Talk:Chewa language

ŵa
Can you explain the meaning of the diacritic sign you have added under the transcription ? Is it necessary? (especially as it isn't found in the sources which were used to compile the article) Kanjuzi (talk) 17:40, 29 June 2024 (UTC)


 * It means that it's an approximant. Without the diacritic, we claim that it's a fricative. Sources are often sloppy about this, if they expect their audience to understand the difference, but we can't make that assumption. The Spanish sound has been argued to lie somewhere between a fricative and an approximant, but AFAICT most sources use the diacritic when they're being precise. From the description, "closely lip-rounded [w] with the tongue in the close-i position", it sounds like it truly is an approximant.
 * I see we don't use the diacritic in our Spanish articles, so we might leave it out of the Spanish transcription here for consistency. It's ambiguous in the case of Spanish anyway.
 * We have another problem. We list /ɽ, ɽʷ/ as approximants, but [ɽ] is a flap. An approximant would be $⟨ɻ⟩$. From the article text, it sounds like they should be in a separate row. It might be a lateral flap [𝼈], but that's not the level of detail you can always expect from sources. — kwami (talk) 17:49, 29 June 2024 (UTC)
 * Kishindo calls it a fricative, but that word is sometimes used for approximants. Atkins calls it a semi-vowel, which would be odd if it actually were a fricative. It would be good to have a recent phonetic description. The sources we're using are for orthography, with the phonetics secondary. — kwami (talk) 18:10, 29 June 2024 (UTC)
 * Downing & Mtenje say it's a fricative. We should switch to that as our source. — kwami (talk) 18:15, 29 June 2024 (UTC)
 * Source that L is reflexive? D&M say it's a lateral flap, others that it's "dark". — kwami (talk) 08:31, 30 June 2024 (UTC)

, ŵ is not a fricative. If you think it is, you haven't heard it. Listen for example to President Chakwera's address (top of the External Links), where he says madera ŵathu (3.21), maiko ŵena (3.28), ŵapolisi (3.45), tiŵapatse (4.34), ŵayamba (5.44). It sounds to European ears almost exactly the same as a w, except for the tongue position, and it melts away entirely to w in front of a /u/ or /o/; so it is no more a fricative than /w/ is a fricative. Downing and Mtenje is not a reliable source for the phonetics of consonants. Neither of the authors is a phonetician, and they have not cited or conducted any phonetic study based on real data. For example, on page 46 they describe the sound of ps and bz as "whistled", a nonsensical and inaccurate label. The main thrust of their book is to describe the tonal patterns of tenses, which is very useful, but even in this part there are a number of errors. I'm happy for l/r to be described as a flap or liquid rather than an approximant. I don't think any of the sounds in Chichewa are retroflex. Again, listen to Chakwera's address and see if you think so. Kanjuzi (talk) 08:46, 30 June 2024 (UTC)


 * I'm only going off available sources.
 * Is Chakwera speaking a dialect with /ŵ/? (From the vid it looks labiodental, maybe [ʋ].)
 * Do we have a phonetic source for Chewa? — kwami (talk) 08:54, 30 June 2024 (UTC)

Yes, Chakwera is from the older generation who still use ŵ, although younger people have mostly dropped it: they substitute w in some words, and omit the sound altogether in others. Moreover, unlike the last three presidents, he is actually a native-born Chewa from the Central Region of Malawi; he speaks it as a native language, not a second-language lingua franca. Another good example is onseŵa ŵalunjika (7.34), where there are two ŵ's in a row. However, he uses ŵ sporadically; it can't be heard in every place. Again, I repeat, I would not rely on Downing and Mtenje. Their table of consonants on page 11 is incomplete (they mention explosive d, and ps and bz only later in the book).

The terminology used in the book seems to be idiosyncratic to say the least. For example, in the tense names, ndadya 'I have eaten (just now)' is called the "perfective" tense; ndimapíta 'I was going (just now)' is called "past habitual" (past habitual is one of its meanings but not the basic one); the ná tense, as in masíkú ano zinthu zinásintha 'these days things have changed', is described as the "simple past", which it can be, but really it is more of a perfect; the -nga- tense, as in mungagunde mténgo 'you might crash into a tree', is called "permissive"; the -ta- tense, as in ndítápíta 'after I went' or 'having gone', is called the "sequential perfective", a name not used by any other linguist I believe.

The idea that ŵ is the same as the Spanish b/v derives from some descriptions by people who were not trained linguists. Atkins' description of ŵ as "a closely lip-rounded with the tongue in the close-i position" is much more accurate, although watching President Chakwera it doesn't seem all that lip-rounded. The main difference with the ordinary w is that the tongue is a little bit more forward in the mouth. Kanjuzi (talk) 10:11, 30 June 2024 (UTC)


 * That makes it sound like [ɥ], which is clearly not what he's saying. — kwami (talk) 10:43, 30 June 2024 (UTC)

The point I'm making is that when Mtenje calls it a fricative, he may not be being very accurate in his terminology. I think you should add your little diacritic again and place ŵ alongside w and y, since they are all similar sounds, as you can hear. It's different, at any rate, from the Spanish b/v. It also isn't like [ɥ] (presumably that is the sound in French huit): the lips aren't so rounded. I don't know anyone who has described it really well. It also occurs in Chitumbuka and other languages of northern Malawi, especially Chilambya. Kanjuzi (talk) 11:16, 30 June 2024 (UTC)


 * You just told us that we should “not” add the diacritic. Stop contradicting yourself. And what makes you state that Downing & Mtenje (2017) is “not” a good source? It is published by Oxford University Press in the department of linguistics. I truly don’t think that that’s an “unreliable” source. Fdom5997 (talk) 14:19, 30 June 2024 (UTC)
 * The word "fricative" often includes approximants, when ppl aren't being precise. We also have a source that it's a semivowel. I put it in the fricative row of the table because that's where D&M put it. But given the contradiction of our sources, I don't have a problem with following Kanjuzi's testimony that it's an approximant. — kwami (talk) 19:55, 30 June 2024 (UTC)
 * Until we find RS's to settle this, I made /ŵ/ span the fricative and approximant rows in the chart, as we do with the IPA chart, and added a note that sources are contradictory. Does that work for people? It's not ideal, but hopefully temporary.
 * Kanjuzi, do you have a source for Chilambya? That might be helpful. — kwami (talk) 20:02, 30 June 2024 (UTC)
 * I say just ignore the contradicting sources. Whichever sound (fricative or approximant) is cited the most should be what goes in. Fdom5997 (talk) 21:45, 30 June 2024 (UTC)
 * Facts are not subject to a vote. This isn't COMMONNAME. 'Fricative' can be ambiguous, and we have a knowledgeable native speaker saying it's an approximant. We also have a source saying it's a semivowel, which is a lot less ambiguous than 'fricative'. — kwami (talk) 21:52, 30 June 2024 (UTC)

Why are you messing about with the table of consonants? Why have you moved [w] to be a "labialised velar/palatal?" It's neither velar nor palatal. Kanjuzi (talk) 21:14, 1 July 2024 (UTC)

Secondly, why have you put m. p, ph, mph in a separate column from f, mf, v, mv? Obviously the first set are labial and the second set are labio-dental, but from the point of view of structure they all seem to belong in the same column.

At any rate I don't think that w should be placed in the column "velarised labio-palatal". In most linguistic descriptions, w is regarded as simply a labial glide. In the structure of Chichewa, it could be a velarised labial, or a plain labial, but not a velarised labio-palatal. Do you have any source for that? Kanjuzi (talk) 21:35, 1 July 2024 (UTC)


 * I never even did any of that. Ask @Kwamikagami Fdom5997 (talk) 21:44, 1 July 2024 (UTC)

Oh yes, so it is! I beg your pardon. , why have you put w in with the velar/palatals in both tables? Surely it doesn't belong there. If you are following Downing and Mtenje (p. 11), they put it under "labial". Kanjuzi (talk) 06:25, 2 July 2024 (UTC)


 * Is it labial? If so, it's not [w]. [w] is labialized velar. In the full table, that's easy, because we have a labialized velar column. For the short table, it's either create another column just for /w/ (like we do for /j/), or fudge it and put it somewhere else. I usually see it under velar; some ppl put it under both labial and velar. If we're going to be precise, though, it should have its own column, it's just that most sources don't bother because they assume everyone knows what a 'w' is.
 * As for the separate columns for labiodental and bilabial, that makes sense if /ŵ/ and /v/ are only distinguished by place. — kwami (talk) 06:41, 2 July 2024 (UTC)

Let me think about that. Meanwhile, as for Chilambya, the only information I have is that found in the article Lambya language. The University of Malawi Language Mapping Survey for Northern Malawi (2006) cites sentences such as: Po ŵovyawa ŵunuŵu ŵosi ŵaŵiri ŵakiza ŵapangana ukuti ŵaye kwa mwene aye aŵalonganie 'Since this was the situation, they both decided that they should go to the chief so that he could decide between them', which is taking the use of the sound to extremes! Kanjuzi (talk) 06:46, 2 July 2024 (UTC)

Now as for "ps", it should be placed in the aspirated row. Structurally it is the palatised form of "ph" i.e. . It is pronounced a little bit like, that is, p followed by an ich-laut. Anyway, it is definitely aspirated. The palatalised form of unaspirated "p" is "py" (which is rather rare and only used in one or two words such as -pyola '(to) exceed'). Kanjuzi (talk) 07:10, 2 July 2024 (UTC)

But a more serious objection to your rearrangement of the table is that you have placed ch, tch, ntch, j, nj in the same column as k, kh, nkh, g, ng. They don't belong there. The whole point of the table is to show how, structurally, each of the three places of articulation (lips, alveolum, palate/velum) can be plain, palatalised (i.e. followed by ), or velarized (i.e. followed by ). This applies to the velar stops as well. When k is palatalised, it doesn't become ky (which doesn't exist in Chichewa), but ch; similarly palatalised g is not gy but j, and so on. So we have (for example): ph, ps, phw; th, thy, thw; kh, tch, khw. There should be three columns for "velar/palatal". Kanjuzi (talk) 07:31, 2 July 2024 (UTC) (By the way, when I wrote "labialised labial" above, this was a slip for "velarised labial". I have corrected it.) Kanjuzi (talk) 07:37, 2 July 2024 (UTC)

Or perhaps I do mean that "w" is a "labialised labial", though that is a bit absurd. The question is, should it be placed in the same column as "m", the same column as "mw", or the same column as "ng'w"? Do you have a good reason for placing it in the same column as "ng'w"? Kanjuzi (talk) 08:38, 2 July 2024 (UTC)

ŵ was presumably formerly a fricative and before that it was presumably a stop. Compare Zulu abantu ababili 'two people', Luganda abantu babiri, Shona vanhu vaviri, Old Chichewa ŵanthu ŵaŵiri, Modern Standard Chichewa anthu awiri, Swahili watu wawiri. To me it seems that in the speech of those people who still use it, it has partially or completely lost its fricative quality. It is also difficult to know which column to place it in, since its articulation isn't quite like any other sound in Chichewa. Kanjuzi (talk) 08:38, 2 July 2024 (UTC)

I don't say that the Phonology of Chichewa is entirely unreliable. It is a huge advance on what was published before. But there are certain things in it which might be corrected. For example, is it correct to say that ndadya 'I have eaten' is the "perfective" tense? (This is presumably Mtenje's terminology, since he uses it in earlier publications written by himself alone.) There are many other questionable things in the book, such as his statement (pp. 61, 118) that lálánjé 'an orange' is borrowed from English (it isn't). So while most of the book is reliable, not everything in it is correct. Kanjuzi (talk) 09:10, 2 July 2024 (UTC)


 * Also, if the unitary analysis isn’t based on Atkins (1950), then which is it? I believe you added then phonology back in 2016 or so (based on the page’s history). Where did you get that info from? Fdom5997 (talk) 15:24, 2 July 2024 (UTC)

The way I had it back then, before it was completely destroyed and changed, was much neater and more logical. It looked like this:

The table doesn't come from anywhere. It is simply an attempt to list all the possible combinations of consonants found in Chichewa and to arrange them on a grid according to their features: (a) their place of articulation and (b) whether they are voiced, unvoiced, aspirated, nasalised, or nasalised aspirated and so on. As you see, it all fits together very nicely. I don't understand why anyone had to change it all round and put the horizontal lines vertical and vice versa. This is how the logic of the language dictates it; we shouldn't try to make it fit the model of a European language, and separate "ch" from "k" on the grounds that the former is affricated. You will also notice that I have transcribed "ny" as not as. This is one of Mtenje's errors, I believe. In Malawian Tonga both and  exist, e.g. nyama 'animal', with ; kuny'a 'it is raining', with. They are separate phonemes. Chichewa nyama 'animal' is pronounced with the first of these,, just as in Tonga, not with ; i.e. it is a palatalised dental, not a palatalised velar consonant. I also added a vowel after each consonant or consonant combination, since the Chichewa consonants are always followed by a vowel, never found at the end of a word. This also is dictated by the logic of the language: most of the consonants cannot be pronounced unless a vowel follows. Arranging the consonants this way also makes it easy to see where there are gaps; for example, there are no words in Chichewa with "nwa" (although it is found in Chitumbuka). The table as it stands at present in the article is a mess, and it doesn't fit together well. Kanjuzi (talk) 22:46, 2 July 2024 (UTC)


 * To answer your claim, it is much more organized now than it ever was. Yes, there are rules to making the phonological charts in this sense, because they are easier for most viewers to read. You cannot make your own rules here, and what you consider "organized" is never always going to agree with the majority. Yes we should separate "ch" from "k" if one of them is an affricate. And how do you know what Mtenje's errors are? If it's palatal, it's palatal. That's not an error. Fdom5997 (talk) 23:23, 2 July 2024 (UTC)
 * "If it's palatal, it's palatal." No, that's not the case. As I said above, Malawian Tonga distinguishes the palatalised alveolar /nʲ/ from the palatalised velar /ɲ/. They are both palatal, for sure, but they are different sounds, the first being articulated further forward in the mouth than the second. To call them both simply /ɲ/ is being inexact. Kanjuzi (talk) 08:06, 3 July 2024 (UTC)

But, you still haven't explained why you have put /w/ in the velar column. Perhaps it is velar, I don't know, but you have cited no source for this. Downing & Mtenje call it simply a labial, and if you are following them, it should be put in the labial column. It is also a very common sound, akin to /m/, and alternating with /m/ in the concords of class 1 and class 3, whereas labialised velar sounds like /ngw/ are much less common. Perhaps phonetically it belongs in both columns. – And what is your source (if it was you that wrote this) for the statement that /l~r/ is a retroflex flap? Downing & Mtenje say nothing about that, but simply state (p. 44): "Unfortunately, there is no phonetic study of this Chichewa liquid consonant, to allow us to more closely pin down its articulatory quality." If that is so, the statement about the quality of /l~r/ should be omitted from the article. – I am also a bit dubious about putting /ŵ/ in the same column as /m/, /b/ etc. Its articulation seems to be a little different from the other consonants in this column. However, it does seem to alternate with /b/ and /w/ (e.g. bére 'breast', maŵére or mawére 'breasts'). Note that Mark Hanna Watkins (1937:13) pointed out that before /u/ and /o/, /ŵ/ is indistinguishable from /w/. He writes: "Before the vowels u and o, this sound is changed to w. In a little experiment, I had my informant underscore the bilabial fricative sound in order to distinguish it from the semi-vowel w. (He wrote w for both sounds, as the missionaries did not distinguish between the two.) He underscored the sound before the vowels mentioned, but could not pronounce it in that position in spite of his feeling that it was the same as in other positions." (Watkins' informant was Kamuzu Banda, who later became the first President of Malawi.) – Incidentally, Watkins also considered "ny" (which he wrote ñ) to be prepalatal like "y" rather than postalveolar like "ch", so you're in good company. He writes (p. 14): "ñ has the value of ny in English canyon or ñ in Spanish año." Kanjuzi (talk) 09:18, 3 July 2024 (UTC)

I suppose the difference between /nʲ/ and /ɲ/ is similar to the distinction in English between due and Jew (or between Tuesday and choose a day). Acoustically, they're very similar, and I suppose some people confuse them. But (at least as I pronounce them) due has a palatalised alveolar /dʲ/ while Jew has a palatal sound /d͡ʒ/; the tongue position is slightly different. My informant, a competent and well-educated linguist, who speaks Tonga, Chichewa and also fluent English and French, was quite clear on this point: in his view, Chichewa nyama 'animal' has the same palatalised alveolar sound as Tonga nyama, not the palatal /ɲ/ of Tonga kuny'a 'it is raining'. Possibly in the south of Malawi, where Professor Mtenje comes from, the pronunciation is different. However, since nothing has been published on this question, we had better go with the flow and copy what Downing & Mtenje say in their book. Kanjuzi (talk) 19:01, 3 July 2024 (UTC)

Out of interest I thought I would check with an expert from the University of Malawi Centre for Language Studies. He writes: "For me the Chichewa 'ny' is more of a palatalised alveolar. I also understand Tonga and the 'ny' in kunya (rain) is palatal. However, I know of some varieties of Chichewa where nyama is pronounced as the Tonga kunya. But these are localised varieties I found in some parts of Ntchisi and Salima. Interestingly, these pronunciations seem to be associated with youth language in those areas." Kanjuzi (talk) 10:11, 5 July 2024 (UTC)