Maybe you’ve wondered about why speakers of a certain native language, say Portuguese, always seem to have the same specific stumbling blocks when pronouncing certain sounds within a second language, like English. This can be true even decades after they have learned the new language and achieved functional fluency. In fact, it has a lot to do with why we are able to identify charming accents between language pairs. Only a very deliberate study and practice with a language coach can normally overcome these pronunciation tendencies. All this relates to electric eels… how?.
It’s been four decades since I was burying myself within anything related to theoretical linguistics that I could find. It was an exciting and concentrated study for me, and an interesting moment for the subject. Ten years after Chomsky’s syntactical revolution, it had become the fertile interdisciplinary crossroads between numerous other fields like cognitive psychology, philosophy, anthropology, computer science, language universals, epistemology, sociolinguistics, and theories of child development. The core competency within linguistics proper hasn’t changed much since those days, being thought to consist of three central elements: phonology, syntax, and semantics. This article focuses upon phonology, which is the study of the sounds within world languages.
English, especially American English, is known for its less than obvious correlation between the way words are spelled and pronounced. Compared to French, for example, there are many more puzzling ‘exceptions to the rule’ pronunciation tips to learn and memorize when a non-native speaker is first trying to advance from reader to speaker. French may have more difficult sounds to produce, but the sound of a word follows fairly well from it’s spelling, once you know the rules. The causes behind this disparity between orthography (written representation) and phonetics (sound representation) have to do with two historical factors, for languages are constantly changing. The first factor is that modern English has resulted mostly from a collision of a dialect having Germanic roots with strong Latin influences via French about 1000 years ago. (On top of this English has always borrowed vocabulary liberally from numerous other languages, with entirely foreign sound structures.) The other factor is that strong pronunciation shifts have taken place within English itself over time, and spellings which may have made more sense before the Renaissance do not really look sensible now.
English is not alone in this. Many languages have some sort of difficulty in representing all possible combinations of their sounds within the limits of their chosen alphabets. Phonologists have developed their own set of symbols to indicate language sounds, known as the International Phonetic Alphabet (IPA). As implied by the name it is intended to be flexible and universal enough to describe any sound encountered by a field linguist anywhere, whether they are researching hackney accents in Britain or previously unknown tribes in Papua New Guinea. The unifying requirement is that human physiology defines the vocal tract.
You can imagine a continuous stream of air travelling upwards from the lungs passing the various potential points of articulation enroute to the outer world. In spatial order, these locations are: pharynx, uvula, velum, palate, alveolar ridge, teeth, and lips. The tongue is used to ‘stop’ the air stream at (or in between) one of these points, thus producing a characteristic consonantal sound. Numerous variations are possible. The tip of the tongue may be used, or the middle surface of it, or the sides in case of L-like sounds. The obstruction can be definite or mushy as with sibilants (S) or fricatives (V), and can be accompanied or not by vibration of the vocal cords. When some type of obstruction like this occurs, what is produced is a consonant. When no obstruction at all occurs but the sound is merely molded by the shape of the mouth and oral cavity and allowed to resonate freely, we get vowels. There are other more exotic speech sounds which do not rely upon an air stream originating from the lungs at all, such as clicks and ejectives. These occur in African and aboriginal Australian languages. (If you’ve never heard them you’re in for a treat.)
The point is that there is a very large number of possible distinct sound points within the phonological space afforded by the human vocal tract. If you realize that the exact point and manner of articulation (or vowel cavity shape) can differ minutely between different languages, you will see that in fact there are an infinite amount of unique potential human vocal sounds. Linguistic convention involves speakers and hearers perceiving that a characteristic narrow range of variation of relevant articulatory factors forms the acceptable boundary line of what is and is not a K, for example. In India, a Marathi K will sound slightly different than a Spanish K in Mexico City. Sometimes the difference between similar sounds will be enough such that a trained phonologist would use different IPA symbols to designate the sounds they recognize across languages. But a native speaker might not even notice the difference or consider it an important distinction because no words in his language are ever distinguished by this difference. A difference which is crucial in one language can often be absent in another language; the speaker’s ear has become psychologically deaf to it. A truly hilarious parody of this situation is given by Steve Martin in his infamous “Damburger” scene in one of his Inspector Clouseau films.
Phones, Allophones, and Phonemes
At this point it is useful to define some terms. First, at the most universal and indeed most physiological level, we have the basic concept of a specific sound produced by the human vocal tract during an utterance, according to how and where and to what degree the sound is obstructed, i.e. articulated. This is known as a phone. Think of it as a raw language sound phenomena being heard for the first time in some language you are completely unfamiliar with before hand. A trained phonologist could identify this by placing it within the universal phone space, describe its articulatory parameters, and determine which commonly accepted IPA symbol most closely corresponds to it. (This fascinating and recommended interactive wiki page will let you place yourself in this position and hear spoken examples of phones organized by how they are articulated.)
The next step in the field analysis would be to painstakingly determine the total set of phones which occur within the target language. On the way you would naturally collect a list of known words, what they mean, and what the acceptable variations for pronouncing them are among the speakers of the language. This is actually a process of eliminating numerous phones from the universal set of possible human spoken sounds, so that only the set belonging to the language in question remains. But now you will still likely have more phones than needed to understand the language, because some spatially adjacent clusters of sounds will contain universal phones (sometimes uttered in the language) which do not really register upon the native speaker’s inner radar when distinguishing sounds. This is nicely illustrated in this chart of English consonant sounds. Consider the first two items, the English words part and place. The initial P sounds are actually different from a linguistic (phonological) point of view. The first p is aspirated, meaning it is accompanied by an audible puff of air, or what is known as a plosive. But the second p lacks this aspiration. The English speaker pays no attention to this distinction, but within Hindi it becomes highly relevant because two words in Hindi with completely different meanings often differ only by the presence of this aspirated or unaspirated p sound.
Allophones then, are two or more phones within a language which although distinct in terms of articulation are nonetheless heard as the same by native speakers. This must be true in all cases within the language. Thus, in English, aspirated and unaspirated p are merely allophonic variants of a deeper phonological entity within the language, which all speakers recognize, namely the phoneme /p/. One language’s allophone may be another language’s phoneme. A phoneme is a speech sound that is capable of changing the meaning of a word. Linguists use brackets when transcribing speech sounds at the ‘raw’ level of phones and allophones. They use slashes when transcribing speech at the phonemic level.
A given language’s phoneme set, abstracted from the universe of possible human speech sounds, defines its character and quality in much the same way as does the semantic choices made by the speaker community to abstract a set of vocabulary words (lexemes) from the space of universal meanings. The phonology or sound of a language is like its physical body or external manifestation, whereas the meanings and expressions are like its soul or inner aspect. There are further complexities connected with the way a language sounds, such as its tones and shifting accent patterns, but the point holds: the sounds of a language are extremely characteristic of it. (I knew someone years ago while a student who used to purchase unfamiliar language learning albums, Japanese and Swahili for example, and listen to them in the evenings as music.)
The English language has an alphabet of 26 letters and a phoneme set of about 44 items; there are 24 consonantal phonemes and around twenty vowels depending upon how one analyzes the diphthongs. According to one contemporary definition: ‘A phoneme is the smallest unit of sound which is significant in a language’. There are two aspects of this significance, namely what is perceived by the speakers aurally and what differentiates meaning (semantic content). It is easy to show why some pairs of English sounds are in fact phonemes, but for other pairs it takes some thinking and hunting. For example, consider the English pair of alveolar voiced and unvoiced stops, commonly known as d and t:
/d/ doom /dum/
/t/ tomb /tum/
Thus, the phoneme sequence for these two entirely different words are identical except for the initial /d/ or /t/. Or consider the pair of velar voiced and unvoiced stops, commonly known as g and k:
/g/ ghoul /gul/
/k/ cool /kul/
A more difficult phoneme pair to validate in English are the voiced and unvoiced post-alveolar fricatives, sounds sometimes represented by zh and sh respectively; their IPA symbols are /ʒ/ and /ʃ/. These are described as post-alveolar because their articulation points are located slightly behind the points in the mouth where the alveolar fricatives, /z/ and /s/ are sounded. They are called fricatives because there is friction, evident in the hissing of air, between the tongue and the alveolar ridge. While the unvoiced sh sound is very common in English, the voiced zh is much more atypical. The zh sound in English occurs in the middle of the word pleasure; it is very similar to the initial consonant sound in the French word jolie. The trick is to find a pair of words identical in pronunciation except for sh and zh — which is exactly why the following question appeared among many others on one of my linguistics finals back in the day.
Cite an example to show that /ʃ/ and /ʒ/ are distinct phonemes in American English (5 points).
And being cute, I was able to go one better than merely citing the two words; instead I included them within a sentence:
/du əlɜktrɪk ilz ʃɑk ʒɑk kʊsto/
Do electric eels shock Jacques Cousteau?
I got extra points for humor with this answer, but to be fair, I’d admit that some of us tended to thumb our noses up at phonemic justifications which depended upon proper names, especially when borrowed from French. For this reason I’ll mention the word pair mesher and measure. (I had to look it up too; it’s a type of medical device used in skin grafting. Have a look.)
I leave you with a few phonology tidbits. The Polynesian language family is known for having extremely curtailed phoneme sets. Hawaiian, as an example, has only 13 phonemes, although there is some dispute as to whether or not to count doubled vowels as separate phonemes. Yet, it sounds gorgeously mellifluous — so variety isn’t everything! Among world languages having the highest phoneme counts are those occurring in the Caucasus mountains, where one called Ubykh had 89. Imagine trying to learn that as a second tongue? Lithuanian is also up there, having 77 phonemes. Finally, an optional puzzle. Consider theta and thorn, the symbols for the voiced (/θ/) and unvoiced (/ð/) dental fricatives, which are normally spelled with a th in English words. These sounds have a rich history stretching back to Old English and Beowulf is liberally sprinkled with them. Interestingly, they tend to present considerable difficulties to native speakers of languages besides English. Can you find a word pairing which establishes them, voiced and unvoiced, as phonemes?
If interested, you can read another Linguistics-oriented article here. Hint, from an imaginary Shakespeare line: “Reveal not more to me, of thy thigh, fairest lady. Unless you wouldst control my heart and make of me your brain-fogged servant.”
► Handy INDEX — scan through all available articles