Douglas Taylor is perhaps the one anthropologist who has collected the most data, done the most analysis, and published the most on what he calls Island Carib and Central American Island Carib (Garifuna).  His work is generally quite thorough although it must be noted that his analyses are more often than not diachronic in nature. However, with regard to the language as it is today spoken in Central America generally, or even Hopkins Village specifically, his work should be viewed with caution, partly because his knowledge of earlier forms in the language keeps intruding on his perceptions of what the people were actually saying, and partly because of what I believe are errors in his transcription.


Taylor lists the consonantal phonemes of Garifuna as follows (Taylor 1977):
m  n  –  –
b  d  –  g
p  t  c  k
f  s  –  h
–  l  r  –

He describes these as being pronounced approximately as in English, except for /c/ which “varies freely between the sound of ‘ch’ in ‘church’ and that of ‘sh’ in ‘shut'”.  Conspicuously absent are /w/ and /y/.  Also absent is the palatal nasal ?.  It is with regard to these three that Taylor departs most dramatically from the other two traditions, and practically all other persons who have tried to write Garifuna.  We will return to the glides when we consider vowels.  As for the ?, there is ample support for Taylor’s position that there is no such consonantal phoneme in Garifuna.  What do exist are nasal vowels and certain vowel sequences namely, ie, ia, iu, and io, which, when nasalised, sufficiently resemble sound sequences containing the palatal nasal to lead most people who write Garifuna to use ?, which they are familiar with from Spanish, and write these sequences as i?e, i?a, i?o, and i?u.


According to Taylor the simple vocalic phonemes include i, e, a, o, u, and v, which is the nasalized counterpart of any of the five oral vowels.  The simple oral vowels /i/, /e/, /a/, and /u/ “have so-called continental values, as in Spanish, but vary from a close to a more open sound according to position and stress, /u/ being occasionally heard as a close ‘o’; /o/ is a high back unrounded vowel ?   ?”.  Taylor’s /o/ is what SIL and S.J.C. write as /?/.  The ten simple oral and nasal vowels are listed below together with some words in which the vowels occur.

i ?giri (nose)   an?gi (heart)
e em?ragua (to rest)  ei?ri (male)
a  ad?lu (porridge)   ar?ri (octopus)
o ?ho (pus)    h?ro (crab)  ?ma (road)
u  ub?u (world)   u?iu (sun)
i h?aru (woman)   her?gi (beach crab)
e gur?tu (plantain porridge) er?ga (say)
a  ba (socks)   ful?su (plank)
u  su (all)    g?fara (cohune leaves used for thatch)
o ir?here (fishing line) gai?giru (she’s still a virgin

In his concern for economy Taylor has failed to include / / (or some other representation of it) in his phonemic inventory, possibly because there are not many instances of its occurrence.  Nonetheless, uh bu, k pu, b nu, f nu, and the ever present demonstrative/relative pronoun t , and names like M d d  and D d must be accounted for by any orthography.

Compound Vowels and Semi-vowels

Taylor lists the following as compound phonemes.  Examples of words containing them are given as well:

ie  ei?ri (male)  fan?e (basket)  as?edu (plate)
ia  iau?ra (cohune palm)  i?iaua (pineapple)
io  ior?do (hummingbird)
iu  hi?ruha (spirit-helper)  saiu (sack)
Iurumai (St Vincent)
uo  uori (woman)  uobu (mountain)
ua  lu?gu (on it)  ?ati (there isn’t)  mas?ua (fish trap)
ue  ueiu (sun)  ueue (tree)  ag?ueriha (to carve)
ui  uira (calabash tree)  hauaui (kingfish)

ei  erei (strength)  agolei (oil, fat, grease)
isubusei (mirror)
ai  sairi (After life)  aiga (to eat)  agai (womb,           container)
ao  h?o (ants)  aronao (arm)  halao (chair)
au  ?uca (to try)  hanau (“old wife” – a fish)
urinauga (yesterday)

The first eight compound vowels listed above are the ones involved in any discussion having to do with the status of [y] and [w] in Garifuna.  Taylor argues that phonemically what we have in Garifuna are /i/ and /u/, and that what may appear to some as [y] and [w] are simply variants of the phonemes /i/ and /u/ respectively (Taylor 1955:241).  But while this is all very well phonemically, in practice it does not make much orthographic sense.  In fact, it creates problems that Taylor himself does not seem to be aware of.  Those eight so-called compound vowels often behave like consonant -vowel sequences, in other words, as ye, ya, yo, yu, wo, wa, we, and wi.  But because Taylor insists on treating /i/ and /u/ as vocalic phonemes at all times he ends up writing words like iaiaua and i?uara such that syllable boundaries are difficult if not impossible for any reader who is not already proficient in the language to distinguish.  The importance of this becomes clear when one considers the rules of accentuation that Taylor uses and are outlined below.  Thus one may rightly ask:  In i?iaua, is the accent marked because ia is the first syllable?  Or is it because ia is di-syllabic and the accent is intended to show that?  In iauara, which is the second syllable?  a?  au?  u?  or ua?  Notice that this would not be a problem if the words were written thus:  yayawa and yawara.  We will return to this matter in Chapter 3 when we will propose a means of reconciling the result of Taylor’s phonemic analysis with regards to /i/ and /u/ on the one hand, and the orthographic usefulness of consonants as syllable boundary markers on the other.

Taylor perceives ei and ai as two separate compound vowels.  This does not seem to be the case, though, for [ei] can be used to replace [ai] in each of its occurrences.  Indeed, it appears that Taylor tends to use/ei/ in word final position and /ai/ elsewhere!

Finally, Taylor missed the /ou/ /au/ distinction.  Because he uses au to represent what SIL writes as ou, he is not able to capture au as in nauba (it will be with me)  as opposed to nouba (my side).


Taylor says that “it (nasality) may adhere to a particular vowel or a given morpheme; or it may in other cases shift or be lost with a change of environment”.  Thus, according to his analysis, nasalisation is constant (i.e., it does not shift) in the first vowel of ?u (high) and hiaru (woman) but not so in io (hymen) or lia (he says).  In gai?giru (she’s still a virgin) and itara lia ti (it is thus), there is a loss of one nasalization and strengthening of another correlated with shift of stress.  As further examples of the elusive nature of nasalization in Garifuna, Taylor gives the following:

1 In u?-ba (whistle!) ? ?uira (to whistle
nasalization moves from stem to prefix.

2 In ug?ie le?  gue le? (today, now) in slow and rapid speech respectively, it clings to the only remaining syllabic vowel.

3 In ?ha (yes, agreed) ? mahati (he is unwilling) and s? (all, every) ? ?sura (to finish) it is completely lost.

4 In n?i (my meat) ? ?i (meat)
bisabadu (your shoes) ? sabadu (shoes), it constitutes a morph.

These examples demonstrate some of the problems in Taylor’s work and should not be left without comment.

1 ?uira (to whistle) can also be written ?uira, ?uira or ?uira, a fact that Taylor seems to have overlooked.
2 ug?ie le? and gu? le? are not identical in meaning.  The former means today and the latter means now.  One is, therefore, not necessarily derived from the other.
3 ?ha is not nasalized on the second vowel.
s? and asura are not related forms, so it should not be claimed that one is derived from the other.
4 n?i and bisabadu do not end with nasalized vowels.  The final vowels are strongly voiced and this is what may have been mistaken for nasalisation.  The final vowels in sabadu and ?i (as in a large number of words in Garifuna) are at best only slightly voiced.

However, the shifting and elusive nature of nasalization in vowel sequences and sometimes across syllables cannot be disputed, and a lot is lost when one tries, as Taylor does, to tie it to particular vowels.  This can be done successfully only when one is dealing with simple nasal vowels unencumbered by proximity to other vowels as in sego (five), sadi (sickness), and arasera (to get ready).

Taylor does not posit the existence of a velar or a palatal nasal consonant, but points out that “a nasalized vowel may approximate, in some environments, to an oral vowel followed by a labial, apical, palatal or velar nasal consonant”.  He also notes that where the nasalized vowel i is followed by semi-vowel off-glide,
as many speakers pronounce it as an oral vowel followed by “an incomplete palatal nasal ?” as there are who pronounce it as a nasal vowel.  (Taylor, 1955).


Stress is not predictable and is therefore phonemic.  Taylor claims that “the functional yield of this opposition (stressed  unstressed) is not great; and, as is the case with English ‘adult  adult’, forms are to be found in whose pronunciation different speakers employ different accentuations”.  With regard to the first claim – that the functional yield of the opposition is not great – there are lots of minimal pairs apart from the few poor examples given by Taylor which suggest otherwise (Taylor 1955).  Consider the following to which a large number of additional examples can be added:
?gura (to throw)   ag?ra (touch)
?g?ra (to bite)   ag?ra (to tie)
?b?r?ha (to fall)    ab?r?ha (to write)
?mura (to defecate)   am?ra (to arrive)

?daha (to push)    ad?ha (to make forms)
?riha (to doze)   ar?ha (to see)
?biri (waist)    ib?ri (sibling)
?bara? (finger or toe nails) ub?ra? (place)
d?ru (offence)    dur? (thick, measles)
f?na (maybe)     fun? (ripe)
h?u (thorn)     hi? (cassava beer)
n?ma (with me)     num? (my friend)

Nor is it accurate to say that “different speakers employ different accentuations”.  What may appear to be arbitrary or idiosyncratic shifts in accentuation may well be grammatically  motivated; for example:

feg?bei bagu (open your eyes)
f?ge lianli lagu (he’s just opened his eyes)

Taylor correctly points out that in most words having three or more syllables, the stress falls on the second syllable and then on every third syllable; e.g. sabadu, nis?badu; war?gabagaba (butterfly), nuw?rigab?gaba (my butterfly).  Words with two syllables are usually stressed on the first syllable, while monosyllabic words pronounced in isolation are invariably stressed.

Unfortunately it has not been possible to obtain the phonemic analysis on which the SIL orthographic tradition is based.  Indeed, it does not appear that such an analysis has ever been published by the SIL, so that our examination of this tradition will necessarily be confined to the orthography itself as employed in some published Garifuna texts and the very brief orthographic note in “According to our Ancestors” edited by Mary Shaw.

Because the SIL fieldworkers did their study of Garifuna in Livingston, Guatemala, they use Spanish pronunciation as a basis for comparison when explaining Garifuna phonology, and the orthography of Spanish has a definite influence on their choice of symbols in writing the language.


The following are the consonants used in the SIL orthography:

b bena (door)
d dar?du (agreement)
g gi?rati (possible)
p pasu (a walk)
t  to (this)   tuei (from her)
ch  ch?l?ti (it arrived)
c,k,q catei (thing), laik (like), quei (like)
m mud?n ( Sheep)
n  nug?feran (my compadre)
f funa (maybe)
s  seru (expensive)
j  jar?ga (tomorrow)
l  luja (it is swollen)
r  ar?ri (octopus)
? ?ei (there)
w weyu (day, sun)
y  yara (over there)

According to the orthographic note (Shaw 1971:308) these consonants conform to Spanish pronunciation with the exception of b, d, and g which are always voiced as in English (bed, dog, go).  Also, n in consonant clusters represents not a consonant but nasalization on the preceding vowel(s).  In the same way, perhaps ? should not be considered a consonant since here again “it indicates a nasalized ‘y’ or ‘i'”.
With regard to the choice of symbols for consonants, the use of j for /h/; k, c, or qu for /k/; and ? for /y/, or /i/, demonstrate the attempt to conform to Spanish orthography.  However, it is difficult to find justification for the use of the different symbols to represent the phoneme /k/.  Even though there may be variation in the pronunciation of the sound represented by these different symbols as in quei, catei and Maskin lus his chipskin, there is certainly no distinctive opposition to justify the use of three different symbols, and the only motivation for this seems to be the Spanish model where the first two are concerned, and the English model in the case of the third.  If the explanation for the use of k in such words as Maskin, chipskin and laik is that these are borrowed and that the spelling should conform to the spelling in the source language, then one would expect the same principle to apply to amicsira from English to mix.

Another puzzling choice of symbol is the use of c instead of ch in a few words like jucu (to throw down) and lajucurunu (his throwing her down).


The SIL list of simple vowels include:

i ?weruja (to steal) lis?misin (his shirt)
e  teb?neri (her doorway) le (this) eg?yeda (to turn)
a  ab?naja (to dance) ?dara (to push)
o  to (this) ?di (or) mosu (must)
u  ?raga (story) murusu (a little)
u  ?ruwa (three) ?mada (on the road) s?g?ti (he passed)

Each of the above cited vowels also occurs as a nasalized vowel in which case it is written as follows:

in ?nwira (to whistle) li?bin (his coming)
en wen (twenty) sensu (money)
an lan (his) ag?nba (to hear) as?nsira (to change)
on g?nwere (basket tie-tie)
un lun (to him) sun (all) mud?n (sheep
?n g?nta (she’s tightened it) lada?run (its striking)

The compound vowels include the following:

ie (ye) g?r?gie (people) yeg? (my pet)
ia (ya) lig?a (he) gi?ra (able, possible) yara (there)
iu (yu) sabiu (wise person) yuga (sweet cassava)
(y?) y?r?d? (hummingbird)

ua (wa) tu?gu (on her)  gar?ragua (turn around) wama (with us)
ui (wi) wan?i (gossip)  wira (calabash tree)
ue (we) gue to (now)    wewe (tree, wood)
(w?) w?ri (female)

ei jaw?i (from them) ?ibuga (to walk)
ai  bad?ira (you find)

au bau (with you)    au (I) tauba (it will be with her)
ou touba (her side)  bougudi (outside)

a? ??ra?ra? (small) lada?ragun (his clothes)

In the orthographic note in “According to our Ancestors” it is clearly implied that the vowels listed above conform to Spanish pronunciation.  Nothing is said about/?/ although there is no such sound or symbol in Spanish speech or writing.  This is most likely due to an oversight on the part of the author so we will not belabour the point.  Suffice it to say that /?/ is an unrounded high back vowel, the same as Taylor’s /o/.

It will also be noted that, unlike Taylor, the SIL tradition allows for the existence of the round mid-back vowel / / for which the phonemic symbol /o/ is reserved.  This tradition, therefore, has a six-vowel system and can therefore adequately represent such words as:

to (this), mosu (must) and fonu (thing), which are problematic for Taylor.
There are not many instances of /o/ (phonetic [o]) but, as shown above, it does exist and must be accounted for in any phonemic analysis of Garifuna.  There are also not many instances of/en/ and /on/, but the investigator has to be careful not to confuse those sounds with the diphthongs /ein/ and /oun/ respectively. ([e] and [o] do not occur in Garifuna except as [ei] and [ou]).  Even though pronunciation is admittedly not uniform in all Garifuna communities and there are consequently some phonetic differences between the Garifuna of Livingston and that of say Barranco or Hopkins Village, it is evident that the SIL investigators did on occasion perceive and record /ei/ as /e/, especially when nasalized as in: wen (twenty) and sensu (money) instead of wein and seinsu.

The SIL orthography captures a necessary distinction between /ou/ and /au/ as in touba (her side) and tauba (it will be with her).  The other two traditions use /au/ in both instances.  However, one wonders about the use of /ai/ as in adaira (to find), for this does not appear to be any more than a variant of /ei/.

Since it was not possible to obtain an SIL phonemic analysis of Garifuna, one cannot say whether this tradition regards [y] and [w] as being identical with phonemic /i/ and /u/ respectively.  Nonetheless, whatever their position in this matter, it is possible, as will be demonstrated in the final section of this paper, to specify the conditions under which Taylor’s /i/ and
/u/ become y and w in the SIL (as well as the S.J.C.) orthographic tradition.


The SIL orthographic note on nasality is reproduced below:

n, when it occurs at the end of a syllable, indicates that the preceding vowel or vowel cluster is nasalized:

lidi is written lidin
nanagu is written nan?ngun
tidou is written tidoun

? indicates a nasalized ‘y’ or ‘i’:

ligiar?g?yei is written ligiar?g??ei
yei is written ?ei
lubie is written lubi?ei (sic)
tariagu is written tari?agun
ji?rura? is written ji??rura?

The first rule, namely that of using n at the end of a syllable to indicate nasality on the preceding vowels is quite reasonable.  This practice has the advantage of being flexible in the sense that it does not try to restrict nasalization to specific vowels, but rather recognises nasalization as a shifting phenonemon within a syllable or even across syllables depending on how “tightly sealed” the syllable boundary is.  My own analysis shows that “hard” consonants seal syllable boundaries tightly, semi-vowels [y] and
[w] less so, and in the absence of either a consonant or a semi-vowel, nasality can shift freely from one syllable to another in different utterances of the same word by the same speaker.
The use of ?, to indicate nasality can be used equally well in at least some of those instances for which SIL uses ?.  For example: tar?angun, lubien and ji?nrura? instead of tar??agun, lubi?e and ji??rura?.  In the other instances where it replaces the nasal palatal glide “y”, the use of n is more defensible on the grounds that the use of the first rule would necessitate the use of two symbols, y and n instead of just one, ?.  For example:  lig?ar?g?yein instead of lig?ar?g??ei.


The only comment in the SIL orthographic note on stress is that “stress, unless marked otherwise, occurs on the first syllable of two-syllable words”.  Nothing is said about words with more than two syllables although a general rule can be formulated so that here again only the exceptions would need to be marked.  (See above to compare with Taylor’s conclusions, and see remarks on stress in Chapter 3 below).  The SIL practice is to mark stress on all words with more than two syllables, and on words with two syllables if stressed on the second syllable.


The Saint John’s College orthographic tradition dates back to the latter part of the 1940’s and Father John Stochl S.J. was the driving force behind it.  About twenty five years later it was revitalized by Father Richard Hadel S.J. with the publication of his enlarged and revised version of the original S.J.C. English – Garifuna Dictionary in 1975.  Like the SIL tradition it does not appear that the S.J.C. tradition is based on the kind of phonemic analysis that Taylor made and published.  The newer edition of the dictionary devotes two pages to explaning the orthography: on the first page, the consonants and their English equivalents are listed, and the second page lists the vowels and provides a brief explication of nasalization and stress.  Thus here again, because of the absence (or unavailability) of a phonemic analysis, our study of this tradition will necessarily be limited to an examination of the orthography as employed in the dictionary and other written texts.

There is a great deal of similarity between this and the SIL tradition.  Consequently, we will be brief in dealing with this tradition, and thereby avoid unnecessary repetition.  The points of convergence and divergence especially with SIL will be mentioned in passing.



Page iii of the revised edition of the S.J.C. dictionary is reproduced below.  It lists the symbols used for consonants in the orthography, their phonetic equivalents, and some English words to illustrate those phonetic values.

Carib    Phonemic Symbol        English
b     /b/      bill
d     /d/      dill
f     /f/      fill
g     /g/      gill
h     /h/      hill
k     /k/      kill
l     /l/      lil
m     /m/      mill
n     /n/      nil
p     /p/      pill
r     /r/      rill
s     /s/      sill
t     /t/      till
w     /w/      will
y     /y/      yet
ch     /c/  /s/     chill shall
j (foreign loan words only) /j/      Jill
?     /?/      –

Notice that like the SIL, S.J.C. includes /w/ and /y/ in its phonemic inventory.  Indeed, except for the S.J.C. use of an addition symbol j to accommodate “foreign loan words” with [dz] and some minor differences in the choice of symbols, the two lists of consonants are identical.

The need for /j/ in the phonemic inventory is, to my mind, highly questionable.  Historically, words containing that sound were changed to conform to Garifuna phonology and in the process
[dz] ? [s].  I do not believe that a sound can be regarded as a phoneme of a language until it is accepted and tolerated in much the same way that a word cannot be regarded as a part of the lexicon of the language until it has been incorporated into the phonology of the language by undergoing the necessary changes.  I cannot think of any word in Garifuna that would warrant the inclusion of /j/ among its phonemes. (Notice that this is not the SIL /j/, which is the Spanish equivalent of the English /h/).

The S.J.C. orthography indicates that the language is analysed as having six simple oral vowels.  These are listed on page iv of the dictionary as follows:

Carib Phonemic Symbol English
a   /a/   father
e   /e/   (mid-front, unrounded)
i   /i/   meet
?   /i/   (lower-high central, unrounded)
o   /o/   boat
u   /u/   boot

With regard to the phonetic value of the vowels, the vowels are all somewhat lower than their supposed English equivalents.  They are more comparable to the Spanish values as both Taylor and the SIL point out.

It will be noted that this vowel system is identical to the SIL version.  The same is also true of the nasalized simple vowels, so nothing further will be said about them.

The following are listed as diphthongs: ai, ei, a?, au, ui, and ua.  However, an examination of the material reveals that the following is a more complete set of compound vowels:

ie (ye) i?rehani (scorn) i?deme (regret) yeg? (my pet)
ia (ya) sur?sia (doctor) ni?rati (my ability) ya (here)
(y?) y?r?d? (hummingbird)
iu (yu) y?di (my meat)
(wu) w?g?ri (man)

ua (wa) ?ati (there is none) w?ti (he called) su?mein             (to admire)
ui (wi) ?i (meat) n?i (my meat) w?ri (our name)
ue (we) du?iti (debt) has?ere (adz) wer?were (housefly)

ei   ab?leigua (to waste) ug?nei (boat)
ai    w?iriti (it is big) ur?ai (king)

au   la?ba (its side) l?uguati (it is not enough)
a?   ir?chau (truth) hur?ra? (game) a?r?da (to spread)

Like the other two traditions, S.J.C. orthography uses ei and ai as two distinct compound vowels.  Since this issue has been discussed, it will be sufficient to repeat here that those two are not distinct in Garifuna and that, as they can be used interchangeably, there is no justification for retaining the two in any orthography for the language.

au is used for the compound vowel that SIL orthography represents as ou.  This has the effect of making the ou au distinction difficult to represent in this tradition which ends up ignoring it.  The words that the SIL tradition would write as touba (her side) and tauba (it will be with her), would both be written as tauba by S.J.C.


The remaining vowels in the list are identical to their SIL counterparts.



Two rules for nasalization of vowels are given and these are again reproduced below (Hadel 1975 :pg.iv):

1 (V) ? V ? (V) ? V (both the vowel preceding and that following ? are nasalized, e.g. ha?a ? ha?a)

2 V N {   } ? V {   } ( a vowel preceding a nasal followed by another consonant or word ending is pronounced as a nasalized vowel; the nasal consonant is not pronounced e.g. nun ? nu     l?mbara ? labara    but, gum?nana and b?mena remain unchanged.

Actually, these are orthographic rules designed to show how nasalization of vowels can be recovered (in reading) and the difference between nasals used as consonants and those used as a means of indicating nasality on vowels.  However, they are also of interest because of their phonological implications, especially in the absence of the phonemic analysis on which the orthography is based.  The first rule, for example, indicates that /?/ is here considered a consonantal phoneme and that nasality on the preceding and succeeding vowels is conditioned; that is, conditioned by proximity to /?/.  This position is markedly different from that of the other two traditions and, in my view, erroneous.  The second rule reflects, and capitalises on, Garifuna word structure which tolerates very few consonant clusters and never allows consonants in word final position.

Even though the S.J.C. use of ? stems from entirely different phonological assumptions, SIL and S.J.C. orthographic conventions with regard to nasalization are virtually identical.  The only difference is that S.J.C. uses both n and m to indicate nasalization on vowels, while SIL employs n even when the following consonant is a bi-labial as in lanbara (wire).  In view of this similarity, the remarks made concerning the treatment of nasalization in the SIL tradition are equally applicable to the S.J.C. orthography.

“Stress must be included among the phonemes of Carib since it is sometimes critical in determining the meaning of a word (compare English c?ntent and cont?nt): ?riha ‘doze’ and ar?ha ‘see’; ?b?r?ha ‘drop, fall’ and ab?r?ha ‘write'”.

The above quotation from the revised dictionary is all that the S.J.C. tradition has to say about stress in Garifuna.  It must now be clear from Taylor’s analysis and, to a lesser extent, from his treatment of stress, that this S.J.C. comment is certainly inadequate.  Also unnecessary is the practice in this tradition of marking stress on every word, in view of the fact that general rules can be formulated to predict accentuation on most words on the bases of syllable structure and number of syllables.  We will return to this in the next chapter when we make proposals for a new orthography.