# Sound system

Wilhelm von Humboldt · developing English translation

English-reading review pending; German fidelity remains provisional.

Edition snapshot: 3e1f30ab374637421861f911f8d42c207b01908b

Source: GS V, Leitzmann 1906. https://github.com/BrettRey/humboldt-grundzuege-translation

## b. The procedure of language in forming discourse.

## α. Sound system.

**56** The influence of the sound system extends throughout language, and a careful study of all its peculiarities is the foundation of every precise investigation of language. Without complete knowledge of the treatment of sounds and of the relations and affinities of the letters, which in each language are subject to different laws and habits, the explanation of grammatical forms and the engagement with the language system as a whole always remain deficient. (29.)[^sound-letters] The intelligibility, emphasis and euphony of discourse rest directly on the sound system, but it also affects the pure and refined intellectual character of language; in this respect, its importance has perhaps not yet been sufficiently appreciated. Yet this importance becomes evident as soon as one recalls that (21.) the whole concept of language lies in that of articulated sound, and that the way each language grasps this concept is expressed in its alphabet, as the system of its articulated sounds.

**57** The first step is to separate sound from all accompanying noise, thereby reducing it to a true speech sound. It should evoke only the concept from which it springs, and must possess no more body than is required for that purpose. It is precisely with the least body that its distinctive character emerges most sharply. As the mind penetrates animal sound, the latter becomes articulated; yet it does not in every language (28.) wholly and purely shed its original nature, but mixes clicking, blaring, hissing, rasping or breathing into the sounds. This is found in great crudity in some languages, particularly African ones, but in finer traces even in highly developed languages.[^sound-clicks] Dullness of the mental impulse towards language and physical roughness of the organs may work together in such cases; but mind and organs refine themselves through their reciprocal influence, and the roughness of the sounds is therefore often gradually worn smooth as languages develop over time.

**58** But even genuine speech sounds, purified of accompanying noise, must be distinctly separated from one another, without one running into another. The letters do not lie, as it were, distinctly separated in the speech organs. Rather, we must conceive of the movement of those organs that brings a particular kind of sound into being as a continuum: at any point in it, suddenly fixing the movement would give the sound a slightly different character, depending, for example, on whether the lips open more or less, the tongue strikes more quickly or slowly, more firmly or lightly, with a narrower or broader surface, and so forth. Each sound thus has a sphere of its own, within which it occupies a definite point, but by no means the only possible one. The habit taking shape within the nation assigns this point to it; this is how the nation’s particular alphabet comes into being, and good or ill fortune may preside over its emergence.

**59** The influence of a lack of clear separation between sounds, however, extends far beyond the mere danger of misunderstanding. Articulated sound is too closely akin to the mind for its correct or faulty formation not to exert a corresponding influence on it time and again. This influence in turn shows itself in the mind’s reaction, giving rise to an interaction that continues to infinity: the mind always brings its purer nature into it victoriously, yet its ascendancy is repeatedly checked to some extent by the resistant material. The pure definiteness, felicitous clarity and penetrating sharpness of a nation’s concepts thus depend very greatly on the clear separation of the basic sounds of its alphabet.[^sound-conceptual-inference]

**60** Even in the use of articulated sounds alone, without any musical or poetic connection, there is an aesthetic feeling that impels one nation to let each sound ring out fully and brightly, giving it its full due, while another nation, lacking this feeling or subordinating it to other considerations, such as the force of accent in English, swallows, fuses or obscures many sounds. With regard to concepts, the same can be said of this as of the lack of proper separation. What then matters is how the quietly shining clarity that concepts lose in this way is made up for by other means.

**61** The use of language develops in the mind a sense for fine distinctions of sound and a delight in them, all the more so to the extent that a particular language correctly employs a well-chosen sound system. Since discourse also needs many distinctions in order to express and evoke concepts in all their relations and shades, this sense serves that need very well. For an inclination and an ability now emerge to mark these differences by changing individual letters, and to recognize them by those changes. In this lies the origin of all inflection, whether one holds it to have arisen at the same time as language or later within it. This brings a double advantage: it gives the idea only as much body as it needs, and, by means of the letter, which is entirely meaningless in itself, avoids every misleading subsidiary concept.[^sound-meaningless] Language contains a striving towards the purification of its intellectual character, a striving that runs through all its functions and continually intensifies. Just as it separates accompanying noise from speech sound (57.), so it excludes from what is thought through language the representation of any material that the thought does not require, and displaces that material with meaningless sound wherever it seeks to appear as a sign of form. Thus all languages formed in a way more appropriate to the mind, when compared with others that are their opposites in this respect, have the property of evoking a smaller mass of material in expressing the same idea, thereby allowing that idea to emerge more freely and in a more purified form.

**62** A natural and almost necessary consequence of the clear separation of sounds[^sound-source-deleted-near] is that it increases the range of the alphabet. For sounds that the ear could easily confuse in indistinct, undeveloped pronunciation now stand distinctly apart. Sometimes, however, differences in the range of the alphabet also arise from particular sound peculiarities and habits of nations, through which individual letters, such as the Polish l with a stroke, and even whole classes of letters, such as the Indian cerebrals or linguals[^sound-source-bopp-62], are added to it.[^sound-cerebral] The position and movement of the speech organs, and the fact that individual sounds fall within the same or different spheres (58.), mean that the letters have mutual connections and relations, and that a more or less complete system of sounds can be discovered in the alphabet. Once writing has come into being, and grammatical study has been added to it, an alphabet can acquire in this respect a more firmly defined and finely developed system than was originally present in the nation’s language use. This may have happened to the Sanskrit alphabet, whose pattern has spread to the languages still living in a large part of South Asia. Yet in its totality such a system always lies in nations’ linguistic dispositions and habits, and the definiteness it admittedly gains through well-chosen written signs acts back upon speaking and becomes the nations’ true possession.

**62a** When several letters are combined into a syllable, new peculiarities of the sound system become apparent in the way languages favour the meeting of some letters, avoid that of others, and exclude a number of syllables. One must therefore attend not only to the range of the alphabet, but also to the range of syllables possible through it. It is noteworthy that it is precisely the languages of savages, which are commonly accused of roughness, that exclude many syllables tolerated in highly developed languages. This may derive in part from the more readily excited sensitivity of these peoples’ senses, already noted above, and in part from the fact that a more developed language often contracts syllables where a language in a more original state lets them follow one another separately.

**62b** Word formation too is subject to the rules of the sound system, since in many languages words do not begin or end with certain letters, and their internal structure is also determined by laws of language, particularly in the changing of consonants that meet.

**63** In these systems lie the grounds and conditions of letter change, which is likewise only partly given by the more general nature of the speech organs and otherwise rests on the particular sound habits of each people. For although the nature of the letters determines the kind of change, a people’s sensitivity to this reciprocal influence of sounds upon contact depends on its ear and speech organs. The closer the sounds stand to one another in a highly systematic alphabet of considerable range, and the more clearly and distinctly their relation is recognized in such an alphabet, the more the inclination towards letter change grows; letter change in turn contributes greatly to finer distinctions of sound and to the systematization of the alphabet. One might suppose that the ear of uncultivated nations would fail to hear and would neglect these subtleties of sound. Yet it is precisely among them that one finds very frequent letter changes; this may be attributable to two causes: the senses of people closer to nature possess a keener sensitivity, and the uncultivated person views discourse, as it appears to the understanding[^sound-understanding], as a continuum whose parts are everywhere in contact, whereas the cultivated person, even in continuous discourse, uses the tongue to check the slurring of sounds in accordance with the pauses of the understanding. Yet the very frequent chaining together of several words by letter change and contraction is probably a true national peculiarity of the Indian peoples, for which no further reason can be given; at least, it is not evident whether other nations have known it to the same extent.[^sound-source-thiersch] At first sight, one is tempted to attribute the many rules that arise from this in Sanskrit, at least in part, to the grammarians’ excessive zeal for systematization. But this conjecture disappears when one compares the living Indian languages that have arisen from Sanskrit. Carey[^sound-source-carey] expressly states of Telinga[^sound-telugu], in which this chaining is especially frequent and difficult to recognize, that a foreigner who were to disregard it would not be understood at all by the natives. In Bengali, by contrast, it is not always strictly observed[^sound-source-haughton], making it still clearer how this point depends on the particular habits of the tribes.

**64** The most immediate influence of this peculiarity is naturally on euphony. It flatters the ear when the harsh meeting of mutually resistant and inharmonious sounds is avoided as far as possible. But the effect also extends to the conceptual force of language and to the proper articulation of discourse for the understanding. This happens, however, only when letter change confines itself to fusing parts of the same word into a whole, and does not, by stringing together whole words that are independent of one another, form a whole for the ear that cannot exist as such for the understanding. I can regard the latter habit as useful only insofar as it provides poetry with one more means of bringing divisions between words into an appropriate relation with rhythmic divisions. For it produces wholes of sound different from, and more frequently larger than, those provided by word division alone. By contrast, it disturbs both the understanding and the ear when a final letter changes in preparation for the one that follows it at a point where the understanding requires a logical pause and the ear a rhythmic one. If, however, this sound change remains within the bounds of the word, it becomes a phonetic means of dividing discourse into clearly separated wholes, and impresses upon the understanding, through the senses as well, that the element of discourse is the word. The logical articulation of discourse suffers no disruption; on occasion, namely whenever such a sound change occurs, it is made still more evident. Nor does Sanskrit fail to recognize this principle of the need to characterize words as wholes through a distinct unity of sound, since different rules apply within words from those that apply when two words meet. But since these rules agree in some respects, and those for several words also apply to the parts of compound words[^sound-source-bopp-64], word division is not made cleanly distinct and clear to the senses by this means.

**65** Quantity belongs to syllables. But since vowels determine the syllable, and can also form one on their own, the feeling for the syllable also affects the completeness of the alphabet, in which[^sound-source-deleted-signs] either all vowel sounds occur clearly distinguished, each in its long or short form, or they do not. Written signs alone, however, cannot decide this. Thus, in the case of Sanskrit, it remains uncertain whether the living language did not also possess a short e and o. Quantity lies wholly in the nature of pronunciation and is least amenable to arbitrary determination. Even individual dispositions and habits can exert comparatively little influence on it, since, for example, the coming together of several sounds must necessarily require more time in pronunciation. By contrast, disposition, habit and intention can favour or restrict the full sway of quantity in a language’s discourse; from this arises a difference of the greatest importance for the euphony and rhythmic beauty of discourse and of language generally, including their effect on character. The expression of thoughts is less fundamentally affected by this rhythmic peculiarity of language. Even excessive favour shown to quantity can never harm the expression of thoughts; unwarranted neglect can do so only insofar as a failure to distinguish long and short vowels carefully brings with it the same disadvantage as obscuring the pure sound of letters in general. On the other hand, suppressing the natural sway of quantity may well give the expression of thought an excessively one-sided prominence. For rhythm and accent stand in a relation that can become hostile to the former wherever proper handling of language fails to bring it into a beautiful balance, since quantity, in its quiet sway, does not always manage to assert its right alongside the vehemence of discourse or the sharply accentuating understanding. The predominance of aesthetic feeling, which can meet with obstacles in one-sidedness just as much as in crudity, is decisive here: it preserves its freedom[^sound-freedom] alongside the emphasis of the connection of thoughts, and assigns accent, even within the word, its place in keeping with rhythmic euphony.

**66** Just as quantity belongs for the most part to the particular nature of sounds, so accent belongs exclusively to the speaker’s intention.[^sound-intention] It is the means by which language joins and divides discourse into different coordinate and subordinate wholes; without it, discourse would consist merely of basic parts strung together. For by accenting one syllable[^sound-source-deleted-word], accent makes those that follow unaccented, thereby drawing them into the domain of the accented syllable and thus joining them to it. The same thing happens when syllables preceding the accent are strung together in such a way that their lack of accent alerts the ear to an accent still to come. Conversely, accent creates different accentual domains and separates the syllables by initiating a new accentuation. Here it follows the direction of thought immediately and serves to make the flow of discourse resemble the flow of thoughts; but it is conditioned by the laws of pronunciation and submits, though to a greater or lesser degree, to those of euphony. Accent in this sense must not, however, be confused with other changes in vowel sounds that are likewise indicated by accent marks and likewise consist in raising or lowering the voice. These are intended only to make possible, through differences of intonation, a greater number of sounds than the nature of those sounds without intonation can yield. This is the case with Chinese accent marks, since in Chinese there are, properly speaking, as many varieties of each vowel as there are intonations the vowel can take. For at bottom, the intonation of a syllable always falls only on its vowel.[^sound-vowel-tone] These intonations thus increase the range of the alphabet.[^sound-source-chinese]

**67** The most important kind of true accent, and the one that belongs most to language, is word accent. Discourse accent naturally varies with the character of the discourse and cannot attach to the parts of language outside connected discourse. But its treatment as a whole certainly belongs to the nature of the language. This treatment does lie in the nation’s characteristic way of forming representations and of feeling, but language can never be separated from that way either. In addition to its, as it were, dead elements, it always also consists in the distinctive character of living delivery that lies in the speakers’ souls; it is only that, in dead languages, this delivery cannot be recognized at all, or at least only imperfectly. For Sanskrit, all information about accentuation has so far been lacking; and although Greek accentuation has been preserved in its signs, many doubts remain about how the accents were actually delivered in living speech, in connection with discourse accent on the one hand and quantity on the other.[^sound-sanskrit-accent]

**68** The word comes into being only through accent, which breathes the unity of the concept into it.[^sound-word-unity] If two entirely equal accents occurred in a simple word, the unity of the concept would not be made perceptible to the ear as well, and the aptness of sound as an accompaniment to thought would be impaired. This would so offend the natural feeling for language that one could not conceive of such a case as possible. Thus, when two accents come together in a single word from two different words, one will be subordinated to the other in pronunciation, unless the concepts of the individual words fail to combine into one; for when they do combine, one element necessarily predominates.[^sound-combined-accent] Since accent, through its fixed position within the word, helps to complete the word’s individuality, different words can arise from the same sequence of syllables through differences of accentuation; nor should it cause surprise that this is found even in the languages of wholly uncultivated nations. It is a very common observation that nothing makes a foreigner so unintelligible to the commonest classes of the people as incorrect accentuation. It is precisely the uncultivated person, in nations and social ranks alike, who is accustomed to taking each word unanalysed as a whole and recognizing its physiognomy, if one may put it so, by its accent.

**69** Although several words are, properly speaking, joined together only by discourse accent, in some cases this also happens through true language accent; in that case, however, the joined parts do not retain their own accents unchanged, as they do in the former case. It is easy to see that I am referring here to enclitic words, which, however, are by no means confined to Greek: if one pays close attention, they exist, and must exist, in all languages. For in all languages there are cases in which words that stand on their own enter into such a relation to others that their independent standing is wholly absorbed in their attachment to those others. With this, their own accentuation also falls away of itself. In the sharply accented English language this can be heard with extraordinary force, indeed among the common people too, and in such a way that the entire sound value of the enclitic words is changed by it. How this may also cause the accent of the word that pulls the other to itself to shift position, or to acquire a second accent beside it, though probably not one of equal strength, depends on the rules and habits of individual languages; so too does whether the kind of words that can be attached is to have two sound forms, one orthotonic and one enclitic.[^sound-source-sanskrit-pronouns] Since Sanskrit added to word accent and discourse accent the exact preservation of quantity and the chaining of words through letter change, its sound system, given the already very great range of its alphabet, must have been especially difficult, but also extraordinarily rich and varied.

**70** Through discourse accent, the speaker can either mark precisely the weight that each word and each sentence has in the meaning of the discourse, or distribute the rises and falls of the voice according to a method that has become habitual in the language and is independent of this weighting; consideration for the sense is not wholly sacrificed to this method, but is nevertheless completely subordinated to it. This difference is most evident in German and French reading aloud and declamation, since the French accuse Germans of singing when they accent French in speech as they would their own language. Besides these two sharply opposed modes, however, there are many less distinctly differentiated ones, all distinguished by the greater or lesser prominence they give to the significance of words and syllables for the understanding, since each people presumably follows a feeling of its own in this. But precisely because the way the sequence of thoughts appears in living delivery has the most important influence, this point is of the greatest significance for intellectual cultivation, especially that of the common people, who are most restricted to living communication.

**71** The introduction of alphabetic writing has some of the greatest and most visible consequences for the development of the sound system. Although alphabetic writing never removes articulated sound from the realm of ever-vanishing audible sound, and gives it nothing but a sign that calls that audible sound to mind, these signs and their comparison nevertheless act upon the understanding and imagination in such a way that speech sounds attach themselves more definitely to their assigned points and present themselves to the soul much more clearly in their character as elements. Through the signs, the pronunciation of one generation passes into that of the next; deviations can be detected by reference to spelling, and speech sounds, together with everything that belongs to them, can become objects of reflection and scholarly study. Alphabetic writing thus brings two great advantages, from which many particular consequences subsequently extend throughout the language. First, all the sound relations of the language are set against one another and ordered more purely and distinctly, and are developed more completely and systematically. Second, as the understanding separates the basic parts of language more precisely and retains them more distinctly, the idea of language’s articulation into parts gains greater life within it, and the representation of its structure, rising from the basic parts upwards, gains in vividness and completeness. On the other hand, scholarly work on the signs for sounds can also lead to an excessively contrived and systematic treatment of them, especially where languages begin to pass from living interaction among the people into exclusively learned use.

## Terminology guide

**Letters and alphabet** (*Buchstabe*, *Alphabet*, *Buchstabenschrift*). A *letter* often means ‘a speech-sound element’, and *alphabet* ‘the inventory or system of such sounds’. They are not automatically written characters or a modern phoneme inventory. Where writing enters in 62 and 71, written signs are identified separately; *Buchstabenschrift* is *alphabetic writing*.

**Sound, articulation and noise** (*Ton*, *Laut*, *Sprachlaut*, *Geräusch*, *Schall*). *Ton* and *Laut* overlap as *sound*; articulated sound is shaped to carry thought. *Geräusch* is the accompanying noise Humboldt wants removed, though his evaluation of particular consonants is qualified in the note on 57. *Schall* in 71 is the transient acoustic occurrence to which the written sign recalls us.

**Mind, understanding and soul** (*Geist*, *Verstand*, *Seele*). These are *mind*, *understanding* and *soul* in this unit. The English keeps the recurring German terms recoverable without turning every lexical difference into a separate modern psychological mechanism. *Understanding* here names an intellectual capacity, not just successful comprehension of an utterance.

**Intellectual character** (*Intellectualitaet*, *Intellectualität*). Here the term concerns language’s intellectual character and its relation to thought. The earlier division used *intellectual life* where the bearer was a nation and its orientation; that contextual change does not establish different faculties. *Intellectual cultivation* in 70 renders a different expression, *intellectuelle Bildung*.

**Changes of sounds** (*Buchstabenveränderung*, *Buchstabenverwandlung*, *Lautveränderung*, *Lautverwandlung*). The German uses several overlapping expressions for alterations of speech sounds. The contexts include inflection, changes inside words and changes when words meet; these are not automatically historical sound laws or merely spelling changes. The German forms remain available in the linked record.

**Word linking and contraction** (*Verkettung*, *Wortverkettung*, *Zusammenziehung*). Words can be linked by changes at their boundaries and by contraction. Humboldt distinguishes changes joining parts of one word from those joining otherwise independent words, criticizes the latter for obscuring conceptual divisions, and allows a poetic advantage. This is his argument, not a modern rule for counting words.

**Quantity and rhythm** (*Zeitmass*, *Länge*, *Kürze*, *Rhythmus*). Quantity concerns timing, including long and short vowels and syllabic duration; it is distinct from accent. Consonants also enter the argument in 65, so syllabic timing should not be reduced to vowel length alone. Rhythm and accent can support or oppose one another in Humboldt’s account.

**Word accent, discourse accent and tone** (*Accent*, *Wortaccent*, *Redeaccent*, *Sprachaccent*, *Intonation*). Accent organizes prominence and groups in speech; word accent belongs to the word, discourse accent varies with delivery. Intonation in the Chinese discussion concerns tone distinctions, not simply sentence melody. These historical categories overlap with modern stress, pitch accent and tone without being identical to them.

**Enclitic and orthotonic** (*enklitisch*, *orthotonisch*). Here an enclitic word loses its independent accent and attaches to another word; an orthotonic form has its own accent. Humboldt’s point is that an independent word’s pronunciation and accent can change through attachment, which complicates the picture of one word as one audible whole.

**Root, word and Taddhita suffix** (*Wurzel*, *Umstempelung*, *Taddhita-Suffix*). In the note on 64, a root is first *stamped* into a word and the word is then reshaped further by Taddhita suffixes. The metaphor distinguishes stages of formation; the note warns that the rules loosely called *internal* do not apply uniformly at every stage.

**Euphony and aesthetic feeling** (*Wohllaut*, *aesthetisches Gefühl*). *Euphony* means ‘pleasing sound’ as assessed in the argument. Roughness, purity and refinement are Humboldt’s aesthetic and developmental evaluations; they are not measurements of a people’s intellectual capacities. Keep the sensory description distinct from the value attached to it.

**Force, standing and sound value** (*Geltung*, *Lautgeltung*, *Verstandesgeltung*). *Geltung* varies with its bearer: the conceptual force of language, the sway of quantity, a word’s independent standing, or its significance for the understanding. *Lautgeltung* concerns audible realization. These are related contributions or kinds of standing, not uniformly logical validity.

## Notes

[^sound-letters]: *Letters* here are speech-sound elements, not marks on a page; see the terminology guide. Writing is introduced separately in 71. (Trans.)

[^sound-clicks]: Modern phonetics classifies clicks as consonants, not as an animal residue added to articulated speech. See the [International Phonetic Association’s chart](https://www.internationalphoneticassociation.org/IPAcharts/common_files/pdfs/pdfs_IPA_charts_archive/IPA_unitipa_2020.pdf), “Consonants (non-pulmonic)”. This classification does not support Humboldt’s scale of intellectual refinement. (Trans.)

[^sound-conceptual-inference]: This moves from the differentiation of sounds to a judgment about concepts. Humboldt offers it as part of his theory; no comparative demonstration is given in this division. (Trans.)

[^sound-meaningless]: *Meaningless* here contrasts with an associated concept carried by the sound. It does not mean that an inflection has no grammatical role: the point is to express the relation without adding unrelated lexical content. (Trans.)

[^sound-source-deleted-near]: Deleted after *der*: “lying close to one another”. (1906 Ed.)

[^sound-source-bopp-62]: Bopp’s *Lehrgebäude der Sanskrita Sprache*, p. 14, no. 20. (Bopp’s *Grammar*.) (Humboldt)

[^sound-cerebral]: *Cerebral* and *lingual* are historical names for consonants made with the tongue tip drawn back, usually called retroflex consonants today. They concern articulation, not a class of mental abilities; see [Whitney, Sanskrit Grammar, §45](https://en.wikisource.org/wiki/Sanskrit_Grammar_(Whitney)/Chapter_II). (Trans.)

[^sound-understanding]: The phrase ‘before the understanding’ is read here as *as it appears to the understanding*, in line with the parallel contrast between ear and understanding in 64. A temporal reading, ‘before understanding intervenes’, is also possible, but is less well supported by that parallel. (Trans.)

[^sound-source-thiersch]: Thiersch maintains that the Greeks joined words, as in Sanskrit. (Humboldt)

[^sound-source-carey]: *Grammar of the Telinga Language*, Preface, p. II. (Humboldt)

[^sound-telugu]: *Telinga* is Telugu, a Dravidian language, not a descendant of Sanskrit. The transition from Sanskrit-derived languages to this example therefore needs qualification: compare [Carey’s catalogue record](https://openlibrary.org/books/OL27882132M/A_grammar_of_the_Telinga_language) with [Glottolog’s classification](https://glottolog.org/glottolog?iso=tel). Carey’s cited preface has not been independently checked here. (Trans.)

[^sound-source-haughton]: Haugthon’s *Rudiments of Bengali Grammar*, p. 147, § 377. (Humboldt)

[^sound-source-bopp-64]: Strictly speaking, the rules for the interior of words are not quite correctly so called. They apply in full only when the root is stamped anew as a word, and in that word’s inflections. Already with the Taddhita suffixes, which further reshape the word thus stamped, the rules for the initial and final letters of different words come into play. Bopp’s *Grammar*, p. 58, § 85. (Humboldt)

[^sound-source-deleted-signs]: Deleted after *dem* in *in dem* (‘in which’): “with or without signs of their own”. (1906 Ed.)

[^sound-freedom]: The referent of *its freedom* is compressed. It may be aesthetic feeling, rhythm or quantity; the English leaves the reference open rather than assigning it uniquely. (Trans.)

[^sound-intention]: The claim of exclusive dependence on intention is Humboldt’s. He immediately adds constraints from pronunciation and euphony, and in 67 treats word accent as belonging to the language; it should not be paraphrased as free choice of the accent on each word. (Trans.)

[^sound-source-deleted-word]: Deleted after *syllable*: “or a word”. (1906 Ed.)

[^sound-vowel-tone]: Tone is not confined to vowels in all Chinese varieties. Cantonese has tone-bearing syllabic nasals, such as *m4* ‘no’ and *ng5* ‘five’, with no vowel; see [Hong Kong Polytechnic University, “Understanding Finals”](https://www.polyu.edu.hk/clc/cantonese/phoneticKnowledge/vowel/recognizingVowel/?lang=en), the syllabic-nasals table. This does not identify the exact varieties covered by Humboldt’s sources. (Trans.)

[^sound-source-chinese]: In Chinese grammars, the intonations are admittedly called modifications of the words, or, what there amounts to the same thing, of the syllables, which is indeed much more appropriate in those grammars. Yet, properly speaking, they rest only on the vowel ending; hence, speaking in the manner of other languages, it is more natural to say that each simple and compound vowel ending can have such and such an intonation, and that the words then take the vowel ending intoned in this or that way according to differences in their meaning. In this way the matter is put on the same footing as the short and long vowels of other languages. (Humboldt)

[^sound-sanskrit-accent]: This records the limits of Humboldt’s information, not a present absence of evidence. Later descriptions document accent marks in Vedic texts; see [Whitney, Sanskrit Grammar, §87](https://en.wikisource.org/wiki/Sanskrit_Grammar_(Whitney)/Chapter_II). Written notation does not by itself recover every detail of spoken delivery. (Trans.)

[^sound-word-unity]: The argument links unity of concept with unity of accent. Compare 69, where Humboldt considers several words joining as one accentual group; a group heard as one whole need not simply be one independent word. (Trans.)

[^sound-combined-accent]: *When they do combine* makes explicit the reading that conceptual union requires one element to predominate. The final ‘in that case’ in the German could instead be attached to the preceding negative alternative; its reference is not explicit. (Trans.)

[^sound-source-sanskrit-pronouns]: The Sanskrit pronouns of the first and second persons do have, for certain cases, a second, more frequently used, abbreviated form alongside the full form, whose fullness is due to its less frequent use. But it does not appear that this second form was used exclusively as an enclitic. (Humboldt)
