PRESUME-tensing: an
analogical account 0. Introduction The present paper deals with a recent development affecting the vowel of word-initial unstressed syllables in words like presume, rejoice, decline, etc. Contrary to Nádasdy (2013), I argue that the possibility of a tense vowel in such positions is not lexically determined, but is conditioned by what I refer to as the ‘separability’ of the syllable in question from the rest of the word. I suggest that this factor correlates with the probability of PRESUME-tensing in a given word. The idea posited here can explain seemingly accidental gaps in the operation of PRESUME-tensing (e.g. secede), while it can also account for its presence in expected (e.g. recall) and less expected places (e.g. December).
1. Why the FLEECE-vowel? We cannot discuss PRESUME-tensing without first considering another development, HAPPY-tensing, which affects historically short word-final and prevocalic /ɪ/ and /ʉ/. As Szigetvári (2017) points out, this can be thought of as a restriction on the set of word-final vowels, from which now all short vowels (expect schwa) have been excluded. Regarding the quality of the resulting vowel, he writes: “When the *V# constraint came to exclude all short vowels, but unstressed ə, words like happy hápɪ had to be amended. The obvious option of simply lengthening the vowel was apparently not available, because this would have been an indication of a historical r, as if the words were happeer.” This explanation, however, fails to take into account that mergers between r-influenced vowels and ‘r-less’ broad vowels have already taken place, as in e.g. spar and spa. The motivation for the glide insertion can be better understood by considering the universal observation that mergers ‘come cheaper’ than splits. If /ɪ/ and /ʉ/ are affected in the same way, the outcome of a simple lengthening would have produced /ɪ:/ and /ʉ:/, thereby creating a new phoneme (/ʉ:/), different from CURE and TOUR, which correspond to historical /u/+/r/ sequences. However, by inserting a homorganic glide after the short vowel, existing sounds are produced in both cases: /ɪ/>/ɪj/ (FLEECE) and /ʉ/>/ʉw/ (GOOSE). Furthermore, as Szigetvári (2017) notes, “the same strategy is applied to loanwords ending in a short vowel in their donor language: e.g. Italian spagɛ́tːi > spəgɛ́tɪj, Polynesian tabu > təbʉ́w, French kafe > káfɛj, The Even Yearbook 13 (2018), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even. © 2018, Márton Jánosy
PRESUME-tensing: an analogical account 23
Italian putːo > pʉ́təw.” The *V# constraint has an important role in the operation of PRESUME-tensing as well: the final vowel of prefixes attaching to free stems (e.g. preheat, prearrange) are subject to the same constraint as the vowel of HAPPY words, although in the latter case it is reinforced by the prevocalic environment.
2. Morphological typology Nádasdy (2013) restricts his analysis to unstressed integrated prefixes based on Wells (2008), whose definition “excludes independent prefixes (meaningful productive morphemes added at word level), such as de#magnetize, re#write, as well as integrated prefixes (meaningless unproductive morphemes present at lexical level, usually attached to bound stems) if they become stressed due to some stress assignment rule, e.g., dé+monstrate, rè+pre+sent” (Nádasdy 2013). The present paper includes both independent and unstressed integrated prefixes, as well as a third category of words where no distinct stem can be isolated. Therefore, I will refer to the following categories: (1) Type A Definition: prefix + a stem that can appear in isolation Example: pre#heat, de#magnetize, re#arrange (2) Type B Definition: prefix + a bound stem (can occur with other prefixes, but not in isolation) Example: pre+sume, de+scribe, pre+empt (3) Type C (quasi-prefixed) Definition: no morphological boundary, no distinct prefix or stem,1 but the phonological shape of the initial syllable is identical to that of an existing prefix. Example: presidium, preamble, December As expected, prevocalic tensing operates “across the board” in all three groups. Thus, rearrange, preempt, and preamble all begin with /prɪj-/ regardless of what kind of boundary (if any) follows. However, in the preconsonantal environment, group membership decides whether tensing is 1
Though many words in this group have etymologically complex morphology.
PRESUME-tensing: an analogical account 24
obligatory or optional. The vowel preceding the strong boundary in Type A words is invariably tense due to HAPPY-tensing. Type B and Type C words show variation. Table 1 summarizes the possibility of variation in the three groups. Type A
Type B
Type C
<pre>#C
<pre>#V
<pre>+C
<pre>+V
<pre>C
<pre>V
example
preheat
prearrange
presume
preempt
presidium
preamble
tense?
always
always
potentially
always
potentially
always
Table 1: word-initial KIT-FLEECE alternations: pre-
As can be seen, most categories always have a tense vowel in their own right, either due to the prevocalic environment or due to the strong boundary and the consequent operation of HAPPY-tensing. The shaded areas show forms that have recently started vacillating.
3. Prefix separability So far, I have referred to morphological boundaries as categorical entities. The analysis presented here, however, requires the adoption of a gradient view of morphology. The approach I propose treats words as sets of concatenated units, much like what Bybee (2010) calls chunks. The strength of the morphological connection between two units is decided by the probability of the cooccurrence of the same two units in other words. In Bybee’s words: “The principal experience that triggers chunking is repetition. If two or more smaller chunks occur together with some degree of frequency, a larger chunk containing the smaller ones is formed.” (2010: 37) This also means that if a given chunk is likely to co-occur with several other chunks (e.g. word-initial pre-), it is less likely to be stored in a bigger chunk together with what it attaches to. In other words, the chunks in presume [pre][sume] are more separable (less likely to be associated on a higher level as pre][sume) than the chunks in secede se][cede as [se] co-occurs only with a handful of other chunks (e.g. seduce, select, secure). In contrast, [sume] is relatively more separable as a unit: it is present in several high-frequency words: e.g. assume, consume, presume, resume. Crucially, chunks are not necessarily associated with any semantic content. Separability is distinguished from the concept of
PRESUME-tensing: an analogical account 25
compositionality2 in that this notion can also be applied to words without an identifiable morphological base. Under this analysis, December consists of at least two more or less separable chunks, as [ember] also appears in September, and November3 and is therefore likely to be stored as an independent unit. On the other hand, pre- in presidium is unlikely to be separable from the rest of the word. Based on this, the following prefix-separability hierarchy can be set up: Type C words < Type B words < Type A words.
4. Tensing and the effects of separability We have seen that the initial elements affected by PRESUME-tensing have varying degrees of independence. Assuming that highly independent and highly frequent elements can influence other elements with similar phonological shape, it is easy to explain why bereave, secede, and elide are less likely to undergo tensing than presume, retain, and describe.4 In other words, the relatively large number of forms in the non-shaded columns of Table 1 (including the open set of Type A prefix # free stem constructions) can cause forms in the shaded areas to behave similarly to them. This means that a highly productive prefix like pre-, which can attach to virtually any free stem (pre-qualify, pre-board, pre-approve, pre-record), will serve as an analogical basis for less separable but similarly shaped forms such as presume and presidium. The more separable the prefix, the more likely it is to undergo tensing. Therefore, the hierarchy suggested in §3 is also an implicational relationship: if a speaker has a tense vowel in the initial syllable of presidium, they will have a tense vowel in the more separable presume as well. On the other hand, a speaker may have /ɪj/ in detract, but not necessarily in December. Taking this idea a step further, we can gain insight into the operation of other types of lax–tense alternations in stressless word initial positions. Thus far, we only looked at alternations involving the KIT and FLEECE vowels. Then 2 3
4
As described in e.g. Hay (2001). Note that November may also have a tense (full) vowel in the first syllable (/əw/ instead of /ə/), indicating that the word is indeed separable. Accordingly, there is reason to believe that September may also be chunked into at least two units; however tensing cannot take place because the first syllable is closed. This is based on the author’s intuition, since the LPD does not provide information regarding the frequency of the different variants, furthermore, as Nádasdy (2018) points out, it is inconsistent in the indication of the possibility of tensing: “Wells includes seamong those to be given with /i/, but in the body of the dictionary there is no trace of this: all se- words continue to appear with /sɪ-/ or /sə-/ (with the regular exception of Seattle /si/, where the prevocalic position triggers HAPPY-tensing).”
PRESUME-tensing: an analogical account 26
again, there are a number of other vowel pairs exhibiting variation of a very similar kind to what has been described as PRESUME-tensing. Tables 2–5 show that vacillating forms (indicated by %) belong to exactly the same categories as seen in the case of PRESUME-tensing in Table 1 (shaded columns). For instance, the prefix bi- seems to be productive enough (and therefore appears in a large enough number of words) to influence less frequent and less separable words (e.g. binoculars). Highly frequent items (e.g. bikini) remain stable as their frequency prevents them from being affected by analogical changes.5 prefix+free stem
example
prefix+bound stem
quasi-prefixed
<bi>#C
<bi>#V
<bi>+C
<bi>+V
<bi>C
<bi>V
bisexual
biannual
binoculars (%)
biathlon
Biafra (%)
biology
bidirectional
bikini (no variation)
bipolar tense?
always
always
potentially
always
potentially
always
Table 2: word-initial KIT–PRICE alternations: bi-
Words beginning with di- show similar variation:
example
tense?
prefix+free stem
prefix+bound stem
quasi-prefixed
<di>#C
<di>#V
<di>+C
<di>+V
<di>C
<di>V
digraph
dioxide
direction (%)
diagonal
dichotomy (%)
dietic
dimorphemic
digest (%)
dilemma (%)
digamma
dilate (%)
dimension (%)
always
always
potentially
always
potentially
always
Table 3: word-initial KIT–PRICE alternations: di-
5
In other words, frequency helps maintain irregularities as is the case with the irregular past tense forms in English. The most common English verbs have resisted analogical levelling.
PRESUME-tensing: an analogical account 27
prefix+free stem
example
quasi-prefixed
<pro>#C
<di>#V
<pro>+C
<pro>+V
<pro>C
<pro>V
pro-life
pro-active
pronounce (?)
–
prosciutto (%)
prooemium
pro-British
pro-American
prokaryote (%)
profound (%)
protrude (%)
Prometheus
pro-choice tense?
prefix+bound stem
always
always
potentially
always
potentially
always
Table 4: Table SCHWA–GOAT alternations: pro
As a prefix, pro- is comparable to pre- in productivity (both can attach to virtually any noun or verb), which means that Type A (columns 1–2) words greatly outnumber members of other types, therefore potentially causing them to resemble them more by having /əw/ in their initial syllable.
5. Conclusion I have argued that PRESUME-tensing can be explained in terms of analogical effects resulting from the influence of highly separable and highly frequent units on less frequent, but similar forms. The likelihood of tensing is dependent on at least two factors: the separability of the target unit and the size of the group of words serving as the analogical basis. The lack of tensing in the case of some prefixes (be-, se-) has been shown to be a natural consequence of their relative inseparability and infrequency, rather than an accidental gap in the operation of PRESUME-tensing. On the other hand, the approach presented here provides an explanation for the emergence of tensing in words whose initial elements resemble a highly productive suffix such as pre-. The problem in the focus of the present paper may involve prosodic, semantic, and orthographic aspects, which have not been examined, nor have been incorporated into the presented analysis.
PRESUME-tensing: an analogical account 28
References Bybee, Joan. 2010. Language, usage and cognition. Cambridge: Cambridge University Press. Hay, Jennifer B. 2001. Lexical frequency in morphology: is everything relative? Linguistics 39: 1041–1070. Nádasdy, Ádám. 2013. “PRESUME-tensing” and the status of weak /i/ in RP. In: Péter Szigetvári (ed.), VLLXX: Papers Presented to László Varga on his 70th Birthday. Budapest: Tinta, 363–381. Szigetvári, Péter. 2017. Strengthening in unstressed position: We happy? In: Péter Szigetvári (ed.) 70 Snippets to Mark Ádám Nádasdy’s 70th Birthday. Budapest: Department of English Linguistics, Eötvös Loránd University. Wells, John C. 2008. Longman Pronunciation Dictionary. 3rd ed. Harlow: Pearson Education. (= LPD3)
Márton Jánosy Eötvös Loránd University, Doctoral School of Linguistics janosym@gmail.com