The Mi’kmaw Automatic Conjugator

This is the culmination of several years of my work, although it only took a few weeks to actually create. Put in any verb, and the conjugator will categorize that verb for you, and put out all (or at least most) of the forms that this verb can take. I am thrilled to be able to bring such a resource to the table, and I hope, given how useful I have found (and would have found) such a resource, that others may find it useful too.

The code for the conjugator is open-source and can be found in my GitHub repository. I’ve done this intentionally to make it possible for other language educators to model a tool off of this one, and to keep the work transparent. It is written in Go, and works using the built-in HTML templating package. The website itself is hosted on DigitalOcean and is set to rebuild every time a change to the GitHub repository is pushed.

I have intentionally made it a stand-alone site instead of rolling it into my personal website (i.e. this website) to avoid unnecessary navigation that may hinder access to the tool. The design of the conjugator is also done with this in mind.

How does it work?

This conjugator works different from other previous tools for Mi’kmaw and other Algonquian languages. I will link SayItFirst’s conjugator for another example of such a tool — another excellent resource. This conjugator (the automatic conjugator, not SayItFirst’s) works procedurally, that is, automatically. This means that it does not rely on a word bank to conjugate verbs. Instead, it looks at what the user enters, and will attempt to classify the verb by the 3rd person singular animate form, the form for nekm/negm. I modelled this on how a reader would use Father Pacifique’s grammar (Francis & Hewson, 1994). You find the model with which the targeted verb matches, and conjugate the targeted verb based on that.

The bulk of the work of the conjugator is done by the parseVerb function. I have included below a visualization of what this function does.

Use the conjugator in English, Mi’kmaw, or French!

The conjugator supports user in both Francis-Smith and Listuguj orthographies. Choose the orthography you are writing in with the radio button option above the entry field, and the conjugator will output the rest of the verb forms in the orthography you have selected. “teluisit” (to be named) is used as an example when first arriving on the page, to show what the output can look like.

All translations into Mi’kmaw and French are my own work, with input from the Mi’gmaq Online Talking Dictionary and personal communication with Gordon Francis. All translations (especially Mi’kmaw) are subject to change as more feedback is gathered.

Because of several elements of Mi’kmaw verbal morphology, the whole word must be parsed. It is not enough just to know the last letter or few letters, nor is it enough to know a broad class of endings. Even within what seems like an uncomplicated conjugation class, phonotactic alterations caused by different verb stems must be taken into account for a procedural conjugator to work. That is to say, doing this is difficult, but it eliminates the need for a large list of known verbs, all of which must be individually verified for all their forms, which may very easily become unwieldy. Instead, a procedural conjugator can make a guess at how any verb should act in Mi’kmaw, and conjugates it according to a model — this way, each verb conjugation and variant only needs to be programmed once, with the caveat that not every verb may be recognized properly. This last point is easy to resolve, though, if the conjugator is built in such a way that certain known verbs that are improperly recognized can be caught and put in their proper conjugations. The automatic conjugator is indeed built with this in mind.

What can it do? What are the limitations?

The conjugator outputs verb forms for:

• First person

• Second person

• Third person (animate, inanimate, obviative)

• Indefinite person (only in singular and plural)

The conjugator outputs, in terms of grammatical number:

• Singular

• Dual

• Plural

Most importantly, the conjugator will give you, for every verb, the following “tenses”. I put tenses in quotes because these categories in Mi’kmaw involve other ideas than just grammatical tense, including aspect and mood. For each of these “tenses”, the corresponding negative forms are also given.

• Present

• Past (affirmative, suppositive, deferential)

• Future

• Imperative

• When-conjunct

• Past when-conjunct

• If-conjunct

• Suppositive if-conjunct

• Deferential if-conjunct

• Conditional

• Attestive conditional

• Counterfactual conditional

What the conjugator cannot do is give you translations for forms, and this is because the conjugator does not know the meaning of what is being entered, only that it recognizes it as a verb. This is a disadvantage of not having a list of known, recognized words. The user must know what verb form they are looking for in the conjugator, though this is no different than what would happen if a reader is trying to match forms in Pacifique’s grammar.

Some “tenses” are also missing compared to what is given in Pacifique’s grammar. In Dr. Stephanie Inglis’ 2002 dissertation Speaker’s experience: A study of Mi’kmaw Modality, Inglis notes that the dubitative and subordinative forms have fallen out of use; as such, I have not oriented the conjugator to output these forms to reduce the overall amount of information being ‘thrown’ at the user. Moreover, Inglis notes several changes in the form of the when-conjunct compared to what is listed in Pacifique’s grammar. Since Inglis’ research is more current, I have deferred to their descriptions of the when-conjunct.

The description of the past and when/if-conjuct forms is informed very heavily by Inglis’ 2002 dissertation, especially with regard to their labels (affirmative, suppositive, deferential, and counterfactual). These are evidential markers, and although technical, they help to convey a better idea of the sense of these “tenses” than Pacifique’s grammar, which notes that one is simply more forceful than the other, which I have found unhelpful.

Because there are always quirks to verbal morphology, and the recognition patterns have flaws, there are a handful of disclaimers that pop up when certain classes of verbs are entered. These are denoted with the little “i” next to the dropdowns.

One major quirk is with verbs in -a’tl, which may have two stems that both converge in -a’tl for the 3SG>3SG.OB forms. One class has stems in -a-, e.g. aptɨsqa’tl (3SG>3SG.OB) is aptɨsqa’q (1SG>3SG); the other has stems in -e-, e.g. pesa’tl (3SG>3SG.OB) is pese’k (1SG>3SG). To avoid the user having to know more than the form for nekm, i.e. 3SG>3SG.OB, forms for both stems are listed (where they differ), and a disclaimer given that the two classes are represented at once.

A large disclaimer: Pacifique’s grammar is very inconsistent with glossing and translation in the later chapters (understandable given the sheer amount of forms), so many of the forms given for VTA verbs (conjugations 6 and 7) are tentative. I would say the forms that have an indefinite subject (e.g. elukwalimk) are especially tentative given the presentation of those forms in the grammar.

I can write in Francis-Smith or Listuguj. Why not Pacifique orthography?

The conjugator relies on the parsing of user input, in short. It is crucial for the conjugator to be able to know where there is a schwa in a verb in order to be able to classify it properly, and then deliver the correct forms. Both Francis-Smith and Listuguj orthography represent schwas well and consistently, which allows the conjugator to work properly. Pacifique orthography represents schwas inconsistently or not at all, which does not allow the conjugator to parse the user input properly.

Once the correct category is found, the Go backend will then go and find all the correct affixes for the “tenses”, and concatenate them onto the stem string. Inspiration for the tabular presentation of verbs is given in the appendix of Dr. Dianne Friesen’s 2022 dissertation A grammar of relationship: How Mi'kmaw verbs indicate the relationship between participants in a sentence.

The stem for the future, imperative, if-conjunct, and conditional “tenses” is contracted: if the vowel is e in the first syllable, remove it, and make any necessary phonotactic adjustments. This is done procedurally by the contractStem function.

Verbs must also be made negative for the negative forms of various “tenses”. For most, this is just the word “mu” beforehand, but it is in the future “ma’”, and in the imperative “mukk” for second persons. The latter is an alteration of an original prefixed form listed in Pacifique’s grammar, e.g. for teluisit, the 2nd person singular imperative negative is mukk tluisi “do not be named … !”, an alteration of Pacifique’s mu ktluisi!. The form that incorporates the original prefix into the negative particle is now by far more common.

Verbs with inanimate objects (most of the 4th and 5th conjugations) have the forms with plural objects done procedurally based on the forms with singular objects. This must consider some phonotactic and morphological allomorphy, but is pretty straightforward.

If you have made it this far in reading my description of the conjugator, I hope that this knowledge and background may be able to help another in their efforts to do something similar. To create the conjugator, I read all of Pacifique’s grammar in excruciating detail, trying to extract all the meaning I could, and present it in the best way possible. Seeing that someone else has done a similar project would, I hope, make the process easier in the future, and ideally even better. If you have any questions/concerns, do not hesitate to contact me through the form linked with the big button at the top of this page.