Proto-Algonquian Phonological and Morphophonological Rules

Last updated: April 25, 2024

This post will collect and exemplify the most important phonological and morphophonological processes of Proto-Algonquian; it represents a substantial revision and expansion of a section that was originally found in my post on Proto-Algonquian. (For an overall introduction to Proto-Algonquian itself, you can find that post here, and for an introduction to verb inflection specifically, including some of the technical terms used here, see this one. I have not finished revising these to reflect my current beliefs and knowledge or for consistency with this post, though . . . )

While I make no pretense to total comprehensive coverage—which would be nearly impossible—I hope this can prove of use as, at least, the most comprehensive such list available, consolidated in one place and carefully ordered. Virtually none of the given rules are “new,” but to my knowledge they have never been assembled together like this before, exactly—only smaller subsets, often scattered throughout a work, and usually without statements on ordering.

In some cases, of course, there will be alternative ways of formulating the suite of rules to produce the same output. (And of course, as discussed below, different researchers don’t agree on the same rules in the first place.) As just noted, although sometimes the order doesn’t matter to give the correct output, many rules are consciously ordered with respect to one another, and such relationships will be stated explicitly for each relevant rule. Incidentally, note that these are synchronic rules, not an attempt to capture any diachronic order or picture. For example, several of the rules are ordered in a way that clearly does not match the order in which the sound changes responsible must have occurred in pre-Proto-Algonquian, and some others, despite being presented as a single rule, encompass the results and morphologization or lexicalization of what were, historically, multiple different sound changes occurring at quite different times.

Several rules, in particular those involving contraction of vowel-semivowel-vowel sequences (*VGV, i.e. *VwV or *VyV), have yet to be completely worked out for PA. These *VGV contractions specifically are addressed in their own separate section at the end.

There remain a handful of areas in this post which require tweaking or flesh out, but I’m just so exhausted from working on it for months that I’m publishing it as it is now. Sorry. I welcome any comments, discussion, critiques, or alerts to errors.

(And yes, I’m aware these sorts of rules have not been in vogue in linguistics for like three decades, but whatever, I’m not writing up an OT treatment of Proto-Algonquian. I’ll leave that to others. Not the biggest fan of most of OT anyway.)

Table of Contents

Introduction Plan and Layout Notation Terminology and Glossing Sources, Transcription, and Abbreviations A Note on Different Reconstruction Models List of Rules [see there for table with more detailed breakdown] Rule 1a, Rule 1b, Rule 2, Rule 3, Rule 4a, Rule 4b, Rule 4c, Rule 5a, Rule 5b Rule 6, Rule 7, Rule 8a, Rule 8b, Rule 8c, Rule 9a, Rule 9b, Rule 9c, Rule 9d, Rule 9e, Rule 10 Rule 11, Rule 12, Rule 13a, Rule 13b, Rule 14, Rule 15 Rule 16, Rule 17, Rule 18, Rule 19, Rule 20a, Rule 20b Rule 21, Rule 22, Rule 23, Rule 24, Rule 25 Rule 26, Rule 27a, Rule 27b, Rule 28, Rule 29 Coda Coda ex. 1, Coda ex. 2, Coda ex. 3, Coda ex. 4 Contraction [see there for table with more detailed breakdown] Note on General Patterns *awe, *aye, *a·we, *ewe Other *VGe *Vwe in Verb Inflection, incl: *‑Vw TA Stems, Within Inflectional Endings, Contraction with Mode Suffixes Other Patterns *ayi, *aya, *eye Summary of Possible Outcomes Sources Used

Plan and Layout

I’ll first provide a prose explanation of each change, then a formal notation, followed by one or more examples in both PA and modern languages when possible. In some instances, in the listing of morphemes in the input of a given example I will leave out some morpheme that is irrelevant to the rule at issue, even though it will then show up in the output, but this should hopefully not cause any confusion. In all examples, the input will be given at more or less the deepest possible level, and often intermediate forms, marked with double asterisks, will be provided before the final output in order to help illustrate the processes under discussion or simply to maintain maximum clarity. I’ve made my best effort to have the rules apply in proper order in such intermediate forms, but have not always succeeded (and occasionally have purposely deviated when this interfered with clarity).

In at least a few cases, the cited PA form likely or certainly did not exist as such; all of the phonological and morphophonemic processes such pseudo-PA and equivalent daughter forms are being used to illustrate are valid, however. A few especially uncertain or problematic reconstructions are preceded by a query “?”. Additionally, in some cases one or more daughter forms used in an example don’t reflect a complete, regular descent from the cited PA form, due to various analogies, the requirement to use additional particles to form a grammatical construction, etc.; these are sometimes simply ignored, while in other cases I enclose the non-cognate material in [square brackets].

Notation

In the formal notation of each rule, the abbreviations and symbols below are used. There is also a special morphophoneme |ẅ|, respelled as surface *w.

/	precedes specification of the environment in which the rule applies
__	position in which the affected segment(s) is/are found
#	word boundary
+	morpheme boundary
~	reduplication boundary
≠	except in the defined environment; OR except for the defined sound or class
{a,b,c}	all or any of a, b, and c; a member of the set comprising a, b, and c
α	a given value out of a set of possible values
Ø	null
C	consonant
G	semivowel
N	nasal consonant
V	vowel
Vː	long vowel
V̆	short vowel
X	any amount of undefined material
POA	“place of articulation”

Terminology and Glossing

For the most part I use the standard repertoire of Algonquianist technical terms and abbreviations—see the end of this section for a few exceptions—and some of these can unfortunately be confusing or opaque to outsiders. Basic introductions to relevant terms, etc., may be found on Will Oxford’s website here and in my posts on Proto-Algonquian (here) and Proto-Algonquian verbs (here); see also a few specific explanations below. For any Algonquianists, note I do depart from standard practice in a few ways. First, I capitalize the terms Initial, Medial, Final, and so one when they are referring to specific portions of the morphology of an Algonquian word, largely to avoid any possible confusion with use of the same terms to describe positions within a word, morpheme, phrase, etc. Second, my glossing practice hews much closer to the Leipzig conventions than traditional Algonquianist practice, especially in the area of person marking, where, for instance, I use “1pl.INCL” where most Algonquianists would use “12.” This sacrifices compactness in order to achieve greater clarity for general linguists. See here for a full list of the glossing abbreviations I employ; for those viewing this page on a computer, all abbreviations can also be moused over to see a pop-up giving the full significance.

A few miscellaneous notes on specifics. As an extremely brief primer on the subject: PA had four major verb classes, depending on the verb’s transitivity and the animacy of its absolutive argument. The names for these classes are of the form “(subject animacy) – transitivity – (object animacy)”:

1. Animate Intransitive (AI) verbs = intransitive with an animate subject
2. Inanimate Intransitive (II) verbs = intransitive with an inanimate subject
3. Transitive Animate (TA) verbs = transitive with an animate object
4. Transitive Inanimate (TI) verbs = transitive with an inanimate object

TI verbs can further be divided into three subclasses, here abbreviated TI(1), TI(2), and TI(3). Verbs could be inflected in any of three major inflectional patterns known as orders: the independent, used mainly in main clauses, the conjunct, used prototypically in subordinate clauses and to form participles (but with further complications), and the imperative, used in commands. The basic TA templates for inflectional elements employed in argument-marking were as below; other verb classes were similar, just with some slots absent.

Prefixes, and central and peripheral endings, marked the person, number, and obviation of the subject and object, but (in the independent, and sometimes in the conjunct) did not specify which of the central or peripheral participant was the subject and which was the object. The theme sign accomplished this by either marking the person of the object, or indicating that the verb was inverse. (In TI verbs, the theme sign just vacuously marked that the object was inanimate and its shape defined the subclass of the verb as TI(1) vs. TI(2) vs. TI(3).) In the independent only, there were four formatives, three of them meaningless submorphemic elements, but one still meaningful, and marking certain third persons.

When glossing inflected verbs, parentheses enclose an indefinite object of so-called absolute verbs, e.g., “I see (s.o.)” stands for “I see an indefinite animate object,” such as “I see [a bear],” and so on. With stems containing a relative root, or otherwise licensing a relative root complement, I follow Goddard’s practice of including in the gloss an indication of the general type of complement licensed by the stem, enclosed in {curly braces} (e.g., “to {somewhere},” “{so}, {thus},” etc.)—the intention being that this gloss is a sort of variable, rather than the more specific translation that would be appropriate given full context.

When the gloss of a reflex in a modern language differs substantially from that of the PA example, this will be provided; however, I omit new glosses when the only difference is that: (1) a modern language has lost distinctive absolute verbs, and so a given verb’s object might be definite even though in PA it was indefinite; or (2) a modern language has lost number distinctions in obviative verb inflections, such that, e.g., an obviative object might be singular or plural even though in PA it was only one or the other.

There were two main (and a few minor) slots in PA verbs containing suffixes with tense/aspectual/modal/evidential/mirative and similar meanings. I think there is utility in using distinctive terminology for a couple of the different slots, and do so in some other places, but for our purposes here only one set of suffixes is really relevant and in need of being identified collectively (viz., chiefly *‑pan “emphatic past,” *‑(e)sahan “emphatic present,” and *‑etoke·h “dubitative”). In this post I therefore follow the general Algonquianist practice of calling all of these suffixes mode suffixes (aka “modal suffixes” or “mode signs”); the categories they mark are modes. The precise functions of the various modes in PA are often not known with certainty or not conducive to a single gloss in the absence of further context.^[1] As such my translation of them is decidedly approximate, and ultimately not important for this post, which is concerned with formal issues.

One notable distinction made here will be between phonological processes I call contraction and coalescence; these have gone by various different names in the literature, most often both being subsumed under the label “contraction.” See the following footnote for more information and justification.^[2]

As a final and more minor note, I refer to the morpheme *‑hk, traditionally called the “prohibitive” following Bloomfield, as a “potential” (cf. Goddard 1979b:102). This, or something like “admonitive,” appears to be a more accurate description of its most neutral function in Proto-Algonquian—essentially, meaning “lest X should happen, maybe X (often undesirable) will happen,” and only functioning as a true prohibitive when combined with some sort of negative or prohibitive particle. Since “potential” is a vaguer label than “admonitive,” I have given it preference.

Sources, Transcription, and Abbreviations

See the following footnote for information on sources. The key works covering PA morphophonology, and providing a great many of the examples used here, are Bloomfield (1946), Pentland (1979, 1999), and Goddard (2001, 2007; also 1979b, 2006), but I’ve drawn on a number of others as well. The footnote also covers constructed examples and how these are indicated.^[3]

My transcription of Proto-Algonquian essentially follows Ives Goddard’s post-1994 system, with traditional *ç, *x, and *we (in surface forms) replaced with *r, *s, and *o. I depart in still writing word-initial *e- as such, since I don’t want to fully commit to the interpretation that it was already *i-. (See Rule 23 for discussion.) Proto-Eastern Algonquian is transcribed following the system laid out here and here.

Because of the historical focus of this post, my transcription of Algonquian languages is much closer to a standardized system for equivalent phonemes across all the languages than I normally employ. They are here written with the Algonquianist version of Americanist notation: letters have their approximate IPA values except for ⟨č, š, y⟩ = /tʃ, ʃ, j/ (⟨c⟩ variably marks /tʃ/ or /ts/ = [ts ~ tʃ]), vowel or consonant length is marked with a following raised dot ⟨·⟩, and an apostrophe ⟨’⟩ is sometimes used instead of ⟨ʔ⟩ for /ʔ/. Marking of pitch, stress, voiceless vowels, etc. follows the traditional notations for individual languages, except for Arapaho, and Maliseet-Passamaquoddy pitch distinctions are generally unmarked. I also use an ogonek to mark nasalization. Capital ⟨E⟩, ⟨N⟩, and ⟨S⟩ (and lowercase ⟨æ⟩) represent morphophonemes in the daughters, while capital ⟨L⟩ in PA represents a segment that was either *r or *θ but either we don’t have sufficient testimony to tell which, or it varied between the two. Departures from this system are given in the following footnote.^[4]

Citations from languages other than English are italicized, except that for most languages which became extinct or dormant prior to being extensively recorded by modern linguists, phonemicizations are enclosed in ⫽double slashes⫽. (/single slashes/ are reserved for actual phonemic forms transcribed in the IPA.) Phonemicizations of these or other languages which are especially tentative or contentious are preceded by a dagger, †. Direct citations of a written record are enclosed in ⟨angle brackets⟩.

A couple of final phonological points to bear in mind with the examples. Massachusett and other New England languages, Mahican, Cheyenne, and Arapaho have all undergone vowel shifts that can partially obscure the fact that they regularly reflect this or that Proto-Algonquian formation. Some of these deviations will be noted again as they come up, but the key ones are: Massachusett/New England/Mahican ⫽a·⫽ and ⫽ʌ̨·⫽ correspond to PA *e· and *a·; Arapaho reflects *o(·) as i(i); and Cheyenne has e, a, o from *i/*o, *e, and *a, and underlying high tone |é| (etc.) from long vowels. Unless noted otherwise, an example from a daughter language regularly descends from the PA or pseudo-PA form being illustrated (in the portions of the forms relevant to the discussion).

Several commonly recurring but longer language names will be abbreviated. These are: MalPass for Maliseet-Passamaquoddy, Mass for Massachusett, and NipAlg for Nipissing Algonquin. Unless otherwise specified, “Unami” refers to Southern Unami.

A Note on Different Reconstruction Models

Inevitably, not all Algonquianists have agreed on all the details of the phonological processes in the protolanguage. The presentation in this post represents my own personal judgment calls and beliefs; they usually correspond to Ives Goddard’s conclusions, where he has given them, though this is not universally true. Some of the more significant points of disagreement among Algonquianists will be noted or discussed at appropriate points when space allows, but plenty of others, especially minor ones, will have to be passed over.

I only just obtained a copy of David Pentland’s enormous Proto-Algonquian etymological dictionary (Pentland 2023) a few days ago, too late to fully integrate information and viewpoints from it into this post. But in a few places I have unsystematically noted some particularly significant insight, or more often a divergence in views on the protolanguage between Pentland and most others, usually including myself.

Two important general differences exist between the views of Proto-Algonquian held by Pentland and most Algonquianists, with cascading effects on a number of specific rules, so I will address them now. First, Pentland reconstructed pre-PA as containing a contrastive set of glottalized consonants which developed a number of morphophonemic differences from the nonglottalized set prior to merging with it. Morphophonemic processes caused by one set but not the other included triggering of umlaut (see Rules 13a, 13b, and 16), inserting epenthetic *‑o- between *‑ekw‑w- “INV”+“FMV” and *‑wa·w “2pl/3pl” (cf. Rule 9c), being lost morpheme-initially in the formation of deverbals (see Coda ex. 4), and so on.

Regardless of the historical validity of Pentland’s explanation for such differences, the situation by the PA period was more complex, and the processes more morphologized/lexicalized, than his presentation often will allow. Pentland’s treatment of processes such as umlaut thus end up being too simplistic and failing to account for complications (and, in the example of umlaut, of other more serious issues, as discussed in footnote 24 in the appropriate section below).

One of the most important specific properties Pentland ascribes to the glottalized/nonglottalized distinction is that “[f]inal syllables beginning with a nonglottalized sonorant were lost . . . except in words of four syllables or less” (Pentland 1999:254-255).^[5] This actually contrasts dramatically with Goddard’s (2007:215-218) rule, adopted here (Rule 28), in which the only loss of final segments is the dropping of any consonants in absolute word-final position. Pentland’s formulation requires significantly more restructuring both in PA and in the daughters, and leads to different hypotheses regarding the relationships among various morphemes (for example, the allomorphs of the N-formative *‑n(ay) or *‑n(e·)—for which see Rule 29—and the collective nominalizer *‑inay and its offshoots—for which see the section on *ayi contraction). Again, I follow Goddard’s position here. For more detailed argumentation, readers may consult the original sources; Pentland and Goddard’s papers, at least, are publicly available (see links in the list of sources at the end of the post).

Second, Pentland’s (esp. 1999) reconstruction of the PA verbal system differs in many ways from Goddard’s (esp. 2007), which is, again, the model I follow for the most part. Some of the differences are minor, but others are quite significant, and often intersect with reconstructions involving the “glottalized”/“nonglottalized” distinction and word-final loss of segments. Proulx’s (esp. 1990) view of the Proto-Algonquian verb was much more radically different than anyone else’s, but also much less defensible, and it and its unique implications are rarely dealt with here.

List of Rules

Table of Contents: Detailed Rules List

Rule 1a — deletion of prefix vowel Rule 1b — epenthetic *t Rule 2 — frozen *t → *s (“Palatalization I”) Rule 3 — *C-t → *C-k Rule 4a — *ChC → *CC Rule 4b — but *θhs → *hs Rule 4c — but *(m)whk remains Rule 5a — *(C1)STOP(G)C2 → *ʔC2 or *sC2 (“Meeussen’s Law”) Rule 5b — but occasionally *tC → *hC Rule 6 — “connective” (epenthetic) *o Rule 7 — *wk → *kw Rule 8a — pluralizers before mode suffix Rule 8b — *‑a·n (and *‑an?) before mode suffix Rule 8c — *‑t before *‑(e)sahan Rule 9a — “connective” (epenthetic) *e and *i Rule 9b — Rule 9a fails to apply Rule 9c — but does apply between *CC and *‑wa·w, or *‑n(ay) and *‑wa·w Rule 9d — when *w‑w inserts *e Rule 9e — epenthesis before *‑pan Rule 10 — *‑a· → *‑Ø before conjunct central suffix Rule 11 — epenthetic *y Rule 12 — second of two vowels lost Rule 13a — umlaut of *a· to *e· Rule 13b — umlaut of *‑ya· to *‑yi· Rule 14 — *Giw → *Gew Rule 15 — *CyaC → *Cye·C Rule 16 — umlaut of *e· to *a or *a· Rule 17 — *aww → *a·w Rule 18 — second of two semivowels lost Rule 19 — *CwaC → *Co·C coalescence Rule 20a — *ye → *i coalescence Rule 20b — *we → *o coalescence Rule 21 — contraction [see also below] Rule 22 — *w lost before *o(·) Rule 23 — initial-syllable *i/*e neutralization Rule 24 — final *k(w)‑i → *k(w)‑e Rule 25 — “mutation” (palatalization) of *t and *θ (“Palatalization II”) Rule 26 — nasal POA assimilation Rule 27a — allomorphy of TI(1) theme sign Rule 27b — allomorphy of TI(2) theme sign Rule 28 — final consonant loss Rule 29 — allomorphy of N-formative Coda Coda ex. 1 — alternate analysis of Rule 8a Coda ex. 2 — complication to Rules 8a/8b, with possible *kw replacing *n Coda ex. 3 — *‑a· + *‑hk = *‑iye·k Coda ex. 4 — deriving deverbals

(1a) The *e of the person prefixes (*ne- “1,” *ke- “2,” *we- “3,” *me- “UNSPEC”) is lost before a vowel-initial dependent noun stem. Rule 1a must precede Rules 1b, 12, 20b, and 22.
e → Ø /#{n,k,w,m}__+V (morphologically conditioned)

• *ne- “1” + *‑i·yaw- “body” → *ni·yawi “my body; myself” (e.g., Miami-Illinois niiyawi)
• *ke- “2” + *‑atay- “belly, stomach” → *katayi “your (sg.) belly, stomach” (e.g., Plains Cree katay, Munsee kătay)

(1b) Before all other vowel-initial stems (= verbs and non-dependent nouns), an epenthetic *‑t- is inserted after the person prefixes. Rule 1b must follow Rule 1a and must precede Rules 2, 12, and 20b.
Ø → t /#{ne,ke,we}+__V (morphologically conditioned)

• *ne- “1” + *api- “sit (AI)” → *netapi “I’m sitting” (e.g., Ojibwe nindab)
• *ke- “2” + *ehkw- “louse” + *‑em “POSS” → *ketehkoma “your (sg.) louse” (e.g., Plains Cree kitihkom)

(2) In PA there was a decent amount of residue of a no-longer-productive palatalization process, termed “Palatalization I” by Pentland (1979:390-392) (as opposed to “Palatalization II” = Rule 25), by which (by the actual PA period) *t became *s before certain specific morphemes beginning in *e, *a·, and rarely *i(·). Rule 2 must follow Rule 1b and must precede Rule 25.
t → s (rare and sporadic, lexically / morphologically conditioned)

• *pye·t- “coming; hither” + *‑a·p “see, vision” [a triggering morpheme] + *‑am “(TA Final)” + *‑a· “3OBJ” + *‑ẅ “3” + *‑a “PROXsg” → *pye·sa·pame·wa “s/he (PROX) sees (s.o. (OBV)) approaching” (e.g., Meskwaki pye·sa·pame·wa)
  • Cf. *pye·t- + *‑a·p “dawn” [etymologically identical with *‑a·p “see,” but a different morpheme by the time of PA] + *‑an “(II Final)” + *‑ẅ + *‑i “INANsg” → *pye·ta·panwi “dawn approaches” (e.g., Meskwaki pye·ta·panwi)
• *went- “RR:SOURCE/REASON” + *‑ehk “by foot/body (TI Final)” [a triggering morpheme] + *‑am “INAN.OBJ(CL1)” → *onsehkamwa “s/he comes to (sth.) from {somewhere}, for {such reason}” (e.g., Menominee ohsε·hkam)
• Cf. *went- + *‑en “by hand (TI Final)” + *‑am + *‑ẅ + *‑a → *ontenamwa “s/he gets, takes (sth.) from {somewhere}” (e.g., Menominee ohtε·nam)
• *ne- “1” + *mi·t- “defecate” + *‑i· “(AI Final)” [triggering in this one case] → *nemi·si· “I defecate” (e.g., Ojibwe nimiizii)
• Cf. *ne- + *mi·t- + ‑t “(TI Final)” + *‑am “INAN.OBJ(CL1)” + *‑n(ay) “FMV” + *‑i “INANsg” = **nemi·tta·ni → **nemi·tita·ni [epenthetic *‑i- by Rule 9a] → *nemi·čita·ni [*ti → *či by Rule 25] “I defecate on it” (e.g., Ojibwe nimiijidaan)
• Cf. also *ne- + *mi·t- + *‑enkwa·m “sleeping (AI Final)” → *nemi·tenkwa·me “I defecate in my sleep, shit my bed” (e.g., Ojibwe nimiidingwaam)
• It’s probable that in PA, Palatalization I had been lexicalized to enough of an extent where other morphemes could intervene between the trigger and the *t affected, since this is the case in some daughters, e.g. (Goddard 2001:202): *na·kat‑aw- “follow” + *‑a·p “see, vision” [a triggering morpheme, see first example above] + *‑am “(TA Final)” → **na·katawa·pame·wa → *na·kasawa·pame·wa “s/he (PROX) keeps h/ eye on, carefully observes (s.o. (OBV))” (Meskwaki na·kasawa·pame·wa)
• Cf., with assibilation evidently leveled out, e.g.: Plains Cree nākatawāpamēw, Ojibwe [o]naagadawaabam[aan]

(3) The animate third-person conjunct suffix *‑t becomes *‑k after a consonant. Rule 3 must precede Rules 7, 8c, 9a, 9b, 11, 25, 26, 27a, and 27b.
t → k /C+__ (morphologically conditioned)

• *takwihθin- “arrive (AI)” + *‑t “3AN.CONJ” + *‑e· “SBJV” → *takwihšinke· “if s/he arrives” (e.g., Ojibwe dagoshing)
• Cf. *pi·ntwike·- “enter (AI)” + *‑t + *‑e· → *pi·ntwike·te· “if s/he enters” (e.g., Ojibwe biindigeed)
• *pya·- “come (AI)” + *‑t + *‑e· → *pya·te· “if s/he comes” (e.g., Munsee páate) vs. *pya·- + *‑w “IRR” + *‑t + *‑e· → **pya·wke· → *pya·kwe· [by Rule 7] “if s/he doesn’t come” (e.g., Munsee páakwe, Meskwaki [me·hi‑]pya·kwe “before s/he came/comes”)
• *na·θ- “fetch (TA)” + *‑ekw “INV” + *‑t + *‑a “PROXsg.PART” + IC → **naya·θekwka → *naya·θekoka “s/he (PROX) who is fetched by h/ (OBV)” (e.g., Munsee ‡náalkwək^[S6]) (see Goddard 1979a:110, n. 15, 1979b:132, 2015b:375)
• But analogically reshaped in nearly all daughters to contain the more common *‑t allomorph (e.g., Ojibwe nayaanigod, Meskwaki ‡na·nekota, Menominee ‡naya·nekot, Arapaho ‡nonooθéit “s/he (OBV) goes and gets h/ (PROX)”) (cf. Goddard 2015b:375)

(4a) *h is lost between two consonants. Rule 4a must precede Rules 5 (at least as a historical sound change), 9a, 26, and 27a.
h → Ø /C__C

• *wa·pant- “look at (TI)” + *‑am INAN.OBJ(CL1) + *‑hk “POT” + *‑an “2sg.CONJ” → **wa·pantamØkani → *wa·pantankani [nasal assimilation by Rule 26 = 27a(iii)] “lest you (sg.) look at it” (e.g., Ojibwe [geego] waabandangeen “don’t (sg.) see it!”)
• *wa·pant- + *‑am + *‑hsi “DIM” + *‑w “IRR” + *‑ẅ “3” + *‑a “PROXsg” → **wa·pantamØsi·wa → *wa·pantansi·wa “s/he doesn’t look at (sth.) even a little bit” (e.g., Miami-Illinois waapantansiiwa “s/he doesn’t look at it,”^[S7] Ojibwe [gaawiin] owaabandanziin “s/he doesn’t see it”)
• *wera·ke·‑e·n- “bowl, dish” + *‑hs “DIM” + *‑i “INANsg” → *ora·ke·nØsi “small bowl, small dish” (e.g., Unami lɔ́·k·e·ns “dish”) (cf. Goddard 1974:326, n. 66, 1979a:92)
• For another probable example of Rule 4a, in the form of an actual sound change operating in pre-PA, see: *name·kw- “*fish” [→ “lake trout (Salvelinus namaycush)”] + *‑hs + *‑a “PROXsg” → **name·kwsa → *name·ʔsa “fish” (cf. Goddard 1979a:92, 2008b:7; mentioned in the discussion on “*xC” and *ʔC clusters in the Proto-Algonquian post)
• Similarly: *e·h- “bivalve mollusk” + *‑hs + *‑a = *e·hhsa → *e·hØsa “bivalve mollusk, clam, etc.; shell” (e.g., Ojibwe ees, Miami-Illinois eehsa, Menominee ε·hs[e·hsεh]; cf. Munsee éehəs ~ éehas and Unami é·həs, reflecting post-PA *e·hehsa, the original stem *e·h- plus later regular diminutive *‑ehs) (cf. Goddard 2008b:7)

(4b) There were at least two clear exceptions to Rule 4a of which I’m aware. The first was that *θhs resolved to *hs, not ^xθs, which was an illegal cluster. Perhaps the rule more broadly was that the *‑h- was deleted when an (intermediate) legal cluster would result, and otherwise the first consonant in a *ChC sequence was deleted, leaving a final legal *hC cluster? Rule 4b must precede Rule 9a.
θ → Ø /__hs   (/__hC ?)

• *wa·pam- “look at (TA)” + *‑eθ “2OBJ” + *‑hs “DIM” + *‑w “IRR” + *‑a·n “1sg.CONJ” → **wa·pameØhsowa·ni “when I don’t look at you (sg.) even a little bit” [opaque due to the seeming absence of the theme sign *‑eθ, and repaired by different daughters by re-inserting *‑eθ:]
  • → pseudo-PA *wa·pamehseθowa·ni (Ojibwe waabamisinowaan “if I don’t see you (sg.)”)
  • → pseudo-PA **wa·pamehsoweθa·ni → *wa·pamehso·θa·ni (Miami-Illinois waapamehsoolaani “I don’t look at you (sg.),”^[S8] Cheyenne [tsés-sáa‑]vóomáhetse |=vóom-aʹhéte| “when I didn’t see you (sg.)”)
• *ne- “1” + pre-PA *‑šiθ- “father-in-law” + *‑hs “DIM” + *‑aki “PROXpl” → *nešiØhsaki “my maternal cross-uncles” (e.g., Menominee nese·hsak) (Goddard 1979a:92)

(4c) The second exception to Rule 4a is that when the irrealis suffix *‑w was followed by the potential/admonitive suffix *‑hk, the cluster remained as such (including when an *m preceded). This *(m)whk cluster is resolved later via the insertion of epenthetic vowels by Rules 6 and 9a (and the *w is lost through the contraction of the resulting *VwV sequence, by Rule 21). Rule 4c must precede Rules 6, 7, 9a, and 21.

• *menah- “give to drink (TA)” + *‑i “1OBJ” + *‑w “IRR” + *‑hk “POT” + *‑e·kw “2pl.CONJ” → **menahiwhke·kwi → **menahiwehke·kwi → *menahi·hke·kwe “you (pl.) should not neglect to give me a drink” (e.g., Plains Cree minahīhkēk “(you pl.) give me a drink later!”)
• *weten- “take (TI)” + *‑am “INAN.OBJ(CL1)” + *‑w + *‑hk + *‑e·kw → **wetenamwhke·kwi → **wetenamowehke·kwi → *otenamo·hke·kwe “you (pl.) should not neglect to take it” (e.g., Plains Cree otinamōhkēk “(you pl.) take it later!”)

(5a) In a sound change of pre-PA date (“Meeussen’s Law” [see Meeussen 1959]), in sequences of a plosive + a true consonant, the plosive lenited, ultimately becoming *ʔ in most cases, but *s before *p and *k (see the discussions here and here); in the operation of the rule, any intervening semivowel and/or consonant immediately preceding the plosive were dropped. Meeussen’s Law is mostly irrelevant to synchronic PA phonology, with the productive resolution of illicit consonant clusters being the insertion of an epenthetic vowel (see Rules 6 and 9a), but there remained some archaic combinations where the effects of the change were still visible, and which may have been analyzed as such by speakers. Rule 5a must precede Rule 9a, and forms a component of Rule 9e(=9a); at least as a historical sound change it followed Rule 4a.
(C){p,t,k}(G) → s /__{p,k} (sporadic, lexically conditioned)
(C){p,t,k}(G) → ʔ /__C[≠G] (sporadic, lexically conditioned)

• Cf. the stems *neʔr- “kill (TA)” ← *nep- “die” + *‑r “(TA Final)” (i.e. “cause to die”) vs. plain *nep- “die (AI)” ← *nep- (e.g., Menominee nepwa·h “s/he dies,” nεʔnεw “s/he (PROX) kills h//them (OBV))
• Cf. the four basic derivationally related stems for “placing,” all derived from the Initial/root *ap- “located ({somewhere}), placed, in place; exist ({somewhere})”:
• AI *api- “sit ({somewhere}); be (in place {somewhere}); exist ({somewhere})” ← *ap- + *‑i “(AI Final)” (e.g., Menominee ape·w “s/he’s in place (etc.),” Munsee ăpə́w “s/he’s there”)
• II *aʔte·- “be (located; placed) {somewhere}, be in place; exist ({somewhere})” ← *ap- + *‑te· “(II Final)” (e.g., Menominee aʔtεw “it’s in place (etc.),” Munsee áhteew “it’s there”)
• TI *aʔtaw- “place, set sth.” ← *ap- + *‑t “(TI Final)” + *‑aw “INAN.OBJ(CL2)” (e.g., Menominee aʔtaw “s/he has it ({somewhere}), places it; she (hen) lays it (an egg),” Munsee áhtoow “s/he puts (sth.) down”)
• TA *aʔr- “place s.o.” ← *ap- + *‑r “(TA Final)” (e.g., Menominee aʔnεw “s/he (PROX) has, places h/ (OBV) ({somewhere}), has h/ (OBV) at h/ (PROX) disposal,” Munsee áhleew “s/he (PROX) puts (s.o. (OBV)) down”)
• *‑t “3AN.CONJ” + *‑pan “EMPH.PAST” → verb ending *‑span “3AN.CONJ.EMPH.PAST”: e.g., *pemwehθe·- “walk along (AI)” + *‑t + *‑i “INDIC” = **pemwehθe·ti → *pemohθe·či “when s/he walks along” vs. *pemwehθe·- + *‑t + *‑pan + *‑i → *pemohθe·spani “when s/he had walked along” (e.g., Moose Cree pimohtēt vs. pimohtēspan “s/he would walk along”)
• Cf. *‑pan with other person suffixes, e.g., *‑a·nk “1pl.EXCL.CONJ” + *‑pan → verb ending *‑a·nkepan (with epenthetic *‑e-) “1pl.EXCL.EMPH.PAST”: e.g., *pemwehθe·- + *‑a·nk + *‑i → *pemohθe·ya·nki “when we (excl.) walk along” vs. *pemwehθe·- + *‑a·nk + *‑pan + *‑i → *pemohθe·ya·nkepani “when we (excl.) had walked along” (e.g., Moose Cree pimohtēyāhk vs. pimohtēyāhkipan “we (excl.) would walk along”) (For a unified account of epenthesis patterns involving *‑pan, see Rule 9e)

(5b) While Meeussen’s Law normally gave *sp, *sk as the outcomes of earlier / underlying *t‑p, *t‑k, in a few cases it resulted in *hp, *hk instead (Goddard 2015b:406, n. 91).
t → h /__{p,k} (rare, sporadic, lexically conditioned)

• *wi·t- “with another” + *‑pe· “sleep” + *‑m “(TA Final)” → stem *wi·hpe·m- “sleep with (TA)” (e.g., Plains Cree wīhpēm-, Arapaho in nenííčoobéθen “I’m sleeping with you (sg.)” [with level high pitch on íí indicating pre-Arapaho stem *ni·čo·m-, with plain *č ← PA *(h)p but not *sp], Cheyenne |véám‑| [with Ø ← PA *(h)p but not *sp])
• Cf. *wet- “pull, draw” + *‑pwa· “by mouth (AI Final)” → stem *ospwa·- “smoke a pipe (AI)” (with expected *sp cluster) (e.g., Plains Cree in derivative ospwākan “pipe”; Arapaho hîičoo- [← pre-Arapaho *iʔčoo-], Cheyenne he’pó-)
• The daughters offer conflicting evidence over the outcome when an II stem ending in *t was followed by the suffix *‑k “INAN.CONJ”: thus, Unami points to *hk, while Gros Ventre points to *sk. Many others are ambiguous. Perhaps there was variation in PA. Thus:
• *tahka·skwat- “be shade, shady (II)” + *‑k “INAN.CONJ” + *‑i “RRC.PART” + *e·ntaθi= “where it is” → ?*e·ntaši=tahka·skwahki “where it’s shady, in the shade” (Unami enta-thá·kɔ |=tahāhkwa|, with loss of pre-Unami final *|‑h| ← PA *hk [vs. Unami |hk| from PA *sk, which would have been retained]) (Goddard 1979b:129)
• *tawat- “be a gap (II)” + *‑k + *‑i “INANsg.PART” + IC → ?*te·waski “that which is a gap, hole” (Gros Ventre tɔ́ɔ́nʔɔ “it is or has a hole,” with /ʔ/ ← PA *sk)

(6) When directly following a consonant, several suffixes beginning with a labial element insert a preceding epenthetic *‑o- (typically referred to as “connective *o”): (i) the irrealis *‑w; (ii) the 2pl imperative *‑kwe; and (iii) the plural augment *‑wa· when it occurs before the third-person suffix *‑t, viz. in a participle or the iterative.

In derivation, epenthetic *‑o- is added before (iv) the nominalizer *‑ẅen and (v) a suffix *‑ẅ which creates Initials from verb stems. (*‑ẅen is etymologically this *‑ẅ + the nominalizer *‑en.) The *‑o- before derivational *‑ẅ and/or *‑ẅen was lengthened to *‑o·- in a number of daughter languages. And it’s plausible that (vi) the relational *‑w also used *‑o-, as in Meskwaki and Ojibwe, but this is difficult to say with confidence; in many varieties of Cree it takes no epenthetic vowel, and Cheyenne shows a third pattern. Morphemes which fail to insert an epenthetic *‑o(·)- include: the formatives *‑w “W.FMV” and *‑ẅ “3” (but see below); the suffix *‑wa·w “2pl/3pl”; and the derivational suffixes *‑w and *‑ẅ. Finally, note that (vii) contrary to its normal behavior, *‑ẅ “3” does insert *‑o- when it is both preceded and followed by a consonant, viz. when it precedes a consonant-initial mode suffix (*‑pan “EMPH.PAST” and *‑(e)sahan “EMPH.PRES”; on this exception, see also Rule 9e).

A number of these morphemes, both those which added epenthetic *‑o- and those which failed to, are etymologically connected. Thus, the plural augment *‑wa· (6(iii)) is cognate with the 2pl/3pl suffix *‑wa·w, while the non-epenthesizing derivational suffixes *‑w and *‑ẅ are etymologically the same as the two formatives. The non-epenthesizing derivational *‑ẅ and formative *‑ẅ “3” are also likely both ultimately the same as the epenthesizing derivational *‑ẅ (and thus the initial portion of the nominalizer *‑ẅen) (6(iv), 6(v)).

In the operation of Rule 6, word-initial *e- (in practice, only the verb stem *e- “say {so} (AI)”) was treated as if it were *ey- (Goddard 2000:126, n. 56; cf. 2003:98, n. 47), so *e- plus a relevant *w became *ey‑o‑w…; several examples will be found below. Rule 6 must follow Rule 4c, must precede Rules 7, 17, 18, 19, 20b, 21, and 22, and forms a component of Rule 9e(=9a).
Ø → o /C__+{w,ẅ,kw} (morphologically conditioned)

• *ne- “1” + *wem- “come from {somewhere} (AI)” + *‑w “IRR” + *‑eʔm “FMV” = **newemweʔm → **newemoweʔm → *no·mowe “I don’t come from {somewhere}” (e.g., Unami nú·mu·wi [Goddard 1979b:168])
• *tahkwen- “hold, sieze (TI)” + *‑am “INAN.OBJ(CL1)” + *‑kwe “2pl.IMPER” → **tahkwenamkwe → *tahkonamoko “(you pl.) hold, sieze it/them (INAN)!” (e.g., Moose Cree tahkonamok)
• *wi·nke·rent- “like, find pleasing, good (TI)” + *‑am + *‑wa· “AUGMT:pl” + *‑t “3AN.CONJ” + *‑iri “ITER” + IC → *wa·nke·rentamwa·tiri → *wa·nke·rentamowa·čiri “whenever they like it/them (INAN)” (e.g., Arapaho nííʔeenêetowunóóθi [u from *i ← *o by vowel harmony])
• *nep- “die (AI) + *‑ẅen “NMLZ” → **nepweni → *nepoweni “death” [→ post-PA *nepo·weni] (e.g., Meskwaki nepo·weni, Ojibwe nibowin [likely with secondarily analogically short o])
• *nep- + *‑ẅ “AI⇒Initial” + *‑e·rem “think, feel (TA Final)” → **nepwe·rem- → stem *nepowe·rem- “think s.o. to be dead (TA)” [→ post-PA *nepo·we·rem-] (e.g., Meskwaki nepo·we·nem-)
• Another example with the same stem: pseudo-PA *nep- + *‑ẅ + *‑a·‑ka‑n “NMLZ” → **nepwa·kani → *nepowa·kani “death” [→ PEA *nəpəwākan] (e.g., Mass ⟨nupꝏonk⟩ ⫽nəpəwʌ̨·k⫽, Penobscot nə̀pəwαkan)
• Cf. also: *e- [→ *ey-] “say {so} (AI)” + *‑ẅen or *‑ẅ‑a·kan → **eyweni / **eywa·kani → *eyoweni / *eyowa·kani “thing said, what is said” (e.g., Meskwaki iyoweni “word, thing said”; Mass ⟨yeuwonk⟩ ⫽yəwʌ̨·k⫽ “word”)
• *mešot- “shoot and hit (TI)” + *‑am “INAN.OBJ(CL1)” + *‑w “RELN” + *‑a· [→ *Ø by Rule 10] “3OBJ” + *‑ak “1sg»3AN.CONJ” → **mešotamwaki → ?*mešotamowaki “that I hit it with my shot in relation to h//them (e.g., I hit h/ item)” (Ojibwe mizhodamowag, Meskwaki ‡mešotamowaki^[S9])
  • But cf. Western Swampy Cree [kā-kī‑]wāpahtamwak “when/then I saw it in relation to h//them” (Cenerini 2014:79), with wāpaht‑am- “see (TI)”
• Formative *‑ẅ normally does not insert *‑o-: *nep- “die (AI)” + *‑ẅ “3” + *‑a “PROXsg” → *nepwa “s/he dies” (e.g., Meskwaki nepwa, Mass ⟨nup⟩ ⫽nəp⫽ |nəpẅ| [Exod. 11:5])
• But does so when between two consonants: *nep- + *‑ẅ + *‑pan “EMPH.PAST” → **nepowpan → *nepo·pa [see Rule 9e] “s/he had died” (e.g., Meskwaki ‡nepo·pani “s/he is surely dead, is dead indeed,” Mass ⟨nupoop⟩ ⫽nəpu·p⫽ “s/he died”)^[10]
• Cf. also: *e- [→ *ey‑] “say {so} (AI)” + *‑ẅ + *‑sahan “EMPH.PRES” → **eywsahan → **eyowsahan → ?*eyowesaha [or ?*eyo·saha] “s/he has said {so}” (e.g., Menominee eyo·sah “so s/he says {so}”) [uncertain example]
• Cf. as well the different treatment before vowel-initial mode suffix: *pankihθen- “fall (II)” + *‑ẅ + *‑etoke·h “DUB” = **pankihθenwetoke·h → *pankihθenotoke· [*‑o- ← *‑we- by Rule 20b] “it must have fallen” (e.g., Ojibwe bangisinodog; cf. variant bangisinowidog with analogically extended *o-insertion, as if from ^xpankihsenowetoke·)

(7) The sequence *wk, when following a vowel, metathesizes to *kw. Rule 7 must follow Rules 3, 4c, and 6, must precede Rules 9a/9b, 11, 19, 20b, and 22, and forms a component of Rule 27b.
wk → kw /V__

• *ki·šiht- “finish (TI)” + *‑aw “INAN.OBJ(CL2)” + *‑t “3AN.CONJ” + *‑e· “SBJV” → **ki·šihtawke· [by Rule 3] → *ki·šihta·kwe· “if s/he finishes it” (e.g., Munsee kiishihtáakwe) [see also Rule 27b(iv)]
• *aθet- “rot (II)” + *‑w “IRR” + *‑k “INAN.CONJ” + *‑e· “SBJV” → **aθetowke· → *aθetokwe· “if it doesn’t rot” (e.g., Munsee alətóokwe)
• *neʔr- “kill (TA)” + *‑i “1OBJ” + *‑w “IRR” *‑t “3AN.CONJ” + *‑e·n “INTERR” + *‑a “PROXsg.PART” + IC → **ne·ʔriwke·na → *ne·ʔrikwe·na “whoever may kill me” (e.g., Ojibwe neesigween, Meskwaki ‡ne·sikwe·na)

(8a) Several morphemes lose their final consonant before a mode suffix. First, the independent order pluralizers *‑ena·n “1pl.EXCL” and *‑wa·w “2pl/3pl” lose their final *n or *w, respectively. (For a few complications to this statement, see the Coda, exs. 1 and 2, below.) Rule 8a must precede Rules 9a, 12, and 26, and forms a component of Rule 9e(=9a).
ena·n → ena· /__+{pan,(e)sahan,etoke·h,…} (morphologically conditioned)
wa·w → wa· /__+{pan,(e)sahan,etoke·h,…} (morphologically conditioned)

• *ne- “1” + *ne·w- “see (TA)” + *‑a· “3OBJ” + *‑ʔm “FMV” + *‑ena·n “1pl.EXCL” + *‑pan “EMPH.PAST” → **nene·wa·ʔmena·npan → *nene·wa·ʔmena·pa “we (excl.) had seen (s.o.)” (e.g., Unami nne·ɔ́həna·p |nənēwāhmənāp|)
• *ne- + *kweʔθ- “fear (TA)” + *‑a· + *‑w “FMV” + *‑ena·n + *‑etoke·h “DUB” → **nekweʔθa·wena·netoke·h → **nekweʔθa·wena·Øetoke·h → ?*nekoʔθa·wena·toke· [with *‑a·‑e- → *‑a·-] “it seems we’re (excl.) afraid of h/” (e.g., Ojibwe ningosaanaadog, Meskwaki ‡nekosa·[pe]na·toke)
• *ke- “2” + *pya·- “come (AI)” + *‑ʔm + *‑wa·w “2pl/3pl” + *‑(e)sahan “EMPH.PRES” = **kepya·ʔmwa·w(e)sahan → *kepya·ʔmwa·saha “you (pl.) have come” (e.g., Menominee kepya·mwa·sah “so you (pl.) have come”)

(8b) Second, as attested by Arapahoan, Cree-Innu (CINA), and Unami, the conjunct 1sg suffix *a·n loses its final *n before the same mode suffixes. Although the same process may also have occurred with the 2sg.CONJ suffix *‑an, it would only unambiguously be reflected in some varieties of Cree (I think?); these probably instead have reshaped the 2sg endings by analogy with the 1sg ones, but there’s room for doubt (cf. Proulx 1990:107). Rule 8b must precede Rules 9a, 26, and possibly 12, and forms a component of Rule 9e(=9a).
a·n → a· /__+{pan,(e)sahan,…} (morphologically conditioned)
an → a /__+{pan,(e)sahan,…} (morphologically conditioned) (possibly [??])

• *pya·- “come (AI)” + *‑a·n “1sg.CONJ” + *‑pan “EMPH.PAST” → **pya·ya·npane· → *pya·ya·pane· “if I had come” (e.g., Unami pa·á·p·ane |pā-yā-pan-ē| ← |pā-ān-pan-ē|)
• *api- “sit ({somewhere}); be placed, located {somewhere} (AI)” + *‑an “2sg.CONJ” + *‑pan → ?*apiyampani “you (sg.) had sat, been there” (e.g., Ojibwe abiyamban)
• But cf. Moose Cree apiyapan, as if from ?*apiyapani [not ^xapiyahpan with ‑hp- ← *‑mp‑]; probably by analogy with 1sg apiyāpān ← *apiya·pani
• *tankam-etwi- “hit e.o. (AI)” + *‑a·n + *‑(e)sahan “EMPH.PRES” → pseudo-PA *tankametwiya·sahane· [as if meaning “if I have, do hit myself”] (e.g., Arapaho tóʔobétinóóhk “if I strike myself”)
  • Note that the ending ‑nóóhk is regular from post-PA *‑y‑a·sapan, with *‑(e)sahan reshaped as *‑(e)sapan; ^x‑y‑a·nsapan would have given Arapaho ^x‑noohʔók (cf. Goddard 2015b:407, n. 102), and ^x‑y‑a·nesapan, with epenthetic or segmental *e, would have given something like ^x‑noonóhk

(8c) Finally, it seems that the 3AN.CONJ suffix *‑t was dropped before *‑(e)sahan “EMPH.PRES.”^[11] This appears to have only applied to the postvocalic *‑t allomorph of the 3AN.CONJ suffix, not the postconsonantal *‑k one. Rule 8c must follow Rule 3, and possibly must precede Rules 9a and 12.
t → Ø /__+(e)sahan (morphologically conditioned)

• *tankam- “hit (TA)” + *‑a· “3OBJ” + *‑t “3AN.CONJ” + *‑(e)sahan → **tankama·t(e)sahane· → *tankama·sahane· “if s/he (PROX) has hit h/ (OBV)” (e.g., Arapaho tóʔowoohók “if s/he (PROX) strikes h//them (OBV)” [‑oohók ← post-PA *‑a·sapan, or theoretically *‑a·hsapan])
• Cf. outcome with other morphemes ending in *‑t: *tankam- + *a· [→ *Ø by Rule 10] + *‑at “2sg»3AN.CONJ” + *‑(e)sahan → *tankamatesahane· “if you (sg.) have hit h/” (e.g., Arapaho tóʔowotéhk “if you (sg.) strike h/” [‑otéhk ← post-PA *‑atesapan])
• *tepe·rem- “judge; manage, control, rule (TA)” + *‑a· + *‑t + *‑(e)sahan → *tepe·rema·sahane· “if s/he (PROX) has ruled h/ (OBV)” (e.g., Old Algonquin ⟨[tiberima]zen⟩ †tipērimāsan [Nicolas 1674:25, “s’il l[’eut ou s’il l’auoit gouuerné]”; Daviault 1994:77; cf. Pentland 1984; equivalent to a modern *dibeenimaazan])
• Cf. Western Abenaki ⟨wajônôza⟩ [wskəbi] wajʌ̨nʌ̨za “perhaps s/he (PROX) had h//them (OBV)” (Laurent 1884:156) with verb wajʌ̨n- “have, possess (TA)” and ‑ʌ̨za /‑ʌ̃sa/ (← *‑a·sa…, not [in this paradigm] /ss/ ← *‑hs- or *‑ʔs-)
• Cf. also Mass ⟨wadchanós⟩ †⫽wačʌ̨·nʌ̨·s⫽ “if he s/he (PROX) did keep, have h//them (OBV)” (Eliot 1666:40; Goddard and Bragdon 1988:566) with cognate verb ⫽wačʌ̨·n-⫽ “keep, have, preserve (TA)” and †⫽-ʌ̨·s⫽ (consistently spelled with single ⟨s⟩ = †⫽s⫽ ← *‑s-, never with an indication of ⫽hs⫽ ← *‑hs- or *‑ʔs-)

(9a) An epenthetic vowel, either *‑e- or *‑i- (normally referred to as “connective *e” and “connective *i”), is inserted to break up most consonant clusters across a morpheme boundary.^[12] It’s often difficult to determine when Rule 9a actually operated, because in many cases it’s hard or impossible to choose whether to segment a given morpheme as, say, *e-initial and losing the *e when following another vowel (by Rule 12 below), or as *C-initial and inserting an epenthetic *‑e- by Rule 9a when it follows a consonant. There are sometimes clues which help solve the question, though. Rule 9a must follow Rules 3, 4a, 4b, 4c, 5a/5b, 8a, and possibly 8b and 8c, must precede Rules 14, 20a, 20b, 21, and 25, and forms a component of Rules 9e(=9a), 27b, and 29.
Ø → e /C+__C (morphologically / lexically conditioned)
Ø → i /C+__C (morphologically / lexically conditioned)

• *šekwah- “crush by tool (TI)” + *‑ike· “ANTIP (AI Final)” → *šekwahike·- “crush (things) by tool (AI)” (e.g., Ojibwe zhishigwa’igee- [reduplicated], Munsee shkwahiikee-)
• Cf., showing that the *‑i- of *‑i-ke· is (historically) epenthetic: *ahšam- “feed, give food to (TA)” + *‑ke· → *ahšanØke·- “feed (people), serve a meal (AI)” with no epenthetic vowel (e.g., Ojibwe ashangee-, Munsee xankee-)^[13]
• *nep- “die (AI)” + *‑t “3AN.CONJ” + *‑e· “SBJV” = **nepte· → **nepke· [*‑t → *‑k after a consonant, by Rule 3] → *nepeke· “if s/he dies” (e.g., Meskwaki nepeke, Miami-Illinois nipeke, Menominee nepε·k, Mass ⟨nuppuk⟩ ⫽nəpək⫽ [Matt. 22:24]^[14])
• Epenthetic status of *‑e- is confirmed by the postconsonantal allomorph of the following consonant (see Rule 3), and cf. as well, with no epenthetic vowel: *nep- + *‑ẅ “3” + *‑a “PROXsg” → *nepØwa “s/he dies” (e.g., Meskwaki nepwa, Miami-Illinois nipwa, Menominee nepwa·h, Mass ⟨nup⟩ ⫽nəp⫽ |nəp‑ẅ| [Exod. 11:5; vs. analogical ⟨nuppꝏ⟩ ⫽nəpəw⫽, with vowel added, in Gen. 48:7])

(9b) Rule 9a failed to apply with certain specific morphemes and morpheme combinations. As the rules are ordered here, the primary exceptions to the insertion of an epenthetic vowel were: (i) between any consonant and a following semivowel; (ii) between any nasal and any consonant (later turned into legal clusters by Rule 26); and (iii) between synchronically legal clusters. If Rule 9a were ordered earlier, as some people have placed it, it would be necessary to specify additional environments where no epenthetic vowel was inserted.^[15] The results of the combination of legal cluster elements were: (iv) *h + *C → *hC (weakly supported, but plausible as a general rule); (v) *š + *C → *šC; and (vi) *θ + *C → *θC (Pentland 1979:363-390; Goddard 1979a:107-108, n. 4, 1997:32). *ʔC and *sC clusters derive from Meeussen’s Law as illustrated earlier (Rule 5a); and nasal clusters, from any *N + any *C, resolve to legal *NC sequences by Rule 26 as just mentioned. The origin of the clusters *rk and *rp, as well as the cluster I write *ʔm—but which might have been *hm (Goddard’s preference) or *mm—is unknown. Putative *čC clusters appear to have had several sources and may not even have been real.

• For examples of *Cw / *Cy taking no epenthesis, see Rules 9c(=9a), 9d(=9a), 17, and 18 (and 6), which resolve any illegal clusters so created.
• *h + *C = *hC (cf. Pentland 1979:371):
• *re·h~ “REDUP [≈REPETITIVE] + *re·- “breathe (AI)” → *re·hre·wa “s/he breathes” (e.g. Shawnee leʔθe, Menominee nε·hnεw)
• Cf. unreduplicated *re·- in *re·wa “s/he breathes,” reflected in archaic Munsee *léew (not directly attested, but derived noun léewan “breath” attested, in the possessed forms ⟨n’lay whun⟩ nléewan “my breath” and ⟨w’laywun⟩ wə̆léewan “his breath”; cf. normal leexéewan “breath” ← léexeew “s/he breathes,” the reflex of the reduplicated form above [Goddard 2013:110])^[16]
• *š + *C = *šC:
• *ki·š- “finish, complete” + *‑pwi “eat, taste, by mouth (AI Final)” + *‑ẅ “3” → *ki·špowa “s/he eats h/ fill, is full, sated” (e.g., Moose Cree kīšpow, Munsee kíispəw)
• *θ + *C = *θC:
• *tankam- “hit (TA)” + *‑eθ “2OBJ” + *‑t [→ *‑k by Rule 3] “3AN.CONJ” → *takameθke· “if s/he hits you (sg.)” (e.g., Penobscot takáməske “if s/he/they hit(s) you (sg.)” [Quinn 2015:[30]])
• With the previous, cf.: *tankam- + *‑eθ + *‑w “IRR” + *‑t [→ *‑k] → **tankameθowke· → *tankameθokwe· “if s/he doesn’t hit you (sg.)” (e.g., Penobscot [ὰta] takáməlokke [ibid.])
• *keθ- “held firmly” + *‑piswi “be tied (AI Final)” + *‑ẅ → *keθpisowa “s/he is tied up” (e.g., Mass ⟨kishpissu⟩ ⫽kəspəsəw⫽ [Matt. 21:2], Munsee koxpíisəw)

(9c) (= 9a) Rule 9a did apply in certain ~exceptions to the exceptions~ captured by Rule 9b(i), however. Epenthetic *‑e- was inserted before *‑wa·w “2pl/3pl” in at least two cases: (i) following a consonant cluster (e.g., after the inverse theme sign *‑ekw + W-formative *‑w); and (ii) per the notation used here, following the N-formative *‑n(ay). In both cases, the resulting sequences of vowels and semivowels coalesced to *o or contracted to *e· by later rules (20b and 21; see also Rule 29(i)). As Goddard (2007:232; see also 2001:170) observes, the first case “can be explained by” the fact that *‑wa·w always takes the same preceding coalescence pattern, etc., as the first-person plural suffixes (inclusive *‑enaw, exclusive *‑ena·n) “in the same paradigm.” Thus, e.g., *‑ekw‑w‑wa·w‑a (PROXsg»2pl.INDEP) becomes **‑ekw‑w‑e‑wa·w‑a → *‑ekowa·wa, just as (PROXsg»1pl.INCL.INDEP) *‑ekw‑w‑naw‑a becomes *‑ekw‑w‑e‑naw‑a → *‑ekonawa. Rule 9c must precede Rules 18, 20b, and 21, and is a component of Rule 29.
Ø → e /CC__+wa·w
Ø → e /n(ay)__+wa·w (morphologically conditioned)

• *ke- “2” + *na·θ- “fetch (TA)” + *‑ekw “INV” + *‑w “FMV” + *‑wa·w “2pl/3pl” + *‑a “PROXsg” = **kena·θekwwwa·wa → **kena·θekwwewa·wa → **kena·θekwowa·wa → *kena·θekowa·wa “s/he fetches you (pl.)” (e.g., Ojibwe ginaanigowaa)
• *ke- + *‑i·θemw- “cross-sibling-in-law” + *‑wa·w → **ki·θemwwa·wa → **ki·θemwewa·wa → *ki·θemowa·wa “your (pl.) cross-sibling-in-law” (e.g., Ojibwe giinimowaa, Cheyenne étamēvo |Ø-étam-evó| [Goddard 2000:87])
• *ke- + *na·t- “fetch (TI)” + *‑n(ay) “N.FMV” + *‑wa·w + *‑i “INANsg” = **kena·tnaywa·wi → **kena·tenayewa·wi → *kena·tene·wa·wi “you (pl.) fetch it” (e.g., Ojibwe ginaadinaawaa [with *e· regularly “recontracted” to aa])

(9d) (= 9a) Another ~exception to an exception~ to Rule 9a was that epenthetic *‑e- was inserted between irrealis *‑w and a following *w (either *‑ẅ “3” or, in theory, the W-formative *‑w—though there are no examples of the latter that survived into the daughters). As with Rule 9c(ii), the resulting *V‑w‑e sequence is contracted to *V· by Rule 21. Rule 9d must precede Rules 18, 20b, and 21.
Ø → e /w+__{ẅ,w} (morphologically conditioned)

• *wi·ʔθenyi- “eat (AI)”^[17] + *‑hsi “DIM” + *‑w “IRR” + *‑ẅ “3” + *‑aki “PROXpl” → **wi·ʔθenyihsiwwaki → **wi·ʔθenyihsiwewaki → *wi·ʔθenyihsi·waki “they don’t eat even a little bit” (Ojibwe wiisinisiiwag “they don’t eat,” Miami-Illinois wiihsinihsiiwaki “id.”)

(9e) (= 9a) A final ~exception to an exception~ (of sorts) to Rule 9a are the cases in which epenthetic vowels were inserted before the mode suffix *‑pan “EMPH.PAST,” which as noted a few times, originally did not take epenthesis, though synchronically in PA it often did. In fact, I will use this opportunity to present here a unified account of the epenthesis processes as they relate to *‑pan, instead of leaving bits and pieces scattered among many rules and examples—though some of these processes are also mentioned elsewhere.^[18]

When the consonant preceding *‑pan was a nasal or the morpheme *‑t “3AN.CONJ,” no epenthetic vowel was inserted; this is likely also true of *‑ẅ “3” and the formative *‑w. The *t became *s by Meeussen’s Law (5a; example given there), while some nasals were deleted (Rules 8a and 8b; examples given there) and any that remained were kept as is, to eventually assimilate to *m (per Rules 9b(ii) and 26). The case of the formative *‑ʔm before *‑pan is complicated, and I don’t have space to discuss it in full here—hopefully in a future post.

Assuming there was no epenthetic vowel inserted after the formatives *‑ẅ and *‑w, these dropped and the preceding vowel was lengthened (or the whole *Vw sequence preceding *p coalesced); see below in the discussion of contraction. Regardless of whether it took a following epenthetic *‑e- or not, *‑ẅ, unlike its normal behavior (see Rule 6, esp. 6(vii)), inserted a preceding epenthetic *‑o- if it also followed a consonant. The final *w of *‑wa·w “2pl/3pl” may also have been lost before *‑pan, though this is not entirely certain (see discussion in Coda ex. 1).

In most other cases, a preceding epenthetic *‑e- was inserted, viz.: after (i) the formative *‑n(ay) (and possibly after the formatives *‑w and *‑ẅ “3”); (ii) the central suffix *‑enaw “1pl.INCL”; and (iii) the conjunct non-singular and portmanteau central suffixes. Note that the synchronic phonological environments for epenthesis would then be, essentially, after any consonant, including any *t that was not the 3AN.CONJ suffix (example below under rubric of 9e(iii)). The *‑Vwe- and *‑aye- sequences that arose from 9e(i) evidently always underwent contraction, and the combination of *V + *‑w/*‑ẅ also always resolved to a long vowel, whether this was by compensatory lengthening, coalescence, historical contraction, or at least synchronic contraction (cf. Pentland 1999:240, 250-251; Goddard 2000:126, n. 56). Rule 9e is composed of Rules 5a, 6, 8a, and 8b (and 9a), and must precede Rule 21.
Ø → e /C[≠t, ≠N]__+pan (morphologically conditioned)
(other elements to *‑pan’s interaction with preceding morphemes are expressed formally under other rules)

• *Ø- “UNSPEC” + *nakaθ- “abandon (TA)” + *‑a· “3OBJ” + *‑w “W.FMV” *‑pan “EMPH.PAST” → ?**nakaθa·wpan → *nakaθa·pa “s/he had been abandoned, left behind” (e.g., Ojibwe naganaaban, Unami nkála·p “s/he was left” [Goddard 2021:90, ex. 4.69m])
• Cf. with preceding: *nakaθ- + *‑a· + *‑ẅ “3” + *‑pan → ?**nakaθe·wpan [with vowel umlaut before *ẅ] → *nakaθe·pa “s/he (PROX) had abandoned (s.o. (OBV))” (e.g., Unami ‡nkále·p “s/he (PROX) abandoned (s.o. (OBV))”^[S19])
• Note also, showing *‑ẅ inserting preceding epenthetic *‑o- [cf. Rule 6(vii) above]: *saLakat- “be difficult (II)” + *‑ẅ + *‑pan = **saLakatẅpan → **saLakatowpan → *saLakato·pa “it had been difficult” (e.g., Ojibwe zanagadooban, Meskwaki sanakato·pani “it sure is hard”; also e.g. MalPass kələwətohpən [Sherwood 1983:232] “it was good,” with verb kələwət- “be good (II)” and /o/ from *o·)
• *we- “3” + *kya·t- “hide, conceal (TI)” + *‑aw “INAN.OBJ(CL2)” + *‑n(ay) “N.FMV” + *‑pan → **okya·tawenayepani → *okya·to·ne·pani “s/he had hidden, concealed it” (e.g., MalPass ‡hkatonehpən “s/he hid, concealed it,” Ojibwe ogaadoonaaban)
• *ke- “2” + *ne·m- “see (TI)” + *‑e· [← *‑am by Rule 27a(i)] “INAN.OBJ(CL1)” + *‑ʔm “FMV” + *‑enaw “1pl.INCL” + *‑pan → **kene·me·ʔmenawepan → *kene·me·ʔmeno·pa “we (incl.) had seen it” (e.g., Menominee kenε·memeno·pah “but we (incl.) saw it”)
• *tahkwen- “seize (TA)” + *‑at “2sg»3AN.CONJ” + *‑pan → *tahkonatepane· “if you (sg.) had seized h/” (e.g., Ojibwe dakonadiban; cf. outcome of *‑t “3AN.CONJ” + *‑pan as *‑span with no epenthesis)
• *po·si- “embark (AI)” + *‑a·nk “1pl.EXCL.CONJ” or *‑ankw “1pl.INCL.CONJ” + *‑pan → *po·siya·nkepane· “if we (excl.) had embarked” or **po·siyankwepane· → *po·siyankopane· “if we (incl.) had embarked” (e.g., Ojibwe booziyaangiban and booziyangoban ~ booziyangiban (excl. / incl.))

(10) The 3OBJ theme sign *‑a· is deleted before a vowel-initial agreement suffix in the conjunct or imperative orders. Rule 10 must precede Rules 11, 12, and 25.
a· → Ø /__+V (morphologically conditioned)

• *wa·pam- “look at (TA)” + *‑a· “3OBJ” + *‑ak “1sg»3AN.CONJ” → *wa·pamØake· “if I look at h/” (e.g., Ojibwe waabamag “if I see h/”)
• Cf., with intervening suffixes protecting the theme sign from deletion: *wa·pam- + *‑a· + *‑hsi “DIM” + *‑w “IRR” + *‑ak → *wa·pama·hsiwake· “if I don’t look at h/ even a little bit” (e.g., Ojibwe waabamaasiwag “if I don’t see h/”)
• *wa·pam- + *‑a· + *‑ehkwe “2pl»3AN.IMPER” → *wa·pamØehko “(you pl.) look at h//them!” (e.g., Meskwaki wa·pamehko, Plains Cree wāpamihk “(you pl.) see h//them!”)

(11) An epenthetic *‑y- is inserted between two vowels in several cases. These are: (i) between any two long vowels; (ii) between the vowel of the reduplicant and of the stem in reduplicated vowel-initial stems; and (iii) between a conjunct central agreement suffix beginning with a vowel other than *e, and any preceding element ending in a vowel (viz. a vowel-final AI stem or the 1OBJ theme sign *‑i). The restriction of the environment of 11(iii) to occurring with all conjunct central agreement suffixes except those beginning in *e is regrettably awkward, but is required in order to avoid a clash with Rule 21 (the contraction of *VGe sequences; see discussion there). In these cases, per Rule 12 below, the *e is subsequently lost. Rule 11 must follow Rules 3, 7, and 10, and must precede Rule 12.
Ø → y /V~__V (technically morphologically conditioned)
Ø → y /Vː__Vː
Ø → y /V+__V[≠e] (morphologically conditioned)

• *akwa·- “out of water” + *‑a·‑ʔθi “be blown by wind (AI Final)” → *akwa·ya·ʔšiwa “s/he is blown ashore” (e.g., Ojibwe agwaayaashi)
• *a·~ “REDUP [≈CONTINUATIVE]” + *a·timwi- “narrate, tell (AI)” → *a·ya·čimowa ≈“s/he narrates a long story, for a while” (e.g., Ojibwe aayaajimo, Menominee aya·cemow, Meskwaki a·ya·čimowa [Dahlstrom 1997:213]; and cf. Munsee aayə̆laachíiməw “s/he’s telling different stories; s/he’s telling stories slowly” [Goddard 2014:144] ← *a·yeθa·čimowa, reduplicated form of *eθa·čimowa “s/he narrates {thus}, tells {such} a story”)
• *api- “sit (AI)” + *‑e·kw “2pl.CONJ” + *‑i “NON3.PART” + IC → *e·piye·kwi “you (pl.) who are sitting” (e.g., MalPass epəyek^w “you (du.) who are sitting”)
• Cf., with no *‑y- between the stem and ending: the consonant stem *pemihθin- “be lying down (AI)” + *‑e·kw + *‑i + IC → *pe·mihšine·kwi “you (pl.) who are lying down” (e.g., MalPass peməssinek^w “you (du.) who are lying down”)
• *wa·pam- “look at (TA)” + *‑i “1OBJ” + *‑an “2sg.CONJ” + *‑e· “SBJV” → *wa·pamiyane· “if you (sg.) look at me” (e.g., Ojibwe waabamiyan “if you (sg.) see me,” Meskwaki ‡wa·pamiyane)
• Cf., with no *‑y- between the theme sign and following suffix, the consonant-initial ending *‑te· in: *wa·pam- + *‑i + *‑t “3AN.CONJ” + *‑e· → *wa·pamite· “if s/he looks at me” (e.g., Ojibwe waabamid “if s/he sees me,” Meskwaki ‡wa·pamite)
• Cf. also lack of application of Rule 11 when the central ending begins in *e: *wa·pam- + *‑i + *‑enk “UNSPEC.CONJ” + *‑e· → **wa·pamienke· → *wa·paminke· [by Rule 12] “if I’m looked at” (e.g., Meskwaki ‡wa·pamike)

(12) In a sequence of two consecutive vowels, the second is lost. Due to the operation of several previous rules, the second vowel is always short.^[20] Rule 12 must follow Rules 1a, 1b, 8a, 10, 11, and possibly 8b and 8c.
V → Ø /V__

• *a·p- “untie” + *‑i “mass, weight (Prefinal)” + *‑ah‑w “by tool (TA Final)” → stem *a·pihw- “untie (TA)” (e.g., Meskwaki a·pihw-)
• *ki·šk- “sever” + *‑kw‑e·- “neck” + *‑ah‑w → stem *ki·škikwe·hw- “chop s.o.’s head off, behead (with an axe, etc.) (TA)” (e.g., Ojibwe giishkigwee’w-, Unami ki·ski·k·wehw-, Miami-Illinois kiihkikweehw- [Old Illinois kiiskikweehw-]^[S21])

(13a) The formative *‑ẅ “3” and a homophonous derivational suffix (and its derivatives, for which see Rule 6(iv)/6(v) above) cause a preceding *a· (either the 3OBJ theme sign *‑a· or a stem-final *a·) to undergo umlaut to *e·. In the case of the formative *‑ẅ, another morpheme may intervene. Rule 13a must precede Rules 14, 17, 18, and 21.
a· → e· /__(+X)+ẅ (morphologically conditioned)

• *pya·- “come (AI)” + *‑ẅ “3” + *‑a “PROXsg → *pye·wa “s/he comes” (e.g., Meskwaki pye·wa, Munsee péew, Menominee pi·w, Miami-Illinois piiwa)
• Cf., with no *‑ẅ to condition umlaut: *ne- “1” + *pya·- → *nepya· “I come” (e.g., Meskwaki nepya, Munsee mpá, Menominee nepya·m, Miami-Illinois nimpya)
• Note an intervening morpheme is transparent to the triggering of umlaut: *pya·- + *‑ri “MARK.OBV.SBJ” + *‑ẅ + *‑ari “OBVsg” → *pye·riwari “s/he (OBV) comes” (e.g., Meskwaki pye·niwani, Unami archaic pé·ləwal [Goddard 2021:63])
• *na·θ- “fetch (TA)” + *‑a· “3OBJ” *‑ẅ + *‑a → *na·θe·wa “s/he (PROX) fetches (s.o. (OBV))” (e.g., Menominee na·nε·w)
• Cf. *Ø- “UNSPEC” + *na·θ- + *‑a· + *‑w “W.FMV” + *‑a → *na·θa·wa “s/he is fetched, someone fetches h/” (e.g., Menominee na·na·w)
• *wetama·- “smoke a pipe (AI)” + *‑ẅ “AI⇒Initial” + *‑api “sit (AI Final)” → stem *otame·wapi- “sit smoking (AI)” (e.g., Meskwaki atame·wapi-)

(13b) Traditionally, a second umlaut pattern is reconstructed as occurring only with a single Final, *‑ya· “go, move (AI Final),” which became *‑yi· before *|ẅ|. As with the much more common *a· → *e· umlaut, another morpheme could presumably intervene before inflectional *‑ẅ. Rule 13b must precede Rules 18 and 21.
ya· → yi· /__(+X)+ẅ (morphologically conditioned)

• *eθya·- “go {somewhere} (AI)” [with Initial *eθ- “RR:GOAL”] + *‑ẅ “3” → ?*ešyi·wa “s/he goes {somewhere}” (NipAlg izhii [Jones 1977:59], Menominee esi·w)
  • Cf. *eθya· + *‑t “3AN.CONJ” + *‑i “INDIC” → *ešya·či “when s/he went {somewhere}” (NipAlg izhaaj [ibid.], Menominee esya·t)

Pentland’s (1999, 2023) view on umlaut was significantly different from the general Algonquianist consensus (which I generally share); he agreed that umlaut existed, but believed that all verbs shared exactly the same umlaut patterns, and that all AI and II verb stems ending in *a· or *e· really ended in *a·, which umlauted to *e· only before *‑ẅ and a few other morphemes. He thus rejected the existence of an equivalent of Rule 13b, of AI verbs in “stable”/non-umlauting *‑a· (see e.g. Rule 17), of an umlaut of *e· to *a(·) in certain forms (see Rule 16), or of an II Final of the shape *|‑yayi| (see Rule 14(iv) and here in the section on *ayi contraction), among other things. Several of these points will be addressed in the relevant sections below.

I think his argument that Rule 13b didn’t exist (and instead *‑ya· “go, move” simply umlauted to *‑ye· per Rule 13a, and was later replaced in many daughters by the more productive AI Final *‑i· “go” [Pentland 2023, e.g. pp. 1971-1972]) is his most plausible one—despite the fact that it would bizarrely involve the three languages in which *ye· fell together with *i· by sound law (Ojibwe, Menominee, and Miami-Illinois) being precisely the languages which didn’t replace the original “go” Final in such verbs with a reflex of the Final *‑i·. But I need more time to consider this proposal before discussing it in further detail here.

(14) In a number of morphologically defined cases, the *i in a sequence *Giw is replaced by *e. Resulting *Vwe and *Vye sequences subsequently undergo contraction to a long vowel, or else a resulting postconsonantal *ye or *we coalesces to *i / *o. (The multiple *VGe contraction processes are subsumed under Rule 21, and are discussed in further detail in their own section below; coalescence of *ye and *we are covered in Rules 20a and 20b.). Specific cases in which this rule applies include when (Goddard 1991:176, 2013:108, n. 33): (i) the intransitive Final *‑iwi “be” is added to a noun stem in *‑w or *‑y; (ii) middle-reflexive AI verbs in *‑wi are inflected with a following *‑w; (iii) the verb *ni·pawi- “stand ({somewhere}) (AI)” or corresponding AI Final *‑ika·pawi “stand” are inflected with a following *‑w; (iv) the II Final *‑ye·yi (from underlying *|‑yayi| by Rule 15 below; further discussion of this Final is below) is inflected with a following *‑w; and (v) the antipassive Final *‑iwe· follows stems in *Cw. Rule 14 must follow Rules 9a and 13a, and must precede Rules 15, 18, 20a, 20b, 21, and 22.
i → e /G__{w,ẅ} (morphologically conditioned)

• *erenyiw- “man” + *‑iwi “be (AI/II Final)” + *‑ẅ “3” + *‑a “PROXsg” = **erenyiwiwiwa→ **erenyiwewiwa → *ereni·wiwa [*iwe → *i· contraction] “he is a man” (e.g. [Goddard 2006:191, n. 41]: Ojibwe ininiiwi, Arapaho hiineníínit, Cheyenne é-hetanéveo’o |=hetanéve-o| “they are men,” Moose Cree ililīwiw “he is a person, human, Cree, Native person; he is alive”)
• *nepy- “water” + *‑iwi + *‑ẅ + *‑i “INANsg” → **nepyewiwi → *nepiwiwi [*‑ye- → *‑i- coalescence] “it’s wet” (e.g., Meskwaki nepiwiwi)
• *ki·wa·mwi- “flee about (AI; middle-reflexive)” + *‑ẅ + *‑a “PROXsg” → **ki·wa·mwewa → *ki·wa·mowa “s/he flees here and there” (e.g., Meskwaki ki·wa·mowa; also e.g. Menominee me·cehsow “s/he eats” with verb |me·cehswi‑| ← PA *mi·čihswi- “eat (AI; middle-reflexive)”)
• Cf., with unmodified *i: *ki·wa·mwi- + *‑ankw “1pl.INCL.CONJ” → **ki·wa·mwiankwi → *ki·wa·mwiyankwe “that we (incl.) flee here and there” (note Menominee mi·cehsiyah “that we eat”; cf. Meskwaki [e·h‑]ki·wa·moyakwe [Thomason 2003:172] with o generalized throughout middle-reflexives)
• Note a non-middle-reflexive in *‑wi lacks this shift: *weLakwi- “be fat (AI)” [= *weLakw- “fat” + *‑i “(AI Final)”] + *‑ẅ + *‑a → *oLakwiwa “s/he’s fat” (e.g., Meskwaki anakwiwa)
• *ni·pawi- “stand ({somewhere}) (AI)” + *‑ẅ + *‑a → **ni·pawewa → *ni·po·wa “s/he stands ({somewhere})” (e.g., Cheyenne é-néé’e |=néé|; most other languages have changed the result of the contraction here to be *a· [Goddard 2000:94] or eliminated it)
• Cf. *ni·pawi- + *‑t “3AN.CONJ” → *ni·pawite· “if s/he stands ({somewhere})” (e.g., Ojibwe niibawid)
• *wa·p- “white” + *‑yayi “be, have quality (II Final)” + *‑ẅ + *‑i “INANsg” = **wa·pyayiwi → **wa·pyayewi → *wa·pya·wi “it’s white” (e.g., Oji-Cree waabaa, Sauk wa·pya·wi “it’s gray,” Arapaho stem |noočoo‑|)
• *sakipw- “bite (TA)” + *‑iwe· “ANTIP (AI Final)” → **sakipwewe·- → AI stem *sakipowe·- “bite (people)” (e.g., Meskwaki sakipowe·- [Goddard 2023:278, ex. 4.289i], Menominee |sakepowε·‑|)

(15) The sequence *‑ya- between two consonants becomes *‑ye·-. Rule 15 must follow Rules 9b and 14, and must precede Rules 16, 18, and 21.
a → e· /Cy__C

• *aʔseny- “rock, stone” + *‑ari “INANpl” and/or *‑aki “AN.PROXpl”^[22] → *aʔsenye·ri ~ *aʔsenye·ki “rocks, stones” (e.g., Ojibwe asiniin ~ asiniig, Potawatomi sənyek [Lockwood 2017:97], Meskwaki asenye·ni, Shawnee [pak]aʔθenyeeki “pebbles”)
• Medial *‑epy- “water, liquid” [regular ← noun stem *nepy- “water”] + Postmedial *‑ak → Medial *‑epye·k- “water, liquid” (e.g., *akwa·- “out of water” + *‑epy‑ak- + *‑i· “move (AI Final)” → *akwa·pye·ki·- “get out of the water, come ashore (AI),” as in Meskwaki akwa·pye·ki·-)
• Irregular initial-changed form of *pya·- “come (AI)”: *pya·- + *‑t “3AN.CONJ” + *‑a “PROXsg.PART” + IC [= infixation of *‑ay‑] → **py<ay>a·ta → *pye·ya·ta “s/he who came” (e.g., Meskwaki pye·ya·ta, Munsee péeyaat, Unami pé·a·t)
• Cf., with “regular” initial change: *pa·hpi- “be playful (AI)” + *‑t + *‑a + IC → *paya·hpita “s/he who is playful” (e.g., Ojibwe bayaapid “s/he who laughs”; [with regular loss of initial change of *a·:] Meskwaki ‡pa·hpita “s/he who performs tricks,” Unami ‡pa·pi·t “s/he who plays”)

(16) Instead of the *a· to *e· umlaut of Rule 13a, several morphemes instead cause umlaut of a preceding *e· to *a or *a·. These include, among others: (i) the nominalizing suffix *‑n; (ii) the formative *‑n(ay); (iii) the archaic nominalizing suffix *‑ʔm; (iv) the suffix *‑ntay which becomes the CINA preterit;^[23] (v) the transitivizers *‑θ and *‑t; and (vi) the Finals *‑s(‑w), *‑s‑w‑i “by heat(ing), burning.” See further below for these last two. Unlike the umlaut patterns covered by Rules 13a and 13b, there’s evidence that for at least some of these, an intervening morpheme does block its application. Rule 16 must follow Rule 15, possibly must precede Rule 19, and forms a component of Rule 27a.^[24]
e· → a(·) /__+{n(ay),n,ʔm,ntay,θ,t,s(‑),…} (morphologically conditioned)

• *pepikwe·- [~ initial-syllable vowel *a] “play a flute (AI)” + *‑n “NMLZ” + *‑i “INANsg” → *pepikwani [~ *pa…] “flute” (e.g., Ojibwe bibigwan, Menominee pepi·kwan, Miami-Illinois papikwani “gun,” Munsee ăpíikwan “musical instrument”)
• Cf. *ne- “1” + *pepikwe·- → *nepepikwe· “I’m playing a flute” (e.g., Ojibwe nimbibigwee, Menominee nepε·pekim [← pre-Menominee *|nεpεpekwε·+m|], Miami-Illinois †nipipikwee, Munsee ntapíikwe “I play a flute, musical instrument”)
• *ne- “1” + *mi·riwe·- “give (AI(+O))” + *‑n(ay) “N.FMV” + *‑i “INANsg” → *nemi·riwa·ni “I give it (away)” (e.g., Penobscot nə̀miləwαn)
• Cf. *ne- + *mi·riwe·- → *nemi·riwe· “I give” (e.g., Penobscot nə̀miləwe)
• *wenike·- “to portage (AI)” + *‑(e)ʔm “NMLZ” → *onikaʔmi “a portage” (e.g., Ojibwe onigam)
• Cf. *ke- “2” + *wenike·- → *ko·nike· “you (sg.) portage” (e.g., Ojibwe g[id]oonigee)
• *we- “3” + *pemwehθe·- “walk along (AI)” + *‑ntay “OPT” [see footnote 23 above] → *opemohθa·nta “may s/he walk along” (e.g., older(/High?) Plains Cree opimohtāh(tay) “s/he walked along”)
• Cf., with intervening morpheme apparently blocking umlaut: *we- + *pemwehθe·- + *‑ri “MARK.OBV.SBJ” + *‑ntay → *opemohθe·rinta “may s/he (OBV) walk along” (e.g., older(/High?) Plains Cree opimohtēyih(tay) “s/he (OBV) walked along”)

Rule 16 results in the only known interconsonantal *‑ya- sequences in PA (which had been removed by Rule 15 above). When an element ending in *‑ye· is followed by a morpheme which triggers *e· → *a umlaut, the result is a surface *CyaC sequence (cf. Goddard 2001:191-192). Such triggering morphemes included (as noted above) the Finals for “heating/burning” beginning in *‑s such as TI(1) *‑s; and the causative Finals TA *‑θ, TI(2) *‑t. For example:

• *wi·ʔsak- “painful; bitter” + *‑nenty‑e·- “hand” + *‑s‑wi “by heat, burning (AI Final)” → **wi·ʔsakinentye·swi- → AI stem *wi·ʔsakinenčyaswi- “burn one’s hand” (e.g., Kickapoo iiθakinecaθo‑)
• Cf. *wi·ʔsak- + *‑nenty‑e·- + *‑hθin “lie, fall (AI Final)” → AI stem *wi·ʔsakinenčye·hšin- “hurt one’s hand” (e.g., Kickapoo iiθakineceesin‑)
• *mo·škine·- “full” + *‑epye· “by water, liquid (AI/II Final)” + *‑θ “(TA Final)” or *‑t “(TI Final)” = **mo·škine·epye·θ/t- → *mo·škine·pyaθ/t- “fill s.o./sth. up with water, liquid (TA/TI)” (e.g., Menominee mo·skenεpyaN- (TA), mo·skenεpyat‑aw- (TI))
• Cf. *mo·škine·- + *‑epye· → *mo·škine·pye·- “be full of liquid (AI/II)” (e.g., Menominee mo·skenεpi·w “s/he/it is full of liquid” = |mo·skenε·-εpy‑ε·‑| + inflectional |‑ẅ|) (Goddard 2001:192)

Of the two possible outcomes of Rule 16, *a or *a·, the former was the more archaic (Goddard 1974:326, n. 66, 2007). For example, the nominalizer *‑n sometimes conditions umlaut to *a· and other times, in nouns that had been created at an earlier period, umlaut to *a, but the homophonous verbal formative ultimately derived from it only conditions umlaut to *a·. (Meanwhile, the nominalizer *‑ʔm conditions *e· → *a umlaut, while the homophonous verbal formative derived from it doesn’t condition umlaut at all.) There were other cases as well in which what were etymologically the same morpheme, but had since become differentiated, condition different umlaut outcomes under Rule 16. For example, the Finals *‑θ (TA) / *‑t (TI) normally serve as vague, general transitivizers or applicatives, and in such use they condition umlaut to *a·, as illustrated above. In rarer cases, however, they are causatives, and in that use they condition the more archaic umlaut to *a (cf. Goddard 2023:208-209, §37.6, 297-298, §§54.11-12 [describing Meskwaki]).

• Nominalizer *‑n = *a in more archaic cases: *e·ške·- “hunt beaver under the ice (AI)” + *‑n “NMLZ” → *e·škani “ice chisel (inan.)” and *e·škana “antler, horn (anim.)” (e.g., Ojibwe eeshkee- → eeshkan, Pessamit Innu esse- /esse‑/ → eshkan /eʃkən/; noun more widely attested, e.g., Mass ⟨askon⟩ ⫽a·skan⫽ [Psa. 18:2])
• But = *a· in most later, more productive / widespread formations: *apwe·- “bake, roast (sth./s.o.) over fire (AI(+O))” + *‑n → *apwa·ni ~ *apwa·na “roasted food item, roasted meat” (e.g., Ojibwe abwee- → abwaan, Menominee |apwe·‑| → apwa·n, Arapaho hočei- → hóčoo “steak; something fried”; in many EA languages verb often lost but noun remains as “bread” (originally “roasted cornbread”): Penobscot àpαn [← verb ape‑], Western Abenaki abʌ̨n, Munsee ăpwáan, Unami ahpɔ́·n |apwān|, etc.)
• And note the following noun, with an originally archaic *a: *ti·me·- “paddle (AI)” + *‑n → *či·mani “canoe” (Arapaho noun = θííw “boat,” pl. θííwonoʔ = |θííwon| [verb lost])
• But with original archaic *a replaced by more productive umlaut to *a· in the other daughters (Goddard 1974:326, n. 66, 2015b:402, n. 39): as if from PA *či·ma·ni (e.g., Ojibwe jiimee- → jiimaan, Meskwaki či·ma·ni [verb lost], Miami-Illinois ciimee- → ciimaani “a paddle, oar” [MyPD])
• Nominalizer *‑ʔm = *a (example repeated from above): *wenike·- “to portage (AI)” + *‑ʔm “NMLZ” → *onikaʔmi “a portage” (e.g., Ojibwe onigee- → onigam)
• But formative *‑ʔm = no umlaut: *ne- “1” + *po·n- “set down, place, cease to hold (TI)” + *‑e· [← *‑am by Rule 27a(i)] “INAN.OBJ(CL1)” + *‑ʔm “FMV” + *‑ena·n “1pl.EXCL” → *nepo·ne·ʔmena· “we (excl.) set (sth.) down, (etc.)” (e.g., Menominee nepo·nε·menaw “we (excl.) put it in the pot”)
• Transitivizers *‑θ/*‑t = *a as causatives (rarer): *pi·ntwike·- “enter, go inside (AI)” + *‑θ “CAUS (TA Final)” or *‑t “CAUS (TI(2) Final)” → *pi·ntwikaθ- (TA) / *pi·ntwikat‑aw- (TI) “bring s.o. / sth. in, inside” (e.g., Ojibwe biindigee- → biindigaN-/biindigad‑oo-, Meskwaki pi·tike·- → pi·tikaN-/pi·tikat‑o·-, Menominee pi·htikε·- → pi·htikaN-/pi·htikat‑aw-)
• But = *a· as applicatives (common and productive): *a·teʔro·hke·- “tell a sacred myth (AI)” + *‑θ “APPLIC (TA Final)” → *a·teʔro·hka·θ- “tell a sacred myth about s.o. (TA)” (e.g., Ojibwe aadizookee- → aadizookaaN-, Meskwaki a·teso·hke·- → a·teso·hka·N-; cf. Unami a·thilu·he·- → a·thilu·ha·l- “tell a sacred myth to s.o. (TA)” [LTD])

(17) The sequence *aw‑w simplifies, with compensatory lengthening, to *a·w, at least in verb inflection and derivation; there are some apparent exceptions in nouns and pronouns, some perhaps representing later analogy by the daughters, and/or perhaps what were not actual *aw‑w sequences at a pre-PA stage. Because it follows the application of umlaut, Rule 17 creates a class of verb stems (all AIs) with “stable” or non-umlauting *a·, which does not shift to *e· or *i· before *‑ẅ. Such verbs originally would have had stem-final *‑a· only in certain inflections, but by the PA period the vowel had been analogically extended to all contexts. Rule 17 must follow Rules 6, 9b, and 13a, must precede Rules 18, 19, 20b, and 21, and forms a component of Rule 27b.
aw → a· /__+w

• *maw- “weep, cry (AI)” + *‑ẅ “3” + *‑a “PROXsg” = **mawwa → *ma·wa “s/he weeps” (e.g., Menominee ma·w, Mass ⟨mâu⟩ ⫽mʌ̨·w⫽)
• Cf. *maw- + *‑rwe “2sg.IMPER” → *mawero “(you sg.) weep!” (e.g., Menominee mo·non [= |maw-ε-non| with vowel contraction], Mass ⟨maush⟩ ⫽mawəš⫽)
• Verbs in stable *a· were for the most part originally the middle-reflexives (formed by the suffix *‑w) of transitive stems in *‑aw, which were often benefactives or other ditransitive verbs (Goddard 1974:326, n. 60, 1979b:65-66), e.g.: *aθoskye·- “work” + *‑ʔta· “act, do (AI Final)” = stem *aθoskye·ʔta·‑, + *‑ẅ + *‑a → *aθoskye·ʔta·wa “s/he is engaged in work, does work” (e.g., Menominee anohki·ʔtaw)
• Cf. the corresponding benefactive TA in *‑aw (with the above verb’s stem originating in *‑aw + middle-reflexive *‑w): *aθoskye·- + *‑ʔtaw “act, do for, in relation to (TA Final)” = stem *aθoskye·ʔtaw-, + *‑a· “3OBJ” + *‑ẅ + *‑a → *aθoskye·ʔtawe·wa “s/he (PROX) works for (s.o. (OBV))” (e.g., Menominee anohki·ʔtawεw)
• The effects of Rule 17 could also be seen in the allomorphy of the TI(2) theme sign; for discussion and an example, see Rule 27b(ii) below.
• But cf, with no operation of Rule 17 (and simplification of *ww to *w by Rule 18 instead): *ke- “2” + *‑i·raw- “(emphatic personal pronoun base)” + *‑wa·w “2pl/3pl” → **ki·rawwa·w → *ki·rawa· “you yourselves, as for you (pl.)” (e.g., Ojibwe giinawaa, Shawnee kiilawa, Woods Cree kīthawāw [Starks 1992:49])

(18) The second of two consecutive semivowels drops. (Pentland 1999:248-249 claims that—at least in inflection—it’s the first semivowel that drops, but his apparent reasons can be explained in other ways.) Rule 18 must follow Rules 6, 9b, 9c(=9a), 9d(=9a), 13a, 13b, 14, 15, and 17, and must precede Rules 19, 20a, 20b, 21, and 22.
G → Ø /G__

• *ke- “2” + *koʔθ- “fear (TA)” + *‑a· “3OBJ” + *‑w “FMV” + *‑wa·w “2pl/3pl” + *‑aki “PROXpl” → **kekoʔθa·wwa·waki → *kekoʔθa·wa·waki “you (pl.) are afraid of them” (e.g., Moose Cree kikoštāwāwak)
• *ne- “1” + ?*paspakam- “hit repeatedly (TA)” + *‑ekw “INV” + *‑w + *‑ena·n “1pl.EXCL” + *‑a “PROXsg” → **nepaspakamekwwena·na → **nepaspakamekwena·na → ?*nepaspakamekona·na [*we → *o by Rule 20b] “s/he hit us (excl.) repeatedly” (e.g., Unami ‡mpəpahkámkwəna)
• Cf., with intervening suffix preventing loss of the second *w: *ne- + ?*paspakam- + *‑ekw + *‑hsi “DIM”+ *‑w + *‑ena·n + *‑a → **nepaspakamekwhsiwena·na → ?*nepaspakamekohsiwena·na “the little one (anim.) hit us (excl.) repeatedly” (e.g., Unami mpəpahkamkwət·í·wəna)
• *taskw- “short” + *‑yayi “be, have quality (II Final)” + *‑ẅ “3” + *‑i “INANsg” → **taskwyayiwi → *taskwa·wi “it’s short” (e.g., Ojibwe dakwaa)

(19) The sequence *wa between consonants in nominal and verbal inflection usually coalesces to *o·, but there are certain exceptions. The pattern in nominal inflection (Goddard 1981a:280-283, esp. 282, 2001:177, 182) was apparently that coalescence occurred: (i) always in inanimate nouns; (ii) always in animate stems ending in a consonant other than the sequence *‑kw; (iii) almost never (perhaps never) in names for animals ending in *‑kw; and (iv) sometimes, but more often not, in other animate stems ending in *‑kw. In verbal inflection, coalescence occurs when a disyllabic peripheral suffix beginning in *a (probably the only possibilities are *‑aki “AN.PROXpl” or *‑ari “INANpl”) is preceded by one of the formatives *‑w or *‑ẅ when these follow a consonant, viz. principally after: (v) the inverse theme sign *‑ekw; (vi) the TI(1) theme sign *‑am; and (vii) an II, AI, or TI(3) stem in *‑C. (Possibly only certain consonants led to coalescence by 19(vii).) Otherwise, coalescence in verb inflection does not occur, such as (viii) with an *a-initial conjunct central agreement suffix following a verb stem in *‑Cw (cf. Pentland 1999:249).

The degree to which *CwaC was retained as such in derivation isn’t entirely clear, but (ix) there are certainly a number of instances where it does coalesce, and this was the older treatment in most phonological environments. Because PA *CwaC is not reconstructible morpheme-internally except when the first consonant is *k (Nilsen 2017:20 citing Goddard, p.c.; cf. Pentland 2023), it’s quite likely that *wa-coalescence was originally a simple pre-PA sound change which was only blocked when the “*w” was really part of a pre-PA unit phoneme *k^w (contrasting with a cluster *kw), following Pentland’s (1999:226; cf. also Goddard 2001:182, n. 34) implicit suggestion. There is also comparative Algic evidence to support part, though not all, of this. As Goddard notes, however, and as the environments detailed in the previous paragraph make clear, the patterns which resulted from this sound change had become morphologized and disrupted by analogy by the actual Proto-Algonquian period. Rule 19 must follow Rules 6, 9b, 9d(=9a), 17, 18, and possibly 16.
wa → o· /C__C (lexically / morphologically conditioned)

• Inanimate noun — shows coalescence:
• *meʔtekw- “stick; tree” + *‑ari “INANpl” → *meʔteko·ri “sticks; trees” (e.g., Ojibwe mitigoon “sticks,” Meskwaki mehteko·ni, Arapaho bêeteíí “bows”)
• Animate noun stem not ending in *‑kw — shows coalescence:
• *we- “3” + *‑i·θemw- “cross-sibling-in-law” + *‑ahi “AN.OBVpl” → *wi·θemo·hi “h/ cross-siblings-in-law” (e.g., Northwestern Ojibwe wiinimoo’, Munsee dialectal wíiləmool [Goddard 2010:19])
• Animal name ending in *‑kw — lacks coalescence:
• *atehkw- “deer, cervid (?)” + *‑aki “AN.PROXpl” → *atehkwaki “deer (pl.), cervids” (e.g., Ojibwe adikwag “caribou (Rangifer tarandus) (pl.),” Munsee ătóhwak “deer (pl.)”)
• Non-animal animate stem ending in *‑kw — sometimes shows coalescence:
• *askehkw- “kettle [anim.]” + *‑aki “AN.PROXpl” → *askehko·ki “kettles” (e.g., Ojibwe akikoog, Meskwaki ahkohko·ki “kettles, drums,” Shawnee haʔkoʔkooki, Menominee ahkε·hkok)
• Non-animal animate stem ending in *‑kw — but usually lacks coalescence:
• *aθa(·)nkw- “star [anim.]” + *‑aki “AN.PROXpl” → *aθa(·)nkwaki “stars” (e.g., Meskwaki ana·kwaki, Shawnee halaakwaki, Unami alánkɔk |alankwak|, Old Algonquin ⟨[aran]gȣak⟩ arankwak [Aubin 2001:2])
• Verb inflection with *‑w/*‑ẅ and appropriate peripheral suffix — shows coalescence when preceded by *‑ekw “INV,” TI(1) theme sign *‑am, or *C-final II, AI, or TI(3) stem:
• *we- “3” + *na·θ- “fetch (TA)” + *‑ekw “INV” + *‑w “FMV” + *‑ari “OBVsg” → **wena·θekwari → *ona·θeko·ri “s/he (OBV) fetches h/ (PROX)” (e.g., Ojibwe onaanigoon)
• *meθk- “find (TI)” + *‑am “INAN.OBJ(CL1)” + *‑ẅ “3” + *‑aki “PROXpl” → *meθkamo·ki “they find (sth.)” (e.g., Kickapoo mehkamooki, Munsee ‡moxkámook)
• *wa~weθet- “be good, attractive (pl.) (II)” + *‑ẅ + *‑ari “INANpl” → *waweθeto·ri “they (inan.) are good, attractive” (e.g., Meskwaki we·weneto·ni, older Unami ɔwə́ltu·l, Menominee [with no reduplication] onε·ton |onεto·n|)
• Verb inflection otherwise — no coalescence:
• *mo·šw- “cut s.o.’s hair (TA)” + *‑a· [→ *Ø by Rule 10] “3OBJ” + *‑ak “1sg»3AN.CONJ” + *‑a “PROXsg.PART” + IC [lost in most daughters] → *mayo·šwaka “s/he whose hair I cut” (e.g., Ojibwe mwaazhwag, Miami-Illinois moošwaka “I cut h/ hair”; cf. Mi’kmaq temsək “I cut, slice, sever (with knife) h/” from *te·mešwaka with the same TA Final *‑ešw “by cutting”)
• Derivation — some evident coalescence, full extent less clear:
• *neʔθw- “three” + *‑api “sit (AI Final) + *‑ẅ “3” + *‑aki “PROXpl” → *neʔθo·piwaki “they sit as three, in a group of three” (e.g., Ojibwe nisoobiwag, Meskwaki neso·piwaki, Cheyenne é-na’héeo’o |=na’héeo| [é regular from *o·] [cf. Goddard 2001:183])

Goddard (2001:183) is equivocal on the antiquity of the coalescence pattern in the last example (quantifier Initials in *w followed by a morpheme beginning in *a), observing that “other languages” than Meskwaki “generally reflect” a putative uncoalesced *wa sequence (e.g., Menominee ni·swapewak; Plains Cree nistwapihkātam “s/he braids it in three”). As the last example above shows, though, there are more languages which at least sometimes show the same pattern, including, e.g., Ojibwe and Innu.^[26]

(20a) The sequence *ye between two consonants coalesces to *i. Rule 20a must follow Rules 9a, 9b, 14, and 18. As a historical sound change it preceded Rule 28, and also applied word-initially and probably word-finally (in those cases preceding the equivalent of Rules 1a/1b). Synchronically in PA, however, it does not apply finally, as in the injunctive (third-person imperative) ending *‑čye.
ye → i /C__C

• *wi·kopy- “basswood (Tilia americana) bark” + *‑ehke· “obtain by effort, use (AI Final)” → *wi·kopihke·wa “s/he gathers or prepares basswood bark” (e.g., Meskwaki wi·kopihke·wa)
• *nepy- “water” + *‑enki “LOC” → *nepinki “in the water” (e.g., Kickapoo nepiki, Shawnee nepiki, Arapaho nečíʔ)
• Does not apply word-finally: *ne·w- “see (TA)” + *‑i “1OBJ” + *‑tye “INJUNCTIVE” → *ne·wičye “may, let h/ see me” (e.g., Miami-Illinois neewice)

(20b) The sequence *we coalesces to *o except when bordered by vowels. Rule 20b must follow Rules 1a, 1b, 6, 7, 9a, 9b, 9c(=9a), 9d(=9a), 14, 17, and 18, and must precede Rule 24.
we → o /{#,C}__{#,C}

• *we- “3” + *kepahw- “shut in (TA)” + *‑ekw “INV” + *‑n(ay) “FMV” + *‑i “INANsg” → **wekepahwekweni → *okepahokoni “it shut h/ in” (e.g., Ojibwe ogiba’ogon)
• *pemw- “shoot (at) (TA)” + *‑a· “3OBJ” [→ *Ø by Rule 10] + *‑ehkwe “2pl»3AN.IMPER” → **pemwehkwe → *pemohko “(you pl.) shoot (at) h//them!” (e.g., Miami-Illinois pimohko)

(21) Many vowel-semivowel-vowel sequences contract to a long vowel. Some can be summarized as *V₁Ge → *V₁ː, but hardly all, and a number of the details are complex, especially for certain combinations. Contraction processes will be discussed in much more detail in their own section further below. In general terms, Rule 21 must follow Rules 4c, 6, 9a, 9c(=9a), 9d(=9a), 9e(=9a), 13a (since its output is not subject to umlaut), 13b, 14, 15, and 17, must precede Rules 22 and possibly 25, and forms a component of Rules 27b and 29.

In reality, however, “contraction” covers numerous patterns, not all of which are part of the ~same rule~, and some of which would need to apply at separate times in a more detailed breakdown than this one. For example, it was noted in the discussion of Rule 11(iii) that because Rule 11 precedes Rule 21 in the ordering used here—which it must, because of a chain of rules partially dependent on one another—it must be formulated awkwardly, such that it applies in the conjunct before central agreement suffixes except for ones which begin in *e, because the actual PA surface forms we see have, e.g., *‑ink “UNSPEC»1sg.CONJ” (= *‑i + *‑enk) instead of ^x‑i·nk (= *‑i + epenthetic *‑y- + *‑enk, with contraction of underlying *iye to *i·). A more fine-grained treatment of contraction rules could place the contraction of at least some *Vye sequences before Rule 11, but some others would still probably have to follow it.

Some of the contraction processes went back to sound changes in the deep pre-PA period, others were very recent, still others lay somewhere in between, and yet others didn’t derive (at least directly) from any historical sound change, but rather from the workings of analogy at various different times; synchronic (morpho)phonology decidedly does not simply recapitulate phylogeny, as it were (cf. Goddard 1983:351). In some cases even very specific subcomponents of Rule 21 would properly be assigned to different places on this list. All of which makes actually bundling all contraction into one “rule,” and coherently ordering it with respect to the other rules, technically impossible. But while I’ve just described the real synchronic situation in PA, the truth is it’s simply too impractical to treat and order all the contraction patterns separately here—so here they are, in “Rule 21.”
VGV → Vː (morphologically / lexically conditioned)

(22) Word-initial and postconsonantal *w is deleted before *o(·).^[27] Rule 22 must follow Rules 1a, 6, 7, 9b, 14, 18, and 21.
w → Ø /{#,C}__o(·)

• *we- “3” + *‑o·hθ- “father” + *‑ari “OBVsg” → **wo·hθari [by Rule 1a] → *Øo·hθari “h/ father” (e.g., Menominee o·hnan)
• *meʔtekw- “tree” + *‑weθ- “boat” → **meʔtekwweθi → **meʔtekwoθi [by Rule 20b] → *meʔtekoši “wooden boat, dugout canoe” (e.g., Plains Cree mistikōsi “wooden boat,” Menominee meʔteko·s) (Bloomfield 1925:139-140)

(23) Pre-PA/underlying *|i| and *|e| are neutralized in word-initial syllables; in almost all cases, by the PA period the underlying vowel was treated as if it were *|e|, when prefixes were added (i.e., it still surfaced as *e when the addition of a prefix moved it into the second syllable of a word). In the examples below, the notation *|E| represents a vowel whose underlying identity as *|e| vs. *|i| is not recoverable. There isn’t full agreement over the phonetic outcome of this neutralization. Traditionally it has been reconstructed as *e in all positions, but more recently Goddard has argued for the outcome being *i- initially and *‑e- after an onset, which is the pattern found in the western Algonquian languages. If one rejects Goddard’s model, these languages, including Blackfoot, thus must be assumed to have shared a post-PA sound change—a development that is possible, but may not be very plausible. For our purposes I continue the traditional notation of *e-/*‑e-, but am relatively uncommitted on what the phonetic reality actually was.^[28] Rule 23 must partially precede Rule 25, though as a historical sound change it partially followed it as well; as a sound change it also followed the equivalent of Rule 20a.
{i,e} → e /#(C)(G)__
    or:
{i,e} → i /#__   AND   {i,e} → e /#C(G)__

• *|eθkwe·w‑| “woman”:
• *|eθkwe·w‑| + *‑a “PROXsg” → *eθkwe·wa OR *iθkwe·wa “woman” (e.g., (western Algonquian) Meskwaki ihkwe·wa / (EA) Mass ⟨ussqua⟩ ⫽əsk^wa·⫽ “young woman” [⫽ə⫽ ← earlier *e], Northern Unami uxkwé·w |əxkwēw| [⟨ochqueuch⟩ LTD, ⟨Ochqueu⟩ Zeisberger 1827:36; |ə| ← earlier *e])
• *|kE‑| “2” + *|eθkwe·w‑| + *‑em “POSS” + *‑a → *keteθkwe·ma “your (sg.) woman; your (sg., male’s) sister” (e.g., Meskwaki ketehkwe·ma / Unami ktuxkwé·yəm |kət-əxkwēw-əm|)
• *|eθkwe·w‑| + *‑iwi [→ *‑ewi → *‑·wi by Rule 14(i)] “be (AI Final)” + *‑t “3AN.CONJ” + *‑a “PROXsg.PART” + IC [= *|e| → *e·] → *e·θkwe·wita “she who is a woman” (e.g., Meskwaki ‡e·hkwe·wita, Ojibwe eekweewid / Unami ‡e·xkwe·i·t^[S29])
• *|i‑| “say {so} (AI)” [combines with Rule 2 = “Palatalization I” to create a synchronically suppletive paradigm (Pentland 1979:391, 403)]:
• *|i‑| + *‑ẅ “3” + *‑a “PROXsg” → *ewa OR *iwa “s/he says {so}” (e.g., (western Algonquian) Kickapoo ia / (EA) Munsee əw^[30])
• *|nE‑| “1” + *|i‑| → **nEti- → *nesi “I say {so}” [epenthetic *‑t- → *‑s- before the *i by Rule 2] (e.g., Kickapoo neθi / Munsee nsí |nəsī| [Goddard 1982:31])
• *|i‑| + *‑t “3AN.CONJ” + *‑a “PROXsg.PART” + IC [= unique shift of *|i‑| → *ye·‑] → *ye·ta “s/he who says {so}” (e.g., Kickapoo yeeta)
• All other languages have leveled out this unique distinction between the initial-changed forms of initial *|e‑| and *|i‑|, generalizing either *e·- or *ye·- (Costa 1996:58-59). Note also the contrast between *|i‑θ‑| “say {so} to (TA)” (= the AI verb *|i‑| + applicative TA Final *‑θ) with initial change in *ye·- and *|eθ‑| “RR:MANNER” with initial change in *e·- (Costa 1996:59):
• *|iθ‑| “say {so} to (TA)” → *ye·θa·ta “s/he (PROX) who says {so} to h/ (OBV)” (Kickapoo yeenaata, but Munsee éelaat; independent indicative *eθe·wa OR *iθe·wa “s/he (PROX) says {so} to (s.o. (OBV))” → Kickapoo inea, Munsee léew |əl-āẅ| [cf. ntə́laaw “I say {so} to h/” |nət-əl-āw|])
• *|eθ‑| “RR:MANNER” + *‑eka· “dance (AI Final)” → *e·θeka·ta “s/he who dances {so}” (Kickapoo eenekaata [form from Costa], Munsee éelkaat; independent indicative *eθeke·wa OR *iθeka·wa “s/he dances {so}” → Kickapoo inekea, Munsee lə́keew |ələkā‑| [cf. ntə́lka “I dance {so}” |nət-ələkā|])

(24) The conjunct indicative/default suffix *‑i and the INANsg/default participial head suffix *‑i are realized as *‑e after the first- and second-person plural conjunct suffixes ending in *k(w) (viz. *‑a·nk “1pl.EXCL,” *‑ankw “1pl.INCL,” *‑e·kw “2pl,” *‑a·kw “3»2pl”/“UNSPEC»2pl,” and *‑akokw “1sg»2pl”; alternatively, the environment could be described as: after plural suffixes not ending in *t (Proulx 1990:109; Goddard 2015b:368). This allomorphy is the pattern found in Meskwakian (Meskwaki, Sauk, and Kickapoo) and Shawnee, but has basically been leveled out in Miami-Illinois. (Other daughter languages lost final short vowels and so provide no evidence.) Rule 24 must follow Rule 20b.
i → e /k(w)+__# (morphologically conditioned)

• *(eh)a·- “go (AI)” + *‑ankw “1pl.INCL.CONJ” + *‑i “INDIC” → *(eh)a·yankwe “that we (incl.) go” (e.g., Meskwaki [wi·h‑]a·yakwe “(that) we’ll (incl.) go” [Thomason 2015:442]; cf. regularized Miami-Illinois iiyaayankwi “we (incl.) go”)
• *se·kih- “frighten (TA)” + *‑eθ “2OBJ” + *‑a·kw “3»2pl.CONJ” + *‑i “INANsg.PART” + IC → *saye·kiheθa·kwe “that (inan.) which scared you (pl.)” (e.g., Kickapoo θeekihenaakwe; cf. regularized Old Illinois teepeerimeraakwi “s/he/they control(s), rule(s) over you (pl.)” [← *taye·pe·remeθa·kwe “that (inan.) which rules you (pl.); when s/he/they/it ruled you (pl.); you (pl.) who are ruled by h//them/it”])

(25) *t and *θ are palatalized (in traditional Algonquian parlance, undergo “mutation”) to *č and *š respectively when preceding *i(·) or *y. This very productive rule is Pentland’s “Palatalization II,” opposed to the fossilized instances of “Palatalization I” (*t → *s, Rule 2). In fact, while it’s convenient to treat the Rule 25 palatalization of *t and *θ together as one process, historically the change palatalizing *θ to *š must have occurred considerably earlier than that palatalizing *t to *č; comparative evidence also shows that it took place in additional environments. Rule 25 must follow Rules 2, 3, 9a, 10, 23 (but, in line with the preceding note, *θ → *š preceded Rule 23 as a historical change), and possibly 21, and must precede Rule 26.
t → č /__{i,i·,y}
θ → š /__{i,i·,y}

• *ne- “1” + *‑sit- “foot” + *‑i “INANsg” → *nesiči “my foot” (e.g., Shawnee niθiči)
• Cf. *ne- + *‑sit- + *‑ari “INANpl” → *nesitari “my feet” (e.g., Shawnee niθitali)
• *werake·weθ- “bark canoe” + *‑i → *orake·ši “[one] bark canoe” (e.g., Shawnee holakeeši “boat”)
• Cf. *werake·weθ- + *‑ari → *orake·θari “bark canoes” (e.g., Shawnee holakeelali “boats”)
• *ketoθ- “speak to (esp. negatively) (TA)” + *‑i “1OBJ” + *‑an “2sg.CONJ” → *ketošiyane· “if you (sg.) speak to me” (e.g., Moose Cree ‡kitošiyanē “if you (sg.) scold me”)
  • Cf. *ketoθ- + *‑eθ “2OBJ” + *‑a·n “1sg.CONJ” → *ketoθeθa·ne· “if I speak to you (sg.)” (e.g., Moose Cree ‡kitotitānē “if I scold you (sg.)”)
• *šenk- “spread out, flat” + *‑hθ “fall, lie (concrete Final)” + *‑in “(AI abstract Final)” → AI stem *šenkihšin- “lie down (flat / spread out)” (e.g., Ojibwe zhingishin-, Meskwaki šekišin-, Menominee sεhke·hsen-)
• Cf. *šenk- + *‑hθ + *‑en “(II abstract Final)” → II stem šenkihθen- “lie down (flat / spread out)” (e.g., Ojibwe zhingisin-, Meskwaki šekisen-, Menominee sεhke·hnεn-)

(26) A nasal assimilates in place of articulation to any following true consonant (i.e., excepting semivowels). The allophonic *[ŋ] which preceded *k is written as *n. Rule 26 must follow Rules 3, 4a, 8a, 8b, 9b, and 25, and forms a component of Rule 27a.
N → [αPOA] /__C[αPOA, ≠G]

• *akim- “count (TA)” + *‑t “(TI Final)” + *‑am “INAN.OBJ(CL1)” + *‑a·n “1sg.CONJ” + *‑i “INANsg.PART” + IC → *e·kintama·ni “what (INANsg) I count” (e.g., Ojibwe eegindamaan, Miami-Illinois eekintamaani “I count it”)
• *wa·pam- “look at (TA)” + *‑t + *‑am + *‑rwe “2sg.IMPER” → *wa·pantanro “look (sg.) at it!” (e.g., Miami-Illinois waapantanto)
• *šek- “urinate” + *‑enkwa·m “sleeping (AI Final)” + *‑t [→ *‑k by Rule 3] “3AN.CONJ” + *‑e· “SBJV” → **šekenkwa·mke· → *šekenkwa·nke· “if s/he wets the bed” (e.g., Ojibwe zhigingwaang)

(27a) This “rule” is essentially just a way of stating the allomorphy of the TI(1) theme sign *|‑am| by conveniently gathering in one place the processes which are involved. In reality, several of the component processes must apply at different times, and most are rules found elsewhere in this list. The only unique process is an otherwise unparalleled *am ~ *e· variation.

The theme sign has the following allomorphs: (i) *‑e· before the formative *‑(e)ʔm; (ii) *‑a· before the formative *‑n(ay); (iii) *‑an before the 3AN.CONJ suffix *‑k [← *‑t by Rule 3] and 2sg.IMPER suffix *‑ro; and (iv) *‑am elsewhere.^[31] In many daughter languages one or another of the allomorphs has been extended to additional environments. Rule 27a is composed of Rules 9a, 9b, 16, and 26, in addition to the unique *am ~ *e· variation. It must follow Rules 3 and 4a and must precede Rules 28 and 29.
am → am, e·, a·, an (morphologically conditioned)

• *ne- “1” + *po·n- “set down (TI)” + *‑am “INAN.OBJ(CL1)” + *‑ʔm “M.FMV” + *‑ena·n “1pl.EXCL” → *nepo·ne·ʔmena· “we (excl.) set (sth.) down” (e.g., Menominee nepo·nε·menaw “we (excl.) put it in the pot”; cf. Ojibwe ninoondaamin “we (excl.) hear it” [verb noond‑am- ← PA no·nt‑am‑], with the *‑a·- allomorph analogically extended)
• *ne- + *po·n- + *‑am + *‑n(ay) “N.FMV” + *‑i “INANsg” → *neno·nta·ni “I set (sth.) down” (Menominee nepo·na·n “I put it in the pot”; and Ojibwe ninoondaan “I hear it” with noond‑am-)
• *po·n- + *‑am + *‑k [← *‑t] “3AN.CONJ” + *‑iri “ITERATIVE” + IC → *payo·nankiri “whenever s/he puts it/them (INAN) down” (e.g., Menominee ‡payo·nahken “whenever s/he sets it/them (INAN) in the pot”; and Ojibwe nwaandangin “whenever s/he hears it/them (INAN)” with noond‑am-)
• *po·n- + *‑am + *‑a·n “1sg.CONJ” → *po·nama·ne· “if I set it/them (INAN) down” (e.g., Menominee ‡po·naman “if I/you (sg.) put it/them (INAN) in the pot”; and Ojibwe noondamaan “if I hear it/them (INAN)” with noond‑am-)

(27b) The exact allomorphy of TI(2) theme sign *|‑aw| is more difficult to reconstruct, since none of the daughter languages apparently reflects the original distribution precisely. It seems to have been realized as: (i) *‑o· before the formatives *‑n(ay) and *‑(e)ʔm; (ii) *‑a· before the formative *‑ẅ; and (iii) *‑aw mostly elsewhere, though not all the details are known with certainty (Goddard 1979b:72-73; Pentland 1999:253-254). (iv) Before the animate third-person conjunct suffix *‑t [→ *‑k], the underlying allomorph was *|‑aw|, but the *w metathesized with the following consonant (by Rule 7) and the vowel was compensatorily lengthened (unlike in other cases of Rule 7), giving surface *‑a·kw. The MARK.OBV.SBJ suffix *‑ri presumably worked the same as in TI(1) verbs and was transparent for the purposes of allomorph selection. See here for more on the distribution of the TI(2) theme sign’s allomorphs throughout the family. Rule 27b is composed of Rules 7, 9a, 9b, 17, and 21, as well as the unique compensatory lengthening of *a before metathesized *wk → *kw. It must follow Rule 3 and must precede Rule 28.
aw → aw, o·, a· (morphologically conditioned)

• *we- “3” + *aʔt- “place, put (TI)” + *‑aw “INAN.OBJ(CL2)” + *‑n(ay) “N.FMV” + *‑i “INANsg” → *otaʔto·ni “s/he places it” (e.g., Ojibwe odatoon)
• *aʔt- + *‑aw + *‑ẅ “3” + *‑aki “PROXpl” → *aʔta·waki “they place (sth.)” (e.g., Plains Cree astāwak)
• *kwetiht- “try (TI)” + *‑aw + *‑kwe “2pl.IMPER” → *kočihtawoko “(you pl.) try it!” (e.g., Mass ⟨qutchehteóꝏk⟩ ⫽k^wəčəhtyawu·k⫽ [1 Thes. 5:21])
• *ankiht- “eliminate; lose (TI)” + *‑aw + *‑k [← *‑t by Rule 3] “3AN.CONJ” + *‑e· “SBJV” → *ankihta·kwe· “if s/he eliminates, loses it” (e.g., Unami ankhitá·k·we |ankəhtākwē| ← |ankəht-aw-t-ē| “if s/he loses it”)

(28) Word-final consonants and consonant clusters are lost. Rule 28 must follow Rules 20a, 27a, and 27b.
(C)C(G) → Ø /__#

• *ke- “2” + *nepa·- “sleep (AI)” + *‑eʔm “FMV” + *‑wa·w “2pl/3pl” → **kenepa·ʔmwa·w → *kenepa·ʔmwa·Ø “you (pl.) are asleep” (e.g., Ojibwe ginibaamwaa, Shawnee kinepaapwa [Luke 22:46]^[32])
• *eškwa·nte·m- “door” [noun stem; to appear as a full noun needing one of the gender/number suffixes *‑i or *‑ari] → *eškwa·nte·Ø (adverbial particle) “at the door” (e.g., Munsee əskwáante, Ojibwe [jiig]ishkwaand “by a door,” Meskwaki [kep]iškwa·te “in the doorway”)
• *ne- “1” + *akwintin- “float (AI)” + *‑eʔm “FMV” → **netakwintineʔm → *netakwinčineØ “I’m floating” (e.g., Old Illinois ⟨nitacȣintchine⟩ †nitakwincine)

(29) The N-formative *‑n(ay) has the allomorphs (i) *‑ne·- before a consonant (= before one of the pluralizers or a consonant-initial mode suffix), and (ii) *‑n- before a vowel. (It surfaces with an initial epenthetic *e when following a consonant—by Rule 9a—and with no initial *e when following a vowel, including the *a· allomorph of the TI(1) theme sign *|am| [see Rule 27a(ii)].) Rule 29 is composed of Rules 9a, 9b, 9c(=9a), 21, and the unique loss of *ay before a vowel. It must follow Rule 27a.
n(ay) → n /__+V (morphologically conditioned)
n(ay) → ne· /__+C (morphologically conditioned)

• *we- “3” + *meθk- “find (TI)” + *‑am [→ *‑a· by Rule 27a(ii)] “INAN.OBJ(CL1)” + *‑n(ay) “N.FMV” + *‑ari “INANpl” → *omeθka·nari “s/he finds them (INAN)” (e.g., Munsee ‡moxkámənal, Ojibwe omikaanan)
• *we- + *meθk- + *‑am [→ *‑a·] + *‑n(ay) + *‑wa·w “2pl/3pl” + *‑ari → *omeθka·ne·wa·wari “they find them (INAN)” (e.g., Munsee moxkamənéewa “they find it/them (INAN),” Ojibwe omikaanaawaan [with *e· regularly “recontracted” to aa])

Coda

As stated up front, this list is not exhaustive, and some of the rules as formulated leave out some relatively minor details. There were various other minor rules that applied to individual morphemes in specific contexts which I have not attempted to catalogue here; in other cases, a full description would involve too much argumentation or discussion for this particular post than the rule itself warranted. There are no doubt also processes I have simply missed or which slipped my mind. And then there are processes which are still not fully understood. To give a few examples of miscellaneous additional, potential rules which are not dealt with elsewhere here, at least in full:

(Coda ex. 1) In Rule 8a it was claimed that *‑wa·w “2pl/3pl” drops its final *w before a mode suffix, just as *‑ena·n “1pl.EXCL” drops its final *n. But an alternative possibility would be that rather that *‑wa·w dropping its final consonant, an epenthetic *‑e- was inserted after it if necessary (Rule 9a), and then the resulting *a·we sequence contracted to *a· (Rule 21; *a· being the regular outcome of *|a·we| contraction following a *w), leaving *‑wa·‑C (cf. Pentland 1999:251).

Three reasons to prefer the analysis as given in Rule 8a over the one just mentioned are: (1) the EMPH.PAST suffix *‑pan evidently didn’t originally insert preceding epenthetic vowels (Goddard 2007:250; see Rules 5a and 9e), and it’s unclear whether *‑(e)sahan did (see Rules 8c and 9e); (2) the final consonants of *‑ena·n and *‑wa·w are treated identically in an obscure alternative to Rules 8a and 8b, discussed below in Coda ex. 2; and (3) historically speaking, *‑wa·w lost its final *w in separate instance, when it was pressed into service as an augment *‑wa·- adding the notion “plural” to a following *‑t “3AN.CONJ” in certain forms (if it had been simply inserted as *‑wa·w‑, we would expect the resulting ending to have been ^xwa·kw… by Rules 3 and 7, not *‑wa·t… [cf. Pentland 2023:76, s.v. *‑twa·w‑]). But in favor of this analysis over the one given in Rule 8a, note that the 1pl.INCL suffix *‑enaw inserted a following epenthetic *‑e- with subsequent contraction, by all appearances including when preceding, e.g., *‑pan. It would not be surprising if *‑wa·w mirrored the behavior of *‑enaw, and the apparent contraction of *‑enaw even before *‑pan weakens the first point above. That said, the full picture of contraction patterns involving the mode suffixes remains decidedly hazy, as partly discussed below.

(Coda ex. 2) There’s another complication to Rules 8a and 8b (deletion of the final consonant from *‑ena·n, *‑wa·w, *‑a·n “1sg.CONJ,” and possibly *‑an “2sg.CONJ,” when preceding a mode suffix) which I omitted from the discussion. While generally the daughters reflect this process or have analogically undone it, a few instead, or as an alternative option, replace the final *n or *w with “a *kw of uncertain origin” (Goddard 2007:251; cf. Goddard 1979b:112-113, 126-127, 149, 2021:89, 96, ex. 4.78c).

• Northern Unami ⟨kpahummoogup⟩ †kpahəmɔ́·k·əp |kə-pā-hm.wākw-əp| for “you (pl.) came” (“ihr kamt” [Grube n.d., sig. 6:[8] = pg. [88]]), as if from PA *kepya·ʔmwa·kopa “you (pl.) had come” ← **kepya·ʔmwa·kwpan ← *ke- “2” + *pya·- “come (AI)” + *‑ʔm “FMV” + *‑wa·w “2pl/3pl” + *‑pan “EMPH.PAST”
• Northern Unami ⟨mpahuummennaaxa⟩ [for *⟨mpahummennaaxa⟩] †mpahəməná·kwsa |nə-pā-hm.ənākw-sa| for “we (excl.) have come” (“wir sind gekom̄en” [ibid.]), as if from PA *nepya·ʔmena·kosaha ← *nepya·ʔmena·kw(e)sahan ← *ne- “1” + *pya·- + *‑ʔm + *‑ena·n “1pl.EXCL” + *‑(e)sahan “EMPH.PRES”
• Northern Unami ⟨payakpanne⟩ †pa·ya·kpáne |pā-yākw-əpan-ē| ← |pā-ākw-pan-ē| “if, when I had come” (Zeisberger 1827:87; Goddard 1979b:126-127), as if from PA *pya·ya·kopane· ← **pya·ya·kwpane· ← *pya·- + *‑a·n “1sg.CONJ” + *‑pan “EMPH.PAST” + *‑e· “SBJV”

(Coda ex. 3) One minor rule of many that I’ve left out but might merit inclusion involves an interesting portmanteau ending. The 3OBJ theme sign *‑a· combines with the potential suffix *‑hk with no intervening suffixes to form a portmanteau suffix *‑iye·k “3OBJ.POT” (Goddard 1979b:101-102, 2006:198-199); the component elements are restored when another morpheme intervenes to block their fusion. To my knowledge this constitutes the only case of a Proto-Algonquian portmanteau suffix which incorporates a virtual theme sign.^[33]

• *neʔr- “kill (TA)” + *‑a· “3OBJ” + *‑hk “POT” + *‑at “2sg»3AN.CONJ” + ‑i “INDIC” → **neʔriye·kati (+ *ka·ta “NEG”) → *ka·ta neʔriye·kači “don’t (sg.) kill h//them!” (e.g., Unami káči nhiliyé·k·ač, Meskwaki ka·ta nešiye·kani [with *‑an “2sg.CONJ” replacing *‑at] [Goddard 2006:195, n. 50, 197], NipAlg gaan nishiyeeg[een] [with analogical further ending])
• Cf., with intervening suffix preventing the portmanteau fusion: *neʔr- + *‑a· + *‑w “IRR” + *‑hk + *‑at + ‑i → **neʔra·wehkati → *neʔro·hkači [contraction of *‑a·we- by Rule 21] “you (sg.) should not neglect to kill h//them” (e.g., Meskwaki nesa·hkani “kill h//them later!” [Goddard 2006:196], NipAlg nisaakan “id.,” both with “recontraction” of *o· to /aː/)

(Coda ex. 4) The creation of derivational elements from stems involves a large and eclectic collection of patterns (cf. Bloomfield 1946:104-105, 107, 111, etc. [§§56, 60, 65, 78], 1962:76-77, 266-272, 321-324, 327-328, 381, etc. [for Menominee]; Goddard 1990:463-469, 2021:117-118, 126-128, 130-131 [for Unami], 2023:178-183, 187-192, 194, 204-205, 214-215 [for Meskwaki]). In a good number of cases a so-called “deverbal” (a non-Initial derived from a stem or stem+theme) is identical in shape to the stem, but other patterns include: the loss of an initial sonorant; shortening of long high vowels or deletion of *e (pre-PA *ə; with subsequent resolution of a resulting cluster by some frozen, unproductive processes, or the rules catalogued in this post); and replacement of initial *e- (original, or after loss of an initial sonorant) by *‑i-. When necessary deverbal inserts a preceding epenthetic *‑i- or occasionally *‑e‑, which has led to varying analyses of deverbals which begin with a surface *i- or *e-. Regardless of the correct synchronic analysis, from a historical perspective such initial vowels can derive both from vowels which were always segmentally present as well as from later epenthetic vowels. In deriving deverbals from dependent noun stems, additional common processes include the loss of the first member of a consonant cluster; and nasal loss or metathesis (*‑C₁VNC₂- → *‑(N)C₁VC₂‑). These descriptions don’t include the frequent addition of obligatory but basically semantically empty “Premedial” (e.g., *‑a·-) or “Postmedial” (e.g., *‑ak-) or similar, since these were not modifications of the shape of the stem⇒deverbal element itself.

I intend to eventually fill in this section with actual examples of common, rare, and idiosyncratic patterns for creating deverbals and Initials from stems, but simply do not have the necessary time or energy at the moment, and crave your pardon.

Contraction

Rule 21 above concerned the contraction of *VGV sequences, but as noted these have frequently proven tricky to reconstruct. It is often the case that for the same underlying sequence, different daughters show different kinds of contraction, or have no contraction at all. Indeed, in some cases the same sequence can be shown to have had multiple contraction outcomes in the protolanguage alone, depending on the phonological or morphological environment: compare, for example, the varying outcomes of the verb stem *no·ntaw- “hear (TA)” below: *ke-no·ntaw‑ekw‑a → *keno·nta·kwa “s/he hears you (sg.)” vs. *ke-no·ntaw‑eθ-eʔm → *keno·nto·θe “I hear you (sg.).” In spite of these complications, in this section I’ll give the best current understanding of the outcome of various *VGV sequences in the PA period as laid out primarily by Goddard (2001), and making heavy use of his examples and insights from that paper. But be aware that there are many aspects of contraction in Proto-Algonquian which remain uncertain.

[I just said it, but I have to reemphasize my significant reliance on Goddard 2001 here. Even though I’ve made use of other sources and have added other observations of my own, much of this section is essentially a distillation of his paper. I attempt to cite it—by just the page number, in parentheses—throughout when it’s the source of any specific insight or the immediate general discussion; I don’t always cite individual examples, which are a mixture of his, some from elsewhere, and some of my own.]

As an aid to navigation, an expanded table of contents covering this section of the post is below.

Table of Contents: Proto-Algonquian Contraction

Introduction Note on General Patterns *awe *aye *a·we *ewe Other *VGe *Vwe in Verb Inflection *‑Vw TA Stems *‑e·w Stems Within Inflectional Endings Realis Non-TA Verbs Contraction with Mode Suffixes Other Patterns *ayi *aya *eye Summary of Possible Outcomes

Note on General Patterns

Almost all contraction occurred more specifically with *VGe sequences, and had two possible outcomes: (1) a long version of the first vowel (e.g., *iwe → *i·, *i·we → *i·); or (2) a different long vowel (e.g., *awe → *o·, *a·we → *o·). The first outcome is more typical for sequences beginning in *i(·)- and *o(·)-, while the second is more typical for *a(·)- and *e(·)-, but there are exceptions. A given *VGe sequence in a certain environment could also, however, remain uncontracted (e.g., *a·‑w‑e = *a·we). (For a summary table of the various possible contraction outcomes for every contracting *VGV sequence in PA, see the end of the post here.)

In turn, a given daughter language can show one of three patterns for a particular contraction pattern in the protolanguage. First, it can continue it unchanged. Second, it can show decontraction, with the original *VGV sequence (or something close to it) analogically restored (e.g., *o· = *|ow‑e| → ‑ow‑e-). Third, it can show what may be termed recontraction (following Goddard and Pentland), with a contracted long vowel analogically shifted to match the quality of its underlying initial vowel (e.g., *e· = *|ay‑e| → a·).

This profusion of patterns and outcomes, plus the fact that some daughter languages, in addition to this, occasionally show various new contractions of |VGV| sequences—both to a long version of the first vowel and sometimes to a vowel of a different quality—is a major part of what makes reconstructing contraction patterns in PA so challenging.

Some other general patterns can be observed here (cf. 166-168). Certain morphemes commonly or always condition contraction, while others seldom or never do. Based on these, we can say that contraction is very common with nominal inflectional suffixes, quite common with verbal inflectional suffixes, somewhat but not terribly common with derivational suffixes used in secondary derivation (i.e., suffixes applied to an already complete stem to create a new stem; but again, only with certain such suffixes), and much rarer with derivational suffixes used in primary derivation (i.e., suffixes used to form primary stems by combining with elements that are not full stems/words on their own). In the examples below, for instance, you’ll note that a great many illustrate some contraction process using the derivational AI Final *‑ehke· “obtain by effort, use,” which is added to fully formed noun stems in secondary derivation. But you won’t note many examples containing Finals used in primary derivation. (Nor will you find any with, e.g., the secondary nominalizing Final *‑wen, which unlikely *‑ehke· never contracted with a preceding vowel in PA.)

Likewise, there are also a few Initials or stems which seem to have consistently resisted contraction, like the dependent noun stem *‑i·w- “wife” (195, 207), TA verb stem *aw- “use” (196, 198, 224-225), or verbal Initial *nay- “carry.” Like these examples, many such noncontracting Initials/stems were monosyllabic, so presumably at least part of the explanation is that they were so short that contraction would have obliterated most of the Initial/stem and made identification of it unacceptably difficult. Other noncontracting initial elements may have that property through various analogical connections to monosyllabic noncontracting elements. (Though there were monosyllables that did contract, notably the TA verb stem *ne·w- “see”; unsurprisingly, the contraction of *ne·w- has been reversed in all but one daughter language, as discussed below.)

A final pattern which can be noted is that the origin and phonological environment of a *VGe sequence plays a role in how likely it is to contract. So for example, *VGe sequences which arise from what at a deeper level are *|VGi| sequences—in which the *|i| has become *e by Rule 14—always contract: e.g., *awe doesn’t always contract, but *awe from *|awi| before *w always does. Similarly, contraction never applies word-finally; it can only apply before a consonant.

*awe

The underlying sequence *|awe| in PA had three different outcomes: uncontracted *awe, contraction to *o·, and contraction to *a·.

*aw‑e contracts to *a· in verbal inflection before the inverse theme sign *‑ekw, and in derivation before the reciprocal suffix *‑(e)twi (208). For obscure reasons, in PEA, *‑aw + reciprocal *‑(ə)tī instead “contract” to *‑āwatī, with an extra *‑wa‑ sequence added after the contracted vowel.

• *ke- “2” + *no·ntaw- “hear (TA)” + *‑ekw “INV” → *keno·nta·kwa “s/he hears you (sg.)” (e.g., Ojibwe ginoondaag, Meskwaki ‡keno·ta·kwa “s/he heard, understands you (sg.)”; also e.g., Munsee kpə́ntaakw with verb pəntaw- ← PA *pentaw-, infl. *kepenta·kwa)
• *a·timoʔtaw- “tell, narrate to (TA)” + *‑etwi “RECIP” → stem *a·čimoʔta·twi- “tell e.o., narrate to e.o., say to e.o. (AI)” (e.g., Shawnee haačimoʔtaw- ⇒ haačimoʔtaati- [Mark 5:16; Luke 24:14])
• Cf. in Eastern Algonquian: pseudo-PA *ši·nkinaw- “hate the sight of (TA)” + *‑etwi → pseudo-PA *ši·nkina·twi- “hate the sight of e.o. (AI)”; PEA *šīnkīnaw- “despise, hate (TA)” + *‑ətī → *šīnkīnāwatī- “despise, hate e.o. (AI)” (e.g., Munsee shiinkiinaw- ⇒ shiinkiinaawătii-)

In other cases when *aw‑e contracts it most often results in *o·. This occurs in verbal inflection before the 2OBJ theme sign *‑eθ (209) or after the 1pl.INCL suffix *‑enaw, and in nominal inflection (e.g., before the locative suffix *‑(e)nki) (205). Another inflectional example involves the AI stem *ni·pawi- “stand” and its AI Final counterpart *‑ika·pawi when before a suffix beginning in *w (→ intermediate **ni·pawe- / **‑ika·pawe; see Rule 14(iii)). *awe sometimes contracts to *o· in derivation as well.

• *ke- + *no·ntaw- + *‑eθ “2OBJ” → *keno·nto·θe “I hear you (sg.)” (e.g., Ojibwe ginoondoon, Meskwaki keno·to·ne “I heard, understand you (sg.)” [197]; also e.g., Munsee kpə́ntool with pəntaw- ← *kepento·θe)
• *ke- + *pemwehθe·- “walk along (AI)” + *‑ʔm “FMV” + *‑enaw “1pl.INCL” + *‑(e)sahan “EMPH.PRES” → **kepemwehθe·ʔmenawesahan → ?*kepemohθe·ʔmeno·saha “we (incl.) have walked along” (e.g., Maliseet kpəmóhsepənos “we two (incl.) must have walked by” [Goddard 1979a:111, n. 18])
  • Also e.g., Menominee kema·cya·menosah “so we (incl.) are setting out” with verb |ma·cya·‑| “get going, go away (AI)” (|kεma·cya·mEno·sa| ← |kε-ma·cya·-mEnaw‑Esa| ← PA *ma·čya·-, infl. *kema·čya·ʔmeno·saha)
• *re·kaw- “sand” + *‑enki “LOC” → *re·ko·nki “on the sand” (e.g., Munsee léekoonk)
• *ne- “1” + *‑i·yaw- “body” + *‑ena·n “1pl.EXCL” → *ni·yo·na·ni “our (excl.) bodies; ourselves (excl.)” (e.g., Miami-Illinois niiyoonaani)
• *wi·yaw- “h/ body [possessed theme]” + *‑ehs “DIM” → *wi·yo·hsi “flesh, meat” (e.g., Miami-Illinois wiiyoohsi, Munsee wə̆yóos, Penobscot wə̀yos)
• *maw- “cry, weep (AI)” + *‑h “CAUS (TA Final)” → **maweh- → *mo·h- “make s.o. cry (TA)” (e.g., Ojibwe moo’-, Menominee mo·h-)

With other inflectional suffixes, contraction evidently does not occur, as detailed below. It is also generally though not completely absent from primary derivation. [Sorry I can’t quickly find more secure examples than those below, but again, the process they reflect is valid to reconstruct for the protolanguage.]

• *a·θ(a)w- “fail” + *‑esi “be, have quality (AI Final)” → pseudo-PA *a·θawesiwa “s/he fails” (Meskwaki a·nawesiwa [204])
• *pi·htaw- “layered” + *‑esi → pseudo-PA *pi·htawesiwa “s/he is layered, in layers” (Munsee pihtawə́səw)

These patterns are subject to frequent restriction, extension, or modification in the daughters (226). For instance, while many languages reflect the original contraction to *o· in nominal inflection, this is decontracted in some others (205), and yet others have recontraction in some or all cases, as though from *a·.

• (Repeated from above): *ne- “1” + *‑i·yaw- “body” + *‑ena·n “1pl.EXCL” → *ni·yo·na·ni “our (excl.) bodies; ourselves (excl.)” (e.g., Miami-Illinois niiyoonaani)
• With recontraction, cf. Meskwaki ni·ya·na·ni “ourselves (excl.), our (excl.) lives” (194)
• With decontraction, cf. Ojibwe niiyawinaan
• (Repeated from above): *wi·yaw- “h/ body [possessed theme]” + *‑ehs “DIM” → *wi·yo·hsi “flesh, meat” (e.g., Miami-Illinois wiiyoohsi, Munsee wə̆yóos, Penobscot wə̀yos)
• With recontraction, cf. Ojibwe wiiyaas, Meskwaki owi·ya·si, Plains Cree wīyās, Mahican ⫽wə̆ya·s⫽ ~ ⫽wə̆ya·θ⫽ (various spellings [Goddard 2008a:276, ex. 26b]; note Mahican ⫽a·⫽ derives from PA *e·, not *a·, proving this is a much later recontraction)
• With decontraction, cf. Shawnee wiyawʔθi (from pre-Shawnee *wiyawehsi with normal deletion of short vowel before a cluster), Loup A ⟨ȣiaux⟩ ⫽wəyawəhs⫽ (Pentland 2013:235), probably Virginia Algonquian with (syncopated) †⫽wəyawhs⫽ (spelled ⟨oiawh⟩ [for *⟨oiawſs⟩ vel sim.?], ⟨wiaaws⟩, ⟨Weyans⟩ [for *⟨Weyaus⟩], and ⟨weghshaughes⟩)

*aye

In most phonological environments *aye originally contracted to *e· in inflection and some derivation, which is reflected in its value in Miami-Illinois and Eastern Algonquian. In most cases in the other languages it recontracted to *a·, except in Menominee and Shawnee, which restructured their nouns in *‑ay. That *e· was the original value is shown by relic/fossilized forms in languages which otherwise point to *a· (214-217; Goddard 1983:369-373, 2007:221-222, 244-245).

• *ne- “1” + *ti·pay- “dead person; soul that continues after death, ghost” + *‑em “POSS” → *neči·pe·ma “my corpse; my dead spirit; my deceased relative” (e.g., Miami-Illinois ninciipeema “my deceased relative”)
• Cf. e.g. Nishnaabemwin njiibaam “my ghost, dead spirit; my shadow; my soul” (Valentine 2001:201)^[34] and Meskwaki neči·pa·ma “my deceased relative” with recontraction; and Mille Lacs Ojibwe injiibayim “my ghost, corpse” (cf. Nichols 1980:55) and Cheyenne nȧséotame “my ghost, departed spirit” |na-séot‑ame‑| (|t| regular ← *y, vowels regular) with decontraction (214)
• *apinay- “group sitting/sleeping platform” + *‑enki “LOC” → *apine·nki “on the resting platform” (e.g., Munsee ăpíineenk “on the bed,” Miami-Illinois (a)pineenki “in bed”; cf. Meskwaki apina·ki “on the family sitting place” with recontraction)
• *ahθay- “skin, hide” + *‑ehs “DIM” → *ahθe·hsi “bit of skin/hide, small skin/hide, etc.” (e.g., Penobscot ‑ahsehs “old robe,” Unami xé·s “skin, hide”)
• Cf. locative (with *‑enki) *ahθe·nki “on/in a skin/hide” (→ recontracted post-PA *ahθa·nki, e.g., Meskwaki asa·ki)
• For this same stem, cf. the fossilized form attesting to the original contraction pattern in: *ahθay- + *‑ehke· “obtain by effort, use (AI Final)” → *ahθe·hke·wa “s/he tans hides” (e.g., Meskwaki ase·hke·wa; and cf. cognate Ojibwe aseekee, where the corresponding noun has been lost from the modern language)
• And cf. more recent recontraction to a· in: *apahkway- “cattail; cattail-reed mat, thatching” + *‑ehke· → post-PA *apahkwa·hke·wa → Meskwaki apahkwa·hke·wa “s/he makes cattail-reed mats”; and decontraction in an even more recent and productive doublet, Meskwaki apahkwayehke·wa “s/he gathers cattails”

Further support for reconstructing *aye as contracting to *e·—although it would date to the pre-PA period and not Proto-Algonquian proper—is offered by the ablaut process known as “initial change.” In initial change, long vowels are mostly changed by prefixing *ay- (*a· → *aya·, etc.), which can plausibly be identified as originating with *Ca-reduplication of vowel-initial stems, plus epenthetic *‑y-. The initial-changed version of *e, meanwhile, is e·, and the most reasonable assumption is that this is from pre-PA *ay‑e.^[35]

However, *aye evidently contracted instead to (stable / non-umlauting) *a· when before *w, at least when it derived from earlier *|ayi| by Rule 14. While some of these cases in PA still consisted of an intermediate-level *aye contracting to *a·, others had been morphologized and can only be described as instances of direct *ayi to *a· contraction. All such processes are therefore discussed in the section on *ayi below.

*a·we

There evidence for an old contraction of *a·we to *o·, which in fact was the regular contraction outcome in most environments, though it has been replaced in various ways by most of the daughters except in Menominee and some fossilized relics elsewhere (220-224).

• *nyi·pa·- “at night” + *‑wehθe· “walk (AI Final)” → *nyi·po·hθe·wa “s/he walks in the dark” (Menominee ni·po·hnεw; cf. Plains Cree nīpāhtēw with recontraction) (220)
• *ma·w- “gather” + *‑en “by hand” → TA stem and derived Initial *mo·n- “dig; pull up/out, pluck; etc.” (e.g., Ojibwe moon-, Meskwaki mo·n-, Munsee moon-, Penobscot mon-, Cheyenne mén-) (222-224)
• Recontracted or decontracted in productive forms except in Menominee: *osa·w- “yellow” + *‑eramon “ocher” [← *oramon- “red ocher”] → *oso·ramona “yellow ocher” (Menominee oso·namon “yellow ocher; vermilion(?)”; cf. Ojibwe ozaanaman with recontraction to aa, and Meskwaki asa·wanamona with decontraction) (220-221)
• Similarly: *maya·w- “directly” + *‑eškaw “by force of foot or body (TA Final)” → *mayo·škaw- “meet head on, directly encounter (TA)” (Menominee meyo·skawεw “s/he (PROX) comes upon h//them (OBV)”)
  • But with recontraction, cf., e.g., Ojibwe omayaashkawaan “s/he (PROX) meets h//them (OBV) head on,” Meskwaki maya·škawe·wa “s/he (PROX) encounters/kicks h/ (OBV) squarely” (200)
• Similarly: *sa·kima·w- or *wekima·w- “chief” + *‑eθkwe·w “woman” → *sa·kimo·θkwe·wa or *okimo·θkwe·wa “female chief; chief’s wife” (Menominee oki·mu·hkiw |okemo·hki·w| ← |wεkema·w‑εhki·w|)
  • But with recontraction, cf., e.g., Ojibwe zaagimaakwee “type of female medicine woman” (and ogimaakwee “female chief; chief’s wife”), Miami-Illinois akimaahkwia, Unami sa·k·i·má·xkwe, Penobscot sάkəmαsk^we

Because *wo· and *yo· were mostly illegal sequences in PA, this *a·we → *o· contraction seems not to have applied in pre-PA if *w or *y preceded; in such cases the outcome is *a· instead (221; Goddard 2007:259, n. 65).

• *wi·ki- “dwell (AI)” + *‑wa·w “2pl/3pl” + *‑eʔm “NMLZ” → *wi·kiwa·ʔmi “lodge, wigwam” (originally literally “(where) they dwell”; cf. Dakotan thípi “dwelling, lodge, tipi” with the same literal meaning) (e.g., Menominee we·kewam, Ojibwe wiigiwaam)
• *‑mya·w “smell (concrete Final)” [← TA stem *mera·w- “smell”] + *‑ekosi “PASS (AI Final)” → AI Final *‑mya·kosi “have a smell” as in: *eθ- “RR:MANNER” + *‑mya·w‑ekosi → **eθimya·wekosi- → *ešimya·kosi- “smell {so}, have {such} a smell (AI)” (e.g., Menominee esi·mya·kose-, Ojibwe izhimaagozi-)

There are some other cases where Menominee shows contraction to *a· rather than *o·, however. Perhaps these reflect later sporadic recontractions, or perhaps the occurrence of this pattern in PA was subject to further conditions.

• *ne- “1” + *sa·kima·w- or *wekima·w- “chief” + *‑em “POSS” → ?*nesa·kima·ma or ?*no·kima·ma “my chief” (e.g., Menonominee n[et]o·kemam |nεt-okema·m| ← |nε-t-okema·w‑εm| [206], Ojibwe n[ind]oogimaam, MalPass nsakəmam, Mass ⫽nəsʌ̨·t^yəmʌ̨·m⫽ [⟨nussontimom⟩ “my master,” Gen. 21:5; ⟨nussonchumom⟩ “my Lord,” Goddard and Bragdon 1988:53, 55; etc.]; cf. Unami nsa·k·i·má·yəm |nəsākīmāwəm| with decontraction)

For instances in which *a·we does not contract in verb inflection, see below.

*ewe

The sequence *ewe, which primarily occurred when the person prefixes were added to a stem beginning in *|we‑|, contracts to *o·. This is generally most obvious in Eastern Algonquian languages, because most western languages mostly came to treat non-dependent *o-initial stems like other vowel-initial stems and thus insert an epenthetic *‑t- by Rule 1b, although the length on the /oː/ when preceded by a prefix usually remains. However, vestiges of this contraction can be found outside the Eastern languages, as in the latter two examples below.

• *ne- “1” + *weL- “good, nice; properly arranged” + *‑esi “be, have quality (AI Final)” → *no·Lesi “I am good [in appearance, etc.]” (e.g., Munsee nóolsi “I am pretty”)
• *we- _ ‑i “possess (AI)” + *wespwa·kan- “pipe” → **wewespwa·kaniwa → *o·spwa·kaniwa “s/he has a pipe” (e.g., Plains Cree ōspwākaniw)
• *ne- “1” + *wekima·w- “chief” + *‑em “POSS” → ?*no·kima·ma “my chief” (e.g., older Plains Cree nōkimām, beside restructured nitōkimām with ni‑t-)

Other *VGe

Most basically (some key exceptions discussed below), most other *V₁Ge sequences contract to *V₁ː (Pentland 1999:251). This occurred in nominal inflection, some verbal inflection, and some derivation, and applied to sequences where the initial underlying vowel was short as well as those in which it was long. Some uncomplicated examples follow.

Contraction in nouns and verbs with stems in long *‑V·w (206-207, 212):

• *ne- “1” + *na·pe·w- “man, male” + *‑em “POSS” → *nena·pe·ma “my husband” (e.g., Plains Cree nināpēm)
• *omyi·myi·w- “passenger pigeon (Ectopistes migratorius)” + ‑ehs “DIM” → *omyi·myi·hsa “young or little pigeon” (e.g., Plains Cree omīmīs(is) “young pigeon, young dove” [Itwêwina; Faries 1938:385], Meskwaki omi·mi·h[e·haki] “doves” [Thomason 2015:320], Munsee míimiish “pigeon”)
• *eškote·w- “fire” + *‑ehke· “obtain by effort, use (AI Final)” → *eškote·hke·wa “s/he makes a fire” (e.g., Menominee esko·tε·hkεw, Meskwaki aškote·hke·wa [194])
• *ne- “1” + *maneto·w- “manitou, god, spirit” + ‑ehka·swi “make oneself (as) such; pretend to be (AI Final)” → *nemaneto·hka·swi “*I make myself (like) a manitou” > “I acquire spiritual power (by my own authority), wield spiritual power, conjure” (e.g., Ojibwe nimanidookaaz, Plains Cree nimanitōhkāson, Old Innu ⟨ni manitȣkassȣn⟩ ⫽nimanitūhkāsun⫽ [“je suis un esprit,” Silvy 1974 cited by OldC], Meskwaki ‡nemaneto·hka·so)
• *mete·w- “shaman, medicine man” + *‑eθwe·w “woman” → *mete·θkwe·wa “female shaman, medicine woman” (e.g., Ojibwe mideekwee, Miami-Illinois miteehkwa, Unami məté·xkwe)

Similarly, noun stems in *‑iy (which were always stems in *‑(C)wiy), contract *iy‑e to *i· in both inflection and derivation (188). There were no surface sequences of *iye or *oye between consonants in PA.

• *asa·twiy- “poplar (Populus spp.)” + *‑enki “LOC” → *asa·twi·nki “at/on the poplar(s)” (e.g., Ojibwe azaadiing, Arapaho hohóótiiʔ “in/on the tree(s)”)
• *apanšwiy- “lodgepole” + *‑ehke· “obtain by effort, use (AI Final)” → *apanšwi·hke·wa “s/he makes or gathers lodgepoles” (e.g., Kickapoo apasiihkea)

*iw‑e and *ow‑e likewise contract to *i· and *o· respectively. (For exceptions in verb inflection, see below.)

• *ne- “1” + *‑htiw- “upper arm, lower leg” + *‑enki “LOC” → *nehči·nki “on my upper arm(s), lower leg(s)” (e.g., Meskwaki nehči·ki “on my upper arms”; also found with recent decontraction to nehčiweki [193, 195])
• *we- “3” + *mi·ti- “eat (TI)” + *‑hsi “DIM” + *‑w “IRR” + *‑en(ay) “FMV” + *‑ari “INANpl” → *omi·čihsi·nari “s/he doesn’t eat them (INAN) even a little bit” (e.g., Ojibwe [gaawiin] omiijisiinan “s/he doesn’t eat them (INAN)”)
• *ke- “2” + *men- “drink (AI)” + *‑hs “DIM” + *‑ow [← *‑w by Rule 6(i)] “IRR” + *‑eʔm “FMV” + *‑enaw “1pl.INCL” → *kemenso·ʔmena “we (incl.) aren’t drinking even a little bit” (Miami-Illinois ‡kimensoomina “we (incl.) aren’t drinking,”^[S36] Cheyenne né-[sáa‑]manéhema |né=sáa=mane-ʹhé-ma| “id.”)

There are plenty of instances in derivation where contraction is lacking in most daughters, though, and this treatment is certainly old. For example, the nominalizer *‑wen does not contract with any preceding vowel. In PEA, however, PA unstressed *i or *o before *w merged with *e as *ə, and a resulting PEA *əwə sequence underwent contraction to *ō (taking on the inherited rule by which PA *ewe contracted to *o·). Thus, even though Eastern Algonquian replaced *‑wen as a productive derivational suffix, it remains in fossilized nominalized forms in various languages with a reflex of *‑ōn, showing PEA *əwə contraction.

• *aspapi- “sit upon (AI)” + *‑wen “NMLZ” → *aspapiweni “seat (n.)” (e.g., Ojibwe apabiwin, Plains Cree aspapiwin “cushion; saddle”)
• Cf. *aspapiweni → *aspapəwən → PEA *aĥpapōn (e.g., Munsee áhpăpoon, Powhatan ⟨Apahpun⟩ †⫽ahpapo·n⫽(?) “a stoole to sit upon” [Strachey 1953:202]; cf. Mass ⟨Ahpappŭonk⟩ ⫽ahpapəwʌ̨·k⫽ [Cotton 1829:17] with replacement of *‑wen by productive PEA *‑wākan)
• For another EA example with the same AI Final *‑api “sit,” cf.: pseudo-PA *te·hθa·skwapi- “sit on a wooden surface (AI)” + *‑wen → pseudo-PA *te·hθa·skwapiweni “wooden seat, bench” → pseudo-PEA *tēhθāĥk^wapōn (→ Western Abenaki dasakwabon “seat, chair”; with the same Initial *te·hθ- “on surface,” cf. e.g. Ojibwe deesabiwin “chair, seat; saddle,” Plains Cree tēhtapiwin “mount, horse”)

By contrast, although Cheyenne, Menominee, Massachusett, and the Delaware languages show a lack of contraction—at least in some forms—of noun stems in *‑V·w and *‑iw, these seem to be later analogical developments (206-207; cf. Goddard 2000:90).

• *watiw- “mountain” + *‑enki “LOC” → *wači·nki “at/on the mountain” (e.g., Ojibwe wajiing) → later decontracted post-PA *wačiwenki (e.g., Cheyenne vȯsēva |vosev‑á|, synchronically |vose‑vá|)
• *eškote·w- “fire” + *‑enki → *eškote·nki “at/in the fire” (e.g., Ojibwe ishkodeeng) → later decontracted post-PA *eškote·wenki (e.g., Menominee esku·tya·h |εskotε·w‑Eh|,^[37] cf. contracted Menominee esko·tε·hkεw “s/he makes a fire” above) (206-207)
• *ne- “1” + *eθkwe·w- “woman” + *‑em “POSS” → *neteθkwe·ma “my woman, my sister (of male)” (e.g., Ojibwe nindikweem “my wife, my woman”) → later decontracted post-PA *neteθkwe·wema (e.g., Unami ntuxkwé·yum |nətəxkwēwəm| “my sister (of male)”)

*Vwe in Verb Inflection

*‑Vw TA Stems

In TA verb inflection, contraction of some sort probably always occurs before the morphemes *‑ekw “INV” and *‑eθ “2OBJ” (unless the stem is one of the few inherently non-contracting ones like *aw- “use (TA)”) (209-212, 212-213; cf. 166-167, 196-198). (For other verb classes, see below.)

As Goddard (212) notes, however, “TA stems in *‑Vw other than those . . . [in] *‑aw are rare and tend to be eliminated in” the daughters, and the same applies to any contraction found with them—greatly complicating reconstruction and making the overall picture murkier. He reconstructs these other stems as probably contracting (normally to a long version of the first vowel) just as *‑aw stems did, which seems likely. Probably this included contraction when such stems preceded the reciprocal *‑(e)twi. (*‑e·w stems contract in the same environments, but at least one has a unique outcome; see below. Presumably *‑a·w stems also contracted to *o· originally, but I’m not aware of any remnants of this.)

• *aθko·w- “come next after, follow behind (TA)” + *‑ekw “INV” + *‑ẅ “3” + *‑a “PROXsg” → *aθko·kwa “s/he (OBV) comes next after, follows behind h/ (PROX)” (e.g., Menominee ahko·k)
• *ke- “2” + *pye·šiw- “bring (TA)” + *‑eθ “2OBJ” → *kepye·ši·θe “I brought you (sg.)” (e.g., Plains Cree kipēsīt[in])
• *aθko·w- + *‑etwi “RECIP” → *aθko·towaki “they come one after another, follow one another” (e.g., Menominee ahko·towak, Meskwaki ahko·ti·waki)

All TA *‑Vw stems evidently fail to contract before the endings *‑ent “UNSPEC»3AN.CONJ” (and other endings beginning in *‑en “UNSPEC.CONJ”), *‑em “MARK.OBV.OBJ” and *‑ehkwe “2pl»3AN.IMPER.”

• *ešinaw- “see, perceive s.o. {so}, in {such} a way (TA)” + *‑a· [→ *‑Ø] “3OBJ” + *‑ent “UNSPEC»3AN.CONJ” → *ešinawente· “if s/he is seen, perceived {so}” (e.g., Ojibwe izhinawind)
• *ne- “1” + *no·hsaw- “chase (TA)” + *‑em “MARK.OBV.OBJ” + *‑a· “3OBJ” + *‑w “FMV” + *‑ahi “OBVpl” → *neno·hsawema·wahi “I chase them (OBV)” (e.g., Cheyenne ná-néhovamóho |ná=néhóv‑am-Ø-óhó|)
• *wi·ntamaw- “tell (TA)” + *‑ehkwe “2pl»3AN.IMPER” → *wi·ntamawehko “(you pl.) tell h//them!” (e.g., Shawnee wiitamaweʔko, Eastern Swampy Cree ‡wīhtamawihkok “(you pl.) tell them!”^[S38])

As in noun inflection, many of these patterns are disturbed in individual daughter languages (210-213). For example, some languages, like Menominee and Ojibwe, eliminate some or all contraction with verbs except *‑aw-stems. Some Cree varieties greatly expand the domain of contraction, to apply to essentially every *Vwe sequence (though this is recontraction—i.e., the output vowel is always equivalent to a long version of the first vowel of the underlying sequence). And Menominee and Eastern Algonquian have extended *awe > *o· contraction to additional contexts. For example:

• Plains Cree wīhtamāhkok “(you pl.) tell them!” (Whitecalf 2021:290-291) with recontraction (← *wi·ntamawehko); though see footnote 38 above
• Cf. uncontracted Eastern Swampy Cree ‡wīhtamawihkok above
• Cf. also Innu natutuek^u = /natuːtweːkʷ/ (Ekuanitshit), /ntətwekʷ/ (Pessamit) “(you pl.) listen to h/!” (← *natohtawehko ← *natohtaw- “try to hear, listen for (TA)” + *‑ehkwe), with no contraction but analogical pre-Innu *awē for *awi
• Menominee no·htoh |no·hto·h| ← |no·htaw‑æh| “when s/he is heard” with new contraction (← *no·ntawent- ← *no·ntaw- “hear (TA)” + *‑ent “UNSPEC»3AN.CONJ”)
• Cf. uncontracted Ojibwe noondawind

The phenomenon mentioned previously in which PEA had a sound change of *i to *ə before *w, followed by contraction of resulting *əwə to *ō, applied to TA verbs in *‑iw as well (212-213), sometimes replacing an earlier contraction pattern.

• *ne- “1” + *pye·šiw- “bring (TA)” + *‑ekw “INV” → **nepye·šiwekwa → PA *nepye·ši·kwa “s/he brought me”
• → pre-PEA *nəpēšəwək^w (with decontraction) → PEA *nəpēšōk^w (e.g., Munsee mpéeshookw)
• Cf. Plains Cree nipēsīk with inherited contraction to *i· (cf. also e.g., Meskwaki newi·čawi·k “s/he’s married to me” with wi·čawiw- “be married to (TA)” [← pseudo-PA *wi·č’ayiw- “have as one’s companion (TA)”] [197])

*‑e·w Stems

While TAs in *‑e·w probably contract in the same environments as other *‑Vw stems, there is evidence such *e·w‑e contraction had the unique outcome of *o·, rather than *e· (213-214; Pentland 1999:251). Such an outcome for the verb *ne·w- “see (TA)” is directly reflected in Shawnee, with an additional indirect reflex in CINA.

• *ke- “2” + *ne·w- “see (TA)” + *‑eθ “2OBJ” → *keno·θe “I see you (sg.)” (Shawnee kinoole)
• TA stem *ne·w- “see” + *‑ekwat “PASS (II Final)” → II stem no·kwat- “be visible, seen; appear” (Shawnee nookwat-, Plains Cree nōkwan- [with replacement of the Final *‑at by equivalent *‑an])

Possibly the other *‑e·w stem TA verbs, principally *wi·čye·w- “accompany,” had the same pattern, but if so they have lost it or the verbs have otherwise been restructured in all the daughters, Shawnee included. Outside of “see” in Shawnee (and fossilized in CINA), any retained *‑e·w verbs are treated like other *‑Vw stems, either contracting *|e·we| to *e·, or showing no contraction, thus regularizing the surface stem shape (212-214).

• *ne- “1” + *wi·tye·w- “accompany (TA)” + *‑ekw “INV” → ?*newi·čye·wekwa “s/he accompanies, goes along with me”
• With non-contraction (the ancestral pattern?), e.g.: Ojibwe niwiijiiwig, Pessamit Innu nuitsheuk^u /nwitseukʷ/ |nəwitsewək^w| (with /eu/ ← pre-Pessamit *ēwi), Unami nəwi·č·é·yukw |nəwīčēwəkw| “s/he goes with me; s/he marries, is married to me”
• With later(?) contraction or recontraction, e.g.: Plains Cree niwīcēk

As represented in the reconstruction above, I have doubts whether the PA *e·we to *o· contraction pattern was found with all the *‑e·w stem verbs besides *ne·w‑, and how widely it existed in general. Clearly it was an unusual and probably archaic pattern at best.^[39] And of course not all the instances of the sequence *e·we would have had the same origin; thus, there’s some reason to think that “see” derived from pre-PA *na·w- (among other things, note that its deverbal, with vowel-shortening, is *‑naw)—it would thus probably represent a pre-PA example of *a·we contraction—but the *e· in *wi·čye·w- could have arisen from a different source. It certainly originated at a later date, since the composition of the word is quite transparent and involves Algonquian-specific elements, while *ne·w- can be traced back to Proto-Algic. *ne·w- has always ended in *w, but the final *w in *wi·čye·w- is probably segmentable.

Within Inflectional Endings

Within verb inflectional endings themselves (198-199) it seems that the sequence *a·we that arose when the 3OBJ theme sign *‑a· was followed by the W-formative *‑w and one of the central suffixes *‑ena·n or *‑enaw (*‑a·‑w‑ena·n / *‑a·‑w‑enaw), or was followed by the irrealis suffix *‑w and the UNSPEC»3AN.CONJ ending *‑ent (*‑a·‑w‑ent), did not contract, although Menominee does show contraction to *o· (which it’s possible was the real PA situation), several other languages show contraction of the former to *a·, and Meskwaki in the latter case shows contraction to a· (Goddard 2007:236-237, 266, n. 73; cf. 2000:102 and n. 31).

• *ke- “2” + *ne·w- “see (TA)” + *‑a· “3OBJ” + *‑w “FMV” + *‑enaw “1pl.INCL” → *kene·wa·wenawa “we (incl.) see h/” (e.g., Munsee kəneewáawəna, Mass ⟨nunnauooun⟩ ⫽nəna·wʌ̨·wən⫽ “we (excl.) saw him” [Goddard and Bragdon 1988:519; with prefix ⫽nə-⫽ “1” in place of ⫽kə-⫽ “2”]; cf., with contraction, Menominee kenε·wo·naw and Eastern Mahican ⟨N’wawehána⟩ †⫽nwa·wihʌ̨·nah⫽ “we (excl.) know h/” [Goddard 2008a:296; verb †⫽wa·wih-⫽ “know (TA),” prefix ⫽n-⫽ “1”])
• *neʔr- “kill (TA)” + *‑a· “3OBJ” + *‑w “IRR” + *‑ent “UNSPEC»3AN.CONJ” + *‑e·n “INTERR” → *neʔra·wente·ni “whether s/he was killed” (e.g., Ojibwe neesaawindeen; cf. Meskwaki nesa·te·ni “s/he must have been killed” with contraction [199])

It’s likely that the sequence *iwe which arose in diminutive verbs (with *‑hsi “DIM”) in these same contexts (*‑hsi-w-ena·n, *‑hsi-w-enaw, *‑hsi-w-ent) also failed to contract, though the available data from the daughter languages are so meager that it’s difficult to speak with any certainty. Ojibwe and Miami-Illinois (Costa 2003:347-348, 519) at least furnish evidence for non-contraction of *‑hsi-w-ent, and Unami obliquely suggests non-contraction with the W-formative (Goddard 2021:157, ex. 5.73(w); with replacement of the original diminutive suffix by a new one), assuming its non-contraction of *‑a·-w-en… continues the PA pattern.

• (Repeated from earlier): *ne- “1” + ?*paspakam- “hit repeatedly (TA)” + *‑ekw “INV” + *‑hsi “DIM” + *‑w “W.FMV” + *‑ena·n “1pl.EXCL” → ?*nepaspakamekohsiwena·na “the little one (anim.) hit us (excl.) repeatedly” (e.g., Unami mpəpahkamkwət·í·wəna)
• *tepe·rem- “judge; manage, control, rule (TA)” + *‑a· “3OBJ” + *‑hsi + *‑w “IRR” + *‑ent “UNSPEC»3AN.CONJ” + *‑a “PROXsg.PART” + IC → *te·pe·rema·hsiwenta “s/he who is not controlled, ruled even a little bit” (Ojibwe deebeenimaasiwind “s/he who is not controlled, ruled,” Old Illinois ⟨[teperi]masionta⟩ †tipeerimaahsiiyonta “s/he is not controlled” [Costa 2003:519; from earlier *‑hsiwenta])

I’m a bit less confident on the outcome of irrealis *‑w + *‑enk “UNSPEC.CONJ” when following a vowel-final AI stem, or a TA with the 1OBJ theme sign *‑i.

In such verbs, Meskwaki shows contraction, just as it does with the sequence *‑a·‑w‑ent discussed above; and as with that sequence, its contraction here could be an innovation. CINA offers no evidence, while Delaware again shows no contraction. Ojibwe has the ending ‑Vwiing (e.g., naamiwiingeen “whether there’s dancing, whether someone is dancing,” with verb niimi- “dance (AI)” plus initial change [expected: ^xnaamiwingeen or ^xnaamiingeen, in either case = |aa\niimi‑w‑ing-een| ← PA *|a·\ni·myi‑w‑enk-e·n‑|]). Presumably this is best taken as an originally contracted sequence *ii (with the vowel quality ii generalized from 1OBJ verbs and verbs with stem-final ‑i like niimi-) which was partially repaired by re-adding ‑w-. In other words, the languages mirror their patterning for TA unspecified subject verbs, except Ojibwe which seems to point to earlier contraction. Taken together, this suggests that there most likely was contraction in PA.

• Unami AI: [né·skɔ] xínkwi-mi·tsahtí·yunk “before the big feast (lit. “when there had not yet been a big feast held”) |=mītəsahətī‑w‑ənk| (with verb |mītəsahətī‑| “hold a feast (together) (AI)”), with no contraction [Note: the Unami |ī| here, and in the example immediately below, are from PA short *i]
• And TA with first-person object: [é·li- … máta … ‑]ne·í·yunk “because I will not be seen” |=nēw-ī‑w‑ənk| (with verb |nēw‑| “see (TA)” and theme sign |‑ī| “1OBJ”), with no contraction
• Both identical to TA 3OBJ treatment with no contraction: [é·li- … né·skɔ ‑]mí·lá·wənt “for he had not yet been given it” |=mīl-ā‑w‑ənt| (with verb |mīl‑| “give to (TA+O)” and theme sign |‑ā| “3OBJ”)
• Cf. Ojibwe waakondiwiingeen “whether there’s a feast held” (with wiikondi- ~ wiikwandi- “hold a feast together (AI),” with ‑(i)di “RECIP” as in Unami: |aa\wiikom-di‑w‑iing‑…|); vs. TA 3OBJ form with no hint of contraction maanaawindeen “whether s/he is given (it/them/etc.)” (with |miiN‑| “give to (TA+O)” as in Unami, plus initial change: |aa\miiN-aa‑w‑ind‑…|). (Note Ojibwe eliminated 1OBJ AI irrealis forms with |‑i| comparable to those of Unami etc.)
• Meskwaki AI: peno·ke·ni “one must have gone home” |peno‑·k-e·n-i| ← *|peno‑w‑ek‑…| (with verb peno- “go home (AI)” and INTERR suffix -e·n as in Ojibwe), with contraction
• And TA with first-person object: wi·h-neši·ke·ni “whenever I may be killed” |=neS-i‑·k-e·| ← *|neS-i‑w‑ek‑…| (with verb |neS‑| “kill (TA)” and theme sign -i “1OBJ”), with contraction (199)
• Both identical to TA 3OBJ treatment with contraction: nesa·te·ni “s/he must have been killed” |neS-a·‑·t-e·n-i| ← *|neS-a·‑w‑et‑…| (with theme sign ‑a· “3OBJ”) (199)
• Again, cf. Ojibwe: gaaweewiingeen “whether one went home” (with giiwee- “return home (AI)”: |aa\giiwee‑w‑iing‑…|); vs. TA 3OBJ form with no hint of contraction neesaawindeen (with |niS‑| “kill (TA)” as in Meskwaki, plus initial change: |ee\niS-aa‑w‑ind‑…|).

Realis Non-TA Verbs

Contraction was of limited distribution outside of TA verbs. Beyond such irrealis AIs as those just covered, it occurred frequently in other irrealis forms (across the irrealis suffix *‑w); examples can be found throughout the post. Otherwise, it had very few potential triggering phonological environments. One of these environments is discussed in the next section.

Contraction with Mode Suffixes

An interesting differentiation of contraction outcomes is found in some Eastern languages when a mode suffix follows a vowel plus *‑ẅ “3” or the formative *‑w, resulting in the sequence *|V‑w‑p| → *|V‑w‑ep| with EMPH.PAST *‑pan and *|V‑w‑(e)s| with EMPH.PRES *‑(e)sahan. In a few languages, the emphatic past ending gives a reflex of contracted *V·pan while the emphatic present gives a reflex of uncontracted *Vwesahan, which Goddard (2007:249) argues “probably continue[s] the Proto-Algonquian situation.” This would, indeed, fit with the other evidence that *‑pan originally had some different morphophonemic properties than *‑(e)sahan (see Rules 5a and 9e). Presumably the actual underlying forms would then be *|Vwp| → *V·p vs. *|Vw(e)s| → uncontracted *Vwes, with the EMPH.PAST never passing through an *|Vwep| stage but simply reflecting the outcome of *|Vw| directly preceding a *|p|, with either loss of the *|w| and compensatory lengthening (for which cf. Rules 17 and 27b(iv) = 7), or coalescence.

• *pya·- “come (AI)” + *‑ẅ “3” + *‑pan “EMPH.PAST” → **pye·wpan → *pye·pa “s/he had come” (e.g., Northern Unami ⟨peep⟩ pé·p “s/he came (in the past)” [“er kam,” Grube n.d., sig. 6:[8]; cf. indicative mode ⟨peuch⟩ pé·w “s/he comes” [“er kom̄t,” ibid.] ← PA *pye·wa ← *pya· + *‑ẅ + *‑a “PROXsg”)
• Versus: *pya·- + *‑ẅ + *‑(e)sahan “EMPH.PRES” → **pye·w(e)sahan → ?*pye·wesaha “s/he has come” (e.g., Northern Unami ⟨peuchsa⟩ pé·wsa |pē‑w‑əsa| ← |pā‑ẅ‑əsa| [“er ist gekom̄en,” ibid.])

Most languages, however, have contraction before both suffixes. Some examples:

• Meskwaki conclusive mode (< PA EMPH.PRES): [katawi‑]pye·hapa “then she must be about to return” (Goddard 1995:131) (as if from *pya·- + *‑ẅ + *‑(e)sahan [reshaped to *‑(e)sapan] → ?*pye·saha “s/he has come”)
• Cf. assertive mode (< EMPH.PAST): wi·teko·wi·pani “s/he indeed has become, is an owl!” (Goddard 1995:134) (from *wi·nteko·wi- “be, become an owl (AI)” + *‑ẅ + *‑pan → *wi·nteko·wi·pa “s/he had become an owl”)
• Cheyenne mediate mode (< EMPH.PRES): éh-mésėhéhoo’o “s/he sure ate” |=mése-he‑ʹhoon| (cf. Goddard 2000:125), with the mode suffix |‑ʹhoon| inducing high tone (← earlier vowel length) on a preceding vowel (as if from *mi·tihsi- “eat a bit; a little one eats (AI)” + *‑ẅ + *‑(e)sahan [reshaped to *‑(e)sapan] → ?*mi·čihsi·saha “the dear little one has eaten”)
• MalPass htəmess “s/he [might] have smoked” (Sherwood 1983:240) (as if from *wetema·- “smoke a pipe (AI)” + *‑ẅ + *‑(e)sahan → ?*oteme·saha “s/he has smoked”)
• Cf. htəmehpən “s/he (had) smoked” (Sherwood 1983:230) (from *wetema·- + *‑ẅ + *‑pan → *oteme·pa)

In some other languages there is variation, but the one constant seems to be that the EMPH.PAST more often shows reflexes with contraction than other mode suffixes do. Given the totality of the evidence, including the other ways in which the mode suffixes differ morphophonemically from one another, I think Goddard is probably right that in PA they differed with respect to contraction as well. The emphatic present did not contract, but the emphatic past did. The dubitative suffix presumably did not contract either, given its initial vowel and, e.g., its non-contraction in most of Ojibwe (although I may be succumbing to Teeter’s Law here . . . ). But certainly none of this is close to a slam dunk.

Other Patterns

While the vast majority of instances of contraction occurred with underlying *VGe sequences, in certain cases the second vowel was *i instead. Some additional patterns may be discerned in pre-PA, even if they were no longer operable in Proto-Algonquian itself. Furthermore, over the years several people have claimed the existence of other, marginal contraction patterns (e.g., *ewi → *i· [Pentland 1979:39, 109, 408]), though I will not cover all of these here. New contractions which arise in individual daughters (e.g., |i·yi·| → /iː/) are also not covered, with one exception.

*ayi

Several suffixes beginning in *‑i are treated as if they began in ‑e when added to a stem in *‑ay, with resulting contraction of the intermediate *|aye|. There are two different outcomes in such cases, reflecting at minimum two separate ultimate diachronic sources.

When *ayi is followed by a *w which triggers the *i to shift to *e by Rule 14, the outcome is contraction of *aye (← *ayi) to stable/non-umlauting *a·. This contrasts with the contraction of *aye in most other contexts, where as exemplified above, the outcome is *e·.

The most clear-cut example of this *ayi → *aye → *a· contraction involves independent-order II verbs containing the Final whose original/underlying form can be identified as *|‑yayi| (see Goddard 2013:107-108) (briefly noted in Rule 14(iv)). Here the sequence *|‑yayi‑ẅ|, with *‑ẅ “3,” becomes **‑yayew → *‑ya·w. This was inherited as *‑ya·‑w in most of the non-Eastern languages—including Miami-Illinois—which then extended the allomorph *‑ya· to all environments. It was replaced by *‑ye· in all environments in Menominee and, in PEA, by reflexes of *‑ye·yi before *‑ẅ and *‑ye· (mostly?) elsewhere. The allomorph *‑ye·yi would be the expected PA realization of the Final in most environments by Rule 15 (*ya → *ye· between consonants), but the source of the *‑ye· reflected in PEA and Menominee is unclear, as is the appearance of the reflex of the *‑ye·yi allomorph before *‑ẅ in PEA, which is precisely where it would not have occurred in PA. A complete reconstruction of the distribution of the allomorphs in Proto-Algonquian at this stage is perhaps not possible.^[40]

• Independent INAN.PROXsg: *wa·p- “white” + *‑yayi “be, have quality (II Final)” + *‑ẅ “3” + *‑i “INANsg” = **wa·pyayiwi → **wa·pyayewi → *wa·pya·wi “it’s white”
  versus
  Conjunct INAN.PROXsg: *wa·p- + *‑yayi + *‑k “INAN.CONJ” + *‑i “INANsg.PART” + IC → **w<ay>a·pyayiki → ?*waya·pye·yiki [*CyaC → *Cye·C by Rule 15] “that which is white, a white one (inan.)”
• → most non-Eastern *wa·pya·wi / *waya·pya·ki, e.g.:
• Arapaho: ‡[hoowu]noočoo “it’s not white”^[S41] / nónoočóóʔ “it’s white”
• Cheyenne (verb |heóvó| “be yellow (II)” ← *osa·wayi- *|wesa·w‑yayi‑|): [é‑]heóvo / [tsé‑]heóvo [both |=heóvó-Ø|, reflecting intermediate *‑osa·wa·‑] (Goddard 2000:98-99)
• Eastern Swampy Cree: wāpāw / [kā‑]wāpāk (Ellis 1983:415, 1995:128)
• Meskwaki: wa·p[ešk]ya·wi / wa·p[ešk]a·ki (Thomason 2003:92; I don’t know why the y is missing in the participle [cf. kesiya·wi “it’s cold” vs. we·či-kesiya·ki “(in the) north,” lit. “where the cold comes from,” Goddard 1991:173])
• Shawnee (verb tawaa- “be open (II)” ← pseudo-PA *tawayi- *|taw-yayi‑|): tawaawi / teewaaki (Luke 14:21; Mark 2:2)
• Miami-Illinois (verb tahkyaa- “be cold (II)” ← *tahkyayi‑): tahkyaawi / teehkyaaki “it’s cold”
• → pre-Menominee (verb *osa·wye·- “be yellow (II)” recomposed from PA *osa·wayi‑ = *|osa·w‑| + *|‑yayi|): *osa·wye·wi / *we·sa·wye·ki
• Menominee osa·wi·w / wε·sa·wi·k
• → PEA *wāpēyəw *|wāpēyī-ẅ| / *wāpēk (this allomorphy is further leveled or adjusted in many daughters), e.g.:
• Mi’kmaq (with long /eː/ ← *ēyə): N/A(?) / wape·k “it’s white”
• Maliseet: wapeyo / wapeyik (Goddard 1983:379)
• Mass (with loss of final *əw): ⟨wompi⟩ ⫽wʌ̨·pa·y⫽ / ⟨wampag⟩ ⫽wʌ̨·pa·k⫽ (Exod. 16:31; Job 6:6)
• Munsee: wáapeew / wáapeek
• Nanticoke ⟨Waap-pay-u⟩, ⟨waap-paya⟩, ⟨Wapeju⟩, ⟨woà-pe-ju⟩ ⫽wa·pe·yəw⫽ / [no attestation] (Heckewelder 1785:3; Jefferson 1808:[14]; Cunningham 2022:60)
• Virginia Algonquian: ⟨Opaivwh⟩ ⫽wa·pe·yəw⫽ (Strachey 1953:206, “white”) / [from *osa·wayi- *|osa·w-yayi‑|]: ⟨Oussawaik⟩(?)^[42] †⫽wəsa·we·k⫽ (Strachey 1953:207, “yellow”)

Pentland (2023:82, s.v. *‑ya·‑) rejects this reconstruction of the *|‑yayi| Final, sticking instead with the traditional *‑ya·. I don’t know how he would attempt to explain the forms found in Eastern Algonquian; while under any hypothesis there are some developments that are very challenging to explain, a starting point of *‑ya· can explain little in PEA beyond the origin of the *‑ye· allomorph. Pentland’s rejection also fails to account for Proulx’s (1980c:291, n. 12) observation, eventually concurred with by Goddard (2013:108, n. 33), that reconstructing an umlauting PA *‑ya· requires postulating that multiple daughters in which umlaut is otherwise highly productive dropped it from this highly productive suffix.

While the *|‑yayi| Final can still be analyzed as containing an underlying *|ayi| that becomes intermediate-level *aye before *w, which then contracts to stable/non-umlauting *a·, other cases in PA existed of *|ayi| seemingly contracting to stable *a· but these had already been morphologized/lexicalized and were no longer triggered purely by a following *w. Despite originating in contraction of intermediate *aye, synchronically they must be described as direct contractions of *ayi to *a·.

For example, verbs of possession with the AI Final *‑i, when built on noun stems in *‑ay, contract *ayi to stable *a· (cf. 215, 218-219), an *a· which occurs in all inflections—synchronically it is simply the stem-final vowel of the verbs in question. This *a· must have been analogically extended from the irrealis and third-person absolute independent forms (before *‑ẅ “3” or *‑w “IRR”) in which it had first arisen by sound law in pre-PA. Like verbs of possession, verbs with the *‑i allomorph of the AI Final *‑iwi “be” (itself evidently created via Rule 14(i)), when built on noun stems in *‑ay, have contraction to *a·, and can be reconstructed as showing the same analogical extension of such contraction throughout all paradigms.

• *we‑htawakay- “h/ ear” + *‑i “possess (AI Final)” + *‑ẅ “3” + *‑a “PROXsg” → **wehtawakayewa → *ohtawaka·wa “s/he has ears” (e.g., Ojibwe otawagaa, Moose and older(? and/or High?) Plains Cree ohtawakāw)
• Hence e.g.: *we‑htawakay‑i- + *‑t “3AN.CONJ” + *‑a “PROXsg.PART” + IC = **we·htawakayita [→ expected ^xwe·htawakayita, but by analogy:] → *we·htawaka·ta “s/he who has ears” (e.g., Ojibwe weetawagaad, Moose Cree ‡wēhtawakāt; and cf. Meskwaki [e·h‑]ohtawaka·či “where h/ ear is; that s/he wore (them as) earrings” [215] with word-final *‑i instead of *‑a)
• *a·pi= “restored, undo (prwd.)” + *ti·pay- “dead person” + *‑i “be (AI Final)” + *‑ẅ + *‑a → **a·pi=ti·payiwa → **a·pi=ti·payewa → *a·pi=či·pa·wa “s/he is revived, comes back to life” (e.g., Ojibwe aabijiibaa; cf. Meskwaki a·pi-či·pe·wa, with shift from non-umlauting to umlauting stem, and cf. also the related Meskwaki stem či·payiwa “s/he’s a ghost” with decontraction [215])
• Hence e.g.: *a·pi=či·pay‑i- + *‑t + *‑a + IC = **aya·pi=či·payita [→ expected ^xaya·pi=či·payita, but by analogy:] → *aya·pi=či·pa·ta “s/he who came back from the dead, has been revived” (e.g., Ojibwe ayaabijiibaad, Meskwaki a·pi-či·pa·ta [215])

Pentland evidently denied that contraction of *ayi (→ *aye) had the outcomes just given. See the following footnote for more.^[43]

In several other cases, *‑ay-stem nouns appear to have unexpectedly contracted with *i-initial inflectional or derivational suffixes which did not provide environments for contraction by Rule 14; instead, the *i was simply treated as if it were *e for no reason we can now discern. (Pentland identifies the environment as before any of his proposed pre-PA glottalized sonorants.) These cases also differed in that the original outcome was *e· (the same as *aye contraction in most environments), rather than the *a· found before *w (217-218). The most prominent suffix exhibiting this trait is *‑inay, which among other things served as a nominal inflectional suffix marking an unspecified possessor, and a suffix in the secondary derivation of collective nouns. In its collective sense, it was frequently followed by the locative *‑enki to give *‑ine·nki “among the [noun]s, in the land/country of the [noun]s.”^[44]

• *we-htawakay- “h/ ear” + *‑inay [as if *|‑enay|] “UNSPEC.POSSR” + *‑ari “INANpl” → **wehtawakayenayari → *ohtawake·nayari “one’s ears” (e.g., Meskwaki ohtawaka·n[in]awani with recontraction [Goddard 2023:110, ex. 3.77])
• *ka·ka·kiw- “raven (Corvus corax)” + *‑irenyiw “person, man” + *‑inay “COLL” + *‑enki “LOC” → **ka·ka·kiwirenyiwenayenki → pseudo-PA *ka·ka·kiwirenyi·ne·nki “in the land of the raven people” (Cheyenne Óoetanéno |óóe-tane‑ʹnó| “Crow person place = Crow Agency, Montana”)
• *arehθirenyiw- “ordinary person” + *‑inay + *‑enki → pseudo-PA *arehširenyi·ne·nki “among humans; in this world” (e.g., Plains Cree ayīsiyinīnāhk [Wolfart 1973:31])
• *maneto·w- “manitou” + *‑inay + *‑enki → *maneto·ne·nki “in the land of the manitous, among the manitous” (e.g., Meskwaki maneto·na·ki)

All contraction of *‑inay has, to my knowledge, been eliminated from Eastern Algonquian and is only found in western Algonquian, in languages which have undergone later recontraction of *e· (from *aye) to *a·. However, as with *aye there exist relic forms which show that the original outcome of contraction with stems in *‑ay was *e·, as illustrated by the forms based on *či·pay- “dead person” below.

• Decontraction in Eastern Algonquian:
• *mataneto·w- “dangerous or evil manitou” + *‑inay + *‑enki → pseudo-PA *mataneto·ne·nki “among, in the land of the bad manitous” → pseudo-PEA decontracted *matanətōwīnēnk (e.g., Munsee mătan’toowíineenk “in Hell,” i.e. “in the land of the devils”)
• Mass |wə-na·mʌ̨·nay‑…| “h/ son” + |‑ən(yay‑)| [← *‑inay] → ⫽wəna·mʌ̨·nayən⫽ “a son,” ⫽wəna·mʌ̨·nayənyʌ̨·h⫽ “the son (obv.)” (Gen. 22:10 ⟨wunnaumon‑⟩; Goddard and Bragdon 1988:491 ⟨wunnaumoniin⟩/⟨wunnaumonain⟩, 492 ⟨wunnaumóniineuh⟩; Cotton 1829:18 ⟨wunnaumonien⟩)
• *ti·pay- “dead person” + *‑inay → *či·pe·nay- “the dead (collective),” e.g. (217, modified):
  • *či·pe·ne·nki “in the Land of the Dead” with *‑enki “LOC” → Cheyenne séáno |séá-nó| (/á/ regular ← *e·; cf., with recontraction, Meskwaki či·pa·na·ki)
  • As if from ?*či·pe·nayesiwa ≈“*s/he is in amongst the dead” with *‑esi “be, have quality (AI Final)” → Meskwaki či·pe·na·wesiwa “s/he is bothered by a ghost”

Goddard (218) points to an even broader “archaic pattern” in many of the daughters, in which *‑ay-stem nouns “mak[e] derived [I]nitials (or derivational bases) in PA *‑e·”; as he notes, “this must continue old contraction but is nowhere analyzable as such synchronically.” However, nearly all the examples he provides involve deverbal Medials or Finals beginning in *i, e.g.:

• *maškimotay- “(woven) bag, sack” + *‑iwaθ “bundle, pack” [dvbl. ← noun stem *‑i·waθ-] → pseudo-PA *maškimote·waši (Innu massimuteush “bag, sack, pocket; scrotum,” e.g. Pessamit /məssəmŭtewʃ/, Ekuanitshit /mahtʃɨmuteːwah/)
• *maškimotay- + *‑i‑kwa·te· “be sewn (II Final)” [with epenthetic *‑i-] → pseudo-PA *maškimote·kwa·te·wi “it’s sewn up as a bag” (Ojibwe mashkimodeegwaadee)
• *ne- “1” + *‑spikay- “rib, ribcage” [dependent noun stem] + *‑i‑kan “bone” [with epenthetic *‑i-, dvbl. ← noun stem *‑θkan-] → *nespike·kani “my rib” (e.g., Plains Cree nispikēkan, Menominee nεhpe·kε·kan, Mass ⟨muhpeteog⟩ ⫽məhpət^ya·k⫽ “the rib” [with ⫽mə-⫽ “UNSPEC” in place of ⫽nə‑⫽] [Gen. 2:22])
• *ti·pay- “dead person” + *‑iθkanaw “path” → pseudo-PA *či·pe·θkanawi “the Path of the Dead” (Ojibwe jiibeekana “Path of the Dead; the Milky Way” [= the road to the Land of the Dead])

With all the above, cf. the following example, which makes it clear that at least in some daughters this pattern has been extended to environments from which it was originally absent, i.e., a new allomorph of the Initial has been created through reanalysis and then generalized further.

• *ti·pay- + *‑askehkw “kettle” [dvbl. ← noun stem *askehkw-] + *‑e· “(AI Final)” → pseudo-PA ^xči·pe·skehkwe·wa (Meskwaki či·pe·hkohkwe·wa “s/he celebrates a memorial feast”; clearly now |či·pe·‑ahkohkw-e·‑|, with |či·pe·‑| an allomorph of |či·pay‑|)

But what Meskwaki shows here is simply an especially obvious example of what happened in PA (or pre-PA) itself. These variant allomorphs in *‑e· of derived Initials in *|‑ay| began in one phonological environment, and were extended to others. As observed above, the examples of this process that are well attested or actually reconstructible to PA almost always involve a following element which begins in *i (either underlying, or epenthetic by Rule 9a); furthermore, the following element tends to be deverbal from a noun or verb stem, or has the distinct appearance of originating as one even if its stem counterpart can no longer be identified. Finally, there are certain following elements (including *‑iwaθ “bundle, pack” and *‑(i)kan “bone”) that are much more commonly found in the more archaic of these constructions.

Goddard writes that: “This pattern . . . is not easily explained phonologically,” since “[e]ither contraction . . . take[s] place with vowels other than PA *e, or the apparent contraction . . . reflect[s] the treatment of *ay directly before a consonant, with no connective vowel” (218). He later (225-226) repeats that the possibility is these “contractions” to stem-final *e· in derivation could represent the “original preconsonantal treatment of *ay.”

It is, indeed, now difficult or perhaps impossible to determine the original environment from which this pattern was generalized. For example, “pack” *‑iwaθ begins in an etymological *i, derived from the stem *‑i·waθ- via archaic shortening of long vowels outside initial syllables (see Coda ex. 4; note this is my opinion of how deverbals like “pack” were formed, I don’t know if everyone would agree), while “bone” *‑(i)kan begins etymologically in a consonant (with inserted epenthetic *‑i-), derived from the stem *‑θkan- via loss of the initial member of the cluster. I would say that *ay-stem noun contraction to *e· is simply too morphologized to tell whether it first began with deverbals in initial *i like “pack” or those in an initial consonant (or epenthetic vowel initially treated differently from an inherited segmental vowel) like “bone.” Wherever it began, no doubt its development was reinforced by the multiple cases of contraction to *e· which these noun stems underwent in inflection before suffixes beginning in *e, like possessive *‑em.

All we can say is that we probably have examples here of pre-PA contracting *ayi to *e· or else coalescing preconsonantal *ay to *e·, reinforced by *aye contractions with the same stems in inflection, but the process had been morphologized and their distribution disturbed by the actual PA period. By then it’s probably best to already reconstruct synchronic allomorphs in final *|e·|.

*aya

In Proto-Eastern Algonquian, *aya unexpectedly contracts to *ā when the animate plural suffix *‑ak follows the suffix *‑inay discussed above (Goddard 2007:260-261). Note, however, that this contraction did not apply with peripheral suffixes other than *‑ak, namely the obviative suffixes *‑ar or *‑ah.

• *we-ta·n- “h/ daughter” + *‑inay “UNSPEC” (+ *‑aki “PROXpl”) → *ota·ninaya “one’s, a daughter” vs. *ota·ninayaki “one’s daughters” (Meskwaki ‡ota·ninawa [with *y → w] vs. ‡ota·ninawaki)
• → PEA *wətānīnay vs. *wətānīnāk (e.g., Mass ⟨wuttaunin⟩ ⫽wətʌ̨·nən⫽ [with loss of final *ay] vs. ⟨wuttauninneunk⟩ ⫽wətʌ̨·nənyʌ̨·k⫽)
• Cf., with no contraction: *eθkwe·w- “woman” + *‑inay “COLL” + *‑ari “OBVsg” → pseudo-PA *eθkwe·nayari → pseudo-PEA *əθk^wēwīnayar (e.g., Munsee oxkweewíinayal “women (obv.)”)
• But *‑inay contracts normally with appropriate suffixes, e.g. (repeated from earlier): Munsee mătán’toow “devil” + ‑iinay “COLL” → *mătán’toowiinay- + -ənk “LOC” [← PA *‑enki] → mătan’toowíineenk “in Hell” (“among, in the land of the devils”) (reflecting the PA contraction of *aye → *e·)

The same *aya → PEA *ā contraction is seemingly reflected in Massachusett and most of Delawaran when *‑ak followed the N-formative (PEA *‑n(ē), from what is here given—at least notationally—as PA *‑n(ay); the expected outcome here would be PEA ^x‑n‑ak, not “contracted” *‑nāk) (Goddard 2007:261-262), e.g., dialectal Unami kəmi·lí·na·k “you (sg.) gave me them (anim.)” |kə-mīl-ī-nāk| ← |kə-mīl-ī-n(ē)‑ak| (← apparent PEA *kəmīlīnāk ← PA *ke- “2” + *mi·r- “give to (TA+O)” + *‑i “1OBJ” + *‑n(ay) “N.FMV” + *‑aki “AN.PROXpl”).

Goddard reconstructs the contraction to *ā after the N-formative as arising later, in the languages in question, and being modeled on the contraction after *‑īnay. This seems the most likely possibility, given the former’s somewhat restricted distribution (cf., e.g., MalPass kmilinək “you (sg.) give me them (anim.)” [LeSourd 1988:22], not ^xkmilinak) including philological evidence that its optional presence in Unami is a recent development. But it should be noted that the question is clouded by the somewhat restricted distribution of the latter as well (as far as I can tell, not all EA languages have a reflex of *‑īnay, outside of some frozen cases in nouns, MalPass being one); by the plausible analogical pathway Goddard once proposed for N-formative contraction being the source of *‑īnay contraction in Mass (Goddard and Bragdon 1988:491); and by the odd shape/morphophonemics of the N-formative, which could be vaguely appealed to in trying to explain how *aya contraction ever arose in the first place—whereas *‑īnay lacked the same sort of weirdness in its ending *ay. Except that noun stems ending in *‑ay are highly unstable across the family, quite frequently being eliminated, merged with other stem classes, reshaped, membership of the class expanded or reduced, or having the *ay dropped from certain inflectional categories. Such instability (which, to be fair, derives in part from the fact that *ay contracts so often) would seem a decent place to start in trying to explain the origin of *aya contraction.

Pentland (1999:251) suggests that the contraction *aya contraction found with the N-formative matches a similar pattern in a dialect of Nipissing Algonquin (e.g., gidadaaweenaag “you (sg.) sell them (anim.),” AI+O verb adaawee-) and can be reconstructed back to PA, but in my view Goddard (2007:262-263) convincingly refutes this. The *‑āk allomorph of the PROXpl suffix after the N-formative in Eastern Algonquian can be explained as modeled after the same allomorph following *‑īnay (← *‑inay), where it did, apparently, represent one aberrant instance of contraction of PEA *aya to *ā. But as noted, this contraction is not found anywhere else, including even before the other peripheral suffixes one might expect to trigger it, *‑ar “OBVsg” and *‑ah “OBVpl”; I’d say it’s reasonable to question whether it was ever, from a historical sound-change standpoint, a case of contraction, rather than some obscure analogical reshaping or contamination of some sort. In any case, the Nipissing AI+O ‑naag ending (instead of the ‑n‑ag [← PA *‑n‑aki *|‑n(ay)‑aki|] or various other endings found in other dialects) is best explained as arising by proportional analogy with TA verbs (Goddard 2007:263):

TA	ni—aa   “1sg»PROXsg”		ni—aag   “1sg»PROXpl”
	gi—aa   “2sg»PROXsg”		gi—aag   “2sg»PROXpl”
AI+O	ni—naa   “1sg»+PROXsg”		X = ni—naag   “1sg»+PROXpl”
	gi—naa   “2sg»+PROXsg”		X = gi—naag   “2sg»+PROXpl”

And so on.

To sum up, there is some evidence that PEA had contraction of *aya to *ā (as if from PA *aya to *a·) in one quite limited context, but the diachronic origins of the synchronic process are obscure, and it cannot be reliably projected back to Proto-Algonquian.

Entirely separate from all this, there is evidence from initial change that pre-PA *aya contracted to *e·. As noted earlier, initial change in most cases originally consisted in the insertion of an *‑ay- sequence before the first vowel of a word (including *e → *e·, presumably from earlier *aye, as noted above), but the initial-changed form of *a is *e·. This would of course not be an active rule in PA, and it also could have another explanation, such as analogy with the *e → *e· IC shift.

*eye

There were no noun or verb stems or derivational elements in *‑ey, and, with one possible but problematic exception, no nouns, verbs, or derivational elements in initial *ye‑, which would allow us to determine the outcome of an underlying *|eye| sequence in PA; indeed, in PA proper neither surface ^xeye nor underlying ^x|eye| existed, except perhaps in reduplication. Some writers have speculated about pre-PA *eye giving PA *i· (e.g. Pentland 1979:408; Proulx 1980a, 1984:193, 194), and while this is plausible enough given the contraction of *|ewe| to *o·, actual good evidence for it strikes me as sparse. (The comparative Algic evidence offered for it in the past has mostly been wrongly interpreted or not really cognate.) In any case, since it would date to a period before Proto-Algonquian, it’s irrelevant for this post.

Summary of Possible Outcomes

Sequence	Can Contract To	Is Non-Contraction Also Found?
awe	o·, a·	yes
aye	e·, a·	yes
ayi	(→ aye →) a·, (→ aye →) e·; (e·)	yes [normal outcome]
aya	(e·); (PEA *ā)	yes [only outcome]
a·we	o·	yes
ewe	o·	no [always contracts]
eye	N/A [no longer exists]; (i·)	[N/A]
e·we	e·, o·	yes
e·ye	e·	yes? [only outcome?][45]
iwe, i·we	i·	yes
iye	i·	no [always contracts]
owe, o·we	o·	yes
Vyi [cf. ayi]	(→ Vye →) V· [cf. ayi]	yes [normal outcome]

Except in some cases of reduplication, the following sequences did not exist as either underlying or surface forms in Proto-Algonquian, and so play no role in contraction: *a·ye, *eye (but see above), *i·ye, *oye or *|weye|, and *o·ye.

Sources Used [click to expand]

(“AIL” = Algonquian and Iroquoian Linguistics)
(“AIL-M” = Algonquian and Iroquoian Linguistics Memoir)
(“AL” = Anthropological Linguistics)
(“APS-M” = Memoirs of the American Philosophical Society)
(“APS-T” = Transactions of the American Philosophical Society)
(“BAE-B” = Bureau of American Ethnology Bulletin)
(“CESP” = National Museum of Man, Mercury Series, Canadian Ethnology Service Paper)
(“HNAI” = Handbook of North American Indians, series ed. William C. Sturtevant)
(“IJAL” = International Journal of American Linguistics)
(“MHSC” = Collections of the Massachusetts Historical Society)
(“NMCB” = National Museum of Canada Bulletin)

• Alford, Thomas W[ildcat] (1929). The Four Gospels of Our Lord Jesus Christ in Shawnee Indian Language. Xenia, OH: W. A. Galloway/Nay-Nah-Ko-Nah-Kah.
• Aubin, George F. (2001). “The Algonquin-French Manuscript ASSM 104 (1661).” In Actes du Trente-Deuxième Congrès des Algonquinistes, ed. John D. Nichols, pp. 1-16. Winnipeg: University of Manitoba.
• Baraby, Anne-Marie, and Marie-Odile Junker, eds. (2023). Conjugation of Innu Verbs. 6th edition. Algonquian Dictionaries Project. Accessed March 20, 2024.
• Beach, Jesse (2004). The Morphology of Modern Western Abenaki. Honors thesis, Dartmouth College.
• Bloomfield, Leonard (1925). “On the Sound-System of Central Algonquian.” Language 1(4): 130-156. DOI: 10.2307/409540.
• Bloomfield, Leonard (1927). “Notes on the Fox Language [Part II].” IJAL 4(2/4): 181-219. DOI: 10.1086/463767.
• Bloomfield, Leonard (1946). “Algonquian.” In Linguistic Structures of Native America by Harry Hoijer et al., ed. Cornelius Osgood, pp. 85-129. Viking Fund Publications in Anthropology 6. New York: Viking Fund. Online version prepared by Will Oxford available here.
• Bloomfield, Leonard (1962). The Menomini Language, ed. Charles F. Hockett. New Haven, CT: Yale University Press.
• Bloomfield, Leonard (1975). Menominee Lexicon, ed. Charles F. Hockett. Milwaukee Public Museum Publications in Anthropology & History 3. Milwaukee: Milwaukee Public Museum.
• Bybee, Joan, Revere Perkins, and William Pagliuca (1994). The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: University of Chicago Press.
• CDO = Fisher, Louise, Wayne Leman, Leroy Pine Sr., and Marie Sanchez (2017). Cheyenne Dictionary Online. Chief Dull Knife College. Accessed March 22, 2024.
• Cenerini, Chantale (2014). Relational Verbs: Paradigm and Practice in a Manitoba Dialect of Swampy Cree. MA thesis, University of Regina.
• Costa, David J. (1996). “Reconstructing Initial Change in Algonquian.” AL 38(1): 39-72.
• Costa, David J. (2001). “Shawnee Noun Plurals.” AL 43(3): 255-287.
• Costa, David J. (2003). The Miami-Illinois Language. Lincoln: University of Nebraska Press.
• Costa, David J. (2008). “New Notes on Miami-Illinois.” In Papers of the Thirty-Ninth Algonquian Conference, ed. Karl S. Hele and Regna Darnell, pp. 123-165. London, ON: University of Western Ontario.
• Cotton, Josiah (1829 [1707–1708]). Vocabulary of the Massachusett (or Natick) Indian Language, ed. J[ohn] P[ickering]. Cambridge, MA: E. W. Metcalf and Company. [Repaginated offprint, originally published in MHSC 3(2): 147-257.]
• Cowell, Andrew, and Alonzo Moss, Sr. (2008). The Arapaho Language. Boulder: University Press of Colorado.
• Cunningham, Keith (2022). “A New Look for Choptank: An Updated Phonological Analysis of the William Vans Murray Vocabulary.” Ms., Academia.edu. Last accessed July 18, 2022.
• Cuoq, J[ean-]A[ndré] (1886). Lexique de la Langue Algonquine. Montréal: J. Chapleau & Fils.
• Dahlstrom, Amy (1997). “Fox (Mesquakie) Reduplication.” IJAL 63(2): 205-226. DOI: 10.1086/466319.
• Dahlstrom, Amy (2015). Meskwaki Syntax. Ms.
• Dahlstrom, Amy (2021). “The Historical Semantics of Past Tense and Irrealis Marking in Meskwaki.” In Thomason et al. (2021), pp. 101-119. Preprint version available here.
• Daviault, Diane (1994). L’Algonquin au XVII^e Siècle: Une Édition Critique, Analysée et Commentée de la Grammaire Algonquine du Père Louis Nicolas. Sainte-Foy: Presses de l’Université du Québec.
• Day, Gordon M. (1994). Western Abenaki Dictionary. Volume 1: Abenaki-English. CESP 128. Hull, QC: Canadian Museum of Civilization.
• ECOD = Junker, Marie-Odile, Marguerite MacKenzie, Luci Bobbish-Salt, Alice Duff, Linda Visitor, Ruth Salt, Anna Blacksmith, Patricia Diamond, and Pearl Weistche, eds. (2018). The Eastern James Bay Cree Dictionary on the Web: English-Cree and Cree-English, French-Cree and Cree-French (Northern and Southern Dialects). Algonquian Dictionaries Project. Accessed March 30, 2024.
• Eliot, John (1666). The Indian Grammar Begun: Or, An Essay to Bring the Indian Language into Rules, for the Help of Such as Desire to Learn the Same, for the Furtherance of the Gospel Among Them. Cambridge, [MA]: Printed by Marmaduke Johnson.
• Eliot, John (1685). Mamusse Wunneetupanatamwe up-Biblum God, Naneeswe Nukkone Testament kah wonk Wusku Testament, with John Cotton. Cambridge, [MA]: Printed by Samuel Green.
• Ellis, C. Douglas (1971) “Cree Verb Paradigms.” IJAL 37(2): 76-95. DOI: 10.1086/465143.
• Ellis, C. Douglas (1983). Spoken Cree: West Coast of James Bay. Revised edition. Edmonton: Pica Pica Press.
• Ellis, C. Douglas (2016). “Cree Verb Paradigms.” Hosted by the Algonquian Dictionaries Project. [Revised and updated version of paradigms in Ellis 1971.]
• Faries, R[ichard], ed. (1938). A Dictionary of the Cree Language as Spoken by the Indians in the Provinces of Quebec, Ontario, Manitoba, Saskatchewan, and Alberta, Based upon the Foundation Laid by Rev. E. A. Watkins, Revised . . . by . . . J. A. Mackay . . . [et al.]. Toronto: Church of England in Canada.
• FOD = Weshki-ayaad [= Svetlana Pedčenko], Charles Lippert, and Guy T. Gambill (2022). Freelang Ojibwe Dictionary. Computer program, version of January 1, 2022.
• Gatschet, Albert S. (1887–1891) [in catalog as “1887”]. [Odawa texts and vocabulary given by John W. Early, Baxter Springs, Kansas, 1887–1888. Further annotated or revised through at least 1891 with input from Jean (John) B. Bottineau.] Manuscript 237, National Anthropological Archives, Smithsonian Institution.
• Goddard, Ives (1973). “Delaware Kinship Terminology (With Comparative Notes).” Studies in Linguistics 23: 39-56.
• Goddard, Ives (1974). “Remarks on the Algonquian Independent Indicative.” IJAL 40(4): 317-327. DOI: 10.1086/465328.
• Goddard, Ives (1979a). “Comparative Algonquian.” In The Languages of Native America: Historical and Comparative Assessment, ed. Lyle Campbell and Marianne Mithun, pp. 70-132. Austin: University of Texas Press.
• Goddard, Ives (1979b). Delaware Verbal Morphology: A Descriptive and Comparative Study. Outstanding Dissertations in Linguistics. New York: Garland Publishing.
• Goddard, Ives (1981a). “Against the Linguistic Evidence Claimed for Some Algonquian Dialectal Relationships.” AL 23(7): 271-297.
• Goddard, Ives (1981b). “Massachusett Phonology: A Preliminary Look.” In Papers of the Twelfth Algonquian Conference, ed. William Cowan, pp. 57-105. Ottawa: Carleton University.
• Goddard, Ives (1982). “The Historical Phonology of Munsee.” IJAL 48(1): 16-48. DOI: 10.1086/465711.
• Goddard, Ives (1983). “The Eastern Algonquian Subordinative Mode and the Importance of Morphology.” IJAL 49(4): 351-387. DOI: 10.1086/465800.
• Goddard, Ives (1990). “Primary and Secondary Stem Derivation in Algonquian.” IJAL 56(4): 449-483. DOI: 10.1086/466171.
• Goddard, Ives (1991). “Observations Regarding Fox (Mesquakie) Phonology.” In Papers of the Twenty-Second Algonquian Conference, ed. William Cowan, pp. 157-181. Ottawa: Carleton University.
• Goddard, Ives (1993). “Songs in Fox (Mesquakie) Texts: Linguistics and Philology.” In Papers of the Twenty-Fourth Algonquian Conference, ed. William Cowan, pp. 212-239. Ottawa: Carleton University.
• Goddard, Ives, ed. (1994). Leonard Bloomfield’s Fox Lexicon: Critical Edition. AIL-M 12. Winnipeg: Algonquian and Iroquoian Linguistics.
• Goddard, Ives (1995). “Notes on Fox (Mesquakie) Inflection: Minor Modes and Incompletely Described Morphemes.” In Papers of the Twenty-Sixth Algonquian Conference, ed. David H. Pentland, pp. 124-150. Winnipeg: University of Manitoba.
• Goddard, Ives (1997). “Addenda and Corrigenda for Leonard Bloomfield, ‘Algonquian’ (1946) [= Bloomfield 1946].” AIL 22(4): 31-36.
• Goddard, Ives (2000). “The Historical Origin of Cheyenne Inflections.” In Papers of the Thirty-First Algonquian Conference, ed. John D. Nichols, pp. 77-129. Winnipeg: University of Manitoba.
• Goddard, Ives (2001). “Contraction in Fox (Meskwaki).” In Actes du Trente-Deuxième Congrès des Algonquinistes, ed. John D. Nichols, pp. 164-230. Winnipeg: University of Manitoba.
• Goddard, Ives (2002). “Grammatical Gender in Algonquian.” In Papers of the Thirty-Third Algonquian Conference, ed. H.C. Wolfart, pp. 195-231. Winnipeg: University of Manitoba.
• Goddard, Ives (2003). “Reconstructing the History of the Demonstrative Pronouns of Algonquian.” In Essays in Algonquian, Catawban and Siouan Linguistics in Memory of Frank T. Siebert, Jr., ed. Blair A. Rudes and David J. Costa, pp. 37-102. AIL-M 16. Winnipeg: Algonquian and Iroquoian Linguistics.
• Goddard, Ives (2004). “Meskwaki Verbal Affixes.” In Papers of the Thirty-Fifth Algonquian Conference, ed. H.C. Wolfart, pp. 97-123. Winnipeg: University of Manitoba.
• Goddard, Ives (2006). “The Proto-Algonquian Negative and Its Descendants.” In Actes du Trente-Septième Congrès des Algonquinistes, ed. H.C. Wolfart, pp. 161-208. Winnipeg: University of Manitoba.
• Goddard, Ives (2007). “Reconstruction and History of the Independent Indicative.” In Papers of the Thirty-Eighth Algonquian Conference, ed. H.C. Wolfart, pp. 207-271. Winnipeg: University of Manitoba.
• Goddard, Ives (2008a). “Notes on Mahican: Dialects, Sources, Phonemes, Enclitics, and Analogies.” In Papers of the Thirty-Ninth Algonquian Conference, ed. Karl S. Hele and Regna Darnell, pp. 246-315. London, ON: University of Western Ontario.
• Goddard, Ives (2008b). “Quauhaug.” AIL 33(1): 6-7.
• Goddard, Ives (2010). “Linguistic Variation in a Small Speech Community: The Personal Dialects of Moraviantown Delaware.” AL 52(1): 1-48.
• Goddard, Ives (2013). “The Munsee of Charles Halfmoon’s Translations.” In Papers of the Forty-First Algonquian Conference, ed. Karl S. Hele and J. Randolph Valentine, pp. 81-119. Albany: State University of New York Press.
• Goddard, Ives (2014). “Reduplication in the Delaware Languages.” In Papers of the Forty-Second Algonquian Conference, ed. J. Randolph Valentine and Monica Macaulay, pp. 134-158. Albany: State University of New York Press.
• Goddard, Ives (2015a). “Arapaho Etymologies.” Ms.
• Goddard, Ives (2015b). “Arapaho Historical Morphology.” AL 57(4): 345-411. DOI: 10.1353/anl.2016.0010.
• Goddard, Ives (2016). “The ‘Loup’ Languages of Western Massachusetts: The Dialectal Diversity of Southern New England Algonquian.” In Papers of the Forty-Fourth Algonquian Conference, ed. Monica Macaulay, Margaret Noodin, and J. Randolph Valentine, pp. 104-138. Albany: State University of New York Press.
• Goddard, Ives (2021). A Grammar of Southern Unami Delaware (Lenape). Petoskey, MI: Mundart Press.
• Goddard, Ives (2023). A Grammar of Meskwaki. Petoskey, MI: Mundart Press.
• Goddard, Ives, and Kathleen J. Bragdon (1988). Native Writings in Massachusett. 2 volumes. APS-M 185. Philadelphia: American Philosophical Society.
• Goddard, Ives, Charles F. Hockett, and Karl V. Teeter (1972). “Some Errata in Bloomfield’s Menomini [= Bloomfield 1962].” IJAL 38(1): 1-5. DOI: 10.1086/465177.
• Goddard, Ives, and Lucy Thomason (2014). A Meskwaki-English and English-Meskwaki Dictionary, Based on Early Twentieth-Century Writings by Native Speakers. Petoskey, MI: Mundart Press.
• [Grube, Bernhard] (n.d. [ca. 1765]). Einige Dellawarishe Redensarten und Worte. MS Am 767 (15), Houghton Library, Harvard University. Accessed February 6, 2024.
• Harrington, John P. (1955). “The Original Strachey Vocabulary of the Virginia Indian Language.” In Anthropological Papers, Numbers 43-48 (No. 46), pp. 189-202. BAE-B 157. Washington, D.C.: Government Printing Office.
• Heckewelder, John ([ca. 1785]). “Comparative Vocabulary of the Delaware, Minsi [Munsee], Mohican, Natick, Chippeway, Shawanoe [Miami] and Nanticoke Languages, by the Revd. John Heckewelder.” Ms., American Indian Vocabulary Collection item 21/22b (Mss.497.V85), American Philosophical Society. Accessed April 27, 2022.
• Inglis, Stephanie, and Eleanor Johnson (2001). “The Mi’kmaq Future: An Analysis.” In Actes du Trente-Deuxième Congrès des Algonquinistes, ed. John D. Nichols, pp. 249-257. Winnipeg: University of Manitoba.
• Itwêwina = Arppe, Antti, Eddie Antonio Santos, Matt Yan, Atticus Harrigan, Katherine Schmirler, Kwabema Amoh, and Arok Wolvengrey, eds. (2019–). Itwêwina: Plains Cree Intelligent Dictionary. Alberta Language Technology Lab, University of Alberta. Accessed March 31, 2024.
• James, Deborah (1991). “Preterit Forms in Moose Cree as Markers of Tense, Aspect, and Modality.” IJAL 57(3): 281-297. DOI: 10.1086/ijal.57.3.3519721.
• James, Deborah, Sandra Clarke, and Marguerite MacKenzie (2001). “The Encoding of Information Source in Algonquian: Evidentials in Cree/Montagnais/Naskapi.” IJAL 67(3): 229-263. DOI: 10.1086/466459.
• Jefferson, Thomas (1808). “Comparative Vocabularies of Several Indian Languages.” Ms., Comparative Vocabularies of Several Indian Languages, 1802-1808 (Mss.497.J35), American Philosophical Society. Accessed March 26, 2022.
• Jenness, Diamond (1935). The Ojibwa Indians of Parry Island, Their Social and Religious Life. NMCB 78, Anthropological Series 17. Ottawa: King’s Printer.
• Jones, David (1977). A Basic Algonquin Grammar for Teachers of the Language at Maniwaki, Quebec. Maniwaki, QC: River Desert Band Council.
• Junker, Marie-Odile, and Marguerite MacKenzie, eds. (2014a). East Cree (Northern Dialect) Verb Conjugation. 3rd edition. Algonquian Dictionaries Project. Accessed March 15, 2024.
• Junker, Marie-Odile, and Marguerite MacKenzie, eds. (2014b). East Cree (Southern Dialect) Verb Conjugation. 3rd edition. Algonquian Dictionaries Project. Accessed March 15, 2024.
• Lacombe, A[lbert] (1874). Grammaire de la Langue des Cris. Montreal: Beauchemin & Valois.
• Laurent, Joseph (1884). New Familiar Abenakis and English Dialogues . . . . Quebec [City]: Léger Brousseau.
• LeSourd, Philip S. (1988). Accent and Syllable Structure in Passamaquoddy. Ph.D. dissertation, Massachusetts Institute of Technology.
• LeSourd, Philip S. (2011). “Enclitic Particles in Western Abenaki: The Syntax of Second Position.” AL 53(2): 91-131. DOI: 10.1353/anl.2011.0009. Preprint version available here.
• LeSourd, Philip S. (2015). “Enclitic Particles in Western Abenaki: Form and Function.” IJAL 81(3): 301-335. DOI: 10.1086/681577. Preprint version available here.
• Lockwood, Hunter Thompson (2017). How the Potawatomi Language Lives: A Grammar of Potawatomi. Ph.D. dissertation, University of Wisconsin-Madison.
• LTD = Rementer, Jim, Bruce Pearson, Jan Brown, Nicky Kay Michael, April Daniels, and Ray Whritenour, eds. (2002–). Lenape Talking Dictionary. Accessed April 1, 2024.
• MCOD = Brousseau, Kevin, ed. (2014–). Moose Cree Online Dictionary. Moose Factory’s Community Language Project, Moose Cree First Nation. Accessed April 4, 2024.
• Meeussen, A. E. (1959). “Algonquian Clusters with Glottal Stop.” IJAL 25(3): 189-190. DOI: 10.1086/464527.
• Michelson, Truman (1919). “Two Proto-Algonquian Phonetic Shifts.” Journal of the Washington Academy of Sciences 9(12): 333-334.
• Michelson, Truman (1920). “Two Phonetic Shifts Occurring in Many Algonquian Languages.” IJAL 1(4): 300-304. DOI: 10.1086/463729.
• MIDA = Baldwin, Daryl, et al., eds. (2012–). Miami-Illinois Digital Archive. Myaamia Center, Miami University. Accessed March 10, 2024.
• MqOD = Haberlin, Sean, Eunice Metallic, Diane Mitchell, Watson Williams, Joe Wilmot, and Dave Ziegler, eds. (1997–). Mi’gmaq-Mi’kmaq Online [Dictionary]. Accessed March 11, 2024.
• MyPD = Miami Tribe of Oklahoma (2019–). Myaamia-Peewalia Dictionary. Miami-Illinois Indigenous Languages Digital Archive. Accessed March 14, 2024.
• Nichols, John D. (1980). Ojibwe Morphology. Ph.D. dissertation, Harvard University.
• Nicolas, Louis ([ca. 1674]). Grammaire Algonquine, Ou des Sauvages de l’Amerique Septentrionalle, Avec la Description du Pays, Journaux de Voyages, Memoires, Remarques sur l’Histoire Naturelle, &c &c. [Listed in catalog as “Grammaire algonquine ou des sauvages de l’Amérique septentrionale, avec la description du pays, journaux de voyages, mémoires, remarques sur l’histoire naturelle”]. Ms., NUMM-109480, Bibliothèque Nationale de France. [= Hanzeli MS #14.]
• Nilsen, Campbell (2017). The Historical Development of the Menominee Independent Indicative. Unpublished senior essay.
• NOD = Naokwegijig-Corbiere, Mary Ann, and J. Randolph Valentine, eds. (2015–). Nishnaabemwin Online Dictionary. Algonquian Dictionaries Project. Accessed March 10, 2024.
• OID = Ambroise, Jérémie, Marie-Odile Junker, Marguerite MacKenzie, and Yvette Mollen, eds. (2023). Online Innu Dictionary. Algonquian Dictionaries Project. Accessed April 3, 2024.
• OldC = Brousseau, Kevin (2022–). Old Cree. Accessed March 8, 2024.
• O’Meara, John (1990). Delaware Stem Morphology. Ph.D. dissertation, McGill University.
• O’Meara, John (1996). Delaware-English, English-Delaware Dictionary. Toronto: University of Toronto Press. [2014 paperback edition.]
• OPD = Livesay, Nora, and John D. Nichols, eds. (2012–). Ojibwe People’s Dictionary. Department of American Indian Studies, University of Minnesota. Accessed March 20, 2024.
• Oxford, Will (2015). “Patterns of Contrast in Phonological Change: Evidence from Algonquian Vowel Systems.” Language 91(2): 308-357. DOI: 10.1353/lan.2015.0028.
• Oxford, Will (2016). “Proto-Algonquian Phonotactics.” Ms.
• Oxford, Will (2023). “Concise Proto-Algonquian Verb Paradigms.” Ms., version of June 15, 2023, Algonquian Linguistics Materials. Accessed March 25, 2024.
• Oxford, Will (2024). The Algonquian Inverse. Oxford Studies of Endangered Languages. Oxford: Oxford University Press. DOI: 10.1093/oso/9780192871800.001.0001.
• PbD = Penobscot Dictionary (2015). Penobscot Cultural & Historic Preservation. Accessed March 12, 2024.
• Pentland, David H. (1979). Algonquian Historical Phonology. Ph.D. dissertation, University of Toronto.
• Pentland, David H. (1984). “New Modes in Old Ojibwa.” AIL 9: 11-17.
• Pentland, David H. (1988). “More New Modes in Old Ojibwa.” AIL 13: 47-51.
• Pentland, David H. (1999). “Morphology of the Algonquian Independent Order.” In Papers of the Thirtieth Algonquian Conference, ed. David H. Pentland, pp. 222-266. Winnipeg: University of Manitoba.
• Pentland, David H. (2023). Proto-Algonquian Dictionary: A Historical and Comparative Dictionary of the Algonquian Languages. 4 volumes. AIL-M Winnipeg: Algonquian and Iroquoian Linguistics.
• PMLP = Francis, David A., Robert M. Leavitt, Margaret Apt, et al., eds. (2008–). Passamaquoddy-Maliseet Language Portal. Language Keepers. Accessed March 3, 2024.
• Proulx, Paul (1977). “Connective Vowels in Proto-Algonquian.” IJAL 43(2): 156-157. DOI: 10.1086/465473.
• Proulx, Paul (1980a). “Algonquian *ye and *we.” IJAL 46(1): 46-47. DOI: 10.1086/465630.
• Proulx, Paul (1980b). “The Linguistic Evidence on Algonquian Prehistory.” AL 22(1): 1-21.
• Proulx, Paul (1980c). “The Subordinative Order of Proto-Algonquian.” IJAL 46(4): 289-300. DOI: 10.1086/465664.
• Proulx, Paul (1982). “The Origin of the Absolute Verbs of the Algonquian Independent Order.” IJAL 48(4): 394-411. DOI: 10.1086/465749.
• Proulx, Paul (1984). “Proto-Algic I: Phonological Sketch.” IJAL 50(2): 165-207. DOI: 10.1086/465826.
• Proulx, Paul (1985). “Proto-Algic II: Verbs.” IJAL 51(1): 59-93. DOI: 10.1086/465860.
• Proulx, Paul (1990). “Proto-Algonquian Verb Inflection.” KWPL 15(2): 100-145. DOI: 10.17161/KWPL.1808.440.
• Proulx, Paul (2003). “The Evidence on Algonquian Genetic Grouping: A Matter of Relative Chronology.” AL 45(2): 201-225.
• Proulx, Paul (2005). “Reduplication in Proto-Algonquian and Proto-Central Algonquian.” IJAL 71(2): 193-214. DOI: 10.1086/491634.
• Quinn, Conor M. (2015). “Recovering the Penobscot TA Conjunct: Documentary, Analytical, and Revitalization Considerations.” Slides for paper presented at the 47th Algonquian Conference, September 24, 2015.
• Rhodes, Richard A. (2021). “The Case for Core Central Algonquian.” In Thomason et al. (2021), pp. 305-345.
• Rountree, Helen C. (n.d.). “Dictionary of the Powhatan Language.” Ms., Box 14, Blair Rudes Papers (NAA.2009-16), National Anthropological Archives, Smithsonian Institution.
• Schaefer, Carl (2017). [Supplementary materials for Schaefer 2019], December 12, 2017: Readme; Suggested Corrections to Alford (1929); Concordance of Alford (1929); Lexicon of Alford (1929); Gospel of Matthew; Gospel of Mark; Gospel of Luke; Gospel of John. Algonquian Conference, Supplementary Materials. Accessed March 30, 2020.
• Schaefer, Carl (2019). “Alford’s Shawnee Translation of the Gospels.” In Papers of the Forty-Eighth Algonquian Conference, ed. Monica Macaulay and Margaret Noodin, pp. 221-238. East Lansing: Michigan State University Press.
• Sherwood, David F. (1983). Maliseet-Passamaquoddy Verb Morphology. Ph.D. dissertation, Yale University.
• Siebert, Frank T., Jr. (1975). “Resurrecting Virginia Algonquian from the Dead: The Reconstituted and Historical Phonology of Powhatan.” In Studies in Southeastern Indian Languages, ed. James M. Crawford, pp. 285-453. Athens: University of Georgia Press.
• Silvy, Antoine (1974 [ca. 1684]). Dictionnaire Montagnais-Français, ed. Lorenzo Angers, David E. Cooter, and Gérard E. McNulty. Montréal: Les Presses de l’Université du Québec. [Not seen personally, cited by OldC.]
• Starks, Donna J. (1992). Aspects of Woods Cree Syntax. Ph.D. dissertation, University of Manitoba.
• Strachey, William (1953 [1612]). The Historie of Travell into Virginia Britania, ed. Louis B. Wright and Virginia Freund. London: Hakluyt Society.
• Thomason, Lucy (2003). The Proximate and Obviative Contrast in Meskwaki. Ph.D. dissertation, University of Texas at Austin.
• Thomason, Lucy (2015). “On Editing Bill Leaf’s Meskwaki Texts,” with “Bill Leaf’s Story of Red-Leggins” by Bill Leaf, ed. Lucy Thomason. In New Voices for Old Words: Algonquian Oral Literatures, ed. David J. Costa, pp. 315-452. Lincoln: University of Nebraska Press.
• Thomason, Lucy, David J. Costa, and Amy Dahlstrom, eds. (2021). Webs of Relationships and Words from Long Ago: A Festschrift Presented to Ives Goddard on the Occasion of His 80th Birthday. Petoskey, MI: Mundart Press.
• Valentine, J. Randolph (1994). Ojibwe Dialect Relationships. Ph.D. dissertation, University of Texas at Austin.
• Valentine, J. Randolph (2001). Nishnaabemwin Reference Grammar. Toronto: University of Toronto Press.
• Voegelin, C. F. (1938–1940). Shawnee Stems and the Jacob P. Dunn Miami Dictionary. 5 parts. Prehistory Research Series 1(3): 63-108, 1(5): 135-167, 1(8): 289-341, 1(9): 345-406, 1(10): 409-478. Indianapolis: Indiana Historical Society.
• Voorhis, Paul H. (1971). “New Notes on the Mesquakie (Fox) Language.” IJAL 37(2): 63-75. DOI: 10.1086/465142.
• Voorhis, Paul H. (1974). Introduction to the Kickapoo Language. Indiana University Publications, Language Science Monographs 13. [Bloomington]: Indiana University.
• Voorhis, Paul [H.] (1979). Grammatical Notes on the Penobscot Language from Frank Speck’s Penobscot Transformer Tales. University of Manitoba Anthropology Papers 24. Winnipeg: Department of Anthropology, University of Manitoba.
• Whitecalf, Sarah (2021). Mitoni Niya Nêhiyaw – Nêhiyaw-Iskwêw Mitoni Niya: Cree Is Who I Truly Am – Me, I Am Truly a Cree Woman: A Life Told by Sarah Whitecalf, ed. and trans. H.C. Wolfart and Freda Ahenakew. Publications of the Algonquian Text Society. Winnipeg: University of Manitoba Press.
• Whittaker, Gordon (2005). A Concise Dictionary of the Sauk Language. Stroud, OK: Sac and Fox National Public Library.
• WOD = Valentine, J. Randolph, and Patricia M. Ningewance Nadeau, eds. (2023). Western Ojibwe Dictionary. Computer program, version of January 23, 2023.
• Wolfart, H.C. (1973). Plains Cree: A Grammatical Study. APS-T n.s. 63, part 5. Philadelphia: American Philosophical Society. DOI: 10.2307/1006246.
• Wolfart, H.C. (1996). “Sketch of Cree, an Algonquian Language.” In Languages, ed. Ives Goddard, pp. 390-439. HNAI vol. 17. Washington, D.C.: Smithsonian Institution.
• Zeisberger, David (1827). Grammar of the Language of the Lenni Lenape or Delaware Indians, trans. Peter Stephen Du Ponceau [= Duponceau]. APS-T 3. Philadelphia: James Kay, Jun.

---

1. What I label here as the “emphatic present” and “emphatic past” are especially uncertain and have gone by various names; they probably marked perfect aspects in addition to other semantics, if any—i.e. present vs. past perfect? The emphatic past is often called the “preterit.” I’ve gone with “emphatic” largely because it is usefully ambiguous, though I translate them as perfects. (An alternative could be to translate them using “do” and “did”: “I did see them” and “I do come” instead of “I had seen them” and “I have come.”) These terms have been used before, or close to it, with Bloomfield in his “Sketch” (1946:99-100, §42) calling them the “emphatic present” and “emphatic preterit,” though he evidently had somewhat different reasons for doing so than I do. ↑
2. PA had a number of important processes by which monosyllabic *GV or disyllabic *VGV sequences were reduced to a single vowel. As just noted, there has been some degree of terminological inconsistency regarding these processes in the past. Thus, for example, Bloomfield (1962:84-87) called the one-syllable processes “merging” and only used “contraction” to apply to the two-syllable processes; Pentland (1979) called the one-syllable processes “coalescence” while using both “contraction” and “coalescence” to refer to the two-syllable processes; Valentine (2001) calls the one-syllable processes “coalescence” but doesn’t give the two-syllable processes a name; Oxford (2015) calls the one-syllable processes both “coalescence” and “contraction”; and Oxford (2024) calls both types of process “coalescence”. However, by far the most common practice is to refer to both types of process as “contraction,” as Michelson (1919:334, 1920:304) and Bloomfield in his earlier work (e.g. 1946) did; this is continued by Goddard, Pentland in his later work (e.g. 1999), and many others—with Goddard distinguishing between “one-syllable” and “two-syllable” contraction when necessary.
   

   Examples of Terminology Used

   Source	*GV → *V(ː)	*VGV → Vː
   Most common: Goddard, Pentland (1999), Rhodes, Costa, Wolfart, Proulx, etc., etc.	contraction	contraction
   Bloomfield (1962)	merging	contraction
   Pentland (1979)	coalescence	coalescence, contraction
   Valentine (2001)	coalescence	—
   Oxford (2015)	coalescence, contraction	—
   Oxford (2024)	coalescence	coalescence

   While most Algonquianists today opt for “contraction” (or more rarely, “coalescence”) to cover both types, I would contend that there are important distinctive properties of the two-syllable processes which justify treating them as a group contrasted with the one-syllable processes.

   Despite the fact that doing so means adding yet one more pattern of nomenclature to the existing mix, in this post I will therefore refer to two-syllable processes as contraction, and to one-syllable processes as coalescence. “Contraction,” beyond being time-honored, is certainly an appropriate label for the *VGV group, while “coalescence” is the standard linguistic term to describe the phonological process by which two adjacent sounds combine to form a single unit with merged features (a description which matches most of the *GV processes). ↑

3. As indicated, while I don’t always end up following their precise formulations or conclusions, by far my greatest debt in assembling this post is to the work of Ives Goddard, David Pentland, and Leonard Bloomfield; a very substantial portion not merely of the broad rules but of more subtle insights—and many of the specific examples used—derives from their writings. Unfortunately, for the core of this post, originating as a section in my post on Proto-Algonquian, I did not inline cite the source of individual rules, insights, and/or examples, and it proved too impractical to substantially alter that when expanding it here. The most important individual sources I’ve drawn on are, roughly in descending order of importance, Goddard (2001 [esp. for contraction], 2007), Pentland (1979, 1999), Goddard (2006 [esp. for negative and irrealis verbs], 2015b), Bloomfield (1946), and Goddard (1990, 2000). Examples and insights from any other sources are generally explicitly inline cited, but I’ve probably missed some cases, for which I offer my sincere apologies (as I do for the haphazard citation structure of this project in general).

   Out of necessity, I’ve constructed most of the PA examples myself, following the principles of the language as I understand it—as reflected in these rules themselves, as well as elsewhere on this blog—which will usually conform to standard views, especially Goddard’s, but not in every instance. To increase diversity, I’ve often tried to add more daughter languages to the roster of those illustrating the reflex of a given rule or formation, beyond what was provided in an original source I’m taking an example from—or if I’ve created the example myself. If the daughter form is not found in one of the key sources listed above, then unless otherwise cited in the text, my sources for such forms are:

   Cheyenne (CDO), Cree – Moose Cree (MCOD), Cree – Plains Cree (Itwêwina; personal knowledge), Cree – Southern East Cree (ECOD), Innu (OID), Maliseet-Passamaquoddy (PMLP), Massachusett (Goddard and Bragdon 1988), Menominee (Bloomfield 1962, 1975), Meskwaki (Goddard 2023; Goddard and Thomason 2014), Miami-Illinois (Costa 2003; MyPD), Munsee (Goddard 1979b; O’Meara 1996), Ojibwe (personal knowledge; OPD; WOD), Penobscot (PbD), Sauk (Whittaker 2005), Shawnee (Costa 2001), Unami (Goddard 2021), and Western Abenaki (Day 1994). I’ve especially heavily mined Goddard (1979b, 2021, 2023) and Bloomfield (1962).

   In a few cases, in order to most effectively illustrate some point or comparison, or simply because no relevant examples existed in my sources, I have constructed an example word in a daughter language. All such constructed words, unless they involved merely replacing one person prefix with another, or are from Ojibwe (which I know well enough to make such constructions with high confidence), are marked with a double dagger, ‡. In constructing such forms, I made use of the following resources:

   Arapaho (Goddard 2015b), Cree – Moose Cree (Ellis 1971, 2016), Kickapoo (Voorhis 1974), Maliseet-Passamaquoddy (Sherwood 1983; PMLP), Menominee (Bloomfield 1962; Goddard et al. 1972), Meskwaki (Goddard 1994, 1995, 2004, 2023; and for glossing, also Thomason 2003), Miami-Illinois (Costa 2003), Munsee (Goddard 1979b), and Unami (Goddard 1979b, 2021).

   While it was not always necessary, in some of these cases I have also added a footnote giving related and other forms in support of the constructed form. Such footnotes are marked with a preceding “S”: ^[S10] and so on; this way readers who feel no need to check such supporting forms can ignore those footnotes.

   Finally, I have also at times added underlying forms for daughter languages which are not present in the original source, but have not considered it necessary to note this. ↑

4. I retain Bloomfield’s practice of writing Menominee /æ(ː)/ as ⟨ε(·)⟩.

   Long vowels are marked by doubled characters in Arapaho, East Cree, Gros Ventre, Kickapoo, Miami-Illinois, Munsee, Ojibwe (including Nipissing Algonquin), and Shawnee, and by macrons in western Cree and Old Algonquin. This includes indicating Ojibwe and Southern East Cree long /ɛː/ with doubled ⟨ee⟩ instead of the practical orthographies’ single ⟨e⟩. Penobscot ⟨α⟩, which represents /ɤ/ (← historical long *a·) is retained from the practical orthography. Menominee short diphthongs are written, following Bloomfield, as ⟨ya, wa⟩, but the long (falling) diphthongs are written ⟨ya·, wa·⟩ instead of Bloomfield’s ⟨ia, ua⟩. Even though its pronunciation is usually lower, I write ⟨u·⟩ for the Unami vowel which Goddard transcribes ⟨o·⟩, but which is paired with short ⟨u⟩ /u/ and is written with ⟨u⟩ in the practical orthography.

   East Cree final /j/ and /w/ are written ⟨y⟩ and ⟨w⟩ instead of the practical orthography’s ⟨i⟩ and ⟨u⟩.

   The lenis-fortis obstruent systems of Western Abenaki and modern southern Ojibwe are written using voiced and voiceless characters (applied here to all dialects of Ojibwe except Old Algonquin). Ojibwe is also written with the practical orthography’s ⟨z, zh, s, sh⟩ for lenis /s, ʃ/, fortis /sː, ʃː/; while Munsee and East Cree /ʃ/ and /tʃ/ are also written ⟨sh, ch⟩. The Western Abenaki affricates are written ⟨j⟩ (lenis) and ⟨c⟩ (fortis). I follow Goddard in distinguishing Munsee primary from secondary medial nasal clusters via an apostrophe in the latter, rather than adopting the practical orthography’s use of voiced characters (thus /nt/ [nd] is ⟨nt⟩ and /n’t/ [nn̥t] is ⟨n’t⟩).

   Innu is written entirely in the pan-dialectal practical orthography, with the actual phonemic form following this.

   Arapaho vowels and pitch are written basically following the system of Goddard (2015b), modulo using doubled letters for long vowels: monosyllabic long vowels or diphthongs (⟨aa⟩, ⟨áá⟩, ⟨êe⟩, ⟨êi⟩, ⟨ei⟩, etc.) are distinguished from heterosyllabic vowel clusters (⟨a.a⟩, ⟨á.á⟩, ⟨ée⟩, ⟨éi⟩, ⟨e.i⟩, etc.).

   I do not write the Munsee diaphonemes ⟨ĭ⟩ /i/ ~ /ə/ or ⟨ĕ⟩ /ɛ/ ~ /ə/ distinctly from regular /i/ ⟨i⟩ and /ɛ/ ⟨e⟩. ↑

5. Proulx (1982:402-403) proposed a similar process, though in this case as a sound change affecting pre-PA, consisting of the “loss of [a] nasal or semivowel plus short vowel” word-finally. ↑
6. Cf. míilkwək “*s/he (PROX) who is given sth./s.o. by h/ (OBV)” (based on Goddard 1979b:185, §6.39) from miil- “give to (TA+O)”; cf. also Unami pa·tamá·k·uk[⸗č] “the one (OBV) who will pray to him (PROX)” (Goddard 2021:87, ex. 4.65n) = |(IC)\pāhtamaw-əkw-ə‑k| = REL\pray.to.TA-INV-EPTH-3AN.CONJ (with omission of peripheral suffix marking the obviative participial head). ↑
7. Reconstructed as such by Costa (2003:336); cf., e.g., ⟨kĭkalĭndansĭwa⟩ kihkeelintansiiwa “s/he doesn’t know it” (Costa 2003:456), from kihkeelint-am- “know (TI).” ↑
8. Reconstructed as such by Costa (2003:345); cf., e.g., Old Illinois ⟨teperimesȣrani⟩ tipeerimehsooraani “I don’t control you (sg.)” (Costa 2003:507), from tipeerim- “control, rule (TA).” ↑
9. Cf. mešotamoweči “they (INAN) (in relation to h//them, e.g. belonging to h//them) are hit by shots” (‑(o)weči = UNSPEC»3INAN.CONJ.RELN) in e·h-pwa·wi-[…]-iši-mešotamoweči owi·kewa·wani “their houses were not hit by the shots [at all]” (Goddard 1995:143). ↑
10. Here and elsewhere in the post I reconstruct many verbs that contain *‑pan or *‑(e)sahan “EMPH.PRES” as ending in final *‑pa or *‑(e)saha, as here with *nepo·pa (by Rule 28, which drops final consonants). However, Goddard (2000:125, 126), Pentland (1999:240, 251), and Proulx (1990:106, 135) all reconstruct the endings *‑pani and *‑sahani [or, in their terms at the time, *‑sapani], with Pentland and Proulx (cf. 1990:104) explicitly or implicitly indicating a division in these forms, *‑pan‑i etc. What they take this final *‑i to be is never made clear; it wouldn’t functionally match any of Pentland’s peripheral suffixes, and Proulx barely even mentions independent-order peripheral suffixes. Goddard (2007:249) seems to have changed his reconstruction to the one I’m using here, at least sometimes? At any rate, it seems to me that the earlier reconstructions are wrong. A detailed technical justification would take up too much space; if you’re really interested, you’re welcome to leave a comment I guess. ↑
11. Another possibility is that this combination resulted in *‑hsahan by (the rarer pattern of) Meeussen’s Law (= Rule 5b), which was then modified in some daughters—or does Rule 8c represent a partial cleaning-up of the result of Meeussen’s Law in this ending, which had made it synchronically opaque? Full argumentation for either of the first two possibilities is too complex to get into here, but see Goddard (2015b:407, n. 102) for a brief comment on the greater likelihood of the scenario I’ve laid out in the body of the post. (In any case, it’s clear that the ending was not just PA ^x‑tesahan or similar.) ↑
12. Pentland (1999:246, 2023) reconstructs the epenthetic vowel as *‑i- before Medials and *‑e- everywhere else, while Proulx (1977, 1984:197-198, n. 11, 1985:61) argues that it was *‑i- in primary derivation and *‑e- in secondary derivation and inflection, though he admits (1977:156, 1984:198, n. 11) that this account “leaves some irregularit[ies]” in “a few archaic combinations”; Goddard (1990:451, n. 9) simply denies that the environment where one vowel was chosen over the other is fully predictable. However, note that this statement is more true for the daughters than for Proto-Algonquian itself—it’s certainly the case that *‑e- is almost the only epenthetic vowel found in inflection, with just a couple exceptions, and *‑i- is overwhelmingly the most common one before Medials, and deverbals (derivational elements derived from stems) in general. I have a theory on the explanation for this pattern, but will save it for another post. ↑
13. This isn’t really a proper example since by the actual PA period there were several antipassive suffixes of similar shape. While most no doubt derived from a single morpheme at some pre-PA stage, they should not be analyzed as such in later stages; *‑ike· with a segmental initial *i, given in the first part of my example, is a general antipassive that doesn’t necessarily have an implied object, while *‑ke· with no initial vowel (and epenthetic *‑e- inserted in some cases), given in the second part of my example, has an implied human object. (There are also *‑kye·, which takes no epenthetic vowel and has an implied human object, and the unrelated *‑iwe·, with segmental *i and an implied human object, among others.) I’m too exhausted and busy at the moment to find a better example illustrating a connective *‑i- which can be proven to be epenthetic. ↑
14. All biblical citations for Massachusett are to Eliot (1685); the corresponding English text would be (roughly) the 1611 KJV. These citations have often been noted by a previous source such as Goddard and Bragdon (1988) or Goddard (1981b), though I have checked Eliot’s original in all cases. ↑
15. Depending on exact ordering, such environments might have to include: between clusters that were affected by Meeussen’s Law (see Rules 5a/5b: a heterogeneous group, though following some patterns, such as, e.g., the emphatic past suffix *‑pan taking no preceding epenthetic vowel [see Rule 9e]); following the 2OBJ theme sign *‑eθ (cf. Rule 4b); and between *w and any following consonant (cf. Rules 4c, 6, and 7, among others). ↑
16. Despite Pentland’s (2023:1826) later rejection (repudiating Pentland 1979:371), perhaps there is also a connection with the verb *re·we·- “exhale dangerous/poison breath (AI)” and derived noun *re·(we·)wa “hognose snake (Heterodon spp.)”(?) (cf. common English names for the hognose like “puff adder” and “blow snake”)? E.g., Shawnee leewe “she (menstruating woman) has poison breath” and leewa “cottonmouth (Agkistrodon piscivorus); copperhead (A. contortrix),” Menominee nε·wε·w “hognose” [“blowing snake”], Old Illinois areewa “bullsnake (Pituophis catenifer sayi)”(?) [“serpent qui siffle b[eaucou]p”] (original gloss from Pentland 2023, suggested species identification following him and Costa 2001:261), etc. ↑
17. *wi·ʔθenyi- is only the main AI term for “eat” in the Core Central languages, but it does have a few cognates elsewhere, e.g. Unami |wīxənī‑| “cook (AI+O)” (Rhodes 2021:319-320, 343, n. 29). The precise original meaning is unclear to me. ↑
18. Most of these points apply as well to *‑(e)sahan “EMPH.PRES,” except that it drops a preceding *‑t “3AN.CONJ” entirely (Rule 8c), rather than retaining it with no epenthesis (cf. the third example under Rule 5a). But some daughter languages appear to point to *‑(e)sahan having an underlying, not merely epenthetic, initial *‑e-, and/or having different vowel contraction outcomes (see section on contraction with mode suffixes below), hence the difference in notation employed here. ↑
19. Cf. Unami nkála·p from previous example, and wenči·má·p·ani·k “they were invited” (Goddard 2021:90, ex. 4.69n) with verb |wēnčīm‑| “invite (TA),” implying ‡wenčí·ma·p (|wēnčīm-ā-w-əp| → |wēnčīm-āp|), versus wenčí·me·p “he (PROX) invited [many people] (OBV)” (ibid., ex. 4.69a) (|wēnčīm-ā-ẅ-əp| → |wēnčīm-ēp|) (cf. Goddard 1979b:173). ↑
20. At least, this is the case given how I’ve chosen to structure the rules here. Alternatives are possible. In fact, the more common presentation is to treat all vowel hiatus rules together, and say that epenthetic *‑y- is inserted in several contexts, including between any two long vowels (Rule 11 here); a short vowel is lost next to a long vowel in either order (here such a process is covered in multiple rules, 1a and 12); and in any remaining sequence of two short vowels, the second one is lost (essentially Rule 12 here), except in the person prefixes of dependent stems, where the first short vowel is lost (cf. Bloomfield 1946:93, 101, and [for Menominee] 1962:82-83 §§4.16-21, 97-98 §§4.71-72; Pentland 1979:39, 401, 406; Oxford 2016:10; Goddard 2023:17-18 [for Meskwaki]). (Proulx’s [2003:222, n. 4, 2005:197 and n. 5] version is very different.) Pentland is the only one to refer to the necessity of a rule deleting the prefix vowel in dependent nouns even when the following vowel is short (Menominee and Meskwaki no longer have such a rule). In an attempt to combine it with the process that, everywhere else, deletes the second of two short vowels, his formulation of the rule is somewhat strained: when two short vowels are in hiatus, “the one farther from the centre of the word is lost” (Pentland 1979:406), where the “center of the word” just means the word’s Initial (cf. Pentland 1979:38).

    I prefer, however, to capture the processes which apply to the person prefixes in a single, coherent (set of) rule(s) (1a and 1b), even if this comes at the expense of nowhere having a single rule of the sort “V̆ → Ø /__Vː, Vː__”. I believe the rules as presented here are elegant and parsimonious enough. Pentland’s alternative requires one to treat the equivalent of my Rule 1a as essentially two completely unconnected processes: (i) a short vowel is lost next to a long vowel, and (ii) a short vowel is deleted before another short vowel if and only if the first vowel is in a person prefix added to a vowel-initial dependent noun stem. Which strikes me as less than ideal.

    Separately from this question, note that some daughter languages extended the Rule 12 pattern (V₁V̆₂ → V₁) to apply to the person prefixes as well, when these precede dependent stems beginning in a short vowel. In essence, such languages drop the equivalent of Rule 1a, thus further regularizing the shape of the prefixes and making the relationship between the underlying and surface forms more transparent. For an example, cf. Plains Cree katay “your (sg.) belly” (|ki‑atay| ← PA *katayi *|ke‑atay-i|) with dialectal Moose Cree kitay (now simply underlying |ki‑tay|, cf. the 2sg prefix ki- found elsewhere in the language). ↑

21. Note ⟨käkĭkwäwaka⟩ keehkikweehwaka “I cut (off) his neck, as with an axe” (David Costa, p.c.), Old Illinois ⟨nikiskicȣehȣa⟩ †nikiiskikweehwaa “[je luy coupe le col] avec une hache, un sabre.” ↑
22. “Stone, rock” is one of several words or word classes which different Algonquian languages, and even dialects within the same language, assign to different genders, so it’s not entirely clear if it was animate or inanimate or dialectally both in PA. Bloomfield was actually inconsistent in his reconstructions: he gave the plural form as inanimate ⟨*aqsenyali⟩ = (properly) *aʔsenye·ri (1946:86, #8) but the singular as animate ⟨*aqsenya⟩ = *aʔsenya (1946:93, #87). Perhaps, as in some daughter languages like Meskwaki (Goddard 2002:214, 216), “stone, rock” was generally inanimate, but was animate when it referred to special, unusual, important, and/or more specific kinds of stones/rocks, and this latter usage was extended to treating the noun as animate in all instances in some other daughters. ↑
23. This suffix *‑ntay is in the same slot as the formatives, and like them presumably derives from a pre-PA nominalizer—perhaps in part *‑n, or whatever nominalizer the formative *‑n‑ay was once equivalent to, since its *‑ay- is sometimes absent (see Rule 29) (Pentland 1988:47, 49)—plus extra material; though Goddard (2007:263-264; cf. 2015b:404, n. 71) compares it with the nominalizer *‑(e)ntwiy.

    While it seems evident that it already existed as a verbal suffix in Proto-Algonquian, as with the “emphatic present” and “emphatic past”/“preterit,” reconstructing the original functions and determining an appropriate gloss for *‑ntay is a challenge. It appears likely to have been some sort of deontic or irrealis with (at least situational) future reference. See James (1991) for some discussion (and cf. Goddard 2007:263, n. 70; and Goddard and Bragdon 1988:568ff; Goddard 1993:231-233, 2004:121). For our purposes I will gloss and translate it in PA reconstructions as an optative, but this should be considered very uncertain. ↑

24. As already alluded to, Pentland (1999 passim, esp. 246-247) denied that the *e· → *a(·) umlaut of Rule 16 existed. In his view, all stems reconstructed by everyone else as ending in *‑e· really ended in *‑a·, and underwent umlaut to *‑e· in both (1) the environments noted in Rule 13a, and (2) the opposite of all the environments defined by Rule 16. He thus would view the transitivizers *‑θ/*‑t, nominalizer *‑ʔm, and so on, as conditioning no change in the preceding vowel, but all of the suffixes before which *e· appears (MARK.OBV.SBJ *‑ri [Pentland 1999:238], etc.) as conditioning *a· → *e· umlaut. While historically this is probably largely correct, it would only apply to a rather ancient stage of the language; it is hardly correct as a synchronic description of the PA facts. See Goddard (2007:255-257) for a refutation of Pentland’s position. ↑
25. [Deleted, but number kept to preserve ordering.]
26. Pentland (1999:249-250) gives *eθo·hikani “pointer, indicator; index finger” as an example in which coalescence has not been reversed by languages like Cree which normally do so (e.g., Plains Cree itōhikan, Menominee eno·hekan), but this is problematic. The word breaks down as some Initial meaning “point (out)” + *‑ah “by tool (TI Final)” + *‑ikan‑i “NMLZ”; Pentland identified the Initial as being of the shape *eθw-, hence *|eθw‑ah‑…| → *eθo·h‑…, but while this is almost surely correct on a deep enough time level, I don’t see the evidence for reconstructing underlying *eθw- by Proto-Algonquian. Instead, all of the cognates Pentland (2023) has cited for this Initial, some of which I find of dubious status, begin in *eθo·-. Synchronically by PA, it is highly likely that speakers would have considered the Initial to simply be *eθo·‑, with any following short vowel being dropped by Rule 12.

    One significant confounder for this general topic is that most Algonquian languages lengthened an *o in the second syllable of a word if the vowel in the first syllable was short, and the resulting patterns were then greatly disturbed by analogical adjustments (see this post). So, for example, the long *o· that appears to be reflected in “they sit in a group of three” could be analogically extended to this word because in many other words the Initial *neʔθw- “three” preceded an element beginning in *e, which coalesced to an *o that was eventually lengthened in most languages, resulting in a common allomorph of the Initial as post-PA *neʔθo·‑.

    This is not an issue for the specific example of *eθo·h-, since CINA was not affected by the *o-lengthening change. And since Innu reflects a long Common Innu *ū with “three” (e.g., Pessamit nishtussin /nəstussən/ “three pairs of shoes” ← Common Innu *ništūščin ← PA *neʔθw- “three” + *‑askesen “shoe”), it seems reasonable to project *wa-contraction with “three” + following *a back to PA.

    (“Common Innu” is used here to mean what can be reconstructed using the modern Innu varieties, not the oldest common ancestor of all attested Innu varieties including the Old Innu of Tadoussac—such a Proto-Innu would be significantly older and much more like western Cree [with, e.g., *ništūsk^yisini]; and strictly speaking, Common Innu as I define it never existed, since it contains some later innovations diffused among all modern Innu.) ↑

27. PA also lacked *Cyo(·) and initial *yo(·)- sequences, but to my knowledge there are no examples of synchronic inputs that would violate such sequences, which are then resolved in some manner. The constraint against initial *yo(·)- could also be described as a subset of the constraint against word-initial *y- in general (with only a couple of exceptions, always before *e·). ↑
28. Though another possibility, of course, could be dialectal variation in the protolanguage. Additionally, as Oxford (2015:320-321) points out, because under any reconstruction this was a total phonological neutralization of the two phonemes, some amount of variation in surface phonetic realizations would have been possible for a speaker; he represents the merged vowel as simply *[i ~ ɛ] in all environments, later resolving to *e → *ə everywhere in PEA, *i- vs. *Ce- in Meskwaki, Arapaho-Gros Ventre, etc., and so on.

    (Note that there are issues with this view, though. The variation in the daughters is not ultimately as extensive as Oxford implies. All Eastern languages point solely to *e, and all non-Eastern languages are consistent with *i-/*Ce-, which is either preserved as such, modified slightly, or obscured by later mergers but can be recovered by the dialectological picture. If the neutralized vowel ever actually did surface as *C[i]- in PA, there’s also the chance this would have triggered palatalization of initial *t- and *θ- by Rule 25, which didn’t happen.) ↑

29. Cf. 2sg e·xkwé·ian “you who are woman” (Goddard 2021:87, ex. 4.66a; a “representative singular”) = |ē\axkwēwī-yan‑Ø| = REL\be.a.woman.AI-2sg.CONJ-NON3.PART (|axkwēwī‑| ← *|əxkwēwī‑|). ↑
30. Note that the Munsee ə here is not significant. Due to several sound changes in PEA, it, like other EA languages, inherited a large number of verb stems in an |‑ī| which umlauts to ə before any suffix of the shape w. At least synchronically, that is the case with “say {so}” too: the stem is |ī‑|, and umlauts to ə before third-person ‑w in this example—even though the regular Munsee reflex of a PA or post-PA *ewa would also be əw (under certain prosodic contexts). ↑
31. The treatment of *|‑am| when the MARK.OBV.SBJ suffix *‑ri followed is not totally clear, but most likely an epenthetic vowel was inserted by Rule 9a, as discussed here. ↑
32. All biblical citations for Shawnee are to Alford (1929), via the redaction of Schaefer (2017). The corresponding English text is the English Revised Version. ↑
33. Many of the other portmanteau suffixes could appear to incorporate 3OBJ *‑a·, but when another morpheme intervenes, the result is theme.sign+X+portmanteau, showing that the theme sign is simply deleted whenever directly preceding those portmanteau suffixes, as it is before all conjunct central suffixes (see Rule 10). (Though it’s impossible to test this for certain imperative endings, which are therefore ambiguous on the question—though I treat them as having a deleted underlying separate *‑a·.) ↑
34. And cf. Kansas Odawa ⟨utchíbam⟩ “spirit” = ojiibaam “h/ spirit, soul, etc.” (Gatschet 1887:55; with o- “3” instead of n- “1”); and Wasauksing Eastern Ojibwe ⟨udjibbom⟩ “a shadow” (= the same word) (Jenness 1935:18). ↑
35. Incidentally, this points to another environment in which PA *aye did not contract. *Ca-reduplication of stems in initial *e- (or *i- [see Rule 23], but the outcome would have been the same regardless of how initial *|e| ~ *|i| was realized) was the same pattern as that which underlay much of initial change: *a- with no onset was added to the stem, and an epenthetic *‑y- was inserted to break up the resulting vowel hiatus by Rule 11(ii). For example: Munsee ayəlíixsəw “he speaks the same language [as his father]” (from *ayereniʔsesiwa ← *ereniʔsesi- “speak the normal, ordinary language (AI)”) (Goddard 2014:142), Ojibwe ayinaabi “s/he looks about” (from *ayeθa·piwa ← *eθa·pi- “look {somewhere} (AI)”). Such reduplicated forms in *ay~e- contrasted with initial change in the identical environments: *eθa·pi- → reduplicated *ayeθa·piwa “s/he looks about” (no contraction) vs. → initial-changed (iterative) *e·θa·pičiri “whenever s/he looked ({somewhere})” (fossilized, though not synchronic, contraction of *|a‑y‑e|). In PA itself, there was never contraction over reduplication boundaries. ↑
36. Cf. ⟨kinepesȣmina⟩ kinepihsoomina “we (incl.) aren’t dead, don’t die” (Costa 2003:453), from nep- “die (AI).” ↑
37. Menominee underlying |ε·w‑E| is realized as ya· (= roughly [iːa̯ ~ iːə̯]) in long syllables, but this is a more recent development and doesn’t reflect any old contraction; as Goddard (2001:206, n. 73) notes, “within a comparative context it is a case of non-contraction.” ↑
38. Cf. wīhtamawihk “(you pl.) tell h/!” (Ellis 1983:386); but note MCOD’s contracted Moose Cree wīhtamāhk “id.” and wīhtamāhkok “(you pl.) tell them!”, plus some analogical forms (citations slightly modified). Likewise, Plains Cree, or at least older and/or “High” Plains Cree, had or sometimes has recontraction to ā here, as it generally does almost everywhere, although sources (e.g., Itwêwina) lack it in this case. So Cree decontracted awi is conceivably a later innovation from an intermediate recontracted form, not an unbroken continuation from PA *awe? Because (re)contraction is the norm in Cree, however, this would be a bit surprising. ↑
39. The very few examples of putative *e·we → *o· to be found in Pentland’s dictionary mostly don’t strike me as convincing. For example, he reconstructs (Pentland 2023:2481) “it is maggoty” as *o·sko·wi [substituting *s for his *x], derived from *o·skwe·w- “maggot” + *‑iwi “be (II Final)” (→ intermediate *o·skwe·wewi by Rule 14(i)), but of the six daughters he lists reflecting a cognate (more could be added: note Mass ⟨ꝏhquou⟩ ⫽u·hk^wa·w⫽ “it [the manna] bred worms” [Exod. 16:20; Siebert 1975:408]), all reflect *e·we except for Munsee, which has ohkóowəw “there are maggots” (O’Meara 1996:211). But Munsee has also reshaped “maggot” from expected ^xóhkweew (cf. Unami ú·kwe·) to the shorter óhkw for unclear reasons, while the Delaware AI/II Final for “have an attribute” is |‑ūwī| in Unami (Goddard 2021:152-153) and frequently -oowii in Munsee (cf. O’Meara 1990:177-178; -oowə before /w/). Ohkóowəw = ohkw-oowii-w is thus the entirely expected Munsee form for “there are maggots, it’s maggoty, etc.” based on the morphemes and morphophonemic processes within Munsee itself, explainable without recourse to any ancient contraction pattern. Cf., e.g., ponkóowəw “it’s dusty” from pónkw “dust” (← PA *penkwi). Though Pentland does suggest (s.v. *o·xkwe·wa) that the shortened Munsee form for “maggot,” óhkw, could be backformed from ohkóowəw in the first place, which is worth consideration, and if correct would then strongly support reconstructing the verb in PA as *o·sko·wi, with contraction. ↑
40. Note that in this section only, for maximum clarity I’ll be citing PA verb stems ending in this Final as ending in *‑yayi and *‑ayi. Properly, these are underlying forms, which never would have appeared with these shapes on the surface. Elsewhere in the post I cite such stems as ending in *‑ya· and *‑a·, corresponding to their realization when inflected in the independent order, at least before third-person *‑ẅ. (Traditionally, such verbs have been given in the standard citation form, which is 3sg independent, and thus as ending in *‑ye·wi or *‑e·wi, because this was assumed—by most—to be the correct reconstruction until quite recently.) ↑
41. Cf. hoowuniʔoo “it’s not good” with niʔoo- “be good (II)” (*wi·nka·- ← pseudo-PA *wi·nkya·‑) and negative preverb |ihoowu=| (Cowell and Moss 2008:390). ↑
42. Thanks to Strachey’s awful handwriting, the reading of this word is uncertain. I’ve very tentatively given it as ⟨Oussawaik⟩, following Wright and Freund in Strachey (1953:207), but others have read it as ⟨Oussawauk⟩ (Siebert 1975:409), ⟨Oussawack⟩, ⟨Oussaweuk⟩ (Rountree n.d.:[87]), and more. Even if Strachey originally wrote something like ⟨au⟩ or even odder, the likelihood is that the vowel in question was something like /ɛː/. ↑
43. Instead, he reconstructed the stem-final vowel of verbs of possession like “have ears” as *a·, as Goddard and I do, but with this undergoing standard umlaut to *e· before *‑ẅ (e.g., *wehtawake·wa [with *we for my *o] for “he has ears,” p. 3362), rather than remaining as non-umlauted *a·. Without more detailed argumentation than is available from his terse notes, I can’t know what his precise reasoning was beyond his obvious aversion to reconstructing diversity in the forms and properties of stem-final vowels in PA intransitive verbs—in essence, projecting a much earlier pre-PA predictability of umlaut forward into PA itself (cf. footnote 24).

    But he would have needed to explain how some dialects of Ojibwe and CINA acquired non-umlauting *a·, when this is—as Goddard (2007:257; cf. 1979b:65) puts it—a “recessive” pattern, with original non-umlauting stems (in languages and dialects which retain umlaut) tending to expand umlaut to such stems. In the absence of good evidence to the contrary, when an archaic, recessive pattern is found in some daughter languages and a highly productive pattern in others, the archaic and recessive one should be reconstructed for the protolanguage, and this is the case with contraction of *ayi → *aye → stable *a· here.

    Note too that this would seemingly require Pentland to postulate a true, historical direct *ayi → *a· contraction process—to explain why the stem-final vowel in these verbs ends up as an umlauting *a· (*ayi to *a· in most cases but to *aye → *e· before *‑ẅ by Rule 14? or simply *ayi to *a· everywhere, later umlauted before *‑ẅ)—which has no other support. ↑

44. Note that the form of *‑inay has been somewhat reshaped in most non-Eastern languages, at least in many cases, to *‑inaw or *‑ina·w. There are several factors which may have played a role, depending on the language (see e.g. Goddard 1995:125-126 for Meskwaki), and most likely this was in part backformation from the recontracted locative *‑ina·nki as Goddard (2007:260) and Pentland (2023:52, s.v. *‑inay‑) suggest.
    • *o·te·- “dwell as a group (AI)” + *‑inay “COLL.NMLZ” → *o·te·nayi “town, village” (e.g., Ojibwe oodeenaw and oodeena |oodeenaw‑|, Plains Cree ōtēnaw, Southern East Cree uteenaaw; cf. EA: Munsee ootéenay, Loup 6 [= Pojassick?] ⟨ȣtannai⟩ ⫽o·ta·nay⫽ [Goddard 2016:122])
    • But note no reshaping of the ending in a western language in: *api- “sit (AI)” + *‑inay → *apinayi “communal sitting and sleeping platform” (e.g., Meskwaki apinayi, Miami-Illinois (a)pinay “bed”; cf. EA: Munsee ăpíinay “id.”)

    One other apparent innovation in Meskwaki is that ‑inaw “UNSPEC.POSSR” contracts as if it has an initial |e|, but then “the full suffix |‑inaw| also appears uncontracted after the |Vn| that results from contraction,” i.e. it’s actually treated as if it were underlying |‑eninaw| for the purposes of contraction (Goddard 2023:110). This irregularity is not found in the collective locative ending, or in other languages, and is pretty clearly a later repair process to retain the full uncontracted shape of the suffix. ↑

45. Only one noun or verb stem ended in *‑e·y: *mye·y- “excrement, feces.” Cree would appear to indicate that it did not contract (e.g., Plains Cree mēyihk “like dung,” with locative -ihk, as if from PA uncontracted *mye·yenki), but this is hard to say with certainty; not many daughters reflect this stem unmodified. ↑