#statistical maps

Bad Takes in the Welsh tag vol. II- this reblog on a post about the number of Welsh speakers. (I have cropped out the username of OP and as ever, I only focus on the sentiment, not the person. If you know OP's url kindly do not send them anon hate etc.).

So I saw this take a few days ago in a reblog on a post in the Welsh tag and wanted to address this sentiment as well, since it does the opposite of that other bad take that I saw and made a post about the other day. To be clear, I don't disagree entirely with OP, but there's an element of wishful thinking that I sometimes see when it comes to Welsh / other minoritised languages which can end up doing more harm than good.

This screenshot was also discussed in the LGBTQIA+ Welsh Discord I run and the broad consensus from those of us in there who live in Wales is that OP is painting a very inaccurate picture of the status of Welsh, particularly of Welsh in North Wales. It is frustrating when you have people who value Welsh, but don't value Welsh enough to bother with accuracy in their promotion of the language. This post is intended as a gentle reminder that we can fight for the Welsh language without misrepresenting the situation on the ground so to speak.

The post itself has a 'fuck yeah, Welsh!' attitude which I personally love. But sadly this particular post is riddled with misinformation. First of all, we have "Welsh law is that all signs must have We[l]sh text on them but there is nothing in the law that says signs must also have English on them". Now the wording is kinda vague here- but I'm going out on a limb and saying that the OP is likely referencing The Welsh Language Standards Guidelines (which have been updated several times over the years). The guidance has a number of Standards relating to signs in the Welsh language, such as Standard 32, Standards 47-52, Standard 66 and Standards 111-113. The section of Interpreting the Standards also contains relevant text, such as in Part 3- Interpreting the Standards article 15:

Plain text: "For the purposes of the standards a requirement to publish, provide or display any written material in Welsh does not mean that material should be published, provided or, displayed in Welsh only, nor does it mean that the material should be produced in Welsh first (unless that is specifically stated in the standard)"

Of the Standards listed above, Standards 47-52 are specifically designated as Standards relating to signs and notices displayed or published by a body. Which state things like "... if the same text is displayed in Welsh and in English, you must not treat the Welsh language text less favourably than the English language text" - Standard 47 and "You must ensure that the Welsh language text on signs and notices is accurate in terms of meaning and expression" - Standard 49.

Anyway, back to the point. OP is incorrect in stating that there is a loophole by which the Welsh Law forgot to specify that the signs had to have English as well as Welsh and that public bodies can get away with monolingual Welsh signs. This just isn't true. Important to note is that the law is intended for public bodies- so big companies, road signage makers etc. This guidance isn't for random farms in North Wales which have signs that say "wyau <-" pointing up the lane with no English translation.

Now, the next sentence is a little loaded, well-meant, but a little loaded nonetheless. "The Welsh nationalist dominated rural authorities in the North"- it's loadedness comes down to its vagueness I think. While it isn't wrong per se that Welsh Nationalist parties like Plaid Cymru do well in the North West, it is a little skewed to ascribe Welsh speaking status to whichever party is doing the best in a given area. It isn't that clear cut, unfortunately. To get into this issue, we have to talk maps.

So those Welsh speaker maps that have nice gradients and have the West of Wales coloured in dark, gradually getting lighter as you move East? Unfortunately, these maps can be very misleading (especially if, like in the map OP was commenting on, the source of the data was left off). But the long and short of it is- these maps tend to imply that Welsh is exclusively spoken in the NW and that everywhere East of Bangor has had it. But the data presentation is very flawed, since it tends to erase Welsh language gains in places like Cardiff, Swansea and Monmouthshire.

You've all seen maps like this right? NW in the darkest colours and SE in the lightest?

Unfortunately when it comes to these kinds of maps, they can be very misleading from a language revitalisation point of view.

Here's some maps I actually studied at undergrad for this purpose

On the face of it, your eyes zip up to Gwynedd and Môn on the first map and then over to the second and- 'oh no!' you might say, there's been a -2.1 to -4.0 percent decline in Welsh speakers in those areas. And of course, this is something that language revitalisation wants to address. But look at the first map again. Look at, Monmouthshire, Caerphilly, Cardiff and Swansea. Then look at the second map.

Welsh speaking is actually being increased in these areas, between 2001 to 2011.

The misleading nature of a language map like this one is not its borders, its colour or key, but its omission of the sociopolitcal forces at play in language revitalisation. Large population centres like Cardiff, Caerphilly, Newport and Swansea are actively gaining more Welsh speakers. While Gwynedd and Môn are losing some. But Welsh speaking (despite a few wobbles) is on the increase. So where did those Welsh speakers from the North go?

South.

It isn't a hard-and-fast rule, but many rural Welsh speakers (especially those who live in areas with high amounts of holiday homes which drive up rent/cost of staying in villages in North Wales) actually end up moving to more urban areas in the South, meaning that some of the decline of Welsh speaking in North Wales is down to Welsh speakers just, moving to a different part of Wales- which in turn makes those areas see an increase in Welsh being spoken.

Of course, we actually have to address the cause of the exodus of Welsh speakers from rural areas holiday homes raising house prices making them unaffordable for locals and drives them away but the way that our data is represented is not as dire as it looks. Still not great, mind, but not apocalyptic either.

Then there's the other inaccuracies in this post. Small businesses like farm shops, high street businesses and houses can have Welsh-only signage because they are not local authorities and much of the guidance indirectly referenced by OP mostly only applies to local authorities. This is how you have farm shops advertising produce in Welsh only, or shop names in Welsh (such as Siop y Pethe and Broc-Môr in Aberystwyth) or the name of the house my flat is in. Businesses have different regulations for signage inside the shop in different situations. But the guidance indirectly referred to by OP in the screenshot mostly applies to road signage.

Big name brands such as Tesco are definitely not going to have monolingual Welsh stores and it is disinformation to suggest that they do- especially not when they've made gaffes such as "sboncen" to mean squash (the drink). "Sboncen" means squash (the sport), while they should have put "sgwash", meaning the drink.

Or my favourite instance of these "arwyddion gwael", in which instead of offering a free ATM service, this ATM on the Tesco Express in Aberystwyth offered "codiad am ddim" (free erections):

So I dread to think what a fully monolingual poorly translated Welsh Tesco would look like.

I don't disagree with OP on the final part, that we should celebrate Welsh's "punk ass attitude" in surviving despite attempts to eradicate it from existence. But spreading false information is definitely not the way we should be doing that.

Good afternoon, folks, and welcome to a shortened but hopefully interesting run of statistical maps to commemorate Eurovision 2020.  I can’t offer as many maps or as lengthy an analysis as usual because online teaching is occupying pretty much my every waking moment at the minute, but I wanted to ensure this year and its songs aren’t swept under the carpet as the EBU seemingly would have it.

Our first map is a relatively straightforward one, showing gender of this year’s leader singer(s). At the start of the national final season, many people were complaining that we had a case of RotterMAN on our hands with the influence of two men having won in 3 years weighing strongly on both internal selectors and national final voters. Eventually, the historic tendency for the number of female main singers to outweigh males at ESC prevailed - only twice (2010, 2019) in the past eleven years have there been more male main singers than females, and in both cases, the difference was small indeed.

ESC 2020 was to be performed with 22 female main singers, 17 male main singers and two groups. It’s only the second time since 2014 that the number of male main singers has dropped at the contest - with 54% of principal singers female, that’s also the highest proportion since 2014. What’s extraordinary too is the continued slump in groups with mixed main singers - from a high of nearly 20% in 2015 to only 5% this year and last.

Good morning, folks, and welcome to today’s Eurovision statistical map! Imagine if there were a way to gauge interest in the contest per country? Well, other than viewing figures, the closest we can possibly get to that in the run up to a Eurovision is by using Google trends, something I’ve been analysing for several years now and to which I’ll dedicate a few posts this ESC season!

One curiosity is that it allows you to see how popular a search term or topic is across nations. If you fancy doing this kind of research yourself, it’s always better to choose topic, because this then includes not just searches for Eurovision but related terms like Eirovīzija, Eurowizja, Eurovisão and so forth that would otherwise be excluded. It then takes the country with the most per capita searches and gives all other nations a comparative score: a country scored 20 has a fifth of the per capita searches for Eurovision as the country with the most per capita Eurovision searches does.

Ever since I’ve tracked this, that nation has been Iceland, not just when they were big favourites with Daði or racking up headlines with Hatari, but with Svala and Maria Olafs too; even with Ari. Because of how small Iceland’s population is and how widespread the love for ESC is there, it’s rare for any country to come anywhere near its per capita search rate, but this year, Lithuania came the closest - a country reënergised by a fantastic national final and a very popular winner, one which got more love outside of the country than most of their recent submissions. Malta, usually near the top of these lists, again is #3 in terms of most pc ESC searches, whilst Sweden, which is typically top of the list, drops below a bunch of ex-USSR nations such as Russia with the highly popular Little Big and Ukraine and Belarus. This year’s hosts and last - Netherlands and Israel - and Serbia, whose Hurricane are a very big deal domestically complete the top of the list.

Meanwhile, on the flip side, the countries with the fewest interest this year - at least, in terms of searches per capita - have been Czechia, Australia, and some of the Big 5: Italy has the lowest per capita search rate of all, surprising given the huge popularity of Sanremo and the notoriety of this edition in particular, surpassed only barely by France. Of the big 5, only Spain is above average with its per capita searches.

Good morning, folks, and welcome to the beginning of my coverage of Eurovision 2018! We start by looking, not quite yet at gender or languages in the contest as is most traditional, but my taking a quick look back at ESC 2011. 

You may wonder why such an unconventional choice, when most Eurovision fans’ attention is understandably focused on what will be focused on what will be going down in Lisbon in fewer than 40 days from now. My reason is this - in all my years observing and analysing ESC, I have not so strongly felt that the result is in the air since the free-for-all that was Düsseldorf 2011. 

Since the all-conquering victory of Loreen in 2012, it’s become a near general assumption that Eurovision is won by collecting a broad amount of maximum points. In every year from 2012-2017, the winner of the contest overall has been the song with the most 12s or someone who came very close (Denmark, second in the 12s in 2013, and Ukraine, second in the televote 12s but top of the jury 12s in 2016.) In 2011, this was not the case. Azerbaijan picked up a paltry 3 maximum points, behind Italy and Bosnia and tied with Denmark, Georgia, Ireland and Ukraine. The douzes were exceptionally scattered, with almost half the field getting at least one twelve, of which 13 countries got 2 or more maximal points. Compare that to just 7 in 2012 and 2013, 6 in 2014 and 2015, 7 jury/10 televote in the 2016 and 8 jury/9 televote in 2017.

Even when disregarding douzes, Azerbaijan didn’t fare extraordinarily, with Italy outperforming them in terms of numbers of countries where it scored the highest number of points compared to the others in the top 6. And yet they got a surprise victory off a number of middling scores from a wide array of countries - something which I have a strong feeling will repeat in 2018. 

There are a number of things that give me this feeling: * The big favourites all have Achilles heels: At the moment, there are only a handful of songs being given 5% or better odds by the bookmakers, with Israel standing imperiously above them all with around 33% odds of winning according to the current average of odds, followed by the songs from Estonia (16%) and Czechia (12%). These are three Marmite songs that have divided tastes amongst everyone I know, and whilst I may be proven completely wrong, I cannot personally see either of them sweeping both juries ánd televote.

* The field is very musically diverse: We have a surprising number of genres and styles represented in this year’s Eurovision. ESC countries’ jurors and televoters can be quite different from one another in terms of how receptive they are to different genres, leading to a more diffuse distribution of points.

* A boost to linguistic diversity: Harken back to the end of the 00s and start of the 10s and we can see regional and cultural pockets that were often traduced as evidence of “bloc voting.” In fact, they were just reflections of shared language and musical heritage - as evidenced by the fact that these pockets have been nowhere near as pronounced in recent years with the likes of the Balkans switching en masse to English. This year is the most linguistically diverse since 2013, which I feel will lead to a greater dispersal of twelve points.

Good afternoon, folks, and welcome to today’s statistical maps - as we take a look at how this year’s Eurovision songs are faring on Youtube. As with every year, we use the music videos on the official ESC playlist - the reason behind this is because these are the most searched and accessed videos, and numbers on the the artists’ official channels are often highly inflated by domestic fans. They also have the advantage of being uploaded around the same time, so some songs don’t have their video up from January and others from late March. (So, please let’s go one year without people barraging me about how Fulanito Wozzizname has got 600 million views on his page and why didn’t I include them, like I got all the time, most notably with My slowanie.)

Youtube is one of a number of tools that can be useful in looking at how the ESC fan community is reacting to the year’s songs. Whilst number of likes and views are not a reliable indicator of who will win, they do give some indication around the amount of hype a song is receiving, how divisive it is and can be handy as part of a package of indicators to predict the songs that will qualify from the semi-finals. Last year, for instance, many of the songs that were neither top 10 in views nor in likes ended up not qualifying.

In terms of views, it seems that few can hope to keep up the pace with Netta of Israel, who has smashed the records in having got over 10 million likes on the ESC channel alone. This time last year, only a few songs had cracked a million, let alone having reached ten times that - this year, in a reflection of ESC getting bigger and bigger on Youtube with every passing year, there are already eleven songs - more than quarter of the field! - sitting on a million views, with many others following them swiftly. 

The much hyped Bulgaria and Spain, whose Operación Triunfo lit a massive fire amongst the fanbase here, are the only two videos joining Netta in having passed the 2m mark, but perennial foci of interest (like Sweden and Russia) some hotly-tipped countries for this year (Estonia, Czechia, Australia) and some well-regarded critical favourites (especially Belgium and Macedonia) have all broken a million. There are only two areas - the Nordic lands and the eastern Balkans - where almost all the songs have been seen a lot, with the popularity of videos from less-augured ESC nations showing that the contest is really an open ball game after last year’s podium where each country got its best result. On the other hand, countries in central Europe, the western Balkans and much of the former USSR tend to have clocked up scarce views.

If we turn to look at likes, the map looks quite different, with a corridor of 90%+ likes covering much of the centre of the continent and only a few countries in its extremes. Only a few of the highly-viewed countries manage to crack the 90% threshold, with the most likes being received by Greece, Poland, Lithuania, Italy and Switzerland. Only a scant handful of songs have a like rating of less than 80%, and unusually, there are only two countries on less than 60% likes - San Marino on 59% and Russia on just 39%. It will be very interesting to see how these tallies change as more views come in! The full data table is available below :

Good evening, folks! Songwriting is all about balance - when crafting lyrics, you want to give your audience something that, through repetition, is memorable and easy to sing along with. Overdo the repeated lyrics, however, and you find yourself with a song that is memorable for the wrong reasons, from having annoyed the listener. 

Terms like “lyrically dense” and “repetitive” are thrown around a lot by reviewers, but there is an objective way to measure lyrical repetitiveness, by dividing the number of unique words by the total number of words in a song: the lower the percentage of unique words, the more words (and often whole phrases) have been repeated. When it comes to English, at least, even very wordy and unrepetitive songs seldom get over 50-55% unique words - not just because of repeated choruses and bridges, but because of the common words, especially articles, basic verbs, conjunctions and prepositions which are difficult to avoid. Songs with 30% original words or less, on the other hand, are notably repetitive. 

Just like the overall number of words in a song, how repetitive or not a years’ songs are is a strangely cyclical thing. In 2015, the average song had a quite considerable average of 39.41% unique words, with about half the field sending songs that had 40% or more unique words. However, the winner and five of the top 10 sent songs well under the average number of unique words - and in 2016, consequently, the average dropped to 34% unique words, with the number of songs that had 40% unique words or more falling by almost half. 

Whilst in 2015, 6 of the 10 most repetitive songs did not qualify - and out of those that did, only Australia really made a mark at the top of the table - in 2016, 7 of the 10 most repetitive songs qualified, and nearly all of them got in the top 15. So, one would expect an even more repetitive year this year - but instead, the average song’s original word count has increased a little to 35.7%, perhaps because last year’s winner, as well as big hitters like Russia and Sweden, were all less repetitive than the average.

The 5 least repetitive songs this year all have a specific reason for this status. Malta’s song has few words and is delivered slowly with little repetition; Portugal makes use of long instrumental breaks and keeps the lyrics poetically sparse; Croatia boosts its unique word tally by singing in two languages, whilst Belarus and Hungary owe their status, to some degree, to their languages. Belarusian lacks articles (like a and the) and in most instances a copula (I am a teacher is  Я настаўнік, I teacher) - two things commonly repeated in English language songs. Hungarian has both copulæ and a definite article, but can use cases and other inflections where most European languages would use prepositions and possessive pronouns (house is ház, but my house is a házam and in my house is a házamban - you can see how this agglutinative structure makes for a high number of unique words.)

The 5 most repetitive songs are mostly ones that you may well have predicted thanks to how often a phrase is repeated in them: Latvia (where we baby, where we draw the line), Moldova (hey mamma, hey mamma, hey mamma), Cyprus (let me be your gravity) and the most notorious of all, Spain, where “do it for your lover” is repeated 26 times and “just do it” 17 times, meaning that we hear one of these implorations every four seconds on average. Making up the five is Belgium, whose multiple lyrical repetitions owe themselves to the fact that musically the song is built like a hymn, repeating two phases and without a prechorus or bridge.

Most winners of Eurovision have a percentage of unique words that sits firmly in the sweet spot of 35-45%, which is where we find the bookmakers’ top tipped songs, Italy and Bulgaria. Other hotly backed songs, like Belgium and Sweden, fall well below this average. Will we see a more or less repetitive song take the trophy this year? We’re inching ever closer to finding out. Check out the full data table below!