPPL phenomenon itself. How are local strong peaks and drops in density to be explained? Some regional trends are at first glance startling: why are the vast fields of Picardy and the gullies of Cappadocia low in HPPL, while Savoy and Bursa so high?

Visualization can provide some answers. For example, it is not possible to know from the present map if an extrema is due to a local agglomeration of the same placename near one another, or, due to many different closely located placenames with thier respective homonyms located far away. Depending on which option applies one concludes that in the given country local fashions are stronger than national ones, or vice versa. Or that a given population is prone to much HPPL generating or, on the contrary, avoids it.

Distance

The distances separating all pairs of HPPL is shown in Figure 4. The histograms differ from what would be expected from a homogenous distribution (Figure 5).

Country size and linguistic homogeneity are two factors immediately identifiable: Switzerland has much more local homonyms because it is much smaller than Iran and the four linguistic regions (French, German, Italian, Romansh) are more homogenous than the liguistic geography of Iran. Yet there is a supplementary factor possibly shaping the diagrams. In some countries old placenames might have survived much longer than in others, so that the linguistic chronological sediments decrease the quantity of HPPL. Such toponyms pockmark the web of HPPL and thus boost the number of long-distance HPPL.

Presently this study looked at the spatial toponymic diversity, but regions with low spatial diversity could turn out to be highly diverse in the temporal dimension, if one considers the rate of toponymic change.

Figure 4 - Distance between pairs of homonyms.

Figure 5 - Distances between pairs of locations in a ideal case.

Name length

Not surprisingly, HPPL with many characters per word are less frequent than shorter placenames (Figure 6). Most frequent homonyms tend to be two to four syllabes long. More interesting is the maximum length of placenames, which varies considerably across countries, depending on the specific roles played by toponyms. In Iran settelments often are transformed into multiple embodiments in landscape of otherwordly powers (five Shahrak-e Vali-ye 'Asr: Mahdiville), revolutionnary powers (eight Shahrak-e Emam Khomeyni: Imam Khomeyni City) and all sorts of other individuals (380+ shorter name Hoseynabad, two longer Kalateh-ye Mohammad Hoseyn Khan: Markettown of Mohammad Hoseyn Khan).

It would be tempting to call the Japanese HPPL haikus (maximum length 12 charcters) and the toponyms of Great Britain sagas (58 letters). Yet deeper issues might be at work, like the navigational role played by toponyms or a holistic concept about what a place is, as part of a complex environment as opposed to a disconnected entity with arbitrary name.

Figure 6 - Word length of HPPL placenames.

Ranking

The percentage of HPPL at country level exhibits a wide variability bracket, from 15% in Great Britain to 53% in Turkey. To the linguistic factor mentioned before, one can add conceptual differences in how people think a settlement should be named.

The most frequent Iranian PPL are constructed around personal names, many with a dual secular and religious symbolism; Austrian PPL around natural features such as wood, mountain, valley; and British PPL around spatial referents like middle, north, up. The religious factor and the weight of social leaders seems paramount for Iranian toponymy. A consistently far more rugged terrain than Iran, covered by dense woods, has made peasants life in Austria a constant struggle with nature. Naming a place Mountain-village or Valley-village has a considerable impact on the resources involved to reach it, for which reason it is far more informative than naming it after the mayor. Saints are sometimes stronger than mayors and mountains, and - like in Iran - religious toponymy is present in Austria, side-by-side the one inspired by the natural environment.

Country	HPPL
Turkey	53 %
Romania	37 %
Japan	34 %
Iran	32 %
Austria	30 %
France	19 %
Great Britain	15 %
Switzerland	11 %

Figure 7 - Percentage of HPPL in regard to all PPL in various coutries. (Follow the links for complete lists of homonyms.)

Issues

A number of data-related issues have been identified to influence the interpretations. At this stage of the study they were only partially addressed.

Data quality

• data coverage is not uniform across countries and within country regions
• coordinates precision varies (part of Korea coordinates are rounded to xx.xx333 or xx.xx666)
• mistypes ('oseynabad < Hoseyabad)
• transliteration of non-Roman-based scripts is inconsistent ('Aliabad and ‘Aliabad differ in straight and curly quotation mark used to render the 'ain Arabic letter)

Data extent

• spelling variation
• orthography variation
• definition of PPL in Geonames

Data segmentation

• spatial segmentation according to present day boundaries doesn't fit with various applications, like analyzing cross-country migration through toponymy or for historical studies
• semantic families can provide different views on homonymy, for example by classifying all Iranian HPPL derived from personal names and those derived from natural features in two super-classes

Conclusion

This work studied the diversity of names of populated places through their written forms. It has shown that numerous factors, of various natures, influence the phenomenon. There is high cultural variability, as well as some universals.

A fundamental finding is that the homonymity of populated places provides a indicator of cultural diversity, the number of homonyms being inversely proportional to cultural diversity.

Software

Desktop - Semantic processing of placenames, statistics and some geographical analysis and mapping is done in Kanga*, a custom software written in Matlab.

* Kanga is a reference to a common confusion made by English-speakers between the country toponyms Austria and Australia. This lead to a profusion of gift-shops in Vienna selling T-shirts with an Australian "kangaroo crossing" road sign and "No Kangaroos in Austria" printed below. The geographical homonyms analysis software was developped in Vienna.

Figure 8 - Screenshot of the Kanaga software for geographical homonymy analysis.

Online - Figure 9 is an interactive map of HPPL in Iran classified by administrative units. Such a map is very simple to generate. Append the URL of the dataset (http://www.waqwaq.info/atanasiu/studies/homonyms/ir.jsonp) to that of the webservice (http://www.waqwaq.info/ws/main/main.html) into a single GIS URL (http://www.waqwaq.info/ws/main/main.html?url=http://www.waqwaq.info/atanasiu/studies/homonyms/ir.jsonp). Read more at Waqwaq GIS.

Figure 9 - Interactive map of HPPL per administrative unit in Iran, generated with Waqwaq GIS.

Note the difference with the map in Figure 2. The likely causes are misattributed admin ids and place coordinates in the original dataset (admin centroids were computed as a mean of all coordinates belonging to the same admin unit).

What next

The author of this research is seeking funding to pursue the research on placenames homonymity. Please contact Vlad Atanasiu.

Acknowledgements

This preliminary study was made possible through a contract in 2009 with the Commission for Scientific Visualization and the Institute of Lexicography of Austrian Dialects and Names, at the Austrian Academy of Sciences, Vienna.

References

• Douglas R. Caldwell and James A. Shine, An Analysis of Toponymic Homonyms in Gazetteers: Country-Level Duplicate Names in the National Geospatial-Intelligence Agency’s Geographic Names Data Base, Coordinates A: 6.
• Richard R. Randall, Place Names: How They Define the World - And More (Boston: The Scarecrow Press, 2001).
• Naftali Kadmon, Toponymy: The Lore, Laws and Language of Geographical Names (New York: Vantage Press, 2000).

Artwork

This scientific study has inspired the author to create a couple of cartographic art objects. The photo below shows their practical side: the density maps can also be used as beer plates, while the 3D geohomonyms landscape of Austria - here in the process of being engraved by laser - looks nice as a paper weight on the desk of your boss.

Figure 10 - Geohomonyms maps as beer plates.

Figure 11 - Engraving by laser of the 3D homonymy landscape of Austria.

-- VladAtanasiu - 27 Mar 2010