#here have a basic phonology lesson and a brief analysis of the national spelling bee

[acommonrose]

What Makes Spelling Words Hard

This post is 75% in response to the spelling bee’s commentators frequently pointing out the wrong parts of words as difficult, but I figured I’d try to use my actual knowledge of the  bee to describe what makes words hard in general.

For some actual statistics, there was a great fivethirtyeight article recently that ran some numbers. I don’t know if I agree with all the analysis, but it does point out things like it’s provably false that longer words are harder. My thoughts are a lot more subjective than that, but hopefully useful.

A hard word is a word you don’t know.

I’m going to lead with this, and honestly, if you want to stop reading here, you’ll still know the main takeaway. We can talk for forever about what makes words hard and what might trip them up, but once you’re watching the finals, most of it comes down to how many words the speller has studied. If they’ve studied the word, they’re going to get it. If they haven’t, it’s going to be a lot harder.

So how do we gauge how likely a word is to be known by a speller and therefore how hard it is? Well, we can’t really necessarily know, since every speller uses a different list to study from. What we do know is that the bee frequently reuses words, so spellers are likely to know words that have appeared in previous bees. It’s possible to find all the previous bee words from 1997 onwards online if you’re willing to do some poking around the Scripps site and maybe use the wayback machine a little. It’s also the case that almost every bee word spelled up until 2004 (I think) was compiled into a list called the Consolidated Word List (CWL) that is no longer “officially” provided by Scripps but is still very findable. Once you get past the prelim rounds, it’s just the case that a speller is more likely to spell a word right if it’s on one of these previous bee lists, and in general, the words that a speller gets right are the ones they’ve studied. (If you, say, look at the championship list rounds of this year’s bee, it was pretty clear that both spellers already knew basically all the words they spelled right.)

Of course, it’s still possible for a speller to correctly guess a word they don’t know, so putting aside the “already known” factor, let’s get into words that can be guessed.

Root words make things easier.

The most guessable words by far are basically the mix and match version of “words you already know”. Many spellers will study common (and not so common) Greek and Latin roots, and even those who don’t explicitly study them will run across the same roots plenty of times in the words they do study, since they pretty much make up the basis of all those impressive-sounding lengthy scientific terms that the bee loves to use. The nice thing about words that are made up of Latin and Greek roots is that if you know the roots, you can spell the word, even if you haven’t seen that particular word before. In fact, the national bee even encourages this, allowing spellers to ask if their word contains a given root word, as long as they provide the root word’s language of origin and meaning.

Consider, for example, the word achromatopsia, which appeared in the written round of this year’s bee. It’s a long, complicated-sounding word, and it’s never been in a bee before, which theoretically should make it quite hard. As things actually go, though, it’s not too hard a word to figure out, and I imagine most of the top spellers did just fine with it. The prefix a- is pretty common, you can figure out chromat- from knowing relatively common words like chromatic, and the end (which is straightforward anyways) is also related to words optic. Put the parts together, and you have the word.

Some languages of origin are easier than others.

Fivethirtyeight has actually done a breakdown of which languages of origin are the hardest for spellers, and I know some spellers (or coaches/tutors) have done the same, but this is just going to be a subjective take on the matter. One of the crucial parts of the spelling bee is learning the spelling patterns from the many languages English borrows from. The bee makes no secret of this. The Spell It, which is the official study list for the regional bees, even splits up words by origin. What makes a language of origin hard is a combination of factors, most notably both how consistently a language is spelled and how intuitive those spelling rules are. Latin words, for example, tend to be spelled in a consistent, intuitive way even if they’re not from common root words, making Latin a relatively easy language of origin (with exceptions, of course). French, on the other hand, tends to have a number of ways to spell most vowel sounds (and some consonants) and even a lot of its simpler spelling patterns don’t make a ton of sense if you’re not familiar with French, so French words are generally considered pretty hard. Old English, which had no standard spelling, tends to produce words that are either common enough that they’re not worth asking at the national bee or horrifying messes like the word getish in last night’s final (pronounced yeh-teesh). And on top of this all, it means that words of unknown origin or that come from proper names (depending on the nationality of the name) can sometimes basically be impossible to guess (like Klydonograph in last night’s bee).

Of course, this is all complicated by the fact that any of these words could change in spelling or pronunciation when entering English or at any of the intermediary points if they went through multiple languages. To give a firsthand example of how this can trip up a speller, let’s consider the word cacoethes, which was on the written round in 2008, where we were given that the etymology was that it went from Greek to Latin to English. In Latin, the hard c sound is basically always spelled c. In Greek, it’s never spelled c and is either k or ch. My mind latched onto the Greek part and connected it back to some words I’d studied with the kak- root like agathokakological, so I spelled it kakoethes. Obviously, I should have focused on the Latin part instead. (Or just connected it back to, say, cacophonous. We all have slip ups.)

Anglicization of pronunciations can also throw a huge wrench in things. Consider, for example, the ending of ��last night’s winning word marocain. Theoretically, this could be really hard, since the long a sound in French can be spelled a lot of different ways. Of course, those of us who know French know that the French word marocain (which isn’t that uncommon, it just means Moroccan) doesn’t actually have a long a at the end. Instead, the ending is pronounced closer to the English word can but with a nasal vowel at the end. In this case, it’s not so tricky, since the “ain” ending is really common in French borrowings, and a lot of the time it gets pronounced the same way as in marocain, since nasal vowels are uncommon in English. (Think of the significantly more common word legerdemain.) Still, any knowledge of long a sounds in French would not be useful here.

In some cases, the fact that sounds not present in English get anglicized can have even trickier consequences. Notably, some languages that get spelled or transliterated phonetically may have two sounds that are considered different in the language but are indistinguishable to English speakers. You might realize, for example, that in a bunch of words from South Asian languages, things like “b” and “bh” or “t” and “th” seem to be used interchangeably. This is because these languages distinguish between something called aspirated and unaspirated consonants, which basically corresponds to whether there’s a short puff of air after the consonant. Interestingly, we do have aspirated consonants in English. (Try putting your hand in front of your mouth and saying the words “top” and “stop”. You should feel a puff of air after the first but not the second.) However, aspirated consonants can only appear in certain places in English and aspirating a consonant can never change the meaning of a word. In many languages, however, changing, say, an unaspirated t sound into an aspirated t can completely change the meaning of word (the same way that changing a t to a d can completely change an English word). Therefore, it makes sense to transcribe aspirated t sounds as th and unaspirated t sounds as t, but when these words go into English, they follow English patterns for aspiration, and the distinction is lost, making it basically impossible to know whether there’s a (seemingly) silent h after the t without knowing the word. Similarly, doubled consonants in Italian are pronounced differently than single consonants (or at least I’ve heard as much from native Italian speakers), but since English never uses consonant length to make a distinction between words, these distinctions are completely lost when the words enter English, leading to all sorts of tricky Italian words in the bee.

The last wrinkle in this is that languages that are written in non-Roman alphabets can be either very easy or very hard depending on how easy it is to capture that language’s phonology with the Roman alphabet and also the history of that language’s contact with English. Japanese words, for example, tend to be transliterated in a consistent, sensible way and are mostly relatively easy to spell. Hebrew and Yiddish words, on the other hand, tend to be spelled pretty inconsistently (and often have spellings and pronunciations in use that aren’t even in the official dictionary), making them often quite difficult. To a lesser extent, this is also a problem for languages that make heavy use of accents and other diacriticals, since these diacriticals get omitted or transcribed in some weird way, often differently in different ways. For example, the word aracari in last night’s bee is pronounced ar-uh-sar-ee, with the c denoting an s sound, because in the original Portuguese, there was a diacritical mark under the c, indicating its pronunciation, but it becomes much more difficult in English, where we don’t have that diacritical.

Schwas are evil.

If you talk to any speller (or even listen to some of the inane commentary during the bee broadcast), you’ll probably hear about schwas being difficult to spell, and I’m guessing you might not know what that means. Basically, in English, nearly every unstressed vowel turns into a short “uh” sound. This sound is called a schwa and is written as an upside down lowercase e in phonetic transcriptions. (I say “nearly every” because sometimes short i sounds actually turn into a slightly raised schwa sound, indicated by a dotted schwa in Webster’s dictionary. There’s a lot of variation among speakers, and it’s not super important. I mention this only because there was a weird amount of drama in the 2007 National Spelling Bee over a dotted schwa. Also stress is a lot more complicated than I’m making it out to be here, but I’m trying to only sort of give a giant phonology lesson here.) Because schwas are the unstressed form of every vowel in English, they can be spelled using basically any vowel, which makes them really hard to guess unless the word happens to have an alternate pronunciation with a different stress pattern. This is why you’ll occasionally see spellers “schwa fishing” or repeating their words but slightly stressing unstressed syllables with tricky schwas and pronouncing these schwas as other vowels, trying to get the pronouncer to slip up and reveal the spelling of the schwa. (Pro tip to spellers: don’t schwa fish. It doesn’t work.)

Sometimes knowing language of origin information can help. For example, schwas connecting Latin roots tend to be i, and schwas connecting Greek roots tend to be o. Schwas in German words tend to be e, and a lot of languages have similar patterns. Unfortunately, there are plenty of cases where there’s a number of fairly reasonable options for what a schwa could be, and all of the general rules have exceptions. Consider, for example, two of the words from last night’s bee: aubusson and saussurism, both of which have French origins (both from French names, I think), and both of which have schwas denoted by u in the second syllable. These are tricky, because schwas are typically spelled as e in French when they appear, though of course neither of these words was pronounced with a schwa in the original French. I believe both are pronounced with a fronted long u sound (that doesn’t actually exist in English), but because of the resulting stress patterns, when they’re adapted into English, they both become schwas. To make things more complicated, schwas spelled with u in English, while prevalent, usually are near the start (as in suspicion) or end (as in impetus) of a word. That makes the schwas in aubusson and saussurism amazingly difficult to guess unless you happen to be familiar with the specific words or the names they’re derived from.

(Side note: There’s another notable instance where English has different sounds that become the same sound in certain contexts, which is the case where d and t appear in the middle of words, as in water or leader. In these case, both t and d turn into the same very short “tap” sound, which is actually the same as the short r sound in Spanish. The only reason this hasn’t been a problem as of yet in the bee is that because of the way these are denoted in Webster’s Third, if there’s a t that becomes this tap, the pronouncer at the spelling bee is supposed to give a pronunciation where the t is more clearly pronounced. You can read more about how the bee handles these over here.)

Overcomplicating is a constant problem.

Sometimes there’s a word in the bee that just doesn’t look too bad. Maybe you don’t know the specific word or any of its roots, but it’s from a relatively straightforward language like Latin, and there seems to be an obvious way to spell it. Your first instinct is to go with the simple spelling. Unfortunately, your second is that this is a trap, and that there’s some sort of secret trick to this word you need to figure out.

Consider, for example, the word coloboma, which appeared in round 5 of this year’s bee. It’s a little bit of a weird word in that it passed through both Greek and Latin, and the beginning uses the Latin c, but the mid-word schwa is the Greek o, but that’s pretty common in Greek to Latin words (see the aforementioned cacoethes example), and mostly, it’s pretty simple and straightforward, and even though it had never be in a bee before, it seemed guessable.

And I really thought the speller who got the word was going to guess it right after he opened with c-o-l. I could see some ways to overcomplicate the beginning: opening with a ch, doing something weird with the first vowel, maybe doubling the l, but the -oboma part seemed straightforward. The speller could put in the normal Greek schwa, spell the completely straightforward (and hard to overcomplicate) ending, and would be on to the next round.

And then he said y for the next letter. Why? I don’t know. I think attempts to analyze what the spellers are thinking are presumptuous and obnoxious. But this is an example of a speller in a hard round overcomplicating things and misspelling as a result. A common piece of speller wisdom is “keep it simple, stupid”, but I get why spellers overcomplicate. It doesn’t seem all that unreasonable that there would be a surprise y in a word in a round where other words include saccharomycete and pterylosis. In general, keeping it simple is the right call, but at the top levels, it can be hard to know which words have bizarre quirks and which don’t.

The problems are exactly what you think they are.

Here’s the thing. Things like schwas and languages of origin are kind of inside baseball, but at the end of the day, the most common things that make National Spelling Bee words hard to spell are the same things that make normal words hard to spell. Yes, the double letters in panniculitis might be more intimidating than the double letters in unnecessary. The silent k in knock seems more reasonable than the silent z in suivez (swee-vay). Yes, the spelling bee words are intimidating, and success requires a working knowledge of basic linguistics and the spelling patterns of many of the world’s languages, but at the end of the day, it’s a spelling bee. If you’ve ever been in one, you might have a sense of what it’s like.