Fun with Place of Articulation!

In the phonetics post I did a very quick explanation of what a place of articulation was, but I think that each one deserves a little more love. In this post, reference the diagram below:

The exolabial and endolabial places of articulation, formed by pursing or touching the lips together, are collectively called labial, and are never distinguished as separate sounds. However, languages or even particular sounds may have a tendency to be pronounced one way or the other. The distinction is the same as that between two types of rounding in vowels - protrusion (endo) and compression (exo). The essence is that endolabial sounds are pronounced with the lips slightly forward and exolabials with the lips drawn tightly together. Don't worry about it too much. Labial sounds in English include "m", "b", "p", and "w". The similar labiodentals, formed by touching the upper teeth to the lips, include "f" and "v".

The dental, alveolar, postalveolar, and retroflex sounds, collectively called coronals, are pronounced with the tip of the tongue touching the teeth or roof of the mouth. The dentals are the furthest forward, with the tongue touching the teeth themselves. The alveolars are slightly more retracted, with the tongue on the alveolar ridge immediately behind the teeth. The retroflex sounds are pronounced with the tongue curled back against the hard palate. Lastly, the postalveolars are a messy categorization of sounds that usually involve the tongue being somewhere behind the alveolar ridge, but usually mostly in front of the palate. Coronals are unique because the tongue's flexibility in this region allow them to take on additional distinctions. They may be distinguished between laminal (tongue blade) and apical (tongue tip or apex) sounds, though only a few languages, such as Basque, make a distinction between the two. They are also distinguished by sulcalization, a phenomenon in which the center of the tongue is grooved to direct airflow quickly. Sulcalized sounds are commonly called sibilant, and have a distinct hissing sound. Sulcalization is the difference between "s" and "th" sounds. Other English coronals include the alveolar "t", "d", and "n" sounds and the postalveolar "sh" and "ch" sounds.

The velar sounds are pronounced with the dorsal (back) region of the tongue against the soft velum. Dorsal sounds in English include "g", the "ng" in "king", and hard "k" sounds.

Palatals are pronounced with the dorsal region of the tongue against the hard palate, though the front of the tongue is usually fairly close to the roof of the mouth. Palatal sounds may sound like coronals being pronounced at the same time as a "y" sound, and indeed "y" and "ee" sounds are pronounced with the tongue near the palate. Palatals are odd in that they are sometimes considered more like coronals, and sometimes lumped in with velars as dorsal sounds, depending on the phonological context. Palatal sounds in English include "y" and the hissing "hy" sound at the beginning of "huge".

Past the dorsals are regions generally unexplored by English tongues. The uvulars are pronounced further back in the throat than velars, against the uvula, like Arabic "q". Pharyngeals go further back still, with the pharynx (throat) constricted. The glottals are pronounced with the glottis, the opening to the lungs themselves, and include the "h" sound and the catch in the middle of "uh-oh".

Languages tend to have between three and five major places of articulation, at which multiple stops, nasals, or fricatives are found. Almost every language has a major labial, dental or alveolar, and velar place of articulation. It's also common to add one or occasionally two from among postalveolar/palatal, uvular, and glottal. Uvulars, pharyngeals and glottals are less common to see as full developed sets of sounds. More often, languages will include only one or two sounds from them, if any.

Fun with Manners of Articulation! (Or, why what I told you is wrong)

In my post about phonetics I said that the four basic pulmonic manners of articulation are stops, nasals, fricatives, and approximants. Here I'm going to completely recant that and explain why it's not that simple. The first step is to look at what each of those mean.

I defined a stop as a sound in which oral airflow is completely blocked. While this is true, this can hardly tell the whole picture of the sound. What happens after the stop can do as much to characterize the phone as the stop itself. Enter affricates, sounds formed when a stop is fricated for a moment after release, like the "ch" sound /tʃ/. In essence, whatever is blocking the airflow opens enough to let airthrough, but the airstream is still very narrow and turbulent and sounds like a fricative before completely opening before allowing the next sound to be pronounced. The issue is that this can be difficult to distinguish from phenomena that wouldn't make a stop considered anything but a normal stop. Aspiration or positioning of the tongue can make the moment of a stop's release turbulent. For instance, in Danish the aspirated alveolar stop is /tˢ/, pronounced with a slight frication upon release. Similarly, Japanese /t/ is pronounced [ts] before /u/ because the vowel height causes the consonant to be sulcalized, or pronounced with the midline of the tongue depressed, causing a characteristic sibilant, or hissing, release. In practice, affricates tend to be very rare at any place of articulation except coronals for the simple reason that affricates often just don't sound very distinct from stops. Coronals are the exception because of the distinctive sulcalization that they may undergo.

The other thing to consider is that what I call nasals actually fit the definition of stops. That is, nasals also don't allow airflow through the mouth. This is because nasals in the conventional sense of the word are more precisely nasalized stops, where nasalization refers to the fact that the nasal passages are opened. As stated in the phonetics post, each component involved in speech acts in relative independence from others, and nasalization simply refers to the opening of the nasal passages during pronunciation. The reason that nasalized consonants are almost always stops has to do with sounding distinct, as usual. If air is allowed to flow through the mouth as in a fricative, the air flowing through the nose becomes relatively insignificant in its effect upon the sound. However, the difference in sound between a nasalized and non-nasalized stop is tremendous. By allowing for air flow through the nose, a nasal can be pronounced with continuous airflow, unlike a stop where air is completely halted. The most salient sound becomes that of air going through the nose, rather than of air being released after the oral stop. For this reason, nasalized stops are usually retained as a separate series of sounds. Counterintuitively, the only other sounds that are commonly nasalized are vowels.

Lastly, there's the messy fact that the difference between fricatives and approximants, or between approximants and vowels, is not always clear-cut. Each of these refers to a differing level of openness of the mouth, but the fact is that the exact level of openness that corresponds to, say, a fricative rather than an approximant, is rather inexact. The standard definition of frication is turbulence between two articulators situated close to one another. However, frication is often not distinctive within phonological boundaries, and fricatives and approximants may run together into a single phoneme. For instance, in Spanish the approximant /j/ is often pronounced as the fricative [ʝ], and corresponds in the phonology to fricatives at other places of articulation. The difference between high vowels and certain approximants, while usually a strict phonemic distinction in the sense that they are treated very differently in syllables, can be minor. X-ray analysis shows that approximants tend to be pronounced closer than vowels, but only barely so.

The simplification of splitting consonants into stops, fricatives, approximants, and nasals is a pragmatic one, and works to explain phonology most of the time. For good reasons, languages tend to split up their consonants that way, but hopefully this post is an insight into the way consonants really work, and why that simplified model just isn't always so.

Phonology

Phonology is the study of sound systems, and unlike phonetics is concerned with underlying structures and discrete items. Phonology sees the introduction of our first "-eme," of which there will come many: The phoneme! A phoneme is a meaningful sound in a language, and is distinct from a phone in that there is a finite number of phonemes in any given language, and phonemes exist only within the construct of a language.

A major part of the description of a language is the phonological analysis, in which a linguist attempts to figure out the pattern beneath the sounds of a language to come up with the "phonemic inventory," the set of all meaningful sounds. This can be complicated by the fact that phonemes may be pronounced differently in different contexts, a phenomenon known as allophony. Each pronunciation of a phoneme conditioned by a particular set of circumstances is called an allophone.

For instance, Japanese clearly contains the phone [ɕ]. However, an analysis of the phonology shows that [ɕ] can never be followed by [j] (a "y" sound) and that [s] can never be followed by [i]. In fact, [ɕ] is an allophone of the phoneme /s/ (It's worth noting at this point that phones are written in square brackets [] and phonemes in slashes //) before /j/ or /i/. Therefore, we'd say that Japanese has the single phoneme /s/ with at least the two allophones of [s] and [ɕ], which assimilates to the frontness of the vowel or glide after it. That's phonology.

It's worth noting at this point that the graphic representation of a phoneme is somewhat arbitrary. A phoneme exists only as it is realized, and it may not always be apparent what the "typical" or "average" pronunciation of a phoneme is. In general, the notation used tends to chosen based on the most common pronunciation, the most graphically simple symbol, or like the value in prior historic forms.

The other component of phonology is phonotactics, or the acceptable juxtapositions of sounds. For almost every language, the concept of a syllable is key as a building block of words (though the North American Nuxalk language is a notable exception, with such mouthfuls as [xɬpʼχʷɬtʰɬpʰɬːskʷʰt͡sʼ]. Better men than me have thrown up trying to say that one.) It's highly recommended that you start by deciding upon syllable structure.

Syllables are centered around a nucleus, which is generally a vowel for the simple mechanical reason that it's easier to transition between sounds if your mouth is allowed to open in the middle. Other nuclei are acceptable, however, with nasals, approximants, or even fricatives acting as the central component in a syllable. Depending on your dialect of English, you may pronounce the second syllables of button and bottle with no vowel at all. Or a syllable nucleus may contain a transition between several vowels sounds, called a diphthong or a triphthong.

On the edges of the nucleus are the onset (consonants at the beginning) and coda (consonants at the end). Cross-linguistically, onsets tend to be more complex than codas. Languages that allow more complexity in the coda than in the onset don't exist to my knowledge, and languages with heavily restricted codas or no codas at all are fairly common. Both codas and onsets have a tendency to follow the sonority heirarchy, the guideline that more "sonorous" or vowel-like consonants are allowed closer to the nucleus. This generally means that approximants are closest to vowels, followed by nasals, then obstruents (stops and fricatives). Any of these steps may be skipped, but it is very uncommon across all languages to see a stop between the nucleus and an approximant, for instance.

There may be other phonotactic rules that take place across full words. It's common for there to be separate rules for what may come at the end of a word. In Spanish, only a few consonants can finish a word, while in Arabic the end of a word is the only place that a coda may contain two consonants (not altogether very common - ask me sometime why Arabic phonotactics are so weird.) There can also be long distance effects like harmony, by which sounds (usually vowels) assimilate across multiple syllables or entire words. Lastly, there is the mess of prosody, accent, and word tone, which deserves a whole other post and then some.

When it comes to conlanging, figuring out your phonology is certainly key. It can be the first step for many conlangers. It's also one of the biggest steps for aesthetics to play a major role. Start by listening to other languages, exploring new and unusual sounds, and deciding what you like before you settle upon a phonology. There's also much more to phonology than explained in this post - more to come.

A Brief Exercise in Mutual Intelligibility

I wanted to take a break from my walls of texts to just give you all a listen to a couple of gems I found and showed to my students. The following two clips are recordings of people speaking variants of English that are on the cusp of mutual intelligibility. See if you can determine where the speakers are from and what they're saying.

Sample 1

Sample 2

Phonetics (with pictures!)

The last couple of posts have been a sort of fire hose of my personal thoughts on language. I went ahead and put them out first because I think they're good to keep in mind going forward into the actual process, but there's going to be a bit of shift in focus. This post is my first real set of instructions on the nitty-gritty bits of conlanging.

It's a tradition among conlangers to begin their process with phonology. I might question the necessity of this, but in any case it's a fine place to start. Before you understand phonology, however you must understand phonetics. If you're up to speed on your Greek roots you might guess that these both have to do with sound, and you'd be right. That's where the similarity ends, though. In fact, if you understand this distinction you'll be ahead of most conlangers. For whatever reason, amateur conlangers have a tendency to conflate phonetics and phonology, but the truth is that they are very separate and understanding the difference is key for good, realistic conlanging. The knowledge I'm about to impart will put you head and shoulders above many experienced conlangers.

Phonetics is the study of the sounds of human language. Linguists and conlangers in particular tend to focus upon articulatory phonetics, and with good reason, but acoustic phonetics is a thing as well and I may reference it from time to time. Articulatory phonetics studies how the human vocal tract produces sound. In linguistics, we refer to each individual sound as a phone, though this may be an iffy concept. No sound will ever be produced exactly the same way twice, and phonetics attempts to describe sounds in general terms. In phonetics one hears talk of the spectrum of narrow vs. broad description. This is simply a measure of how exactly a sound is described. Perfectly narrow transcription would consist of a high-quality recording and an x-ray of the speaker's vocal tract as they spoke. However, for the sake of convenience, we tend instead to describe a phone in terms of its features, and a more narrow transcription simply contains more information about the features of speech.

This is a diagram of the human vocal tract:

There are many separate components at work here, including the lips, tongue, nasal passages, and vocal cords (larynx). The may act in relative independence to give rise to a whole host of combinatory sounds. People are sometimes surprised to learn that there is a very strict organization among the sounds we make, and every phone may be described by its features. However, we can indeed characterize every sound by the behaviors of each of these components. The most basic distinction among sounds, one you are more than likely familiar with, is between vowels and consonants, though in reality there is a slight bit of gray area even between these.

A vowel sound is one in which the mouth is open and there is minimal obstruction of the airstream. In vowels, as in almost everything in phonetics, the tongue is the most major player. (It's no coincidence that "tongue" is a euphemism for language in so many different languages.) Vowels are chiefly characterized by the position of the tongue in the mouth. The chart below is an abstracted representation of the oral cavity in the diagram above:

It refers to the positioning of the tongue within the mouth, where left is further forward, right is further back, and down lower, with the jaw wider. Note that this corresponds to the model above. Towards the upper left, we have high front vowels, "ee" sounds. In the upper right are high back vowels, "oo" sounds. And towards the bottom are low vowels, "ah" sounds. For a more full set of vowel sounds with recording, click here. (Note that this link makes use of IPA, the International Phonetic Alphabet. The long story short is that in the 19th century, several old white guys sat down and invented a way to represent every sound of human language. It remains the gold standard for linguists, and will be used throughout this and future posts.) Another prominent feature in vowels is rounding, or whether or not the lips are pulled together in an "o" shape. By convention, as in the link above, rounded and unrounded sounds are written in pairs, with unround to the left and round to the right. Additionally, nasalization (opening of the nasal passage) and phonation (qualities of the vocal cords' behavior that I won't discuss here) may characterize vowel sounds.

Consonants are sounds in which the airstream is more obstructed, and are altogether more complex. Consonants tend to be chiefly organized by place of articulation (where it's obstructed) and manner of articulation (how it's obstructed). The chief pulmonic (don't ask - non-pulmonics are a whole other can of worms) manners of articulation include stops, fricatives, nasals, and approximants. (Ask me sometime why this organization is a cop-out, but I don't want to get started now. This post is already super long, I know.) In short, stops are formed when the airstream is for a moment entirely blocked; fricatives when the airstream is closed enough to create a turbulent airstream but not entirely stopped; approximants by obstructing the airstream a little, though not even enough to produce frication; and nasals by redirecting sound through the nasal cavity. Place of articulation is an even more complex matter. The diagram below shows all places of articulation:

In general, sounds will be produced by constriction at one of these places. However, it is perfectly possible to produce sound by blocking or constricting several places. This is called coarticulation. The most common examples of coarticulation include labialization, by which the lips are rounded during pronunciation, palatalization, when the dorsal region of the tongue pulls up toward the soft palate, and pharyngealization, when the pharynx (back of the throat) is tightened. These coarticulations can all be paired with other sounds to create very complex consonants. The link above also provides an organized chart of pulmonic consonants (though coarticulation is disregarded) with recordings.

Hopefully this fly-by of speech articulation mechanics gives you a basic enough idea of phonetics to successfully move forward. I'll post about many other specific phonetics topics at a later time.

Why I make natlangs

When I was younger I liked the idea of engineered languages. I wanted to make languages that broke new ground, or were totally alien. One of my earliest conlangs came about after being so inspired by Rowan Atkinson's touching performance in Blackadder Goes Forth that I decided to make a language in which every word ended in -bble (wibble, jabble, blobble, fubble, etc.). It was weird and I don't remember getting too far. I also made a snake language which made heavy use of sibilants but didn't use any voicing or back vowels. That was also the language where I tried eliminating verbs, after the fashion of Sylvia Sotomayor's Kēlen. I also had a few failed attempts at entirely logical languages. I don't think that I ever thought much about whether my languages would ever be used, though I suspect that I somehow thought they might.

I while ago, though, I had a shift in thinking. As I learned more, I came to conclude that humans speak the way they do for a reason. Such a pristine, inhuman system, would fall apart to an imperfect mess if it were ever to be used. Rather than change humans, though, something which I think was at the core an assumption I made when making such otherworldly langs, I began to focus on the humanity of language itself. A language is a system only in practice, and I had to admit to myself that what I was making wasn't really a language. I suppose that it would have been easier simply to admit that I wasn't making a language and move on, but that realization was accompanied by a shift in motive. My prior projects, I decided, were merely exercises in aesthetics or information theory. There's certainly nothing wrong with such pursuits, but I lost interest in making something that was ultimately unrepresentative of the real world.

I think that a lot of conlangers, particularly new ones, set out with same sorts of goals that I did. This is by no means an attempt to talk such people down. Indeed, I think such experimentation can be fun and useful. Perhaps, if anything, this is more of a reminder. Humans will always slur their words, break rules, and delight in ambiguity because it serves a purpose. To create something that has grounding in human experience, or in fact in the logical rules that language must follow, it's necessary to follow the model of real people. This might lead to a shift in focus, as well. I know that it has for me. No longer am I like an painter trying to invent new colors, or to hope against hope that I might be able to make the canvas perfectly smooth. I'm merely trying to appropriate what I see, and use the materials given to me, to create something that I find beautiful. This makes conlanging a much more personal experience as well - rather than try and create something of objective merit, I try to make something that makes sense and feels good to me. I'm only trying craft my own identity, something which is perhaps more useful anyway.

What's a language?

DISCLAIMER: The following piece reflects my personal philosophy, and is not to be taken as the only way to think about conlanging. However, I think it remains for anyone a useful thought experiment.

Language construction is the art or science of making a language. Sounds simple. But what, exactly, are we intending to accomplish when we say that we first embark upon the creation of a language? To know how to get there, we must first know where we are getting.

The fact is that language is not a simple thing. Language in the truest sense is a behavior, a phenomenon. It's the name we give the complex array of symbolism that people use to convey information, emotion, and ideas to one another. In modern humanity, language most commonly is an act of oral sound production, and as such this tends to be the focus of linguistic description. However, language can mean the conveyance of meaning through hand signs, writing, or indeed any other mode of semiotic representation. What we mean by a language is something separate entirely.

A language most loosely refers to a single mode or method of conveying information. However, it is in reality never than simple. Describing a language is an attempt to generalize a set of specific behaviors and exchanges. In short, language is defined by its usage. This is something that is so often lost among conlangers, and linguistic hobbyists in general. Language, especially in our culture, may be seen as something pristine, a well-defined, ideal system from which usage deviates, as though everyone in a language community sat down and decided upon every rule in the language. In reality, it's much more messy than that. People speak like their peers, like those with whom they interact, and language is in chief a symbol of communicative identity. To speak someone's language is to indicate in a way that cannot be faked that one subscribes to the same cultural identity, exists within the same community and owns that group's mode of communication through extended exposure. Similarity of language is proportional to how much of one's cultural and linguistic background is shared. As such, a single language is not something to be easily defined when one's community, one's interactions, and even identity itself may be in constant flux.

In general, the standard for defining a language is mutual intelligibility (though national and ethnic identity can also play a major role). However, this leaves a lot more room for error than tends to be described in the modeling of a language. Within a language under such a definition, there is room for regional dialects and cultural registers that may each constitute as rich a background as the "language" itself. In addition, speakers may slide unconsciously between and among these without realizing it, meaning that language can never really be the same twice.

It's said that one may take a succession of taxis between Gibraltar and Rome along the Mediterranean coast without ever realizing that the driver is speaking a new language. This is a fact that often surprises people, used as we are to the neatness of the concept of a language. Officially, one goes through Spanish, Catalan, Occitan, Provençal, and Italian along such a journey, but in reality one just gradually slides through a continuum of minor accent changes until the cab lets out at the Trevi Fountain and they realize they've crossed two national borders. 

Simply put, one may never define an entire language, no more than they may define the entirety of all language. To create a perfect system would be to either see it fall apart in the hands and mouths of its users, feeling the pain of a thousand MLA officers, or to hide it away from the world like a precious relic. Either way, one fails to enjoy the communal identity that is, after all, the only lasting or meaningful feature of a language. If one truly endeavors to create language, one must first begin to use it, and embrace its usage. As we move forward into the adventure of language construction, what we must remember is that we are above all creating a code of communication, an identity among users, and that what matters in the end is the exchange. Ennius, the father of Latin poetry, once claimed to have three hearts, one each for his mastery of Latin, Greek, and Oscan. To create a language, then, is to create a heart all one's own.