About a year ago or so I was contracted to write the lessons for a language learning app for Bambara (the principal Manding variety of Mali) by a French firm. The company already has another set of language-learning apps on the marketplace (most notably for African languages is Wolof) but I was really excited to get to contribute to the project for a couple of reasons. First, I want people to learn African languages like Manding. Second, I was glad to have the opportunity to write out some of the knowledge that I have gleaned about Manding over the years and formalize many of the lessons that I have given over the years into what is essentially a learner’s grammar that I’ve written for Manding.
Commercial self-promotion aside, the project also forced me investigate a number of grammatical issues that linguists and Manding-speakers themselves have long debated. While many of these issues have been resolved (that is, linguists claim to have an adequate account of certain grammatical phenomena), a number of them remain up for debate depending on you ask. While these debates can sometimes be attributed to theoretical orientations or preferences amongst scholars, there is also the issue of for whom are the grammatical analyses of linguists correct? Let’s investigate this issue by diving into an example that came up during the review of one of the lessons (by my colleague and collaborator, Antoine Fenayon of Association Donniyakadi that promotes Bambara via regular classes and books out of Paris):
The issue is this: how does one formulate progressive sentences in Bambara?
From studies of Manding and my travels and research in Mali, I am well familiar with the most typical way of formulating progressive sentences in Mali via a word between that subject and the verb that variably marks Tense, Aspect and Mood (as such some linguists call this a TAM marker, though in the Mande linguistic tradition it is most typically called a predicate marker).
This is what Valentin Vydrin refers to as Progressive-1 (PROG-1). As any good student of Bambara, I included this in the textbook. Il n’y a pas de problème. Problems arose however when I included another way of expressing the expressive in Bambara which we’ll call Progressive-2 (PROG-2) for convenience’s sake:
I offered the following example:
N bɛ takisi bolila ‘I am driving (a) taxi’
However my reviewer pointed out that he doubted that this form existed in Bambara and offered the following alternative:
N bɛ takisiboli la ‘I am driving (a) taxi’
Note that based on orthography alone, the only thing that has changed is a word boundary:
…takisi bolila VS. takisiboli la
By what principle was Antoine disagreeing with me? To understand the divergence, we must look at what is called compacité tonale or ‘tonal compactness’ in Manding. But to get there we need to unpack a number of features of Manding grammar.
Manding varieties are tonal languages. What does this mean? In the case of most Manding varieties this means a number of things, but for now we’ll only focus on the fact that it means that words can be distinguished by the relative pitch of one’s voice vis-à-vis other words (linguists call this lexical tone). Here’s an example:
bá ‘river’ ≠ bà ‘goat’
To capture this kind of distinction in Manding, it necessary to have a system for marking tone. Linguists have opted to mark tone by using what are typically known to Westerners as “accents” (that is, the little marks that go above letters) but are more technically referred to as diacritics. Different languages use diacritics for different purposes. In French for instance, è and é refer to distinct phonemes (that is, sound categories) whereas in Manding orthography, these same diacritics would distinguish a low and a high tone /e/ respectively.
Compound Words Meet Tone
Manding varieties also make ample use of compounding to create new words. Compounding refers to two words that are combined to together to form a new word (ex. door + knob = doorknob). Let’s look at an example:
nɛgɛ ‘iron’ + so ‘horse’ = nɛgɛso ‘bicycle’ (lit. ‘iron horse’)
Seems simple enough. But we know that Manding words can be distinguished by relative pitch. What if we were marking tone? What happens when two words with their distinct tones meet in a new compound word? Does each constituent word retain its original tone? Let’s take a look:
What accounts for this outcome? This is where tonal compactness comes in. Tonal compactness refers to the phenomena of the underlying tone of a headword in a compound word propagating across the entirety of the constituent elements of a compound word. Let’s look at the underlying tones of nɛ̀gɛ́ and sǒ:
Both have the underlying pattern of Low to High (LH). As such, when they combine into a compound word, the first constituent’s (or headword’s, which in this case is nɛ̀gɛ) underlying pattern of LH spreads across the entirety of the new word:
Progressive Constructions Meet Tone
With these issues covered, we can proceed to the matter at hand: progressive constructions in Bambara and specifically how one is to write the Progressive-2 form. Let’s compare again the differing options proposed by Antoine and myself:
Given these two orthographic examples, it is clear that the difference is one of word boundary. Does boli belong with -la or does it belong with takisi? By what criteria are Antoine and I arriving at our divergent interpretations?
While the schoolmarm may classify parts of speech by focusing on meaning (e.g., “a noun is a person place or idea”), linguists rely on so-called formal criteria stemming from grammar in the linguist’s sense (that is, phonology, morphology, syntax). Regardless of one’s approach, these classifications are then used to decide upon some of the conventions for writing down a language. While neither Antoine nor I marked tone via diacritics in our discussion of Manding orthography, we nonetheless relied upon it as a formal system by which to judge word boundaries. But who was right?
From Antoine’s perspective, example number 3 was right because it more accurately respected the tonal system of Manding. This view is similarly confirmed in a number of Manding reference grammars such as most notably Charles Bailleul’s. For Bailleul, the tendency to want to mark the progressive via the suffix -la onto the verb of the sentence (as in example 2) is an unfortunate error because tonally this is not correct in Bambara. On this account the proper analysis of the sentence at hand would be parsed as follows:
On this reading, we are dealing not with a suffix -la tacked onto the verb but rather a non-verbal progressive construction in which the postposition lá ‘on, in, at’ follows a compound noun tàkisibóli ‘the act of taxi driving’. While this struck me as a possible utterance and interpretation, my own experience as a Manding speaker (albeit one most familiar with Jula over Bambara) suggested that Bailleul’s conclusion about the inaccuracy of the -la suffix progressive being incorrect was too strong — I felt that I regularly used and heard it. Could it be that Bailleul was wrong or is one of the markers of the dialect differences between Jula and Bambara their ways of expressing the progressive?
To began to respond to these questions, I did a own little investigation with a good friend, who is a native speaker of Jula originally from Bouaké.
Without any prompting I, in Manding, asked my friend how he would say the following sentence which I said in French: Ma mère est en train de piler du mil ‘My mom is grinding millet’. To this my friend initially responded as follows:
To confirm what I this was, let’s look at this utterance under a spectrogram for phonetic analysis of the fundamental frequency or pitch of an utterance. This allows to formally measure the shifts of tone within an utterance.
I then asked him what it would mean if someone altered the tonal scheme as follows (this is a recording of my friend parroting my speech):
This looks as follows under the spectrogram:
For my friend this utterance was also possible and meant “My mother is grinding millet” albeit with an important nuance that we’ll explore further later. For now note two things:
- In green: the tone of the word ɲɔ ‘millet’ is subtly distinct in the
two utterances. In the first one there is a slight U-shape to that segment of the recording signaling a Low-to-High tone. In the second one, the line is simply a downhill curve that never comes back up that signals a true low tone. This is important because a Low-to-High tone on ɲɔ in the first example signals that it is the standalone noun form of the word. This would confirm the hypothesis that example 1 is a verbal progressive construction with ɲɔ̌ being the direct object.
- In the orange box: In both cases there is a progressive drop in toneacross susu to la. That said, it is essential to note in the first example that the one of susu drops markedly following ɲɔ in example 1 whereas in example 2 the tone of susu is higher than the tone of ɲɔ.
Also note that in example 1 the pitch of the second of susu (starting around the second segment of blue line within the orange box) is nearly identical pitch of the following la, where as in example 2 there is a marked drop from the pitch of the second vowel susu to the following la. This can be attributed to the floating low-tone article that typically follows nouns in Manding (it is marked by the presence of the apostrophe and glossed as ART)
The combination of these insights suggests indeed that indeed there is a verbal progressive construction that utilizes the suffix -la in Manding. Keep in mind that Bailleul (and Antoine) however were suggesting that there is no such construction in Bambara whereas my insights come from a Jula-speaker (this ignores the fact that he regularly refers to what the language that he speaks as Bambara). Bracketing off the issue of dialect/register distinctions in Manding for the time being, let’s dive a little further into these two constructions that my friend produced. Are the two utterances in fact identical? Fully glossed and marked for tone after our spectrogram analysis, they look as follows:
The first one (example 5) verbal construction that marks the progressive via a suffix. The second one (example 6) is a non-verbal locative (as in location) construction that uses a noun and the postposition lá. Let’s refer to them as a verbal suffix progressive and a locative progressive respectively. While both can be construed as progressive au sens large, there was an important nuance between the two sentences for my friend. The verbal suffix progressive as a response to a query about his mother would lead to a response like “Oh ok, I didn’t know, I’ll come back later”. The locative progressive on the other hand would lead to a response like “Where?” or “For who?” What semantic nuance is this?
Further probing with my friend and confirmation with another Jula-speaking friend from Bobo-Dioulasso revealed that the distinction is as follows:
However, I think that there is an important caveat. While at first the distinction was clear for my friends and they all proffered the verbal suffix progressive as their natural response when I prompted them for a progressive sentence, the distinction became increasingly tenuous the longer we discussed.
I’d argue that stems in part from the fact that Bambara and Jula are not hermetically-sealed languages or varieties. While most Jula speakers may typically use the verbal suffix construction to express the progressive and the locative construction to express the habitual, they also receptively “know” that the semantic meaning of these constructions varies. Indeed, assuming that Bailleul’s analysis is correct, in Bambara the locative construction is prototypically used to represent the progressive and not the habitual. But speakers of Jula know this. They aren’t confused when Bambara speakers use the locative construction. In fact, under the right conditions (such as my extended elicitation session with a White American friend for instance) it is possible that Jula speakers report the locative constructive as the preferred way of expressing the progressive when in fact their observed behavior is quite distinct. Bambara speakers (such as Bailleul’s research participants) on the other hand may not report that it is possible to express the progressive in the prototypically Jula way (that is, with a verbal suffix construction) because either they are unaware of this variability or because of normative sociological pressure that gives them a sense of what is “good”, “correct” or “appropriate”. In this sense, Bambara and Jula progressive constructions and the tones that distinguish them are not just dialectal variants but also potential tokens of register distinctions in West African society.
While I am unaware of the institutions or forces that may be contributing to this particular register distinction between Bambara and Jula, an exploration of the approach to progressives and these constructions in N’ko (ߒߞߏ) can provide some potentially unique insights. As an orthographic system for Manding that obligatorily fully marks tone, N’ko necessarily has already dealt this issue of how to properly express the progressive. In N’ko grammar the two constructions are distinguished as follows:
Applying this aspect to our example gives the following:
Applying this to our example gives the following:
Note that this is the exact opposite of the what I elicited from my friend from Bouaké and confirmed with my friend from Bobo! N’ko books and structures such as classes and radio shows therefore function as a locus of an alternative linguistic norm for progressive and habitual constructions that differ from that which is typically used by Jula speakers. On the other hand, this N’ko interpretation lines up well with the progressive form that is proscribed by so-called standard Bambara grammar and reinforces its already circling model of usage.
So where does this all leave us? Tentatively…
- Prototypically Jula progressive constructions differ from those of Bambara.
- Jula uses a verbal suffix construction for the progressive
- The locative progressive construction in Bambara also exists in Jula but is typically used to refer habitual actions.
- Jula and Bambara speakers often have receptive or performative competence in both usages and meanings and insisting on their distinctness potentially obscures their actual usage in social life.
- Promoting Manding language literacy and standardization requires focusing on how tone plays into grammatical divergences between varieties and registers.