Edward Sapir’s two-part monograph from 1915 — published as a memoir of the Geological Survey of Canada — belongs to the descriptive linguistics of the early twentieth century, when a handful of researchers were racing to record the indigenous languages of the Pacific Northwest before they changed beyond recognition. The first study examines the “abnormal” registers of Nootka speech on Vancouver Island: the special words and sound-twists speakers use for children, the fat, the lame, the cross-eyed, and other marked classes of person. The second turns to Comox, a Salish language, and works through the reduplicative machinery by which that language builds plurals and diminutives. Both are technical, example-driven, and grounded in fieldwork — Sapir took his Nootka material from named consultants at Alberni Canal in 1910 and again in the winter of 1913–14 — and both are less about exotic curiosities than about how grammar itself can carry social meaning.
“When we say “big dog make bow-wow” instead of “the dog barks,” it is a fair inference that we are talking to a baby, not to a serious-minded man of experience.”
Speech That Points Without Saying
Sapir opens by framing his problem carefully: languages possess devices that signal the status, sex, age, or characteristics of the speaker, the addressee, or the person spoken about, without stating any of it directly. His homely illustrations are “big dog make bow-wow,” which implies one is speaking to a baby, and the Quaker “thee.” Nootka has special vocabulary for obscene matters when women are present, and a set of “baby-words.” Usually it is the speaker or the addressee who is thus signalized, less often a third party. The specialized type that interests Sapir most works not through vocabulary but through grammatical elements, consonantal and vocalic changes, and inserted meaningless sounds — morphologic and phonetic means rather than lexical ones.
He then surveys the two most prominent cross-linguistic types: sex-discrimination and rank-discrimination. Eskimo dialects, following Boas, had women pronounce final p, t, k, q as nasals; in Yana of northern California, women’s forms differ by unvoicing final vowels, by turning final -na into aspiration, and by lengthening final vowels for the interrogative where men suffix -n — and men speaking to women adopt the women’s forms. Rank forms appear in East Asiatic and Indonesian languages, in the Nahuatl reverentials built with -lia, -tia, -ltia and reflexive prefixes (“he causes himself to sleep”), in the European second- and third-person plurals of address, and in Yana siblings who address each other in the plural.
The Nootka Classes: Children, the Fat, the Marked
The heart of the Nootka study is a catalogue of classes whose members are marked in speech by suffixes and by “consonantal play” — altering sibilants or inserting meaningless consonants. The classes run: children, fat or heavy people, abnormally short adults, people with eye defects, hunchbacks, the lame, the left-handed, and circumcised males, with a further group for greedy people. Children take the diminutive -is, added before temporal, modal, and pronominal suffixes, so that qwistci’ becomes qwis’istci’. The form is one of affection and is not used for stranger children; a lullaby example survives in the notes. Dan Watts reported that some use the diminutive to belittle others, and that a chief doing so is “haughty.”
Fat people take the suffix -aq’. The abnormally small combine the diminutive with palatalization of sibilants into c-sounds, a formation extended to sparrows and wrens, sometimes with a meaningless added sound. Eye defects — cross-eye, squint, one eye run out, but not blindness — combine the diminutive with a shift of s and c to lateral stops and spirants, giving forms like qwAl-‘ilma’ from qwisma’; the pattern is also applied to deer and mink, and the mythological Mink carries a corresponding name. Adults may take offence at these forms, which are avoided unless contempt or teasing is intended. Hunchbacks combine the diminutive with thickish c-sounds, and the resulting words hover between contempt and affection — avoided before hunchbacks for fear of giving offence, yet used for a well-known humpbacked child or a good-natured old woman. The lame take the diminutive with a meaningless lc or Lci inserted before it, its position left to the speaker’s whim. The left-handed take a meaningless tcH after the first syllable, forms used to twit acquaintances and by jokers even for bears, which are supposed to be left-handed. Circumcised males take a meaningless cf after the first syllable, and greedy people a tcx in the same position.
Defects of Speech, Mockery, and the Making of Myth Voices
The second part of the Nootka study pushes from description into genesis. A defect called kakdf’win — “to talk as one with missing teeth” — is distinct from the lisping hahdt.Hn: speakers substitute Q for s and c, and other sounds for ts and tc, so that “apples” and “bear” and “canoe” come out distorted. Afflicted persons are imitated when addressed; Alex Thomas, before having caps put on his remaining teeth, was mocked in exactly this fashion. Stuttering is a fifth and not uncommon defect, and stutterers, like all with abnormal speech, are derided by imitation.
Sapir then brings in West Greenland kutafoq, a defect that substitutes ordinary gutturals for velars, evidently because tongue-root contact with the velum is harder. It confuses two etymologically distinct consonant series, producing theoretical ambiguities of grammar and vocabulary that parallel the Nootka forms. Children are apt to be kutdt’oq and generally outgrow it; some adults, particularly women, remain so, and in the Upernavik district the habit has been socialized into a dialectic peculiarity — strikingly analogous, Sapir says, to Nootka forms used for or to ill-favoured classes.
From this he builds his central argument about origins. The use of morphological elements for characteristics of the addressee needs no comment. But contact can be established between speech defects and mocking-forms with consonantal play, and between the latter and the forms used for myth characters. Sapir believes that observation of involuntary consonant substitutions in incorrect articulation set the pace for the conscious, humorous substitutions of mocking and myth-character speech. Nootka mocking-forms combine the diminutive affix — pity and affection — with imitation of a speech defect — contempt and jest. A humorously treated myth character may be relegated to the poor talkers or to nature’s step-children; the “enfant terrible” motive is fairly clear in American mythology, with consonant play possibly symbolizing it. The socialized kutdt’oq suggests children’s speech as a factor in myth-character forms, and the Kwakiutl Mink is offered as a likely “enfant terrible.” Such forms may then extend to ordinary individuals, feeding the American Indian habit of comparing peculiar persons to favourite mythological characters.
A further factor is the comic or novel effect of imitating foreign speech — divergent enough to be funny or impressive, yet intelligible. Mythological characters often speak a neighbouring dialect, as the Nass River TxamsEm and others speak the Tsimshian proper of the Skeena; the Melanesians of Mota, following Codrington, use the dialect of Saddle Island for songs; and in clownish ritual episodes foreigners’ speech is imitated to guaranteed laughter, with accuracy neither attained nor aimed at, the foreign style reducing to conventional displacements of vowels and consonants. The change of anterior palatal k- to ts- in Kwakiutl Mink speech resembles the passage of original Wakashan k- to Nootka tc-; the Mink ts- might imitate northern Nootka, but the two languages are too divergent for many cases to be illustrative, so Sapir prefers a common Kwakiutl-Nootka tendency, perhaps children pronouncing anterior palatals as sibilants — and he wishes for data on Kwakiutl children’s speech.
Tribal speech peculiarities close the study. Nootka tribes imitate one another’s real or supposed peculiarities, stressing intonation or articulation rather than vocabulary or grammar, much as New England has its “nasal twang” and the South its “drawl.” The speech of the TsUcd’atH and Hopdtdas’atH of Barkley Sound and the head of Alberni Canal is arbitrarily taken as normal. The HOutcuq’us’atH of Uchucklesit Harbour are said to have spoken in a rumbling fashion, using the throat more; the Ho’.di’atH of Sarita River and southern Barkley Sound are said to “splutter,” using tc sounds more liberally. The TsUcd’atH have seized on the tc-insert to poke fun at their Ho’.di’atH kinsmen, using it inaccurately — the tribe is humorously renamed in the notes, and the joke itself becomes a small demonstration of how speech caricature works.
Comox: The Sound System and the Ten Shapes of Reduplication
The Comox study begins with phonology. Eliminating b and d as secondary from gʷ and dj, the remaining consonants are etymologically distinct, though k, k!, and x may prove secondary forms of kʷ, k!ʷ, and xʷ — giving thirty-six, or perhaps thirty-three, organically distinct consonants. A secondary series of aspirated surds arises when voiceless stops occur as syllabic finals; q is often released into a weak spirant glide before the following vowel; final vowels, m, and n are often followed by aspiration, though not consistently heard. Long consonants seem of no etymological significance. Lengthening and reduction of vowels matter greatly, but since they are generally grammatical, Sapir treats them under reduplication. Many cases of gʷ descend from original w, which regularly passed into gʷ between vowels and often initially — which explains a class of plurals with o-vowel from stems in internal -gʷ-, such as t!egʷem “sun, moon” becoming t!oH!egʷem, and the suffix -agHl “canoe” from an earlier *-awil. Test cases include q!gʷas “deer,” a loanword from Kwakiutl, and tlgʷx’ “nine.” Some words keep w between vowels, and the relation is unclear; dj and y are probably related, with evidence from the Type VIII plurals and comparisons with KwantlEn and Siciatl.
Many Comox nouns, Sapir observes, always appear reduplicated with no grammatical significance — many are animal names, some onomatopoetic — and he illustrates ten types of this complete or partial reduplication, with examples like ho’mho’m “blue grouse” and q!iq!q!d’adj^uk’ “butter-ball duck.” A footnote pauses over the wooden-ball game qAq’td’amas: two sides, each guarding a goal-pit; all but two guards gather at the center; one throws up the ball, players try to catch it and place it in the opponents’ pit; first to ten goals wins; and after four goals a general free-for-all fight broke out.
Plurals, Diminutives, and the Grammar of Smallness
Most Comox nouns form plurals by reduplication; some use different stems; some form no plural at all. Sapir lays out eleven types. Type I repeats the first and second stem consonants; Type II replaces the stem vowel with inorganic A, with sub-types palatalized to i and labialized to u, and irregular examples from “mouse” and “tongue”; Type III mostly has gʷ as second consonant from original w; Type IV shows syncope of the first stem-vowel, corroborated by KwantlEn; Type V shortens the reduplicated long vowel; Types VI through XI ring further changes, including a Type X that alters the first stem-vowel to I. Irregular plurals include dja’dja’ “tree” becoming djadjid’m and niAl’q’ “fawn” becoming mama’aliq’. A few nouns change stem entirely — “woman,” “girl,” “married man,” “leg” — and many nouns simply lack plurals, some because of a feeling against re-reduplication, some not; for “robin” a plural was explicitly denied, and the word for “many” may be juxtaposed to express plurality.
Diminutives are formed by reduplicating only the first stem consonant, with vowel rules different from the plurals, plus internal stem changes. Sapir lists twenty-two diminutivizing features — reduplication with short or long e, ǐ, ǔ, ū, short or long a, and short o; a glottal stop inserted, breaking a vowel or glottalizing a final consonant; quantitative changes such as lengthening and the insertion of t or a short vowel; qualitative changes such as umlaut of a to e; and vocalic reductions — which combine into types numbered I through XXII, with sub-types and examples like mo’os “head” becoming mimo’os. The classification continues through Types XXIII to XXXII: Type XXIII reduplicates a syllable cf and yields forms like mdmiyofs “raccoon” and g^dgHdi’m “slave,” while tdyac “killer-whale” to tdtlyac is called quite irregular, its long and short vowels the reverse of what would be expected. Type XXIV builds on the unreduplicated stem abstracted from a reduplicated simplex; Type XXV treats a long vowel as the reduplicating element; Type XXXII b, for “house,” involves further stem change.
Beyond reduplication, Comox makes diminutives of animal nouns with the suffix -ol or -ol, attaching the shorter form to stems whose last vowel is short and the longer to those whose last vowel is long, by a phonetic law of rhythmic balance; a couple of forms, the hummingbird and the panther, seem irregular and were perhaps misheard. Plurals of diminutives are as a rule doubly reduplicated — the first syllable expressing diminutive, the second plurality — so that they are “diminutivized plurals, not pluralized diminutives,” combining the characteristics of both. Types I through V are set out with sub-types according to whether the plural of the simplex is unchanged or modified by given diminutive features, and some strange forms are explained by analogy with a set of attested words where the element -xAU- is etymologically justified.
Numerals, Pronouns, and the Wider Salish Frame
The final stretch of the Comox material is miscellaneous but telling. Numerals one through ten, the tens, hundreds, and thousands are given, with two thousand expressible in either of two ways, and numeral classifiers for people, canoes, fathoms, houses, and dollars — the last properly for round objects, including heads and turnips. Body-part substantivals — hand, eye, head, nose, foot — occur only in composition, while “white ears” stands as an independent use. The pronominal data Sapir himself calls fragmentary and unreliable; tA and ku are articles implying visibility and invisibility respectively, and possessive, subjective-suffix, and object paradigms are set out.
A systematic comparative treatment is declined, since the phonetics of the available material — especially vocalic quantity and glottal stops — prevent definitive classification. Still, plural Types I and II are called fundamental Salish processes, illustrated from Bella Coola, Tcil’Qeuk, Shuswap, Okanagan, Thompson River, Nanaimo, and Lower Lillooet; Type VII seems more typical of Interior Salish; Types IX and X are illustrated outside Comox; and an inserted -l(a)- and a prefix -E- are noted as types absent from the Comox material. Sapir suggests that Type II may be a reduced form of Type I, and that secondary types set the pace for analogical forms, so that the same stem is treated differently in different languages — more material is needed. For diminutives, the most consistent difference is that the plural reduplicates the first two stem consonants, often only the first, while the diminutive reduplicates only the first and never the second; vocalic reduction, e-vowel reduplication, breaking, and umlaut are tendencies, and glottal stops are thought likely diminutivizing elements, with Upper Lillooet and Thompson River examples.
What the Two Studies Together Argue
Read side by side, the monographs make a single point in two keys. Nootka shows that a language can encode social judgment — affection, contempt, mockery, deference — directly into its morphology and phonetics, so that a suffix or an inserted consonant says what a sentence never states. Comox shows the same machinery in its purest grammatical form: reduplication as a systematic engine for plurality and smallness, elaborated into dozens of types and sub-types, with analogy and rhythmic balance doing as much work as fixed rule. Sapir’s insistence on the social life of sound — that a speech defect can become a mocking form, that a mocking form can become a myth character’s voice, that a children’s habit can become a dialect — is the thread joining the two halves.
The work’s significance lies in that method as much as in its findings. These are early, meticulous field records of languages that were already under pressure, taken down from named speakers and organized with a taxonomy builder’s patience; the Nootka study in particular is an early argument that “abnormal” speech is not a curiosity but a window onto how communities use language to place one another. The comparative notes at the end of the Comox study, hedged and provisional as they are, mark out questions about Salish reduplication that later workers would take up with better phonetic data. For a reader today, the memoir reads as a compact demonstration of Sapir’s conviction that grammar and social life are not separate subjects — that the way a language bends a word can be a way of looking at a person.
