Version of 2010-06-15

Wersja polska • Bilanguage version • Wersja dwujęzyczna

Grzegorz Jagodziński

How to Construct a Language

1 – 2 – 3 – 4 – 5 – 6 – 7 – 8 – 9

A look at linguistic typology

Sounds

Phones and phonemes

The study of speech sounds is called phonology. This branch of linguistics includes phonetics and phonemics (however, sometimes phonology is understood as phonemics only). Phonetics examines all the speech sounds which are possible in a given the language. The way of producing sounds is the subject of phonetics as well. Phonemics is the study of the sound system instead. One of its subjects is the function of each single sound in the system.

To make it clearer, I will explain two important terms: phone and phoneme. Any single speech sound, considered as a physical event, is a phone. A set of similar sounds which are heard as the same sound by native speakers is a phoneme. Let’s take two English words: pin and spin. Both contain the phoneme /p/. But this /p/ in pin is slightly different than that /p/ in spin. Namely, /p/ in pin has an h-like addition or a breath puff while /p/ in spin is pronounced without such an addition. We call the puff aspiration and hence we can talk about two varieties of /p/: aspirated and non-aspirated. Both are considered to be the same sound – so both are varieties of the same phoneme /p/. But the aspirated [pʰ] (like in pin) and the non-aspirated [p] are two different phones. Of course, phones are the subject of phonetics while phonemes are the subject of phonemics.

If given phones belong to the same phoneme (like [pʰ] and [p] in English), we call them allophones of this phoneme. In typical instances, allophones are positional variants of the phoneme. In our example, the allophone [pʰ] occurs initially (pin) while the allophone [p] occurs after /s/ (spin). Instead, different phonemes must be able and enough to distinguish one word from another. It means that each words which are similar but different must contain at least one different phoneme. For instance, pin and bin are two different words which create a minimal pair – they differ in only one place. The former begins with [pʰ] while the latter begins with [b]. These phones cannot be allophones of the same phoneme. Indeed, they occur in the same position. So, we say that [pʰ] is an allophone of /p/ while [b] is an allophone of /b/.

Of course some phonemes have two or more allophones while others have only one allophone. If we cannot show any minimal pair with two similar, examined sounds, we can suppose that these sounds are allophones and not different phonemes. But while in one language such a minimal pair cannot be found, we will perhaps find it in another language. For example, English phones [pʰ] and [p] are allophones because they cannot distinguish words. But in Old Greek, [pʰ] and [p] were not allophones! Instead, they could form minimal pairs, ex. πως [po:s] ‘somehow’ : φως [pʰo:s] ‘light, luminosity’, and hence these phones belonged to two different phonemes ([o:] means long [o]).

If you are just a language student, take your consideration that the sounds which are only positional varieties in one language can be found completely different sounds in another language. And if you are a language maker, you can use this fact. For example, you can invent a language where [p] and [b] are only positional variants (allophones). The same about more distant sounds, like [l] and [r] (for your interest: actually, there are languages where [l] and [r] are only positional variants). And, if your language is to have many sounds and you do not know which sounds would be good for this, you can promote, let’s say, English allophones (like [pʰ] and [p]) to different phonemes in your language. Finally, you can invent allophones of some phonemes of your language. They would make your grammar really naturalistic.

back to contents

Transcription

I am sure that you have already noticed how we mark phonemes and phones (if not sure, take a look at the vocabulary). Some languages, including English, have rather complicated spelling, and using just letters for writing sounds may be a bad idea. That is why linguists use various types of phonetic and phonemic transcription.

Among phonetic / phonemic transcription, the one using IPA symbols is the most wide. You can acquaint yourself with IPA transcription systems for English and next study Peter Lagefoged’s Sounds of Languages (the chart of IPA sounds together with recordings of each of them may be of your special interest). But you should know that sometimes, in some languages, IPA symbols are not too comfortable, for instance because of lack of single symbols for affricates, and hence the IPA solution may not be the solution you need. Indeed, special transcriptions are in use for Slavic languages, Georgian, Sanskrit, Arabic etc. You must be careful that some of symbols which are used there can have different values than in the IPA transcription.

If you are a language maker, you can use the IPA transcription or a transcription of your own. But in the latter instance, you should use only well-defined and univocal symbols (for example, g like in get and not like in gem, while ǧ = g like in gem). You should also avoid duplication, i.e. do not use two symbols for one sound (g in gem and j in jet represent the same sound, and hence ǧ can be used in both words).

back to contents

Consonants

Consonantal features

Consonants are formed by obstructing – in such or another way – the air stream. All possible consonants can be classified with various criteria:

1. Place of articulation is the place where the obstruction occurs. It depends on both passive and active organs of speech (the tongue). In this dimension, consonants can be:
   • labial – articulated with the lips (p, b, m, w),
   • labiodental – articulated with the lips and the teeth (f, v),
   • linguolabial – articulated with the tongue touching the lips (written t̼, d̼, extremely rare, known only in one or two languages),
   • interdental – articulated between the lower and upper teeth (th /θ/ in think, th /ð/ in this, Spanish z /θ/),
   • dorsal-dental – articulated with the front part of the back of the tongue and the teeth (French, Russian or Polish t, d, s, z, n),
   • apical-dental – articulated with the very tip of the tongue and the teeth (Spanish or German t, d, German s, z, n),
   • dorsal-alveolar narrow – articulated with the front part of the back of the tongue and the gums (German s, z, n, l – variant),
   • dorsal-alveolar wide – articulated like the previous but with wider opening (French ch, j, south German sch /ʃ/, g /ʒ/ in Genie),
   • apical-alveolar – articulated with the tip of the tongue and the gums, behind the teeth (English s, z, t, d, l, n),
   • coronal-alveolar narrow – articulated with the tongue forming like if a spoon and the teeth (Spanish s, German or Polish l),
   • coronal-alveolar wide – with wider opening than the previous (German sch, Russian sh, zh, Polish sz, ż, cz, dż),
   • cacuminal-alveolar – articulated with the tip of the tongue raised toward the palate behind the gums (Spanish n, l),
   • dorsal-postalveolar – articulated with the back of the tongue and the place behind the gums (Russian ch, shch),
   • apical-postalveolar – articulated with the back of tongue and the place behind the gums (English ch, j, sh, zh = s in pleasure),
   • coronal-postalveolar – articulated with the edges of the tongue, behind the gums (English r /ɹ/, Russian and Polish r),
   • retroflex-postalveolar – articulated with the tongue raised and bent slightly backwards toward the hard palate (American r),
   • alveolo-palatal (prepalatal) – articulated yet further back the teeth with the back of the tongue (Polish ś, ź, ć, dź, ń),
   • palatal – articulated with the back of the tongue and the palate (English y, German (i)ch, Spanish ch, y, ll, ñ, Polish j),
   • palato-velar (prevelar) – articulated at the place between the palate and the soft palate (velum: Polish ki, gi, chi),
   • velar – articulated with the back of the tongue and the velum (k, g, ng),
   • velar-labial – articulated as two simultaneous consonants (kp, gb in some African languages),
   • velar-alveolar – as above (Swedish has a sound that can be described as simultaneous sh and German ch),
   • uvular – articulated with the back of the tongue and the uvula (Scottish and German ch, Spanish j, Arabic q, ḫ, ġ, German r),
   • pharyngeal – articulated with the root of the tongue and the back face of the throat (Arabic ḥ, ʕ),
   • epiglottal – articulated with the epiglottis (ʜ, ʢ, ʡ in Agul which is a northeast Caucasian language),
   • glottal – articulated in the larynx (h, glottal stop).
2. Degree of closure lets us divide consonants into:
   • stops – produced by entirely stopping and then releasing the air stream (English p, b, t, d, k, g),
   • affricates – similar to stops but with fricative release of the air stream (English ch, j),
   • lateral affricates – stops with lateral release of the air stream (tl in many Amerindian languages),
   • fricatives – produced by forcing the air stream through a narrow passage which forms a noisy air stream (English f, v, th, s, z),
   • lateral fricatives – with the air stream escaping through a narrow passage along one or both sides of the tongue (some Bantu),
   • lateral approximants (lateral sonants) – with the air stream escaping along one or both sides of the tongue (English l),
   • trills – articulated with the lips, the tip of the tongue, or the uvula set into vibration by the air stream (Polish, Scottish or Spanish r),
   • taps (flaps) – similar to trill but with only one single rapid contact (or: a single vibration: Spanish r),
   • nasal sonants – with the air going out through the nose and blocked from going through the mouth (Eng. m, n, ng),
   • approximants – similar to fricatives but with a little more open passage between speech organs (Eng. y, w, r).
3. Voicing depends on whether the vocal cords vibrate or not
   • voiceless – articulated without simultaneous vibration of the vocal cords (Eng. t, ch, s are always voiceless),
   • voiced – produced with vibration of the vocal cords (Eng. d, j, z are voiced in most positions).
4. Force depends on the degree of straining of the vocal cords. See below for more information on voicing and force
   • strong – articulated with stiff vocal cords (Eng. t, ch, s are always strong),
   • weak – articulated with neutral or slack vocal cords (Eng. d, j, z are always weak).
5. Aspiration depends on whether the opening between vocal cords (glottis) is spread or not
   • aspirated – pronounced with spread glottis, with a h-like puff of breath (Eng. t and p in top are aspirated),
   • non aspirated – pronounced without aspiration (Eng. t in stop is not aspirated).
6. Nasalization depends on whether the air travels through the nose or not
   • nasal – articulated with air going out through the nose (Eng. m, n, ng are nasal),
   • oral – articulated with air going only through the mouth (Eng. b, d, g are oral).
7. Palatalization depends on whether the back of the tongue is raised toward the palate or not
   • palatalized (soft) – articulated with the tongue raised towards the palate (Irish and Slavic languages have palatalized sounds),
   • non palatalized – self-evident (Eng. consonants are non palatalized or their palatalization is not important).
8. Velarization depends on whether the back of the tongue is raised toward the velum or not
   • velarized (hard) – articulated with the tongue raised towards the velum (Eng. w, “dark” l, and also in Irish, Arabic or Russian),
   • non velarized – self-evident (other Eng. consonants are non velarized).
9. Pharyngealization depends on whether the root of the tongue is raised towards the back face of the throat or not
   • pharyngealized – articulated with the tongue raised towards the back face of the throat (e.g. in some Turkish languages),
   • non pharyngealized – self-evident (Eng. consonants are non pharyngealized).
10. Labialization depends on whether the lips are rounded or not during pronunciation of the sound
    • labialized (rounded) – articulated with rounded lips (in some Caucasian or Semitic languages),
    • non labialized – self-evident (Eng. consonants are non labialized).
11. Glottalization describes some types of so called non-pulmonic consonants
    • ejective (abruptive) – a voiceless sound pronounced with the simultaneous glottal stop (ex. in Georgian),
    • implosive – a voiced stop pronounced with the sudden rush of air into the pharynx (in some African languages),
    • click – a stop sound pronounced by touching the tongue to the roof of the mouth, drawing in the breath, and abruptly withdrawing the tongue from the roof of the mouth (in some African languages and in English tsk),
    • breathy voiced – pronounced with excessive emission of breath (in some languages),
    • creaky voiced – pronounced with a harsh or shrill (in some languages),
    • not glottalized – self-evident (all Eng. consonants in the normal speech are not glottalized).
12. Length is the feature of vowels but sometimes concerns consonants as well; long, geminates, or double consonants are present in Italian, Swedish, Finnish, Hungarian, Japanese, Korean, Avar (a northeast Caucasian language) and their exact nature can vary in different languages.

Now it is the time for learning consonantal systems in several languages. If you are a language maker and you have now idea how to start, the best idea for your language is to borrow the given patterns, even as the whole. The next step is to combine elements of different patterns.

back to contents

The consonantal system in English

Note my remarks concerning English consonants:

1. Stops, affricates and most fricatives occur in pairs. The sound to the left is strong (s.) while the sound to the right is weak (w.). It is not true that English distinguishes voiceless and voiced consonants: ex. /d/ is voiced in dog but voiceless in bad. Even if both bat and bad contains voiceless stops in the final position, they are still different. The basic difference is in force, not in voicing then.
2. Strong consonants are always voiceless. The glottal fricative /h/ does not have a counterpart and it can be voiceless or voiced.
3. All sonorants, i.e. nasals and approximants (incl. the lateral /l/) are voiced and weak.
4. IPA phonetic symbols are not always the same as popular symbols based on spelling. Especially, the symbol /j/ denotes y, never j. Actually the letter j marks the sound /j/ (= Eng. y) in many languages like (Medieval) Latin, German or Polish. Some people think that English y is not an approximant but a weak fricative. In such case, the symbol /ʝ/ instead of /j/ should be used.
5. When no IPA symbol is given, it means that the symbol is just the letter.
6. Strong consonants are pronounced with stiff vocal cords while weak consonants – with neutral or slack vocal cords.
7. Strong stops have two basic allophones:
   • aspirated [pʰ tʰ kʰ] occur when /p t k/ stay single before a stressed vowel like in pot, table, appear, or if they are followed by a sonorant like in play, pray, clean, tune (pronounced “tyoon”), twig, queen (the aspiration marked with [ʰ] is weaker and shorter than the glottal fricative /h/); aspirated [pʰ tʰ kʰ] also occur before a pause like in pot, cap, pack (pre-aspirate [ʰp ʰt ʰk] are dialectal),
   • non-aspirated [p t k] occur elsewhere, ex. after /s/ like in spy, stop, sky, in the cluster /tɹ/ like in train etc.
8. Do not think that weak consonants are always voiced! In fact, weak stops, affricates and fricatives have some basic allophones:
   • the fully voiced variants [b d g ʤ v ð z ʒ] (with totally slack and vibrating vocal cords) can occur only between two vowels or a vowel and a sonorant or another weak consonant, like in table, order, ago etc,
   • the voiceless or fully “devoiced” allophones [b̥ d̥ g̥ ʤ̥ v̥ ð̥ z̥ ʒ̥] occur before a strong consonant like in absolutely, big tree etc, in the final position weak stops can also be almost completely voiceless, also in clusters, ex. dogs /dogz/,
   • in other positions, including word-initial, weak consonants are only partially voiced – they are voiceless from the side of a pause or another voiceless element: hence the best way to transcribe bad is [b̥bædd̥].
9. The lateral /l/ has two different allophones, at least in British English:
   • the “light” allophone [l] which occurs before a vowel,
   • the “dark” allophone [ɫ] which occurs before a consonant or word-finally, and which is velarized.
10. The approximant /w/ can be considered to be a velarized labial, a labialized velar, or simultaneously labial and velar.
11. Some English speakers still distinguish /w/ from /ʍ/ which is its strong and voiceless counterpart. The sound /ʍ/ is spelled wh and it should be placed in the table to the cell left to /w/.
12. The strong alveolar stop /t/ can have the allophone [ɾ] (tap) between vowels. The British like pronouncing the glottal stop [ʔ] for /t/ at the end of a syllable. Hence they pronounce cat as [kʰæʔ], not [kʰætʰ].
13. The alveolar approximant /ɹ/ has some varieties, ex. in Scottish English it can be tap – [ɾ] or even trill – [r]. Dialectal variants are a very special types of allophones which occur instead of other allophones in the speech of different speakers.
14. Many sounds can have secondary positional variants ex. m before f (labiodental) or l before th (dental, not alveolar). But these are minor, secondary variants, which can occur in other languages too.

When you are a serious language investigator or maker, do not treat my notices on weak consonants with disregard. Romance and especially Slavic languages use true voiced consonants in many (or all) positions while English knows them only between vowels. In Slavic languages the vocal cords can be either stiff or totally slack and that is why Slavic consonants can be either voiceless or fully voiced. English does not belong to this type of languages. That is why you should term the English p, t, k, etc. strong not voiceless, and b, d, g, etc. weak not voiced.

back to contents

The consonantal system in German

The German language is related to English. English, Frisian, Dutch, Afrikaans, German, Yiddish, Danish, Swedish, Norwegian, Faroese and Icelandic form the Germanic branch of languages.

Remarks:

1. In German there exist strong and weak consonants, the same as in English. All strong consonants are voiceless while weak consonants are voiceless or voiced (in similar conditions as in English). We can even say that in German, esp. in southern dialects, weak consonants are voiced less frequently than in English.
2. Weak consonants cannot occur word-finally, unlike in English. Even if there is a symbol for a weak consonant in spelling, the final consonant must be voiceless and strong in German, e.g. Tag /tɑ:k/ or phonetically [tʰɑ:kʰ].
3. Similarly like in English, strong stops /p/, /t/, /k/ are voiceless and aspirated in the beginning of a stressed syllable (not when preceded by /s/, /ʃ/) before a vowel or /l/, /n/, /ʀ/, as well as word-finally. The voiceless non-aspirated allophones are pronounced elsewhere.
4. German affricates /pf/, /ts/ (spelt z or tz) have no weak counterparts; we can also take them as the clusters of two consonants.
5. The weak labiodental fricative /v/ is spelt w while the letter v denotes /f/ as a rule.
6. German has no interdental consonants; German /t d ʦ s z n/ are dental (or postdental) but not alveolar (like English ones).
7. The strong alveolar fricative /ʃ/ is coronal and hence it sounds different than the English alveolar fricative /s/ which is apical. Coronal sounds are pronounced with the tongue forming a spoon-like shape (they are pronounced with the edge of the tongue) why apical sounds are pronounced with the tip of the tongue. We can also say that coronal sounds are wide while apical ones are narrow. Even if some sources name this consonant alveolo-palatal, it is not correct and German sch is different from English sh. In addition, it is remarkably labialized (it is pronounced with rounded lips).
8. The weak alveolar fricative /ʒ/ is rare in German, it can only happen in foreign words like Genie, Garage etc.
9. The uvular trill /ʀ/ is the most frequent but there also exist different ways to pronounce the letter r. The other variants are the alveolar trill /r/ and the uvular weak fricative /ʁ/. Individual persons may prefer one of these three possibilities (/r ʀ ʁ/). In come contexts, this consonant may be changed into the vowel /ɐ/.
10. The letter j denotes the weak palatal fricative consonant, not an approximant (like English y). We use the symbol /ʝ/ for such a sound.
11. The sequence ch denotes either the strong counterpart of j (the sound is called the ich-sound) or the strong uvular fricative (which is called the ach-sound). It is hard to say whether they are allophones of one phoneme or not. The palatal sound /ç/ occurs after front vowels (ex. ich /ɪç/), l (ex. Milch /mɪlç/), r (ex. durch /dʊʀç/), but also word-initially before a front vowel (ex. China /çi:nɑ/) and is the suffixes -chen /çɛn/ and -ig /ɪç/. The uvular sound /χ/ occurs elsewhere. English speakers who pronounce loch without /k/ but with a special sound are familiar with it. The same about Spanish speakers – the sound spelt with the letter j in this language (or with g in the groups ge, gi) is identical to the German ach-sound.