Thursday, August 25, 2016

Name this creature: How to scientifically name a species

http://www.abc.net.au/news/2016-08-05/how-to-scientifically-name-species/7681634

Photo: Scientific names prevent confusion among taxonomists (Ann Jones)

Putting a name on a new species is one of the most exciting parts of taxonomic research, but also one of the most important because — unlike common names — a scientific name is unique to one species. But how are they decided?

What are the rules?

Let's say that an Australian bird scientist — an ornithologist — is talking with a researcher in Europe about magpies.
But which magpies are they talking about: the Australian magpie (Cracticus tibicen) or the Eurasian magpie (Pica pica)?
By using the scientific name, two scientists on opposite sides of the world can know for a fact that they're talking about the same species.
But naming a new species comes with some big challenges. There are about 8.6 million species in the world and we've only named 1.2 million of them.
That's a lot of creatures and a lot of names to keep track of, so we need some rules of thumb.

Photo: Other than convention, there's no requirement for scientific names to be in Latin (Wikimedia Commons)

The International Code of Zoological Nomenclature dictates how new animal species are named, but their guide is an 18-chapter doorstop.
Here's a much shorter, step-by-step guide to naming this creature.
There are some rules you must follow when naming a new species:
  • The name must be unique. The combination of genus name and species name cannot have been used for any other animal. You're naming a new species of the genus Moridilla, so once you've come up with an idea for a species name try googling it: Moridilla X. If it's entirely new then you're good to go.
  • The name can't be rude. The ICZN states that no name should give offence on any grounds. Historically, some scientists waged war on each other this way, but that's no excuse so just play nice.
  • You can't name the species after yourself. Nobody — and I mean nobody — names a species after themselves. It's tacky!

Where do I find my inspiration?

Great species names tell you something about the creature itself, so get creative.
You can draw on any feature of the species, such as its appearance, behaviour, habitat or geographic location.
Taxonomists draw on all kinds of inspiration when naming new species, for example:
  • Five species of fungus beetle — a beastie that is small and round — were named Gelae baen (sounding like "jelly bean"), Gelae balae ("jelly belly"), Gelae donut ("jelly doughnut"), Gelae fish ("jelly fish"), and Gelae rol ("jelly roll").
  • The spider Aphonopelma johnnycashi was named after Johnny Cash because the the 'species can be found near the area of Folsom Prison in California, and like Cash's distinctive style of dress ... mature males of this species are generally black in colour.'
  • A new genus of sea snail was called Ittibittium, because it is smaller than sea snails from the genus T. Bittium.
  • A giant fossil turtle was named Ninjemys oweni, Owen's Ninja Turtle, with the authors explaining that the name was from "ninja, in allusion to that totally rad, fearsome foursome epitomising shelled success" and "emys" from the Latin for turtle.

Does the name have to be in Latin?

Scientific names are written in Latin because historically this was the language of science.
However, you don't have to worry about this part. We have experts on hand to translate your name into its correct Latin form.
Naming a species is both an incredible honour and an enormous responsibility.
A species name has the potential to last as long as the species itself, and perhaps even longer. With posterity in mind, go name a nudibranch!

Dr Amber Beavis is a senior research officer at The Office of the Chief Scientist, and formerly a taxonomist at the Western Australian Museum specialising in arachnology.

Wednesday, August 24, 2016

Call for Papers: Special Journal Issue of NAMES devoted to Indigenous Names and Toponyms

http://www.americannamesociety.org/call-for-papers-special-journal-issue-of-names-devoted-to-indigenous-names-and-toponyms/



The American Name Society (ANS) is inviting abstracts for scientific papers providing an analysis and discussion of indigenous names and toponyms found in former European colonies in the Americas, Africa, Asia, the Pacific and the Antipodes. Any area of the use of indigenous names may be the subject of analysis. Suggested issues for discussion include, but are by no means limited to the following: the transcription (spelling) of indigenous names and/or determining their meanings, indigenous naming practices, indigenous names as identity markers, the reinstallation of indigenous toponyms, the reclamation of indigenous language and culture through their names, and the appropriation of indigenous names, etc.



Proposal Submission Process:
  1. Proposals should include a précis of no more than 500 words and a 50-word biographical sketch of the author including the author’s name, affiliation, onomastic interests, and email.
  2. All submissions must follow the Journal’s official stylistic and grammatical regulations.
  3. Proposals should be sent via email attachment in a .doc or .docx format to Dr. I. M. Nick at <mavi.yaz@web.de>, with “Indigenous Names and Toponyms” in the subject line.
  4. Proposals must be received by 1 February, 2017.
  5. All submissions will be subjected to a blind peer review process.
  6. Notification of acceptance will be announced on or about 31 March, 2017
  7. Final submissions due for publication 31 July, 2017.
  8. For questions, please email Dr. I. M. Nick at <mavi.yaz@web.de>.



Monday, August 22, 2016

Importance of multiple languages in translation and challenges in onomas...

Zornheimerin erforscht Familiennamen: Mehr als nur Müller und Maier

http://www.swr.de/landesschau-rp/couchgespraech/zornheimerin-erforscht-familiennamen-mehr-als-nur-mueller-und-maier/-/id=5661010/did=17740534/nid=5661010/c984qx/

Manche sind einzigartig, andere wiederum fast schon inflationär, ganz ohne geht es aber in keinem Fall. Einen Nachnamen hat jeder, auch wenn er diesen vielleicht mal wechselt. Rita Heuser aus Zornheim erforscht die Nachnamen der Deutschen in einem Großprojekt.



5.000 Namen haben Rita Heuser und ihre Kollegen schon gesammelt und in ihr "Digitales Familienwörterbuch Deutschlands" aufgenommen. Mehr als 200.000 sollen es einmal werden. Es ist ein Mammutprojekt, an dem die Zornheimerin mit ihren insgesamt 14 Kollegen arbeitet.
"Im Südwesten", erklärt die Namensforscherin, "sind körperliche Charakteristika, die sich im Nachnamen wiederfinden, sehr häufig." So finde man hier viele Familien mit dem Namen Jung, Dürr, Dick und vor allem Klein. Grund zur Beunruhigung sei das aber nicht, denn: Klein könne auch so viel wie "jung" bedeutet und ziele nicht zwangsläufig auf die Körpergröße ab.


Zählen, zählen, zählen

Was Rita Heuser in ihrem Arbeitszimmer fast täglich macht, ist echte Detektivarbeit. "Man muss viel suchen, historische Zusammenhänge - wie beispielsweise Migration - erörtern und sich mit historischen Berufen auskennen", erklärt sie. Ihre wichtigsten Arbeitsmittel sind historische Wörterbücher, denn bei jedem Namen müsse man zuerst schauen, welches mittelhochdeutsche Wort der Ursprung sein könnte.

Als Grundlage für die Datenbank dienen den Wissenschaftlern dabei nicht etwa Archive oder Bibliotheken, sondern das Telefonbuch von 2005. Ab einer Häufigkeit von zehn schafft es ein Nachname in den Online-Datensatz. Für Rita Heuser heißt das: zählen, zählen und nochmal zählen.
Besonders spannend findet die Rheinhessin natürlich die Genese der Familiennamen in dieser Region. So hat sie herausgefunden, dass der Name Dejung hier sehr häufig auftaucht. Nach dem 30-jährigen Krieg seien gezielt Einwanderer angeworben worden, um die entvölkerten Dörfer wieder zu beleben - darunter viele religiös verfolgte Protestanten aus den Niederlanden. Aus den De Jongs wurden dann die deutschen Dejungs. Auch slawische, italienische und griechische Namen fänden sich aus diesem Grund sehr oft in Rheinhessen.


Anders allerdings der Name des wohl bekanntesten Rheinhessen, Carl Zuckmayer. "Der geht auf den Namen Zugmaier aus Bayerisch-Schwaben zurück, erklärt Heuser. Generell sei der Name Meier mit all seinen Variationen in Rheinland-Pfalz eher selten, weswegen man auch vom so genannten "Meierloch" sprechen würde. Im Norden und Süden Deutschlands komme er deutlich häufiger vor.

Wann ist ein Name deutsch?

Unter den bisher 5.000 ausgewerteten Familiennamen sind nicht nur "typisch deutsche". Auch einige türkische Namen wie zum Beispiel Kaya sind in der Datenbank zu finden. "Es ist eben nicht das 'deutsche Familiennamenwörterbuch', sondern das 'Familiennamenwörterbuch Deutschlands'", betont Rita Heuser. Migration sei schon immer ein Teil der Gesellschaft gewesen und spiegele sich in der Sprache und den Namen wider. Und wer könne heute noch genau sagen, ob Schimanski ein deutscher oder ein polnischer Name sei?

Im "Digitalen Familienwörterbuch" kann man zu jedem Eintrag die Herkunft des Namens nachlesen, findet zudem die Häufigkeit und sogar eine Deutschlandkarte, die die Verteilung des Namens zeigt. Noch etwa 20 Jahre wird es dauern, bis das Forschungsgroßprojekt abgeschlossen sein wird. Rita Heuser wird dann schon in Rente sein. Den Spaß an der Onomastik, wie die Namensforschung unter Wissenschaftlern heißt, wird sie aber niemals verlieren, glaubt sie.

Ihr eigener Name verweise im Übrigen auf jemanden, der nur ein kleines Häuschen hat, jemanden, der aus dem Ort Hausen stammt oder den Beruf des Hausverwalters, so die Forscherin. Wie sehr viele der deutschen Nachnamen sei er mehrdeutig - und auf jeden Fall kein typisch rheinhessischer Name.

Das Forschungsprojekt "Digitales Familiennamenwörterbuch Deutschlands (DFD)" wurde an der Mainzer Akademie der Wissenschaften und der Literatur in Kooperation mit der Technischen Universität Darmstadt und der Johannes Gutenberg-Universität Mainz begonnen. Erstmals sollen die derzeit in Deutschland vorkommenden Familiennamen – unter Einbeziehung der fremdsprachigen Namen – lexikographisch erfasst, kartiert und etymologisiert werden.

Deriving age and gender from forenames for consumer analytics

http://www.sciencedirect.com/science/article/pii/S0969698915300540


Deriving age and gender from forenames for consumer analytics

by        Guy Lansley                        and                  Paul Longley


Highlights


Infers consumer age and gender from forenames.
Infers characteristics of individual social media users.
Compares modelled age and gender profiles at different locations.
Links forenames to different geodemographic market segments.



Abstract

This paper explores the age and gender distributions of the bearers of British forenames and identifies key trends in British naming conventions. Age and gender characteristics are known to greatly influence consumption behaviour, and so extracting and using names to indicate these characteristics from consumer datasets is of clear value to the retail and marketing industries. Data representing over 17 million individuals sourced from birth certificates and market data have been modelled to estimate the total age and gender distributions of 32,000 unique forenames in Britain. When aggregated into five year age bands for each gender, the data reveal distinctive age profiles for different names, which are largely a product of the rise and decline in popularity of different baby names over the past 90 years. The names database produced can be used to infer the expected age and gender structures of many consumer datasets, as well as to anticipate key characteristics of consumers at the level of the individual.

Keywords

  • Age;
  • Gender;
  • Geodemographics;
  • Big Data;
  • Social media




1. Introduction and overview

The advent of new sources of consumer data, such as those arising from the use of social media, online shopping and customer loyalty databases, present new opportunities to measure and model the activity patterns of individuals. Such data may be related to detailed functional taxonomies of the locations that consumers visit or the products that they buy, bringing insight into the nature and likely motivations for observed activity patterns. But there has been no commensurate improvement in the detail with which we are able to characterise the individuals themselves, and thus ascertain how representative they are of consumer segments or indeed the population at large.

In this context, our own attempts to understand consumer behaviour have become focused upon the task of front-loading the inferences that may be drawn from consumer names and social media user identifiers, in order to relate new Big Data sources to the wider populations from which they are drawn. In this paper we describe some of the ways in which individual given (fore-) names can be analysed in order to ascribe age and gender characteristics, as part of this task. The work builds upon a commercial classification (Monica: CACI, London) and is part of a wider research programme which explores given and family name pairings in order to infer individual characteristics in consumer research (Mateos et al., 2011 and Longley and Adnan, 2016).

A person's given name can be used to infer a number of key individual characteristics, as a result of the ways in which names are typically distributed according to age, gender and ethnicity. Most forenames are gender specific and many can be traced to ethnic groups through common heritage (Mateos et al., 2011). There has been much research into patterns in names and how they link to cultural heritage and wider society. However, only limited attempts have been made to consider the age and gender distributions of forenames in Britain (Finch, 2008). Names vary by gender and age across time largely because of shifts in popularity of different baby names over time and the influence of migration. This paper summarises work undertaken to model the age and gender structures of different name holders in the Great Britain. From achieving a greater understanding about the demographic characteristics of different name bearers, more information can be inferred from consumer data which include names but no further details about the individuals.

Retailers have benefitted greatly from geodemographic datasets, made available from the government or other businesses, as a means of segmenting and understanding consumers. Such data allow retailers to plan their stock and marketing accordingly to the local population characteristics (Mitchell and McGoldrick, 1994 and O’Malley et al., 1997). However, this conventional approach is based upon the assumed correspondence between night-time residence and consumer behaviour (Harris et al., 2005). The data is also typically aggregated and/or modelled, therefore the traits for each spatial unit may not be entirely representative of every resident (Openshaw, 1984). Consequently, there have also been developments in micro-economic modelling which seek to place the focus upon the individual to arguably provide an inherently superior approach to understanding consumer behaviour as it circumvents any issues of ecological fallacy (Hensher and Johnson, 1981; see also Longley and Adnan (2016)). However, in order to apply such techniques individual level data is required, and often consumer data is absent of demographic information. For instance, the records of an online account associated with large retailer will typically only include a name, address and purchase history. Therefore, inferring demographic traits from names data could allow analysts to harness more consumer insight from many data sources.

1.1. Demographics and consumption

In this paper we take it as axiomatic that gender and age of consumers both heavily influence behaviour and consumption practices. Consequently, much research has been devoted to understanding the influence of demographics on consumption, such as the incorporation of such characteristics in product and brand choice models (Kalyanam and Putler, 1997), and identifying target demographics for new products. Geodemographic segmentations are widely used to segment the population in to distinctive consumer groups from multivariate data (Harris et al., 2005), and age and gender are included in them as key correlates of consumer behaviour.

The gender divide in consumption practices is most obvious amongst certain product types such as clothes and cosmetics, where retailers produce and sell entirely different lines of stock tailored for each sex (Scanlon, 2000). Consumption has become an important means through which individuals construct their gendered identities (Baudrillard, 1998). There are even gendered variations in consumption of products which are not produced exclusively for one gender. For instance, women have been found to be more health conscious as exerted by their food consumption practices and therefore they are generally likely to perceive certain foods differently to men (Wardle et al., 2004). There is even a gender divide in perceptions of shopping behaviour, women being traditionally more likely to perceive shopping as a leisure experience and therefore spend more time visiting high streets (Campbell, 1997 and Lunt and Livingstone, 1992). By contrast, males have been traditionally the most dominant patrons of online shopping websites (Dittmar et al., 2004 and Rodgers and Harris, 2003), and are more comfortable using multiple channels when making a purchase (Blázquez, 2014).

Age is also an influential characteristic of consumer behaviour, in terms of product and brand preference, and also in terms of how individuals shop. For instance, younger consumers are usually more likely to patronise online shopping channels, whilst older individuals are typically less engaged with the Internet and other modern shopping channels such as mobile commerce (Sorce et al., 2005). Consumer behaviour is greatly influenced by the family life cycle and the ways in which disposable incomes are channelled through consumption (Reynolds and Wells, 1977). Amongst the adult population, different cohorts are known to have different consumption practices which link to their life stage, their physical and health characteristics, and their cultural characteristics. Shared experiences during adolescence and beyond that are traceable to societal, cultural and environmental traits can encourage individuals to develop values that they will retain over time and can give rise to “cohort effects” (Harmon et al., 1999). Consequently, those from an age cohort may share distinctive values and this is likely to influence their consumer behaviour (Pentecost and Lynda, 2010).

Understanding the demographic characteristics of consumers is therefore very important to developing a sustainable retail strategy. Consequently, inferring demographic traits of consumers has been a vital area of marketing and consumer research (McDonald, 1995; Carpenter and Moore, 2006). It is also important to adapt to local and national demographic changes. For instance, there is an increasing imperative to understand the consumption practices of elderly consumers in many western countries given their aging populations (Kohijoki and Marjanen, 2012). Even amongst stable populations, previous research has established that age cohorts have unique consumption traits relative to previous generations (Bakewell and Mitchell, 2003). Although every individual may have distinctive tastes, general consumption practices nevertheless vary by age and gender. Therefore, the possibility of estimating the general demographic structure alongside a consumer's distinctive personal characteristics from individual customer records can be a fruitful means of obtaining key information about clients and customers.

2. Names and demographics

There is a wide range of Big Data sources on the population which includes name identifiers, but have little or no additional demographic information. These include electoral registers, customer records and social media data. There have subsequently been attempts to harness information from names by examining how names are distributed through contemporary society. Perhaps the most sophisticated developments have been in the production of cultural, ethnic and language group classifications from forename–surname pairs (Mateos, 2007 and Mateos et al., 2011). Surnames, in particular, can identify bonds between family members, and therefore can be aggregated to represent distinctive cultural groups.

Historically, there has been a range of processes that influence popular naming conventions in the Great Britain (Smith-Bannister, 1997). Over the last century, popular naming practices have become far more erratic (Galbi, 2002), reflecting secularisation of society, migration trends and social mobility. The UK Office for National Statistics (ONS) has nevertheless identified clear trends in baby naming over the years (Matheson and Summerfield, 2000), and names therefore offer a viable means of estimating age structures from larger populations (Scharf, 2005).

We associate names with their bearers, yet forenames more directly manifest the predilections, priorities and preferences of either or both parents (Gureckis and Goldstone, 2009). The choice of baby name is likely to vary systematically between parents from different socio-economic backgrounds and cultural groups. The favourability of names is also influenced by popular trends, giving rise to temporal autocorrelation in name frequencies (Xi et al., 2014). Research has identified that parents in the USA perceive baby names which are growing in popularity to be more desirable than those whose popularity is waning (Berger and Mens, 2009 and Gureckis and Goldstone, 2009). In additional to shifts in societal values, more subtle environmental and internal influences also drive the popularity of names for particular groups (Lieberson, 2000). Whilst the choice of baby name is influenced by various sociological factors, some names are handed down by family members or have remained popular because of links to cultural heritage: such names are much less likely to vary much between age groups (Finch, 2008).

Forenames are subsequently an important part of an individual's identity and can even act as a positive source for cultural capital (Lord, 2002). Observers may associate names with stereotypes, such as a child's likely educational attainment (Harari and McDavid, 1973 and Erwin and Calev, 1984). This may be grounded in truth because of the different forename preferences of parents who themselves have different experience of, and attitudes to, educational capital formation.

A major source of names data for Great Britain is the 44.8 million individual records that make up the 2011 CACI (London, UK) Consumer Register. This is built from a number of sources, including the public version of the national Register of Electors. Our analysis of this data source investigates how forenames are distinctively distributed between the super-groups that make up the 2011 UK Output Area Classification (OAC: Gale, 2014). Similar names associations have been developed in the past for commercial segmentations. The CACI Consumer Register used in this analysis was compiled for the same year as the most recent UK Census, which underpins the OAC classification. The database includes the forename and surname of registered individuals, and also recorded their home postcode so their residential locations could be identified. Forenames from the Consumer Register were joined to residential Census statistics at small area level. The elemental Census Output Areas that make up the OAC classification had an average population of just 309 for England and Wales in the 2011 Census, and are the smallest unit census estimates that have been publicly published under (ONS, 2015). The OAC segments output areas into 8 super-groups by a range of population statistics pertaining to demographics, cultural identity, socio-economics, employment and household characteristics. Table 1 shows some of the results obtained by aggregating the forenames from the Consumer Register to Output Area scale in order to link with the OAC. The table displays the top five most overrepresented names from each OAC super-group relative to the national average.


Table 1. 
       The top five over-represented forenames for each 2011 OAC super-group, based upon.names with a frequency of 10,000 or more in the 2011 Consumer Register (Source: CACI, London).

Rural residents

Cosmopolitans

Ethnicity central

Multicultural metropolitans

PENELOPETOMMOHAMEDMOHAMMED
HUGHNICKAHMEDMUHAMMAD
ALASTAIRHARRIETALIMOHAMMAD
ROSEMARYMAXJOSEABDUL
PHILIPPAALEXABDULAHMED




UrbanitesSuburbanitesConstrained City DwellersHard-pressed living
TOBYHILARYLILLIANKAYLEIGH
PHILIPPAGEOFFREYMAYLEANNE
JEREMYKATHRYNETHELLYNDSEY
KATHERINEJILLKAYLEIGHSTACEY
DUNCANGILLIANELSIEKYLE



The two most ethnically diverse super-groups (namely Ethnicity Central and Multicultural Metropolitans) have a much higher representation of foreign origin first names. The Hard-Pressed Living group contain modern variants of some traditional names. Names popular in Victorian times are more common in the Suburbanites super-group, suggesting there is a link between socio-economic status and naming practices. It is also notable that the two super-groups with high proportions of elderly residents, Rural Residents and Constrained City Dwellers, are over represented by forenames that were more fashionable in bygone times. The Cosmopolitans super-group, which has a higher proportion of young adults, is characterised by shortened names.

Taken together, this analysis suggests two key findings: the popularity of different forenames varies between neighbourhoods; and many of these differences can be accounted for by local neighbourhood characteristics, notably cultural heritage and age. Whilst research into heritage and naming connotations is well established (e.g. Mateos et al., 2011), the influence of temporal popularity of forenames and the subsequent age structure of forename bearers in England and Wales is relatively under-researched. In the following section, we therefore generalise the age and gender structure of the bearers of given names in the UK, using data from birth certificate records to supplement the CACI Consumer register.

3. Enhancing the consumer register

Data were linked from two main sources, with the aim of acquiring a representative register of the UK population at large: the CACI Consumer Register and birth certificate data from the UK Office for National Statistics. The age-sex structure as recorded in the 2011 Census was also used in order to standardise the distribution recorded from the combined name data.

CACI Ltd. (London, UK) provided the original derived data product called Monica, which contains details of age and gender distributions associated with different given names amongst adults. The data were extracted from credit card applications, and pertained to a total of 7,085,617 individuals, who were bearers of over 21,000 individual names (multi-gendered names are counted twice). For reasons of disclosure control only names with a sample size of at least 10 were included in the dataset. However, the dataset had two key limitations because of the nature of its remit, viz. credit card approvals. First, no applicants were aged under 18. Second, certain age groups were underrepresented relative to their known frequencies in the UK, particularly those aged 18 and 19. There is also a possible limitation that some names which are more prevalent amongst more deprived households may be slightly underrepresented because of socio-economic inequalities within the credit card market.

To establish a more representative and inclusive age structure, birth certificate data were obtained from the Office for National Statistics (ONS). Since 1996 the ONS has released data detailing the frequency of births registered in each given name. As of 2012 the birth certificate data accounted for a cumulative population group of 10,412,724 individuals, excluding individuals with names with a frequency of 3 or less in any given year in England and Wales. While these data cannot account for children and teenagers born outside of England and Wales, it nevertheless represents a substantial proportion of the overall UK population for this age range. Combined, the two datasets represent over 32,000 unique forenames, although only 1441 of these have a collective sample size greater than 1000 individuals.

The forenames were not recoded into their most common variants: for instance Matt and Matthew were kept as separate and distinguishable names. Parental choice of shortened baby names peaked in popularity about 25 years ago. Indeed the Consumer Register data from 2011 revealed that Output Areas with higher proportions of young adults had the highest concentrations of shortened names such as Tom, Alex, Nick, Joe and Sam.

Our final names database has been aggregated into 5 year age bands, consistent with the CACI Monica classification. The datasets were reweighted to account for the uneven sample sizes based on their penetration of the UK population. The primary aim was to develop a model of age and gender distributions which was reflective for the entire UK population across each age group. Age and gender distributions were calculated and were found to be very similar to the overall age distribution of the UK, with the exception of ages 18–19 (when looking at the data within the 15–19 age group) and a very slight underrepresentation of elderly age groups. It was therefore decided to reweight the entire dataset by the UK population using official statistics from the 2011 Census using the same age bands.

4. Key trends in naming practices

The combined names database reveals that there are more unique female names than individual male names in the UK. The top 10 most common male names account for 25.13% of the male population, whilst the top 10 female names account for 13.07% of the female population. The frequency distributions of names for both genders are very positively skewed. Lorenz Curves for the frequencies of male and female names revealed that a minority of names represent a majority of the population (Fig. 1). Although female names are slightly more evenly distributed.

Fig. 1. 
Lorenz curves demonstrating the frequency distributions of male and female names. Only names with a projected frequency of 1000 were included in this visualisation.



The most common forenames for both genders are shown in Table 2. The data reveal that John is the most common male name, and Margaret is the most common female name. Other research from the ONS confirms that these names were the most common baby names in the earlier years of the 20th Century (Matheson and Babb, 2002), and it is therefore likely that the bearers of these names from the 2011 Consumer Register derive from older age cohorts.


Table 2. The most common forenames, as identified in the combined database (England and Wales weighting).
Female
Male
RankNameEstimated numberNameEstimated number
1MARGARET555,000JOHN1,003,500
2SUSAN544,600DAVID969,000
3SARAH425,600JAMES614,600
4ELIZABETH363,200MICHAEL612,800
5PATRICIA350,200PAUL546,700
6MARY332,500ROBERT482,800
7CHRISTINE321,400PETER480,600
8JULIE314,600ANDREW435,700
9KAREN313,700WILLIAM415,300
10LINDA298,200MARK370,700



The data also reveal that the choice of baby name has diversified over time. For instance the most popular baby names for the youngest age band, Oliver and Olivia, represent 2.2% and 1.6% of all persons from the 0 to 5 age band for each gender respectively. By contrast, the most popular names for the eldest age band, Margaret and John, represent 5.6% and 8.3% of persons for each gender.

The name data have confirmed that popular naming trends have fluctuated over time, the most popular names for each age band are presented in Table 3. Consequently, certain names can be generally associated with particular age groups. This can be easily demonstrated by taking the average age for each given name. Using this technique, the ten oldest and youngest names for both genders are presented in Table 4.


Table 3. The most common name for each age band.
Age groupFemaleMale
0–4OLIVIAOLIVER
5–9EMILYJACK
10–14CHLOEJACK
15–19SARAHJAMES
20–24SARAHJAMES
25–29SARAHDAVID
30–34SARAHDAVID
35–39SARAHPAUL
40–44KARENPAUL
45–49JULIEDAVID
50–54SUSANDAVID
55–59SUSANDAVID
60–64SUSANJOHN
65–69MARGARETJOHN
70–74MARGARETJOHN
75–79MARGARETJOHN
80–84MARGARETJOHN
85+MARGARETJOHN


Table 4. The oldest and youngest names in the UK (excluding names with a projected population under 10,000).
Female
Male
RankOldestYoungestOldestYoungest
1DORISSIENNACYRILRILEY
2GLADYSAVAHERBERTJAYDEN
3ETHELEVIEREGINALDLOGAN
4EDNASUMMERERNESTFINLEY
5WINIFREDLACEYHAROLDNOAH
6HILDASCARLETTWALTERALFIE
7BETTYGRACIERONALDLUCAS
8MURIELMADDISONLEONARDHARLEY
9PEGGYFREYANORMANFREDDIE
10VERAMADISONDONALDKIAN


However, the average age alone is not an appropriate indicator for understanding the typical age structure for every name. Many name distributions manifest a peak in popularity which gradually diminished over time. For plenty of these names, the rise to popularity was gradual. Yet overall they do not quite share normal distributions across ages due to the age structure of the population and all frequencies must be truncated at age zero. However, generally trends in names seem to fluctuate gradually over time due to shifts in baby name popularity and changes in migration, and most have a distinctive peak.

Some forenames have been more resilient to fluctuations in popularity over time, and these are therefore well represented across a wide range of age groups today. For instance the aggregated age structure of those with the names Patrick, James, Catherine, Ruth and Robert vary little from the national average. When comparing the relative proportion of the age distributions from the names to the national population, Patrick, George and Edward share the smallest standard deviations (0.01). In contrast Oscar, Archie, Jordan and Shannon share deviations above 0.11 – all four have very young skewed age distributions.

Certain names exert bimodal age distributions, perhaps reflecting an inter-generational popularity. For instance the name Guy is most common amongst men in their mid to late forties, and second most popular amongst those in their mid-twenties, a gap of just over 20 years. Madeline is common amongst older children and also those in their 60 s too. There has also been a resurgence in popularity of some names which were also popular in the 1930s as revealed by the data. Notable examples include Clara, Rose, Sidney and Henry.

It is not uncommon for influential celebrities' baby names to gain popularity: for example, over 96% of persons named Brooklyn in the UK are younger than Brooklyn Beckham. However, this is not always the case, for instance there are only 10 instances of the name Apple in our dataset, the moniker given to Chris Martin and Gwyneth Paltrow's first child which caught the attention of the media. Celebrities' own names may also influence baby naming: 63% of those named Rihanna were under five years of age as of 2012, and there are no records of this name in the original Monica file which only represents those aged 18 and older. 46% of those named Beyoncé are in the 5 to 9 age group. The name Rod is most common amongst those in their late thirties to their forties, perhaps this was driven by the influence of Rod Stewart's popularity, whose career peaked in the late 60 s and early 70 s.

4.1. Grouping names

To achieve an overview of how forenames can be generalised based on their standardised age distributions, the names have been clustered by running a k-means clustering algorithm for each gender. The k-means algorithm is an interactive process which clusters data by allocating and reallocating observations to their nearest cluster centroid in a multidimensional space as determined by the variables. The algorithm attempts to minimise the average distance of each observation to its nearest centroid with each iteration by updating (and therefore moving) the centroids based on their existing allocation of observations and then reallocating the data. This process continues until an optimum solution has been achieved and the centroids can no longer be moved (Harris et al., 2005). The number of clusters are specified by the researcher and their centroids are initially randomly located when the algorithm commences. Only names with a frequency of 1000 or more in the original data were included in this analysis and the variables inputted were the proportion of their populations within each 5 year age band. The classification produced five groups for both genders and these have been labelled; youngest, young, middle, old, oldest based on their average ages. Each group generally represents a 20 year generation of popularity. The results of the classification have been visualised as a series of frequency probability histograms. The size (in area) of each name within the histograms corresponds with its projected population size in Fig. 2.


Fig. 2. 
The age distributions of the five name clusters based on five year age bands for each gender. From top to bottom; Youngest, Young, Medium, Old, Oldest.




The classification groups are reasonably well balanced in terms of both the number of names and their estimated populations. However, the youngest name group contains the most unique monikers but is the second smallest group for both genders in terms of expected population of the bearers: this reflects the increased the diversity of names over time, with the oldest groups being the smallest in both the frequency of individual names and total population.

There are some notable distinctions between the male and female classifications in that the oldest group for the male population is smaller in size and has a much flatter distribution than its female equivalent. The youngest age group is well represented by contemporary popular baby names such as Jack, Harry, Emily and Jessica. Bearers of these names typically average 10.4 years of age for males, and 12.5 for females, and 75% of females and 84% of males are under the age of 15 on average for each name within the groups. There is also a higher proportion of names imported from abroad, which can be accounted for by the higher fertility rates amongst their parents' cohort (Zumpe et al., 2012). Mohammed is the 11th most popular male name in the Young group.

The young, medium and old groups all display relatively normal histograms suggesting that the popularity of such names have gradually risen and fallen again. On average 53% of the young group for both genders are aged between 10 and 30, the male equivalent being slightly younger on average. The most popular names in these groups are Emma and Laura for females and James, Thomas and Daniel for males. The medium groups are represented by large numbers of adults aged between the ages of 20 and 50, with peaks in the forties. Names from this generation are well represented by Sarah, Julie and Karen for females and Paul, Andrew and Mark for males.

The old group is most prevalent amongst persons aged between 50 and 70 years of age. Although these names have declined in popularity, this phenomenon has been quite gradual and subsequently about 4% of bearers of these names are aged 25–30. The male names assigned to this category have remained more prevalent in recent years than their female counterparts. This is probably because of the presence of inter-generational use of recurring family forenames, some of which are biblical in origin (such as John, David, Peter and Michael).

The oldest group are very advanced in years. Cyril, the oldest male name in our dataset, has an average age of 73.5 and a median age of 76.5, whilst the oldest female name, Doris, has an average age of 78.4. The ten names with the oldest average ages for each gender are presented in Table 5. Interestingly, the data also identify the recent resurgence in popularity of Victorian names (West, 2012), and there is a subtle rise in the frequency of these names amongst the youngest age cohort too.

Table 5. The names with the oldest mean ages in Great Britain (Excluding those with a projected population below 10,000).
Female
Male
RankNameEstimated ageNameEstimated age
1DORIS78.35CYRIL73.55
2GLADYS77.99HERBERT72.10
3ETHEL77.74REGINALD71.90
4EDNA76.92ERNEST70.84
5WINIFRED76.75HAROLD70.13
6HILDA76.64WALTER68.85
7BETTY76.46RONALD68.61
8MURIEL75.80LEONARD67.24
9PEGGY75.75NORMAN66.95
10VERA75.58DONALD66.02



5. Modelling demographic structures from names

As a test of concept, the names classification has been used to estimate the age structure of georeferenced Twitter users in London using a sample of 2.5 million Tweets from September 2012 to March 2013, from 129,400 individual users. Twitter data has been widely utilised by researchers and marketers to gain an understanding about people and their opinions (Williams et al., 2013). However, the demographic characteristics of Twitter users are poorly understood as no such information is recorded by Twitter. For this study, users' forenames were extracted from the user names using an approach outlined in Longley et al. (2015). To gain an understanding how the demographics of users may vary across space, the data were partitioned for four distinctive places: an entertainments venue (the O2 Arena), a football stadium (the Emirates Stadium), a business district (Canary Wharf) and a shopping centre (Westfield, Stratford). The modelled age pyramids are displayed in Fig. 3, where the grey lines represent the average distribution across all the Tweets in London from the original sample.



Fig. 3. 
Name inferred demographic structure of geotagged Twitter users at four sites in London. a) The O2 Arena, b) The Emirates stadium, c) Canary Wharf and d) Westfield Stratford.


Inference of age from the database identifies the football stadium and business district as places where there are greater proportions of male Tweeters. It also identifies younger age distributions at the shopping centre and entertainments venue. Both would appear to be logical conclusions based on the known activities of these places. Of course, it is very unlikely that persons under the age of 5 are Twitter users. The presence of very young names in the Twitter dataset, which is particularly evident at the football stadium, can be largely accounted for by the use of informal spellings of names. However, future analysis could refit the names dataset to accommodate a minimum age cut-off if the data we are trying to model are known to only represent the adult population. In summary, whilst the findings for the four sample locations are unsurprising, it indicates that the analysis can be conducted in other locations where the demographic composition are not known. It also indicates that gender and ages can be inferred from names from other datasets too.

6. Conclusions

This paper has demonstrated that it is feasible to identify the age-gender distributions of forenames and thence to ascribe demographic characteristics to data where such information is not otherwise available. This approach makes it possible to unshackle geodemographic analysis from an exclusive preoccupation with the geography of night-time residence, and harness value from individual level data (Longley and Adnan, 2016). The modelled name data provide a suitable means of estimating age and gender distributions from British forenames because of trends in the popularity of baby names. Such data could supplement the analysis of population records such as customer datasets, which inherently lack demographic characteristics. The inferential procedure is of course by no means perfect, not least because of the incompleteness of the adult population that are included in the market data and the presence of young people who were not born in England and Wales. In addition, birth certificate records could not be obtained for Scotland or Northern Ireland, and the database required further reweighting to account for uneven sample sizes between the two data sources. It is also important to consider the limitations of individual level data in the consumer context. A person may not always be shopping for his or herself exclusively. For example, a grocery shopping trip could be undertaken by an individual for their entire household.

However, such an approach is a viable means of assigning characteristics to individuals in customer databases which contain few if any demographic attributes but do include names. The database can be used to ascribe the probability of each individual falling into each age band and gender. Of course, each of the names is uniquely distributed across ages and genders and some may be more uniformly distributed than others, making them less effective discriminators. However, most names have been found to be broadly representative of particular age groups. Names are also particularly successful as a means of estimating gender as the vast majority of names are not unisex. Previous research has identified that consumer's product preferences vary considerably by age and gender. Therefore harnessing demographic attributes of consumers at the individual level is of great benefit to retailers, particularly those which are able to invest in micro-level targeted marketing strategies. In this context, the analysis of forenames can empower retailers to take advantage of better insight from their data to inform their future marketing and planning decisions.

Acknowledgements

This research was funded by Economic and Social Research Council grants ES/L011840/1(Consumer Data Research Centre) and ES/L013800/1 (The Analysis of Names from the 2011 Census of Population).

References

    • Bakewell and Mitchell, 2003
    • Generation Y female consumer decision-making styles
    • Int. J. Retail Distrib. Manag., 31 (2) (2003), pp. 95–106
    • Berger and Mens, 2009
    • How adoption speed affects the abandonment of cultural tastes
    • Proc. Natl. Acad. Sci., 106 (20) (2009), pp. 8146–8150
    • Blázquez, 2014
    • Fashion shopping in multichannel retail: the role of technology in enhancing the customer experience
    • Int. J. Electron. Commer., 18 (4) (2014), pp. 97–116
    • Baudrillard, 1998
    • Baudrillard, J., 1998. The Consumer Society: Myths and Structures. Sage, London
    • Carpenter and Moore, 2006
    • Consumer demographics, store attributes, and retail format choice in the US grocery market
    • Int. J. Retail Distrib. Manag., 34 (6) (2006), pp. 434–452
    • Campbell, 1997
    • C. Campbell
    • Shopping, pleasure and the sex war
    • P. Falk, C. Campbell (Eds.), The Shopping Experience, Sage, London (1997), pp. 166–175
    • Dittmar et al., 2004
    • H. Dittmar, K. Long, R. Meek
    • Buying on the internet: gender differences in on-line and conventional shopping motivations
    • Sex Roles, 50 (5) (2004), pp. 423–444

    • Erwin and Calev, 1984
    • P. Erwin, A. Calev
    • The influence of Christian name stereotypes on the marking of children’s essays
    • Br. J. Educ. Psychol., 54 (1984), pp. 223–227
    • Finch, 2008
    • J. Finch
    • Naming names: kinship, individuality and personal names
    • Sociology, 42 (4) (2008), pp. 709–725
    • Galbi, 2002
    • D.A. Galbi
    • Long-term trends in personal given name frequencies in the UK
    • Names, 50 (4) (2002), pp. 275–288
    • Gale, 2014
    • C.G. Gale
    • Creating an Open Geodemographic Classification Using the UK Census of the Population. Doctoral thesis
    • University College London, Department of Geography, London (2014)
    • Gureckis and Goldstone, 2009
    • T.M. Gureckis, R.L. Goldstone
    • How you named your child: understanding the relationship between individual decision making and collective outcomes
    • Top. Cognit. Sci., 1 (4) (2009), pp. 651–674
    • Harari and McDavid, 1973
    • H. Harari, J.W. McDavid
    • Name stereotypes and teachers’ expectations
    • J. Educ. Psychol., 65 (2) (1973), pp. 222–225
    • Harmon et al., 1999
    • H.H. Harmon, R.L. Webster, S. Weyenberg
    • Marketing medium impact: differences between baby boomers and Generation Xers in their information search in a variety of purchase decision situations
    • J. Mark. Commun., 5 (1) (1999), pp. 29–38
    • Harris et al., 2005
    • R. Harris, P. Sleight, R. Webber
    • Geodemographics, GIS and Neighbourhood Targeting
    • John Wiley and Sons, Chichester (2005)
    • Kalyanam and Putler, 1997
    • K. Kalyanam, D.S. Putler
    • Incorporating demographic variables in brand choice models: an indivisible alternatives framework
    • Mark. Sci., 16 (2) (1997), pp. 166–181
    • Kohijoki and Marjanen, 2012
    • A. Kohijoki, H. Marjanen
    • The effect of age on shopping orientation—choice orientation types of the ageing shoppers
    • J. Retail. Consum. Serv., 2 (2) (2012), pp. 165–172
    • Lieberson, 2000
    • S. Lieberson
    • A matter of taste: How names, fashions, and culture changes
    • Yale University Press, New Haven (2000)
    • Longley et al., 2015
    • P.A. Longley, M. Adnan, G. Lansley
    • The geotemporal demographics of Twitter usage
    • Environ. Plan. A, 47 (2) (2015), pp. 465–484
    • Lord, 2002
    • E. Lord
    • Given names and inheritance
    • ,in: D. Postles (Ed.), Naming, Society and Regional Identity, 169−192, Leopard’s Head Press, Oxford (2002)
    • Lunt and Livingstone, 1992
    • P.K. Lunt, S.M. Livingstone
    • Mass consumption and personal identity
    • Open University Press, Buckingham (1992)
    • McDonald, 1995
    • W.J. McDonald
    • Time use in shopping: the role of personal characteristics
    • J. Retail., 70 (4) (1995), pp. 345–365
    • Mateos, 2007
    • P. Mateos
    • A review of name-based ethnicity classification methods and their potential in population studies
    • Popul. Space Place, 13 (2007), pp. 243–263
    • Matheson, 2002
    • J. Matheson, P. Babb (Eds.), Social Trends 32, The Stationery Office, London (2002)
    • Matheson, 2002
    • J. Matheson, C. Summerfield (Eds.), Social Trends 30, The Stationery Office, London (2002)
    • Mitchell and McGoldrick, 1994
    • V.W. Mitchell, P.J. McGoldrick
    • The role of geodemographics in segmenting and targeting consumer markets: a Delphi study
    • Eur. J. Mark., 28 (5) (1994), pp. 54–72
    • O’Malley et al., 1997
    • L. O’Malley, M. Patterson, M. Evans
    • Retailer use of geodemographic and other data sources: an empirical investigation
    • Int. J. Retail Distrib. Manag., 25 (6) (1997), pp. 188–196
    • Openshaw, 1984
    • S. Openshaw
    • Ecological fallacies and the analysis of area census data
    • Environ. Plan. A, 16 (1984), pp. 17–3

    • Pentecost and Lynda, 2010
    • R. Pentecost, A. Lynda
    • Fashion retailing and the bottom line: the effects of generational cohorts, gender, fashion fanship, attitudes and impulse buying on fashion expenditure
    • J. Retail. Consum. Serv., 17 (1) (2010), pp. 43–52
    • Rodgers and Harris, 2003
    • S. Rodgers, M. Harris
    • A gender and e-commerce: an exploratory study
    • J. Advert. Res., 43 (3) (2003), pp. 322–329
    • Scanlon, 2000
    • J. Scanlon (Ed.), The Gender and Consumer Culture Reader, New York University Press, New York; London (2000)
    • Scharf, 2005
    • T. Scharf
    • Recruiting older participants: lessons from deprived neighbourhoods
    • C. Holland (Ed.), Recruitment and Sampling: Qualitative Research With Older People, Centre for Policy on Ageing, London (2005), pp. 29–43
    • Smith-Bannister, 1997
    • S. Smith-Bannister
    • Names and Naming Patterns in England
    • 15381700, Oxford University Press, Oxford (1997)
    • Sorce et al., 2005
    • P. Sorce, V. Perotti, S. Widrick
    • Attitude and age differences in online buying
    • Int. J. Retail Distrib. Manag., 33 (2) (2005), pp. 122–132
    • Wardle et al., 2004
    • J. Wardle, A.M. Haase, A. Steptoe, M. Nillapun, K. Jonwutiwes, F. Bellisle
    • Gender differences in food choice: the contribution of health beliefs and dieting
    • Ann. Behav. Med., 27 (2004), pp. 107–116
    • Williams et al., 2013
    • S.A. Williams, M. Terras, C. Warwick
    • What people study when they study Twitter: classifying Twitter related academic papers
    • J. Doc., 69 (2013), p. 3
    • Xi et al., 2014
    • N. Xi, Z. Zhang, Y. Zhang, Z. Ge, Le She, k Zhang
    • Cultural evolution: the case of babies' first names
    • Phys. A: Stat. Mech. Appl., 406 (2014), pp. 139–144
Corresponding author.