Emily Dickinson's poetic corpus offers evidence that she creates ambiguity and with it a variety of interpretive possibilities for the reader. By simple elision, most of which is licensed by syntactic recoverability rules, she creates indeterminacy. In instances of nonrecoverable deletion, readers are forced, consciously or unconsciously, to supply uncertain details. There also exists a number of instances of recoverable deletion for which alternative unlicensed substitutions are possible. We are better able to read Dickinson's poetry, then, if we understand the ways in which she elides words and phrases.

This understanding shows us the possibilities for recovering or inserting those bits of language that standard grammar allows us to identify as missing. By recovery, I mean the recognition of unique and unambiguous elided words or phrases. Insertion, on the other hand, is the speculative addition of words or phrases to lines in which the syntax gives strong clues of grammatical ill-formedness caused by elision.

This study will look at elision both through an examination of syntactic categories and a consideration of rules that Dickinson appears to have formulated and that operate throughout her corpus. My first contention is that, while specific recovery is often not possible in her poetry, Dickinson's elision accommodates our grammatical expectations. In places where we cannot identify specific elided elements, we can often identify their grammatical categories. A subsequent contention is that Dickinson's 1,775 poems show patterns of elision from which we can derive some guidelines for making guesses as to what should be inserted into syntactic positions that she appears to have left empty.

A number of studies have addressed this question in the past, but the consideration of elision in Dickinson's poetry is a far larger concern than those studies or this can adequately treat in a few pages. Still, for the sake of orientation it is helpful to review the discussion thus far. The question of how to deal with elision is no less interesting now, but our ideas concerning the treatment of poetic syntax are quite different. Few people, for example, discuss grammatical rules for the generation of poetic discourse any longer.

In 1971 Samuel R. Levin, in his article "The Analysis of Compression in Poetry," first proposed that we could recover deleted elements of poems like many of Emily Dickinson's, elements that seem to have been removed by "compression," the apparent elision of words and phrases that we would expect in normal syntax. In 1972 Eugene Kintgen attacked Levin's argument in "Nonrecoverable Deletion and Compression in Poetry"; the article appeared with a reply by Levin. Five years after that, Frank L. Coppay took his turn at Levin in "The Internal Analysis of Compression in Poetry."

Levin begins by assuming the existence and availability of a poetic competence: an ability to read, understand, and possibly even generate poetic utterances, as well as an ability to say what is and is not consistent within a poetic structure ("Analysis of Compression" 39). This poetic competence is akin to linguistic competence, our unconscious knowledge of grammar and the resulting ability to generate acceptable sentences and understand language. Given this poetic competence, readers would be able to recognize deleted constituents and recover them according to the rules of the poetic grammar. This intuition, Levin says, has "presystematic validity," informing the reader whenever the poetic structure includes deleted elements ("Analysis of Compression" 40).

The effect of these deletions induces a reader to believe that compression is present. Levin, however, refuses to be constrained by rules that permit recovery only when the deleted elements can be recovered uniquely and unambiguously. (Such rules are universal in discussions of nonpoetic syntax.) Levin says instead that "the grammar and the speaker must coincide in their ability to interpret sentences" ("Analysis of Compression" 41). Given the sometimes unusual character of poetic syntax, one must be able to account for structures that would otherwise seem uninterpretable by the grammar of the language proper. This notion of a "poetic competence" derives from earlier work by Levin himself, as well as works by Manfred Bierwisch, Roman Jakobson, M. A. K. Halliday, A. A. Hill, and Jan Mukarovsky, a host of the best linguists of the era. All would have supported Levin's assertion that the analysis of poetry ought not to be constrained by normal linguistic rules.[1]

Talk of "poetic grammars" was largely discarded with the recognition that conventional grammatical rules give us most of what we need for interpreting poetry without creating new problems. Yet, Levin was correct about poetic licensing, in this case, for the reader. We have to be able to flesh out compressed lines especially in poetry like Dickinson's. Strict constraints are not desirable when one is reading the work of a literary artist whose hallmark is ambiguity. Recoverability violations are necessary if readers are to explore the possibilities for explicating compressed lines. But again, this is insertion, a form of systematic recoverability violation and an interpretive device that is informed by the grammar but operates outside of it.

In a more recent discussion of Dickinson's poetry and nonrecoverable deletion, Christanne Miller explores ways of recovering "the complete (or deep) syntax" of cryptic lines (Grammar 28-29) by reinstating possible deletions, which s`e encloses in brackets.[2] (This is essentially the method that I will use for marking insertions.) Miller suggests that compression is so frequent in Dickinson's poems as to cause readers to approach all deletion with caution, "expecting indeterminacy and multiple meanings at every instance" (Grammar 29). She suggests that Dickinson uses nonrecoverable deletion to mask logical links between consecutive statements and stanzas (Grammar 29). This latter practice, Miller argues, has the effect of placing abstract and narrative statements together and of juxtaposing metaphorical contexts. The result of this "disjunctive" or `'coordinative" linking is to leave the task of creating thematic and logical associations to the reader, who, in doing-so, inherently exercises the possibility to which Dickinson refers in poem 657.

Instead of inserting speculative deletions, Levin's 1971 article explores a method of characterizing nonrecoverable deletions by a set of semantic features (for example, [Actional, + Contented], [+ Contented, + Sound] as a description of the verbs in the phrase "basks and purrs") ("Analysis of Compression" 43). But Kintgen suggests that we must limit such reconstructions to prelexical units, or else, he says, "we are ipso facto providing another deep structure for the poem, thus making it multiply ambiguous" (101-02). Here I agree, but I would argue that this ambiguity that Kintgen sees as an "embarrassing consequence"(101) is inherent in poetry that evidences deletion; it is in part the purpose of the elision. Levin's argument also falls short in that it fails to recognize that readers have a grammatical mechanism, not just a semantic one, for dealing with deletion. My analysis follows Levin's lead, but I isolate syntactic features within a category frame that names the salient syntactic and lexical features for a nonrecoverable deletion (in Levin's example above both verbs are intransitive and would be noted in a subcategory frame like [V: -transitive]).

These category frames, in keeping with Kintgen's suggestion, show only prelexical structures, and while the syntactic categories should be uniquely identifiable, the insertion of lexical items into these frames can vary as different readers flesh out "compressed" poems. When we start to see similarities and recurrences in the syntactic contexts of Dickinson's poems, we can draw inferences about the syntactic behaviors within her poetic corpus.

To begin with a single poem (1094), then, we can look at the following example from a poem in which elision occurs without license:

Themself are all I have--
Myself a freckled--be--       (lines 1-2; parenthetical references
                              are to lines in Dickinson's poetry)

In the first line we find a reflexive pronoun without an antecedent, a clear violation of binding rules within government and binding (GB) theory and a violation of well-formedness within any description of standard English.[3] (The fact that "themself" is an ungrammatical form, apparently assigned a singular marker by a mechanism to which we have no access, requires its own story, one that I will not attempt to tell here.) GB theory (though any account of standard English would do) dictates that each of these reflexives must be bound to an antecedent in the same clause. We know that this missing antecedent is a nominative plural noun or pronoun, and we can place it before "themself" because antecedents of reflexives must precede them in order to govern them. In the second line of this example, the head (the noun itself) of the second noun phrase, "a freckled -----," is missing; this phrase also violates well-formedness. We know the category, case, and (from the indefinite article) number of the missing word even though we cannot replace it with a unique lexical item. We know further that the missing noun must be coreferential with the pronoun "myself," since this second noun phrase is a complement governed by the copula. As well, the reflexive pronoun "myself" must be governed within its clause by a coreferential personal pronoun, which in this case is clearly 1, the first-person nominative singular form and apparently missing subject of the sentence. This last example is of a deletion that is recoverable even though it is not conventionally licensed.

The point of this analysis is simply that, even though these two lines include at least three deletions, two of which cannot be uniquely recovered (the possible noun antecedent of "themself" and the head of the noun phrase "a freckled ----- "), we are not entirely at a loss to know what might be inserted for the deleted elements. This category and feature recovery allows a certain (and no doubt desired) indeterminacy within the line. With insertions and recoveries in place, we might get the following result:

[NP: +nominative, +plural] Themself are all I have
[1] Myself [a Freckled [N: +nominative, +singular]] be

This much can be done by any account of standard English grammar, our understanding of which tells us that items of these categories must be underlyingly present. In one case, at the beginning of the second line, the analysis leads to unique recovery. In the other two cases, the subject of the first line and "freckled [noun]," we can speculatively insert items that are nonrecoverable. To do so we follow whatever clues are available from context and familiarity with Dickinson. It would be unwise, for example, to insert girl as the missing head of the noun phrase because Dickinson occasionally uses male narrators, but any noun that fits the semantic context of a line such as "[I am] a freckled [noun]" could be attempted.

The next level of analysis, then, is really explication and not recovery. From my reading of the poem I might attempt to flesh the lines out this way:

My freckles Themself are all I have
I, Myself, am a freckled child

I build these judgments on the following assumptions: (1) "Themself" is a nonstandard form, an error perhaps, and so might be accounted either to dialect or to the speech of a child; I choose the latter, partly on the basis that I think of freckles as belonging to children, and insert child for the missing head noun in the second line; (2) the adjective "freckled" means "having freckles," and since "themself" is coreferential with the underlying object of "have" in the lower sentence of the first line, I insert freckles, along with a determiner (My) in the subject position of the upper sentence of the first line and make it the antecedent of "themself." In these judgments I may very well be wrong. I am using grammatical criteria to inform my choices, but these choices hold no certainty of being true. The crucial point is that if another reader perceives that I am wrong, I am not wrong on the grounds of linguistic analysis, but rather wrong because I am not reading Dickinson well.

Still more remains to be said about compression in anticipation of some problems that readers of Dickinson's poetry might encounter. One unanswered question is, "When do we feel intuitively that something has been compressed?" Three things seem to signal compression: first, two parallel structures in which the second is an elided version of the first; second, an instance in which we recognize a necessarily incomplete phrase structure, as in "Myself a freckled--be--"; third, a semantic context that indicates strongly that some constituent is missing. Normally, the first of these cases has general license, as in parallel constructions that end with "too." The second is allowed less generally in everyday speech but is evidenced, for example, in substantive adjectives, of which "freckled" in the lines above could be an example, albeit an unusual one. The third is the one with which we are most likely to find trouble. What happens, for example, when our intuitions tell us that something is missing, but we cannot find syntactic evidence of its deletion?

This question is precisely the one that Levin attempted to answer in his first article when he examined "When Etna basks and purrs" (1146). In short, Levin undermines a good argument for compression by choosing a poem that gives no real evidence of deletion. Coppay points this out, and he is correct in suggesting that this intuition is semantic and not syntactic, but this possibility does not dismiss all (especially better-founded) arguments that hinge on syntactic competence in those cases where evidence of deletion can be shown. For this reason much of the original discussion among Levin, Kintgen, and Coppay is not about the question of recoverability at large; rather, it revolves around Levin's intuitions and his discussion of poetic intuition.

Levin begins his article by looking at the following poem and saying that he feels by his poetic intuition that something has been compressed:

When Etna basks and purrs
Naples is more afraid
Than when she shows her Garnet Tooth--
Security is loud--

Levin comes to the end of the poem and thinks that he has missed something; in fact, he has not, but he assumes that something has been deleted: "a fourth term, to make the comparison symmetrical" ("Analysis of Compression" 42) ("basks and purrs," and "shows her Garnet Tooth" [and ----- ]). He uses his semantic-features analysis to suggest that Dickinson's line might read, ". . .shows her Garnet Tooth and roars" ("Analysis of Compression" 44). However, both Kintgen and Coppay argue well that symmetry is in no way essential to our understanding of the poem. Kintgen points out that "basks and purrs" functions unitarily; the two conjoined words seem to be natural concomitants, just as the phrase "rich and powerful" opposes simply "poor" (100). Coppay, too, notes the functional equivalency of "basks and purrs" (18). Levin's response concerning symmetric and asymmetric as well as unitary and distributive conjuncts fails to provide syntactic evidence of deletion in this poem, nor does recovery of deletion seem essential to its interpretation ("Reply" 107-08).

In looking at parallelism Levin was on the right track, but we should appeal to parallelism only when we see conclusive evidence--a necessary and unfilled syntactic category--and such an environment may be difficult to identify. To illustrate this problem we can consider two other poems by Dickinson, one in which parallelism seems obvious and one in which it does not; in the first we can recover the deletions reasonably quickly and unambiguously, while in the second we need to stretch a bit. The primary difference between the two involves a question as to what it is that tips us off to the presence of deletion.

Consider first, then, the final four lines of"I can wade Grief-" (252):

Give Balm--to Giants--
And they'll wilt, like Men--
Give Himmaleh--
They'll Carry--Him!

Here we find two recoverable deletions licensed by syntactic parallelism. We see immediately that the first two lines are parallel with the third and fourth. There is no need to recover categories because we can recover the unique lexical items that are missing. After "Himmaleh" in the third line we can recover to Giants, the deletion of which is licensed by a parallel in the first line. We can recover And before the fourth line because its deletion is licensed by a parallel in the second.

Consider next a stanza from poem 495 that from its construction can be read as a single sentence:

It's thoughts--and just One Hean--
And Old Sunshine--about--
Make frugal--Ones--Content--
And two or three--for Company
Upon a Holiday--
Crowded--as Sacrament--

Our awareness of the rules for existential sentences allows us to insert "that" between the complements in the first two lines and the verb they govern, "Make" in the third line. Next, we might look at "about" in line 2; if it is a preposition, then an object noun phrase is missing. But if"about" were an adverb, there would be no missing noun phrase. Here, then, we know how to recover a category, but we do not know whether or not we should. An awareness of ambiguity suggests that we must make a choice. Likewise, in the fourth line "Two or three" can be read as pronouns. But if we read them as adjectives, they would suggest a missing noun, leaving us yet another choice. And still more ambiguity exists: does "Crowded--as Sacrament" modify "Holiday," or is it parallel with "Content," describing a feeling? A choice between these possibilities brings about a change in meaning.

What would happen, then, if we assumed that the last three lines were parallel with the first three? In that case, the stanza might be fleshed out with some certainty (though we still have to make choices):

It's thoughts--and just One Hean--
And Old Sunshine--about--[oneself]
[That] Make frugal--Ones--[feel] Content--
And [it's] two or three [hearts]--for Company
Upon a Holiday--[that make frugal ones feel]
[As] Crowded--as Sacrament--

Here I have chosen to insert a pronoun as an object for "about," but the sense of the lines would change little if "about" were read as an adverb. The rest follows directly from the assumption that the first three lines are parallel to the last three, an assumption that I made based on the frequency of elision licensed by parallelism in Dickinson's poetry.

In the face of such ambiguity readers have two choices: they may focus on syntactic (and subsequently, though less certainly, on lexical) expectations, or they may focus on possible interpretations (a set of semantic expectations) to tell them what to expect, since before they have a "sense" of the lines they may have an incomplete understanding of the sentence structures necessary to create that sense. A set of possible paraphrases may provide clues to the syntactic pattern that arises most naturally out of the lines. In either case, readers may be guided by an aesthetic sense.

While context makes recovery possible, it does not always lead to unique recovery. Separating syntactic from semantic context, though, helps us avoid the kind of problem that Levin found with reading "When Etna basks and purrs." A poem like the one below may have either, both, or neither of the two characteristics previously examined: parallelism and syntactic elision. Because recovery here is dependent on thematic considerations this poem demonstrates the limits of semantic context in solving problems of insertion. The meaning of the following poem (1206), for example, seems to hinge on definitions of"Show" and "Play":

The Show is not the Show
But they that go--
Menagerie to me
My Neighbor be--
Fair Play--
Both went to see--

If we say that the first instance of "Show" means a presented program, but the second means the real, the desired spectacle, then we can speculate as to the meaning of the lines. This assumes that the ambiguity here is lexical rather than syntactic. Either definition can be substituted for either occurrence of "Show" in the first line. This choice, then, leads us to an interpretation of the poem. The word "But" in line 2 provides a solid hint of syntactic parallelism.

If we make this assumption we can uniquely recover a parallel deletion in the second line (although "but" demands that we omit "not" in the recovery). The rest has to follow solely from semantic expectation:

The Show is not the Show
But [rather] they that go-[are the Show]
My Neighbor[s] [are]
[A] Menagerie to me--
[But it is all] Fair Play--
[We] Both went to see-[each other]

(I have reversed the third and fourth lines to normalize the syntax.) As in "I can wade Grief," Dickinson declines repetition. We might note that "Neighbor" in the original fourth line is used collectively, as one would find it in the Biblical setting "Love thy Neighbor," but I have made it plural so that it better agrees with "Menagerie," an argument to which it is bound by a copula.

Anything inserted in the fifth and sixth lines must depend on semantic intuition because of the ambiguity of the noun phrase "Fair Play," which could mean either "clean sport" or `'an attractive drama." I selected the former and made the choice to depict this noun phrase as the complement in an existential sentence. I could also have chosen the latter to provide a contrast, informed by what could be contrastive syntactic parallelism with the first line: the show is not the show itself, but what we see is an attractive spectacle. The first insertion of the last line involves an insignificant choice; one could choose freely among "[We] Both," "Both [of us]," or simply "Both," standing independently as a pronoun. The final insertion, [each other], is based on my impression that "see" is a transitive verb in this line. One could also read the lines this way:

[It was a] Fair Play--
[That we] Both went to see--

"Both" went to see something, and presumably the same thing, but because of lexical ambiguity we cannot say with certainty what they saw. This example indicates that possibility does not depend purely on "compression."

There is also a contextual question: to what extent does the pattern of discourse determine our processing of the message? This question brings us to a discussion of what Bierwisch would call "textual microstructure" and "macrostructure" (112-13). He equates macrostructure roughly with genre, though this equivalence is never overtly expressed, giving as examples the construction of fables and the interlacing of episodes. The textual microstructure, he says, includes the structural qualities that lie in the recording area of short-term memory; during the process of understanding, he says, they can be reconstructed. Perhaps a better distinction would be that the microstructure (in this case, the poem) includes that which is unique to individual texts, while the macrostructure (a poetic corpus) is a collection of those features that are found throughout a collection of related texts.

A question could then be posed concerning Emily Dickinson's poetry: are 1,775 poems sufficient to comprise a macrostructure? Given this number of texts linked to each other by common features-elision and ambiguity, for example-we might expect one poem by Dickinson to inform our reading of another. Certainly those who have read the corpus of her poems have had a greater opportunity to discover the features of Dickinson's poetry (and possibly intuit rules that Dickinson has invented for her poetic syntax) than those who have read only a few poems. Subscribing to a notion like macrostructure would allow us to describe Dickinson's poetic idiolect in terms of a feature set, which would be more accurate than describing it in terms of a grammar. Through the experience of a specific task such as reading Dickinson's poems, we intuitively identify a set of macrostructural features.

An example of this intuition might be found in a pair of poems that appear to be constructed in parallel fashion with each other. "I can wade Grief--" bears a strong resemblance to "Give little Anguish--"(310) Each poem exhibits recoverability by parallelism. Here again is the recovered version of the former:

Give Balm--to Giants--
And they'll wilt, like Men--
Give Himmaleh--[to Giants]
[And] They'll carry--Him!

Compare these lines with the first five of the latter poem:

Give little Anguish--
Lives will fret--
Give Avalanches--
And they'll slant--
Straighten--look cautious for their Breath--

Both poems use the same construction toward the same end. A comparison between the giving of relief and pain achieves the same result in each case: relief brings wilting and fretting while a heavy load brings strength and straightness.

If we want help in recovering or inserting deletions in the second poem, we can look at the first for clues. For example. we sense from their syntactic similarity that in the second poem a prepositional phrase is missing in each of the first and third lines, and a conjunction is missing in the second. Since "Lives" replaces the pronoun "they" in each of the other lines parallel to it, we can assume that to lives or something equivalent to it should be this missing prepositional phrase. With their deletions recovered and some speculative insertions added, the lines of the second poem might look like this:

    Give little Anguish-[to lives]  [to Giants]  [to
      people]  [to them]
    [And] Lives will fret--
    Give Avalanches [of Anguish] [to lives]  [to Giants]
        [to people]  [to them]
    And they'll slant-[themselves] land then]
    Straighten [themselves] - [and] look cautious for their
      Breath--

If we subscribe to the notion of macrostructure? the insertion of to lives or to people in line 1 is in some sense recoverable by parallelism with the earlier poem in which the syntactic object, "to Giants," is overt. The syntactic categorization frame would look like this: [PP: + [NP] + [Pro]]. Then, to use a feature set like Kintgen's, any noun phrase that has semantic characteristics like [+human-like, +large, +strong] is contextually appropriate and could be suggested for insertion. The contrast of "fret" ("ripple," "interlace") in line 2 with "straighten" in line 4 gives further evidence of parallelism within the stanza. The rest of the insertions are superfluous.

Another problem arises here. What happens, for example, when insertions are made into the first structure of a parallel string? Can they license recoveries in later parallel structures? Of course, the answer is no, but this is precisely the effect that insertion in parallelism has. The "recovery" of of Anguish and to people in the third line is motivated by parallelism with prepositional phrases in the first line, but neither is actually licensed. Even though "Anguish" is overt in the first line, its insertion in a prepositional phrase is based on its speculative association with "Avalanches."

Another macrostructural comparison involves lexical ambiguity in Emily Dickinson's curious use of if clauses. A short poem (1653) will demonstrate that identical clauses can have different meanings in different contexts:

As we pass Houses musing slow
If they be occupied
So minds pass minds
If they be occupied

In the second line of the poem "If" has a conjunctive function (whether), while in the fourth line "If" works to establish the subjunctive mood (when), a conditional situation. This example points out a problem for assumptions built on parallelism: here, two syntactically parallel elements are not semantically parallel. Thus, while syntactic parallelism might yet provide clues, we should be aware of the necessity of interpretative insertion where items appear to have been compressed.

Based on this assumption, that some other information has been compressed into the "if" terms and that the information they contain differs, we can attempt speculative insertions:

As we pass houses musing slow
[Whether or not] they be occupied
So minds pass minds
[In the event that] they be occupied

If this interpretation is well founded, we might take this poem to be a clue to interpreting other poems in the corpus as well as a warning that, throughout the corpus, we need to look for shifts in the definition of terms set parallel syntactically. To see if this key fits, we can look at another poem (1665) with similar if clauses in parallel positions:

I know of people in the Grave
Who would be very glad
To know the news I know tonight
If they the chance had had.

'Tis this expands the least event
And swells the scantest deed--
My right to walk upon the Earth
If they this moment had.

Here the previous ordering of the if terms (conjunctive, subjunctive) is reversed, but one possible reading of this poem would allow the same interpretive insertions as in "As we pass houses" to be substituted in place of"If":

[Assuming that] they the chance had had
[Whether or not] they this moment had

If an examination of macrostructural features can provide a clue in an example such as this, then it could yield similar results in other cases involving both recovery and speculative insertion throughout the corpus. This case also indicates that certain lexical items or syntactic constituents can be collapsed in a way that leads us to "feel compression" when in fact there is nothing to be recovered. This situation should cause us to question what "compression" in poetry really is, as well as to speculate on the possible poetic effects of compression.

At the most general level we can say that a poet will use language with attention to the response that he or she wishes to elicit from the reader. As Michael Riffaterre says, "The author's consciousness is his preoccupation with the way he wants his message to be decoded" (157). While Riffaterre describes the ways in which the author limits the reader's freedom of perception, we can say conversely that Emily Dickinson controls the medium in such a way that it cannot unambiguously be decoded. She forces the reader to try out various decodings and thus to see multiple interpretations and opportunities of perception.

When elements that cannot be uniquely recovered are elided from the poems, possibility and ambiguity become an inherent part of attempts at decoding. Moving to the more specific effect, Jakobson says, "Ambiguity is an intrinsic, inalienable character of any self-focused message, briefly a corollary feature of poetry" (370-71). The layering effect of similarity and continuity, the levels of analogy, the overlap of schemes or tropes, the layers of discourse (poet, persona, and indirect speech) all are causes of ambiguity as a poetic effect. Thus, if ambiguity exists and the poet's function is to control (or decontrol) the decoding of the poem, ambiguity can be controlled (or decontrolled). Dickinson decontrols her poetry through "compression": the apparent deletion or conflation of selected lexical and syntactic elements within her poems to allow the possibility of perception.

Finally, to impose a more subjective view of poetic language, we can consider the function of poetic language that Hill describes. To distinguish between poetic and ordinary language, Hill differentiates between (a) an individual's microlinguistic or preliterary world; (b) style, the microliterary world itself, and (c) the metalinguistic, metaliterary world of things: what we call objective reality (387). Between what is objectively real and what has reality in ordinary language is style, a poet's unique set of features for creating a world. The poet's hope is that through poetic style reality can extend beyond the microlinguistic world. Dickinson's compression could be seen as an attempt to disrupt the microlinguistic consciousness and thereby force readers to reexamine their preliterary reality; for a poet like Dickinson, who has a strong concern for metaphysics, this suggestion seems entirely reasonable.

Emily Dickinson knew how some discourse deletes and recovers elements almost unconsciously. In another poem ( 1035) she imitates a prosaic letter and lampoons our habits of correspondence:

You'll get my Letter by
The seventeenth; Reply
Or better. be with me--
Yours, Fly                    (9-12)

These lines suggests an awareness of the compression that occurs when writers attempt to get the largest return from the fewest words or perhaps try to save money on telegrams. She might also be suggesting that readers routinely recover elisions without any awareness that they do so. Another poem (494, version 2) shows that Dickinson was aware of her economical habits. Again in a letter, she admits her habit of incompleteness and owns the puzzles it sometimes creates:

Happy-Letter! Tell Her--
Tell Her--The page I never wrote!
Tell Her, I only said-the Syntax--
And left the Verb and the Pronoun-out!      (2-5)

Everyone who reads Dickinson's poetry must eventually find the page she never wrote.

[1] Such discussions were numerous during the decades of the sixties and seventies. For the best examples, including those of the critics mentioned here, see Seymour Chatman and Samuel R. Levin's Essays on the Language of Literature and Donald C. Freeman's Linguistics and Literary Style.

[2] Also see Miller's "Dickinson's Language: Interpreting Truth Told Slant."

[3] For an explanation of binding rules see Noam Chomsky's Lectures on Government and Binding (183-84); a more simple account can be found in Andrew Radford's Transformational Syntax (362-95).

As there is not sufficient space to argue each of these examples, I will make only summary comments about each group of examples. A few of these elisions are uniquely recoverable, but as a rule I have not included grammatically licensed elisions. Many more elisions are recoverable from discourse context if not from syntactic context. Readers may disagree with some of these examples, since at times they reflect the way I read Dickinson's poems, but I believe that a grammatical justification exists for each one. I would emphasize again, however, that anything substituted for these category and phrase markers would be insertion, not recovery.

ELIDED NOUNS

These are primarily examples of noun phrases from which the noun itself has been elided. In some cases a good argument can be made that these phrases include substantive adjectives, but I include them because a macrostructural argument suggests that the ambiguity that we find elsewhere in Dickinson's poetry could be present here as well.

163  Tho' she wear a silver apron--
     I, a less divine [N] --                    (3-4)

163  Still, my little sunburnt bosom
     To her Rosier [N],                         (7-8)

218  Are there two [N]                            (2)

366  A sweeter [N] to obey,                      (18)

476  But since the last [N]--included
     both [N]--

476  As Children-swindled for the first [N]               (27)

483  A Wonderful [N]--to feel the Sun                      (7)

483  The Single [N]--to some lives.                       (16)

495  And two or three [N]--for Company--                   (4)

ELIDED NOUN PHRASES

Most of these are examples of missing objects for transitive verbs. In a few cases there are clauses without overt subjects. Some may argue that these lines read well enough without any speculative insertion of noun phrases. I would argue that these examples help show the range of possibilities for different readings.

172  And yet, [NP] as poor as I,
     Have ventured all upon a throw!                      (3-4)

172  Defeat means nothing but Defeat,
     NO drearier [N], can befall [NP]!                  (11-12)

267  [NP] Charged us to forget Him--                        (3)

280  And finished knowing--[NP] then--                     (20)

281  'Tis so appalling--it exhilarates [NP]--
     So over Horror, it half captivates [NP]--            (1-2)

286  A second more, [NP] had dropped too deep               (4)

289  Inviting to [NP]--                                     (5)

339  From flasks--so small--
     You marvel how they held [NP]--                    (14-15)

341  The stiff Hears questions was it He, that bore [NP]    (3)

347  That frightened [NP]--but an Hour--                    (8)

355  For deeming [NP]--Beggars--play--                      (8)

355  To lack [NP]--enamor Thee--                            (9)

393  A few--and they by Risk--procure [NP]--                (3)

433  [NP] Knows how to forget!                              (1)

438  Because she breathed against [NP]                      (3)

443  We came to Flesh--upon [NP]--                         (16)
     (or [Pro] = which w/ antecedent Errand?)

456     So well that I can live without [NP]--                 (1)

459     The Peace cannot deface [NP] (or [S]?)--               (2)

ELIDED VERBS

For the most part I have not included examples of missing copulas, but one may cite instances within Dickinson's canon for which more than one linking verb could be inserted.

163  Tho' she wear a silver apron--
     I, [V] a less divine [N] --                          (3-4)

196  Tim-shall [V]--if I--do--
     I--too--if he--[V]                                 (23-24)
     (both [V]s = die, 1. 21)

212  Least Rivers--[V] docile to some sea.
     My Caspian--[V] thee.                                (1-2)

281  How easy, Torment, now--                               (7)

284  As I--[V or VP] toward Thee--                          (3)

427  That every sigh--may lift you
     Just as high--as I [VP]--                         (18- 19)

438  Deny! Did Rose her Bee--
     For Privilege of Play
     Or Wile of Butterfly
     Or Opportunity--[V] Her Lord away?                   (4-8)

446  And could she, further [V], "No"?                     (12)

448  Himself [V]--to Him--a Fortune--(or) [V]           (15-16)
     (second preferred by scansion)

ELIDED AUX

Some of these are tense markers for perfect constructions. Most, though, are modal auxiliaries. Dickinson uses a great number of subjective verbs without models, and so some of these insertions might be unnecessary.

293  Without that forcing, in my breath--
     As Staples--[AUX] driven through                     (11-12)

307  His Name--[AUX] remain--                                 (9)
     ([AUX] = would)
     (there are many subjunctives of this type)

322  Deposed--at length, [AUX] [V:linking] the Grave--       (26)

327  The Meadows--[AUX] [V:linking] mine--
     The Mountains--[AUX] [V:linking] mine--               (9-10)

334  And [AUX] just sipped--me--                              (8)

373  [AUX] One accost me--                                   (24)

384  Bind One--The Other [AUX] fly                            (8)

469  The Territory Argent--that
     Never yet--[AUX] consumed--                            (7-8)

ELIDED ADVERB

These do not constitute a particularly interesting or numerous class of elisions, but they occur occasionally.

263     So [Adv] greater than the Gods can show,               (21)

445     And would it blur the Christmas glee
        [Sub. Adv.] My stocking hang too high               (17-18)

ELIDED PRONOUN

In this class it is most often the case that we are only able to spot a missing pronoun because we know from context which one to expect. Still, most of these instances are found in the same context as that of elided noun phrases.

242  No Lightning, scares [Pro] away--                     (8)
242  As if an Axle, held [Pro] --                         (16)
293  [Pro] Could dimly recollect a Grace--                (13)
360  A Book I have--a friend gave [Pro]                   (13)
400  To earn a Mine--[Pro] would run                       (5)
411  When [Pro] that you met it with before--             (23)
435  'Tis the Majority
     In this, as All, [Pro] prevail--                    (4-5)
466  Or Gold--[Pro] who am the Prince of Mines--           (5)
487  An Ample Letter--How you miss [Pro]--
     And would delight to see [Pro]--                    (5-6)

ELIDED DETERMINER

We know from any standard English grammatical paradigm that determiners must occur in these contexts. In most cases they are articles, but we do not know whether they would be definite or indefinite.

209  [Det] Leopard breathes--at last                      (4)
285  Winter, were [Det] lie--to me--                     (14)
400  Say--[Det] last I said--was This--                  (22)
427  Wear you on [Det] Hem--                              (9)
433  [Det] Globe did not teach it                        (11)
491  Show me [Det] Division [Pro] can split or pare--     (6)

ELIDED PREPOSITION OR PREPOSITIONAL PHRASE

I would guess that there are few of this type of elision because of the importance that prepositions have in marking grammatical function. An abundance of such elision might create too much possibility even for Dickinson.

398  The other side [Prep] the Block--                      (4)
410  And tho' 'its Years ago--[Prep] that Day--            (15)
439  But proves [Prep] us                                   (6)
473  Nor introduce--my soul--[PP]                           (6)
     (depends on definition of introduce)

AMBIGUOUS ELISION

Here are a few more examples for which it is difficult even to say what has been elided.

200  I stole them from a Bee--
     Because--[Prep] Thee--
     Sweet plea--
     He pardoned me!                                      (1--4)
     (because the phrase or clause containing Sweet plea is
     incomplete,
     we do not know whose it is)
243  [NP] [V] No Trace. . . (or)
     No Trace. . .dissolved as utterly. . .
     [NP] discloses just a Hue                            (9-15)
284  if All--is All--
     How [can] [NP or Pro] larger--BE                      (5-6)
325  All these--[NP] did conquer--(or)
     All these--did conquer [NP]                             (5)
339  Thy flower--be gay--
     Her Lord--away!                                    (21-22)
     (missing subordinator? missing [V:linking]? what explains
     the exclamation point?)

Elision, Recoverability, Insertion in Dickinson

352  But smaller bundles--Cram [NP or PP]. (or)           (8)
     But smaller bundles--[NP] Cram [PP].                 (8)
     Industrious until--[S]
     (or) Industrious until--
     The Thimble weighed too heavy--                    (8-9)

413     --but they say
     Himself--[Prep] a Telescope
     Perennial [-ly] beholds us--                      (11-13)
     (Telescope = apposition or means? Perennial = adj or adv?)

MISSING EXISTENTIAL (IT IS....)

Like missing linking verbs, these pose little trouble to our understanding.

433     (3)
439     (3)
452     (16 (I am . . .?)
470     (8) (That is. . .)
483     (13)

AMBIGUOUS CONTRACTION

306     Eternity's disclosure                                (13)
        (pose. or copula?)

[*]I owe a great debt to Joseph Moldenhauer and Mackie J. V. Blanton for their insightful criticism and copious commentary upon this manuscript.

Bierwisch, Manfred. "Poetics and Linguistics." Trans. Peter H. Salus. Linguistics and Literary Style. Ed. Donald C. Freeman. New York: Holt, 1970. 96-115.

Chatman, Seymour, and Samuel R. Levin, eds. Essays on the Language of Literature. Boston: Houghton, 1967.

Chomsky, Noam. Lectures on Government and Binding. Dordrecht: Foris, 1981.

Dickinson, Emily. The Complete Poems of Emily Dickinson. Ed. Thomas H. Johnson. Boston: Little, 1955.

Coppay, Frank L. "The Internal Analysis of Compression in Poetry." Style 11 (1977): 19-38.

Freeman, Donald C. Linguistics and Literary Style. New York: Holt, 1970.

Halliday, M. A. K. "The Linguistic Study of Literary Texts." Chatman and Levin 21723.

Hill, A. A. "Poetry and Stylistics." Chatman and Levin 385-97. Jakobson, Roman. "Linguistics and Poetics." Style and Language. Ed. Thomas A. Sebeok. Cambridge: MIT, 1960. 296-321.

Kintgen, Eugene R. "Non-recoverable Deletion and Compression in Poetry." Foundations of Language 9 (1972): 98- 104.

Levin, Samuel R "The Analysis of Compression in Poetry." Foundations of Language

------. "Reply to Kintgen." Foundations of Language 9 (1972): 105-12.

Miller, Christanne. "Dickinson's Language: Interpreting Truth Told Slant." Approaches to Teaching Dickinson's Poetry. Ed. Robin Riley Fast and Christine Mack Gordon. New York: MLA, 1989. 78-84.

------. EmilyDickinson: A Poet's Grammar. Cambridge: Harvard UP, 1987.

Mukarovsky, Jan. "Standard Language and Poetic Language." Chatman and Levin 24149.

Radford, Andrew. Transformational Syntax. Cambridge: Cambridge UP, 1981.

Riffaterre, Michael. "Criteria for Style Analysis." Word 15 (1959): 154-74.