BASIC ASPECTS OF MARKEDNESS THEORY
IN DEVELOPMENTAL LINGUISTICS

© 2000 by Orchid Land Publications

[updated 20001501, 20050809]

i

PRELIMINARIES

(see also L30.html [and L92.html])

     Marking theory is based, not on what is normal (i.e. statistically preponderant) but on what is natural in a clearly conceptualized sense:  What is natural (and in fact a default form or usage) in an unmarked context (where context can involve any component of the grammar, including stylistic matters like emphasis), and what is natural in a marked context), and what is natural in an overmarked context is defined implicationally:  

  Thus, in any position in a syllable in which we find //k// in a language, it naturally implies //t// in the same environment.  (Where this is not so—Hawai'ian lacks //t// altogether, as //t// changed to //k// when //k// became a glottal stop--we have to look for a higher-level phenomenon (see below) that overrules the implicational prediction.   Feature values in markedness theory in the developmentalist framework are three-valued (ternary); there is usually a neutral value between the extremes.  (The values are given in an Appendix to the Reference Minigrammar in the volume referred to in the box below.  There are no phonemes in developmentalist linguistics; the units, phonetemes, are conceived of differently.   Vowels have three degrees on the frontness-backness parameter and also three on the high-low parameter; each of the latter positions is subdivided into closer and opener.  Consonants are labial, apical, or velar (with subfeatures for prevelar, neutral, and postvelar)--with more-fronted, neutral, and less-fronted subvalues for each.  There is also a sulcal-neither-lateral feature for liquids--//r// and //l//, respectively.  The phonetological rules are ordered according to the usual criteria. 
     While child language can play a róle in studying naturalness--less-natural sounds and syllable shapes are acquired later than more-natural sounds and syllable shapes, one should be aware that vowel-space and consonant-space are more restricted in the earliest phases of language-acquisition.  Thus, labials are first acquired, though after consonant space has expanded, labials are not the most-natural consonants in any of the positions discussed below.  Similarly with vowel space.  A child begins with some kind of [a] and progresses to [e] and then to [o]; but as vowel space expands, the least-marked positions after the low vowel are [i] and [u]—in the upper corners of fully developed vowel space.  So much for child language. 
     There are two aspects of markedness theory—MARKEDNESS-REDUCTION (regularization of forms and syntactic patterns) and REVERSALS in marked contexts and even somewhat exotic reversals in the markedness of an item or context itself.
     Before proceeding, it will be helpful to help the reader conceptualize markedness itself.  The most useful criterion for ascertaining the markedness value of any unit of any component in the grammar is natural change—"natural" being qualified by reversals in different environments discussed under iii below and with a few changes (standing somewhere between connatural and abnatural) that overrule the usual directionality of change (listed on p. 293 of Essays; see the box below).  An unmarked unit or context can be thought of as the default unit or context.  The usual pattern for a given environment is encapsulated in the formulas:

 i. If x changes connaturally to y, then y is more natural/marked than x.
ii. If x implies y (as the result of i), then y is more natural/marked than x.

Time obviously is an important ingredient in developmentalist linguistics.  The foregoing principles apply to the special situations of deletion and insertion if y is regarded as deletion or, respectively, as a virtual (epenthetic or parasitic) consonant.  Reversals occur in contrasting environments; e.g. while heavy (unaspirated voiceless) obstruents are usually less marked, in the assimilatory environment between vowels (under given stress conditions in English) a heavy //t// changes to a the corresponding light //d// (a light consonants is either voiceless or unaspirated or both) in city (in most kinds of English other than in parts of the British Isles), thus reversing the default pattern.  We see here that assimilation reverses and overrules the otherwise natural pattern; of course assimilation is a very natural development, and its results are natural in the relevant contexts.

     Even vowel length is three-valued, at least in North American r-less varieties of English.  Consider North American r-less English:  card (where //r// is replaced with not phonetological and phonetic lengthening of the preceding nucleus) : cart (with //r// replaced by phonetological but not phonetic lengthening of the preceding nucleus)  and cod  (where "o" is phonetically lengthened before a tautosyllabic light consonant) : cot.  Note that nuclei are less long before tautosyllabic heavy obstruents than before other tautosyllabic consonants:

Parallels with the foregoing are heard in curd : curt/cud : cut and in cord : sort/sawed : sought.  The word-final nucleus in mar is longer than the nucleus in ma in the varieties of English under scrutiny; and a similar relation holds for the nuclei in fur and duh and in (stressed) for and paw.  The situation is one in which language-user hears only a binary distinction between long and short, while phonetic instrumentation record three lengths for a given vowel.  (Of course, the higher a vowel, the louder and longer it must be in order for it to be heard as having the same loudness and length as vowels lower than it.)

ii

    It is reasonable to think that MARKEDNESS-REDUCTION originates with child learners of a language.  A language that is too marked is harder to acquire than some other kind, and efficiency requires getting rid of some of the baggage.  The replacing of fewer by less (which we deduce from the change to be the less marked of the categories represented) in much current English ignores a category that uselessly complicates the language.  However, the distinction between all nouns and all of the nouns or between both nouns and both of the nouns is so well-established with other quantifiers—e.g. some, no/none, certain, many—that it simply regularizes all and both, thus ignoring grammarians' concern for uses of French and German.    When and how much is discarded is not known, though languages with little contact with other languages (those of the Pacific Ocean) exhibit radical simplification in phonology and inflection.  This naturally increases the burden of morphology (e.g. different words for possessive "of") and syntax.  
     So much having been said, let us consider some problematic examples from English.  The regularization of know in I knowed it and She just growed and growed are easy to understand.  But why do many speakers replace the past verb form with the participial form in I seen it, He give it to me, I done it, It come early.   

    Since confusions of transitivity like lay for lie and set for sit are isolated forms    (I've even encountered await for in a newspaper!) are supported by an entire system of such differences, it is easy to see how they turn up.  English has no trouble in using a transitive verb intransitively (the so-called middle diathesis, more accurately the contraponent diathesis), as in the system has now stabilized and thousands of additional examples.  Saying lay for lie is simply an example of this property of transitive verbs in English. 

iii

PROPORTIONALITY

     As a result of the unidirectionality of change, patterns tend to be asymmetrical, with fewer categories in the more-marked contexts where no reversing or other factor intervenes.  Change works to this end in both directions.  Many languages have fewer differences in plurals than in singular nouns or verbs.  And when a developed category disappears, it yields a similar pattern:  

 More, less, and fewer are further marked as comparatives—i.e. with an extra marking for comparative forms of already marked parallel non-comparative items.  For some reason, English has got more comparative forms on the "less" side than on the "more" side.  However, the violation of expectations found in  there being more comparative forms on the "less" side than on the "more" side causes  fewer to get replaced (by less) in the usage of many native users of English.  Note that, like one instance of more, fewer is marked both on account of its being a comparative form and because of its use with plurals, which are more marked than singulars.  Indeed, it is the replacement of fewer by less in the speech of many that shows us that fewer is more marked than is less when it is an adjective.  As for morphology (the investigation of form[ative]s and formations), note the difference between "You mustn't've failed to do that" (which is deontic, i.e.  laying down a requirement, and can be augmented with "by the time they arrive") and "You must not've failed to move that" (which is epistemic, here:  deducing a fact from some such evidence as a following "in view of its now being in the right position").  Linguists have noted syntactic and semantic distinctions between elided and non-elided forms--like gotta, hadda, useta, be sposeta, be bounda--and their non-elided forms.  Their markedness varies according to the usage, though they are not markered (see the following box) by elision --i.e. they are neutralized with full forms like got to--when directly followed by a deleted infinitive (as in "They've got to [depart]").

     English distinguishes forms in various syntactic positions.  R, drunken, agëd, and burnt are preferred as attributives preceding the nouns they modify, whereas rotted, drunk, ag(e)d, and burned are preferred otherwise.  (The situation is more complex than presented here; see the forthcoming volume on CD-ROM.)  Contrast a rotten apple with an apple rotted by the sun, a drunken sailor with a sailor drunk on gin, an agëd teacher with wine aged in an oaken barrel, and burnt toast with toast burned by having been left too long in the oven.  Note the form and position of the participle after a noun in the contrasts rotten apple : apple rotted . . . ,  drunken sailor : sailor drunk . . . , aged teacher : wine aged . . ., and burn toast : toast burned . . . :   In English, an adjective has got to be placed after the noun it modifies to prevent the (disallowed) intervention between of a modifier or predicate complement of the adjective of itself between it and the noun it modifies  (note that we say apple rotted by the sun rather than rotted/rotten by the sun apple).  In a long known to be drunken sot, the speaker treats drunken as the main adjective modifying the noun sot; one would be more likely to say a person long known to be a drunken sot, where known is treated as the main adjective modifying the noun person.  What one cannot say is:  a long known to be a drunken sot person.  In summary, both the forms and word order get reversed under the specifiable marked or non-default condition; the attributive order, with adjective preceding the noun, is the unmarked or default order in an unmarked context, but the form and order are switched in the marked context.   Incidentally, compound nouns are more marked than simple nouns and sometimes employ marked forms of participles not used otherwise; e.g. store-boughten,  guilt-laden, conscience-stricken, well-girt, misshapen, misspelt, overwrought, and others. 
     To take an example from phonetology, we determine which segment is more or less marked in a given phonetological context by noting--in languages of the world--what it implies and is implied by in the same position and by ascertaining which segments change to it in that position and which segments it changes into in that position in languages of the world.  The relevant positions of a sound segment in a syllable are:  (a) syllable-initial, (b) ante-pre-nuclear, (c) nuclear, (d) satellite (of a diphthong or triphthong) (e) post-nuclear, and  (f) post-post nuclear; note that prenuclear is the default position, since some languages have consonants only in this position and not elsewhere.  Other positions found in different languages are progressively more marked environments.  (Note that sibilants sometimes don't match other fricatives consonants in the patterning.  "Sh" is more natural than "s" after a tautosyllabic nucleus.)  In terms of articulatory position (only), the consonants that are most- and least-natural in (a) are the same as in (f) but the converse of those in (b) and (f)--which agree.  But if (b) and (f) reverse (as), (f) doubly reverses, and ends up just like, (a), while  (f) doubly reverses, and ends up just like, (b).  (See iii below on reversals.)  If we ignore vowel differences,  tuck or dug is thus more natural than could or good, respectively, while pup and bub stand in the middle.  Incidentally, a liquid final fricative (with qualifications for "th" vs. a sibilant) is more natural than a non-continuant after the nucleus; and a sibilant stands between the two possibilities.  Thus, Classical Greek orthotonic words could end only with "n," "r," or a sibilant, though  proclitics like the unstressed preposition, ek could have another obstruent at the end.    Back rounded vowels tend to become more fronted and in time, like other non-back vowels, become unrounded; thus, English put in many varieties exhibits the sequence of umlauting and then unrounding, so as to sound just like the "i" following "w" in wit and wizard.  All else being equal, heavy vowels tend get higher.       
     The glides, pre-nuclear //h  y  w//, constitute a real problem.  (Note that diphthongs with satellite u are more marked than otherwise similar ones with i.  Note also that some languages treat a given satellite as a consonant (Greek satellite u became a labial fricative), though most languages treat it as vowel-like (and not a separate segment but part of a single-segment diphthong); this is one of the rare cases when it is legitimate to let a phonetic transcription--[au] vs. [aw]--be influenced by the phonetology of the language.  Some languages have //y//, with or without //h// being present, but no //w// (which has become //v// more often than //y// has become an obstruent)--e.g. German, Russian.  Hawai'ian has got //w// and //h// (from older //f//; heavy fricatives often become [h] over time) as well as the glottal stop (from earlier //k//)--but no //y//--while some Polynesian languages lack all three glides.  In Micronesia, Marshallese  has got only [w].   All three glides are heard in Classical Hebrew (which also had the glottal stop and a pair of pharyngeal fricatives) as well as in English, Japanese, and Malay--though //y// is marginal in Malay.  Glides obviously present problems for markedness theory.

iv

REVERSED USAGES AND VALUES
IN MARKED CONTEXTS

     It has already been noticed that the naturalness of sounds can be reversed in marked positions, e.g. complementary positions in the syllable.  The writers' Essays on time-based linguistic analysis offers many confirmations in English of W. Mayerthaler's theory of reversals in marked contexts--i. e. in marked forms (including anterior, exochronous, and posterior modalities as well as in imperatives and prohibitives and the modalities formed on modal verbs, marked usages, or marked linguistic environments), marked word order, and other kinds of marked contexts--notably negation, true interrogation, comparison, etc.  Most forms can be reversed in a marked context.  For verbs, such marked environments include negation, interrogation, and comparison, and certain marked subordinate clauses embedded in a posterior main clause that are called the surrealis environment, as well as the following verb modalities--anterior [e.g. have run], exochronous [e.g. runs], and posterior [was/is gonna run, will run].  Be gonna gets reversed to will (and vice-versa) in the marked contexts, and the same is true of the passivizing auxiliary verbs--be and get.  See the early pages of Ch. 6 of my Essays on time-based linguistic analysis.  It should be pointed out that will and would have three non-posterior uses that differ in their effects from the posterior uses (see pp. 203-204 in Essays.)  We need not know which uses are less-marked and more-marked in order to predict their switches.  Note the reversal of be and get in marked posteriors:    

A double reversal occurs in a marked modality in the marked category of posteriors--was to:  "It was to be watched the next day" and "It was to get copied on the following day."   Where there of more than one marking, each additional marking causes a  reversal in the places where reversals occur.  (Being over-marked is the same as being doubly reversed and ending up looking like an unmarked item; this is true of ever even-numbered reversal, as shown in the literature on this subject.)   Hence, in an environment that is both a true question and negative, a verb usage will be like the default usage.      
    It should not be thought that reversals occur only with segmental phenomena (including word order) and not with suprasegmental prosodic phenomena (which can be either segmental, like vowel length or tone, or suprasegmental [intonational], like lengthening for emphasis, reversing high and low tones and degrees of loudness.  Contrast the dissyllabic noun cónvèrt  with the verb convért or the noun óverwith the verbs and flòw óver as well as the noun únderbìd with the verbs ùnderbíd and bìd únder.  There is, however, no reversal when a verb is copied from an identical noun (cf. the verb détàil) or when a noun is copied from an identical similar verb (cf. the noun disdáin).  There are three or four special formations that are subject to their own special rules (e.g. sýndicate [noun] : sýndicàte [verb]).  P. 198 of Essays on time-based linguistic analysis lists special categories in which unmarked falling stress is reversed to rising stress:  
--names; notice that Worth doesn't rise in <Fort Worth, Texas> but does rise when Texas is left off and Worth ends the name); T. S. Eliot and T. S. are parallel; 
—copulative or twofer compounds like stùdent-prínce; 
—quantifier compounds like fìrst-áid, dòuble-héader, and  òne-tímer; 
and 
—compounds beginning with old like òld-máid, where the stress is the same as in the phrase, òld máid.    
Some examples, like Fòurth Ávenùe, combine more than one category; and certain compounds with predictable and/or deletable second elements [e.g. Fóurth Strèet] constitute exceptions, since the predictable second element is always on a lower tone when not standing in contrast with some other item.  On the same page just referred to, ex. 67 provides interesting examples (analysed from Bolinger's data) of intonational contrasts in which the tones of unstressed syllables--the definitive contour of an intonational accent are changed, say, from high tone to low tone.  The unstressed syllable -gle in single roll falls or does not fall, depending whether it is a compound noun or a phrase; and the unstressed syllables, -ican, in American history teacher fall or do not fall, depending on whether the first two items are linked more closely than the last item in a double-compounded word (i.e. American-history teacher) or whether the last two items are linked more closely than the first part, as in the phrase exhibiting American modifying history-teacher.   
     Double reversals are not rare in syntax or in prosodics.  An example is found in the way in which subject-verb order and the normal falling cadence at the end of a statement in English--cf. "They're ready by now"--get reversed at the end to a rising cadence in the true yes-no question, "Are they ready now?"--with a rising cadence.  (Note that the extent of the rise or fall is gradient in all cadences; a truncated rise or fall conveys less finality in a fall or less non-finality in a rise than a longer fall or rise, respectively.)  The cadence of the true question, "Are they ready by now?" gets re-reversed--back to a falling cadence--in a WH-question (one beginning with Where?, When?, Why?, How?, or How [much]? as well as Who?, What?, and Which?) like "Where have they gone?"  WH questions are more mark(er)ed than non-WH-questions:  A non-subject nominal (i.e. an object or predicate complement) as well as a subject comes at the beginning (as in "What did they say?" and "Who is that?").   
     Let's now look at several kinds of reversals in special (non-true) questions.  Both the word order and rising cadence of a true question get reversed in an indirect question like "I'm inquiring whether they are ready by now."   Reclamatory questions have rising cadences but reverse the word-order of a question to unmarked order:   "They're ready by now?" and "They said what?"   The echo questions, "Are they ready by now?" and "What did they say?" both have rising cadences (a re-reversal in the WH-question?) but leave the interrogative word order unaffected--a double reversal?  The tag questions at the end of  "They're ready by now, aren't they?" and "They aren't ready yet, are they?" reverse the negation of the main sentence, the relation (order) of subject and verb, and the direction of the final cadence--which falls.  In the rhetorical (marked) tag question, "They're ready by now, are they?" both the negation of the tag and the rising cadence of a true question get re-reversed--back to a default non-negative expression with falling cadence; the word order also gets re-reversed to that of an unmarked question.  There is a marked connotation in such examples; here, it happens to be the connotation of disbelief.  
     Reversals can result from emphasizing and also from over-correcting utterances as well as simply for the purpose of conveying a marked meaning.  In Bolinger's example, after one has been invited to have a seat, one responds--not with unmarked "What on?" but--with reversed and marked  "On what?"  The effect of this is to  convey the  marked sense of, say, one's dismay at not perceiving any furniture to sit on--except perhaps a table or the floor.   A person corrected for saying lay instead of intransitive lie (the verb for reclining--not the verb for speaking an untruth) may then turn around and say lie for transitive lay.  (Of course, contraponent lay [without an object] in "The puppy laid down" is no more incorrect than saying contraponent [i.e. the passive or reflexive use of a basically transitive, almost always causative, verb] translated in "That book translated easily.")  
       While emphasis and other stylistic oddities trigger reversals, social situations do this.  In the later sixties, Robin Lakoff pointed out that we say

—to a child:  "You may have some of this cake."
—to a person viewed as equal:  "Have some of this!" or "Taste this!" in the imperative form  

But to someone treated deferentially as more than equal, we say:

—"You must try some of this!"  

Note, how an increasingly peremptory tone--permissive may (which would be rude to a honored guest), a direct or oblique imperative, and the necessitarian modal verb, must--is used for increasing deference, reversing the literal usages of what is polite!   In the third instance, must sounds like invoking some more general principle than the host(ess)'s desire--though it could imply, "You must do so to avoid making me cry, hurting my feelings, being haughty," or whatever. 
     Let's say that one has three helpers or workers.  To a paid helper (one's secretary or a paid technician), one makes a polite, even deferential-sounding, request in order not to seem to be domineering:

—"Could you please have this finished by noon?"

To one's buddy, one says:  "How 'bout getting this finished by noon, man" or perhaps, "We hafta get this finished by noon."  (It would be patronizing, or treating a friend like an employee, to use too many "please" expressions other than in special contexts.)  To a volunteer--in the extreme case, former-President Jimmy Carter and his team--one would be very deferential, of course, and say something like:

—"Would it be feasible to have this finished by noon?" or "Would you mind having this done by noon?"

To all three kinds of helpers, we might make our request reasonable by adding a reason why the request is in order--e.g. "We need to get this done by then so that the workers for the next phase can get started when they arrive about lunch time."
     Note that one needs to be polite to non-equals on reversed grounds:  To those who work at our bidding, we wish to avoid coming across as domineering or flaunting our authority; to those we cannot order around, we need to be deferential.  What is notable is that deontic or permissive may, which is polite for offering something to a child, would be horribly rude and inappropriate for someone we cannot order around.  It is indeed strange that that must comes across as deferential in the situation indicated.  But that's the way language works.
     

[PAGE UNDER CONSTRUXION]

APPENDIX

The dots in 4 are meant to indicate that without some word, phrase, or clause in that place--e.g. "once," "only," "for my sake," "to impress her," "when I was present"--the sentence is not acceptable.   Clause-initial neither (a conjunction) is like the adverb never.   Examples with clause-initial no, none, and not  modifying the determiner or other modifier of a non-subject are infrequent:  

We do not find the exact reversal of 3a (the asterisk indicates a non-existent example):

     Note other asymmetries in our examples:

     a. Only 1 lacks the inserted crutch word--a form of do--in the absence of another auxiliary or a form of non-auxiliary be or have; though never doesn't take the crutch in default world order, it does do so in the marked order triggered by any clause-initial negative-- other than a negative modifying the subject of a clause (as in 3b).
     b. Clause-initial never, but not not, can be directly followed by the predicate.

    The basic rule is that whenever a clause-initial negative--which always happens except when the negative is a quantifier (determiner) modifying the subject of the clause--creates a marked word order, the crutch word do is inserted.
     That never lacks the crutch word do in default order is a glaring anomaly--or else, the glaring anomaly is that not does takes this auxiliary in unmarked word order.   The latter is the anomaly from the historical point of view (see below on early Modern English.)  The conclusion that the lack of the crutch word with never in unmarked word order is less likely, since unmarked   phenomena--in this example, default word order--are generally not less complex than the corresponding marked phenomena.  Negation is of course a marked environment in any order, and its marked position at the beginning of a clause is doubly marked, except in the instance of a negative quantifier (determiner)'s modifying the subject of the clause.
    Why then does crutch do accompany not in unmarked word order?   Perhaps it markers the marked context caused by the  negative; then, the more-marked never would reverse this situation.   Does this explain too much?  Since ex. 2 with more-marked never at the beginning of the clause reverses ex. 1, already accounted for, there is no problem with it.  If any problem exists for this analysis, it will be found in connection with ex. 4.  The predicted reversal of 3a in 6a does not violate the theory's predictions in any positive way:  It just doesn't exist.  What we find in  4 and 5 are objects, negated in some other way than with determiner-not, triggering marked word order triggering the crutch auxiliary do.  If the negatives in 4 and 5 are either less marked or more marked than over-marked never in the clause-initial position, the theoretical prediction would be 4 and 5--both being, like a double reversal of 3a and a single reversal of 1.     
     On further reflection, perhaps an example comparable with 6a does exist:

While me is a nominal, indeed a pronoun like that in 6a and perhaps many in 5c, it functions, not unexpectedly, more like a noun; and nouns are normally not modified by not except in possibly in predicates like: 

But in view of the overall patterning, we should treat not as modifying the copular verb--not foreigners.  This approach means that not does not modify a directly following noun--only no serves to negate a directly following noun--in environments other than possibly cleft sentences like this example:  

But since we don't say, "Not foreigners lived in that house" (for "Foreigners did not live in that house"), it is like that not is not an adverb in either 7a or 7b.  In 7b, not would have to be regarded as a negative adverb moved to the clause-initial position--just as in 4a,b (not as in 5c).   This usage may have arisen through truncating a longer cleft sentence:

     While not entirely satisfying, the foregoing account at least indicates that, with the help of W. Mayerthaler's theory of reversals in marked contexts, linguists have at their disposal the potential for eliminating the unsystematic anomalies (noted at the beginning of this Appendix) that we would otherwise have to put up with.  For my part, that would be a good deal more unsatisfactory than the surmise of a cleft-sentence origin just hazarded.  After all, cleft sentences exhibit phenomena that would be ungrammatical in other contexts--e.g. the use of It was (contrast "There were") with a plural complement.  Note that even in the cloven sentence,  "What was at issue were the errors that they had committed," we find were--not was--before the plural predicate:  the errors.

     As has been shown in the author's Essays on time-based linguistic analysis, grammar is vastly more complicated and requires more mental resources than could be imagined before W. Mayerthaler's markedness theory explained the otherwise unexplained reversals that permeate grammars.  Reversals can pile up considerably in multiply marked compound verbs, e.g. may have been going to be striving to have begun to lay [the groundwork by the time they came on the scene]--with seven irrealis modalities.  Somehow, the human brain keeps track of each reversal and comes out with the predicted result.

      Middle English did not use a form of do with not (except, irrelevantly for present purposes, for emphasis) with past verbs in the default word order--except in poetry where a syllable was needed to fit the rhythm of the verse, and where inverted word order could be used for the same purpose.   (It is a parody of Elizabethan-Jacobean English to replace a past form with did(st) plus the infinitive, as imitators of early Modern English are wont to do; although it does avoid the hard pronunciation of believedst--even when the last "e" is pronounced.)  Object-first word order was used more than today, where we have other ways to highlight a non-subject (e.g. "It was the dark that the child feared" and "What the child feared was the dark"); but in prose, both the older and newer usages are infrequent.  Even so, inverted word order and did were not required when no negative or word like so or thus (cf. today's "So did we") begins the clause; here is an example:  "For thy sake we are killed all the day long."   Like our current optative-may, marked word order is found in wishes; e.g. "[May] Grace be unto you." 
     What early English did do was to use marked order when the object preceded the predicate and a negated and a negated subject follows it:   "For other foundation can no many lay than is laid, . . . "   Today, we would prefer "For no other foundation can anyone lay that is laid, . . . "