Americanist phonetic notation

Americanist phonetic notation, also known as the North American Phonetic Alphabet (NAPA), the Americanist Phonetic Alphabet or the American Phonetic Alphabet (APA), is a system of phonetic notation originally developed by European and American anthropologists and language scientists (many of whom were students of Neogrammarians) for the phonetic and phonemic transcription of indigenous languages of the Americas and for languages of Europe. It is still commonly used by linguists working on, among others, Slavic, Uralic, Semitic languages and for the languages of the Caucasus, of India, and of much of Africa; however, Uralists commonly use a variant known as the Uralic Phonetic Alphabet.

Despite its name, NAPA has always been widely used outside the Americas. For example, a version of it is the standard for the transcription of Arabic in articles published in the Zeitschrift der Deutschen Morgenländischen Gesellschaft, the journal of the German Oriental Society.

Diacritics are more widely used in Americanist notation than in the International Phonetic Alphabet (IPA), which seeks to use as few diacritics as possible for phonemic distinctions, retaining them only for the dental–alveolar distinction. Americanist notation relies on diacritics to distinguish many other distinctions that are phonemic in the languages it transcribes. On the other hand, Americanist notation uses single letters for most coronal affricates, whereas the IPA requires digraphs. Otherwise Americanist notation has grown increasingly similar to IPA, and has abandoned many of the more obscure letters it once employed.

Summary contrast with the IPA alphabet
Certain symbols in NAPA were once identical to those of the International Phonetic Alphabet, but have become obsolete in the latter, such as ⟨ι⟩. Over the years, NAPA has drawn closer to the IPA. This can be seen, for example, in a comparison of Edward Sapir's earlier and later works. However, there remain significant differences. Among these are:


 * $⟨y⟩$ for, $⟨ñ⟩$ for , $⟨c⟩$ or $⟨¢⟩$ for , $⟨ƛ⟩$ for and $⟨ł⟩$ for
 * Palato-alveolar $⟨č ǰ š ž⟩$ and sometimes alveopalatal $⟨ć ś ź ń⟩$
 * Advancing diacritic (inverted breve) for dentals and palatals (apart from non-sibilant dental $⟨θ ð⟩$), and retracting diacritic (a dot) for retroflex and uvulars (apart from uvular $⟨q⟩$)
 * $⟨r⟩$ or $⟨ř⟩$ for a flap and $⟨r̃⟩$ for a trill
 * Ogonek for nasalization
 * Dot over vowel for centering, two dots (diaeresis) over a vowel to change fronting (for front rounded vowels and unrounded back vowels)
 * Acute and grave accents over vowels for stress

History
John Wesley Powell used an early set of phonetic symbols in his publications (particularly Powell 1880) on American language families, although he chose symbols which had their origins in work by other phoneticians and American writers (e.g., Pickering 1820; Cass 1821a, 1821b; Hale 1846; Lepsius 1855, 1863; Gibbs 1861; and Powell 1877). The influential anthropologist Franz Boas used a somewhat different set of symbols (Boas 1911). In 1916, a publication by the American Anthropological Society greatly expanded upon Boas's alphabet. This same alphabet was discussed and modified in articles by Bloomfield & Bolling (1927) and Herzog et al. (1934). The Americanist notation may be seen in the journals American Anthropologist, International Journal of American Linguistics, and Language. Useful sources explaining the symbols – some with comparisons of the alphabets used at different times – are Campbell (1997:xii-xiii), Goddard (1996:10–16), Langacker (1972:xiii-vi), Mithun (1999:xiii-xv), and Odden (2005).

It is often useful to compare the Americanist tradition with another widespread tradition, the International Phonetic Alphabet (IPA). Americanist phonetic notation does not require a strict harmony among character styles: letters from the Greek and Latin alphabets are used side-by-side. Another contrasting feature is that, to represent some of the same sounds, the Americanist tradition relies heavily on letters modified with diacritics; whereas the IPA, which reserves diacritics for other specific uses, gave Greek and Latin letters new shapes. These differing approaches reflect the traditions' differing philosophies. The Americanist linguists were interested in a phonetic notation that could be easily created from typefaces of existing orthographies. This was seen as more practical and more cost-efficient, as many of the characters chosen already existed in Greek and East European orthographies.

Abercrombie (1991:44–45) recounts the following concerning the Americanist tradition: "In America phonetic notation has had a curious history. Bloomfield used IPA notation in his early book An Introduction to the Study of Language, 1914, and in the English edition of his more famous Language, 1935. But since then, a strange hostility has been shown by many American linguists to IPA notation, especially to certain of its symbols.

An interesting and significant story was once told by Carl Voegelin during a symposium held in New York in 1952 on the present state of anthropology. He told how, at the beginning of the 1930s, he was being taught phonetics by, as he put it, a "pleasant Dane", who made him use the IPA symbol for sh in ship, among others. Some while later he used those symbols in some work on an American Indian language he had done for Sapir. When Sapir saw the work he "simply blew up", Voegelin said, and demanded that in future Voegelin should use 's wedge' (as š was called), instead of the IPA symbol.

I have no doubt that the "pleasant Dane" was H. J. Uldall, one of Jones's most brilliant students, who was later to become one of the founders of glossematics, with Louis Hjelmslev. Uldall did a great deal of research into Californian languages, especially into Maidu or Nisenan. Most of the texts he collected were not published during his lifetime. It is ironic that when they were published, posthumously, by the University of California Press, the texts were "reorthographized", as the editor's introduction put it: the IPA symbols Uldall had used were removed and replaced by others.

What is strange is that the IPA symbols seem so obviously preferable to the Americanist alternatives, the 'long s' to the 's wedge', for example. As Jones often pointed out, in connected texts, for the sake of legibility diacritics should be avoided as far as possible. Many Americanist texts give the impression of being overloaded with diacritics.

One may wonder why there should be such a hostility in America to IPA notation. I venture to suggest a reason for this apparently irrational attitude. The hostility derives ultimately from the existence, in most American universities, of Speech Departments, which we do not have in Britain. Speech Departments tend to be well-endowed, large, and powerful. In linguistic and phonetic matters they have a reputation for being predominantly prescriptive, and tend to be considered by some therefore to be not very scholarly. In their publications and periodicals the notation they use, when writing of pronunciation, is that of the IPA. My belief is that the last thing a member of an American Linguistics Department wants is to be mistaken for a member of a Speech Department; but if he were to use IPA notation in his writings he would certainly lay himself open to the suspicion that he was."

Consonants
There is no central authority. The Western Institute for Endangered Language Documentation (WIELD) has recommended the following conventions since 2016: Note however that WIELD is designed specifically for Native American languages, whereas NAPA, despite its name, is widely used elsewhere, e.g. in Africa.

Advanced is ⟨C̯⟩ and retracted is ⟨C̣⟩. Geminate is ⟨C꞉⟩ or ⟨CC⟩. Glottalization is e.g. ⟨č̓⟩ or ⟨m̓⟩ (ejectives are not distinguished from other types of glottalization). Palatalization is written ⟨Cʸ⟩. Labialization, velarization, aspiration, voicelessness and prenasalization are as in the IPA. Pharyngeals, epiglottals and glottals are as in the IPA, as are implosives and clicks.

Differences from the IPA fall into a few broad categories: use of diacritics to derive the other coronal and dorsal articulations from the alveolar and velar, respectively; use of c j λ ƛ for affricates; y for its consonantal value, and r for a tap rather than a trill.

Notes:
 * Among the dental fricatives, $⟨θ ð⟩$ are slit fricatives (non-sibilant) while $⟨s̯ z̯⟩$ are sulcalized (sibilant).

Rhotics table
About 90% of languages only have one phonemic rhotic consonant. As a result, rhotic consonants are generally transcribed with the $\langler\rangle$ character. This usage is common practice in Americanist and also other notational traditions (such as the IPA). This lack of detail, although economical and phonologically sound, requires a more careful reading of a given language's phonological description to determine the precise phonetics. A list of rhotics is given below.

Other flaps are $\langleň\rangle$, $\langlel͏̌\rangle$, etc.

Common alternate symbols
There are many alternate symbols seen in Americanist transcription. Below are some equivalent symbols matched with the symbols shown in the consonant chart above.


 * $\langle¢\rangle$ may be used for $\langlec\rangle$.
 * $\langleč̣\rangle$ may be used for $\langlec̣\rangle$.
 * $\langlej\rangle$ may be used for $\langleʒ\rangle$.
 * $\langleǰ\rangle$ may be used for $\langleǯ\rangle$.
 * $\langleƚ\rangle$ may be used for $\langleł\rangle$.
 * $\langleɸ\rangle$ may be used for $\langleφ\rangle$.
 * $\langleG\rangle$ may be used for $\langleġ\rangle$.
 * $\langleX\rangle$ may be used for $\langlex̣\rangle$.
 * ʸ may be used for fronted velars (e.g., kʸ = k̯, gʸ = g̑).
 * Some transcriptions superscript the onset of doubly articulated consonants and the release of fricatives, e.g. $\langleᵍɓ\rangle$, $⟨t̓ᶿ⟩$.
 * There may be a distinction between laminal retroflex $⟨č̣ ṣ̌ ẓ̌⟩$ and apical retroflex $⟨c̣ ṣ ẓ⟩$ in some transcriptions.
 * The fronting diacritic may be a caret rather than an inverted breve, e.g. dental $\langlet&#x032D;\rangle$ and palatal $\langlek&#x032D;\rangle$.
 * Many researchers use the x-caron (x̌) for the voiceless uvular fricative.
 * The use of the standard IPA belted l (ɬ) for the voiceless lateral fricative is becoming increasingly common.

Pullum & Ladusaw
According to Pullum & Ladusaw (1996), typical Americanist usage at the time was more-or-less as follows. There was, however, little standardization of rhotics, and $\langleṛ\rangle$ may be either retroflex or uvular, though as noted above $\langleṛ\rangle$ or $\langleṛ̌\rangle$ may be a retroflex flap vs $\langleṛ̃\rangle$ as a uvular trill. Apart from the ambiguity of the rhotics below, and minor graphic variants (ȼ g γ for c ɡ ɣ and the placement of the diacritic in g̑ γ̑), this is compatible with the WIELD recommendations. Only precomposed affricates are shown below; others may be indicated by digraphs (e.g. $⟨dz⟩$).

Ejectives and implosives follow the same conventions as in the IPA, apart from the ejective apostrophe being placed above the base letter.

Pike
Pike (1947) provides the following set of symbols:

Voiceless, voiced and syllabic consonants may also be C̥, C̬ and C̩, as in IPA. Aspirated consonants are C or C̥ʰ / C̬ʱ. Non-audible release is indicated with superscripting, Vꟲ.

Fortis is C͈ and lenis C᷂. Labialization is C&#814; or Cʷ; palatalization is Ꞔ, C⁽ⁱ⁾ or Cʸ; velarization is C⁽ᵘ⁾, and pharyngealization is C̴.

Other airstream mechanisms are pulmonic ingressive C←, ejective Cˀ, implosive Cˁ, click C˂, and lingual ejective (spurt) C˃.

Vowels
WIELD recommends the following conventions. It doesn't provide characters for distinctions that aren't attested in the literature:

No distinction is made between front and central for the lowest unrounded vowels. Diphthongs are e.g. ⟨ai⟩ or ⟨ay⟩, depending on phonological analysis. Nasal vowels are e.g. ⟨ą⟩. Long vowels are e.g. ⟨a꞉⟩. A three-way length distinction may be ⟨a a꞉ a꞉꞉⟩ or ⟨a aꞏ a꞉⟩. Primary and secondary stress are e.g. ⟨á⟩ and ⟨à⟩. Voicelessness is e.g. ⟨ḁ⟩, as in the IPA. Creak, murmur, rhoticity et al. are as in the IPA.

Pullum & Ladusaw
According to Pullum & Ladusaw (1996), typical Americanist usage at the time was more-or-less as follows:

Pike
Pike (1947) presents the following:

Nasalization is V̨ or Vⁿ. A long vowel is V꞉ or Vꞏ; half-long is V‧ (raised dot). Positional variants are fronted V˂, backed V˃, raised V˄ and lowered V˅.

Bloch & Trager
Bloch & Trager (1942) proposed the following schema, which was never used. They use a single dot for central vowels and a dieresis to reverse backness. The only central vowels with their own letters are $\langleɨ\rangle$, which already has a dot, and $\langleᵻ\rangle$, which would not be distinct if formed with a dot.

Kurath
Kurath (1939) is as follows. Enclosed in parentheses are rounded vowels. Apart from ⟨ʚ, ꭤ⟩ and some differences in alignment, it is essentially the IPA.

Chomsky & Halle
Chomsky & Halle (1968) proposed the following schema, which was hardly ever used. In addition to the table, there was $\langleə\rangle$ for an unstressed reduced vowel.

Tone and prosody
Pike (1947) provides the following tone marks:


 * High: V́ or V¹
 * Mid: V̍ or V²
 * Norm: V̄ or V³
 * Low: V̀ or V⁴

Stress is primary ˈCV or V́ and secondary ˌCV or V̀.

Short or intermediate and long or final 'pauses' are |, ||, as in IPA.

Syllable division is CV.CV, as in IPA, and morpheme boundaries are CV-CV.

Historical charts of 1916
The following charts were agreed by committee of the American Anthropological Association in 1916.

The vowel chart is based on the classification of H. Sweet. The high central vowels are differentiated by moving the centralizing dot to the left rather than with a cross stroke. IPA equivalents are given in a few cases that may not be clear.

Notes:
 * surd = voiceless; sonant = voiced; intermed. = partially voiced
 * In the glottalized stop column, the phonetic symbol appearing on the left side (which is a consonant plus an overhead single quotation mark) represents a weakly glottalized stop (i.e. weakly ejective). The symbol on the right side is strongly glottalized (i.e. it is articulated very forcefully). Example: = weakly glottalized,  = strongly glottalized. (Cf.  = [k] followed by glottal stop.) This convention is only shown for the glottalized stops, but may be used for any of the glottalized consonants.
 * "Laryngeal" refers to either pharyngeal or epiglottal.

Anthropos (1907)
The journal Anthropos published the alphabet to be used in their articles in 1907. It is the same basic system that Sapir and Boas introduced to the United States. Transcription is italic, without other delimiters.

Variation between authors
Following are symbols that differ among well-known Americanist sources.

Encoding
The IETF language tags register fonnapa as a subtag for text in this notation.