Morphemic and Syntactic Phrasemes

2.2.1 Morphemic idioms

Morphemic idioms are found only in diachronic word-formation (= diachronic derivation and diachronic compounding) and diachronic affix-formation, that is, in the historical, at present non-productive creation of new lexemes and new affixes. This is so because synchronic derivation and compounding as well as inflection are by definition too regular to allow for the idiomaticity, i.e., the semantic non-compositionality, of produced complex signs.

To put it differently, a morphemic idiom is necessarily stored as a whole (= as a simplex) in the lexicon, if it is a lexeme, or in the grammar, if it is an affix.

Morphemic idioms can be subdivided into the same three subclasses as lexemic idioms: strong idioms, semi-idioms and weak idioms.^[2]

Terminological remark

A morpheme {M} is implemented on the surface by one of its morphs m_i: thus,

      {leaf} ⇔ leaf-, leav-,

where the stems leaf-(Ø) and leav-(es) are morphs.

A morphemic idiom ˹{M₁+M₂}˺ is implemented by a quasi-morph; thus,

      ˹{for+get}˺ ⇔ forget,

where the stem forget- is—from a diachronic viewpoint—a quasi-morph.

The components of a quasi-morph are

1. either morphoids, if they have semantic links with morphs of the language,

2. or submorphs, if they do not (Mel’čuk 1993–2000: vol. 4, 249–251).

For instance, in the wordform conceive, the elements con- and -ceive are morphoids; in forget, the elements for- and -get are submorphs.

This analysis is valid only with respect to diachrony. In the modern language, the diachronically complex stems and affixes illustrated below can be simple stems and simple affixes. Thus, in a formal synchronic description of English, {forget} and {conceive} are simple morphemes and forget- and conceive-, simple morphs.

☛  m stands for “morph”; within a morphemic idiom, m denotes the corresponding morphoid/submorph; morphoids/submorphs are separated by pluses: m₁+m₂; in the examples, lexical stems are printed in caps, and affixes, in boldface lowercase characters.

■ Strong morphemic idioms

The meaning of a strong morphemic idiom ˹{M₁+M₂}˺ does not include the meaning either of {M₁} or of {M₂}: ‘˹{M₁+M₂}˺’ ⊅ ‘{M₁}’ and ‘˹{M₁+M₂}˺’ ⊅ ‘{M₂}’.

Strongly idiomatic diachronically complex stems

In Modern English, all the stems in (1) and (2) are simplexes. But there also exist strongly idiomatic diachronically derived and diachronically compound stems that remain complex in modern language; for instance, in Dene-Suline (Athabaskan) (Holden 2009):

Strongly idiomatic diachronically complex affixes

Diachronically, the circumfix ge-…-en consists of the prefix ge-, which is not productively used in modern German with verbs, and the suffix -en, which, outside of this circumfix, marks a verb as either infinitive or 1pl and 3pl.

In Modern Russian, the prefix za- does not combine with nouns at all; the suffix -j- derives depreciative collective nouns, such as babʹ+j(-o) ‘women of the type the Speaker dislikes’ (from baba ≈ ‘woman of the type the Speaker dislikes’) or duračʹ+j(-o) ‘fools’ (from durak ‘fool’)’.

■ Morphemic semi-idioms

The meaning of a morphemic semi-idiom ˹{M₁+M₂}˺ 1) includes the meaning of one of the morphemes {M₁} or {M₂}, which is not its semantic pivot,^[5] and 2) does not include the meaning of the other:

‘˹{M₁+M₂}˺’ ⊃ ‘{M₁}’, and ‘˹{M₁+M₂}˺’ ⊅ ‘{M₂}’, and

{M₁} is not the semantic pivot of ˹{M₁+M₂}˺

Semi-idiomatic diachronically complex stems

Semi-idiomatic synchronic affix complexes (uninterrupted and interrupted)

■ Weak morphemic idioms

The meaning of a weak morphemic idiom ˹{M₁+M₂}˺ 1) includes the meanings of both the morphemes {M₁} and {M₂}, 2) but neither of these is the semantic pivot, and 3) it includes an additional meaning ‘A’:

‘˹{M₁+M₂}˺’ ⊃ ‘{M₁}’, and ‘˹{M₁+M₂}˺’ ⊃ ‘{M₂}’, and ‘˹{M₁+M₂}˺’ ⊃ ‘A’, and {M₁}/{M₂} is not the semantic pivot of ˹{M₁+M₂}˺.

Weakly idiomatic diachronically complex stems

Morphemic idioms and suppletion Klammer nicht fett (Mel’čuk 1993–2000: vol. 4, 403)

A morphemic idiom is defined by its semantic non-compositionality—that is, its signified is not a regular (i.e., compositional) union of the signifieds of its component morphemes. However, formally it is quite regular, that is, compositional. At the same time, there are complex morphological signs of, so to speak, an inverse nature: they are semantically compositional, but formally non-compositional. Thus, the English wordform has is semantically compositional, since its signified is regularly constructed out of the signifieds of its virtual components:

                    ‘has’ = ‘have’ ⊕ ‘ind.pres’ ⊕ ‘3’ ⊕ ‘sg’.

But its signifier is not regularly constructed: it “should” be *haves, but it is has. This is a well-known linguistic phenomenon: the wordform has is а strong megamorph suppletive with respect to the stem have-. I am not in a position to develop this point any further here (on suppletion, see Mel’čuk 2006: 405–467), and I will limit myself to emphasizing the following fact:

Either the signified of a phraseologized morphological complex sign can be a non-regular (= compositional) combination of the signifieds of its components, or its signifier can be a non-regular combination of the signifiers of its components.

The first case gives us morphemic idioms, and the second, pairs of suppletive units—a strong megamorph and the corresponding stem. In one of his last papers, Weinreich (1969: 43) defines an idiom as a phraseme whose meaning is “suppletive [sic!—IM] with respect to the sum of the meanings of its components,” i.e., Weinreich treats phraseologization as suppletion in the domain of meaning. Inversely, we can say that suppletion is phraseologization in the domain of form.

The properties of morphemic idioms and those of strong megamorphs can be presented in parallel:

These close ties between suppletion and phraseologization are theoretically important.

2.2.2 Morphemic collocations

☛  The collocation base is printed in small caps, and the collocate (in this case, the derivational affix or dependent compound component), in boldface lowercase characters.

Morphemic collocations are found in diachronic and synchronic derivation and compounding, as well as in inflection.

Collocational complex stems

  Collocational diachronically derived stems

Here are two stock examples of diachronic derivational collocations.

One of the suffixal morphemes {person who lives in L} is selected by the Speaker as a function of the stem, which is the name of a city or a town.

Now, three more complex examples of diachronically derivational collocations.

Russian has two homonymous derivational suffixes -išč-:

1. the semi-productive augmentative suffix -išč¹- ‘extremely big’, as in nos+išč(-e) ‘extremely big nose’, komnat+išč(-a) ‘extremely big room’, etc.;

2. the unproductive, but frequent locative suffix -išč²- ‘place where …’, as in požar ‘fire’+išč(-e) ‘place where there has been a fire’, strelʹb ‘shoot’+išč(-e) ‘shooting range’, etc.

The suffix -išč- carries the meaning ‘handle’ only in the lexeme toporišče. (knut+ovišč(-e) ‘whip handle’ contains a different, albeit similar, suffix: -ovišč-.)

Collocational synchronically complex wordforms (inflection)

In a language with rich non-agglutinative morphology, where nouns of different types have formally different declensions and verbs of different types have different conjugations, each declined or conjugated wordform (that is, each inflectional form) is a morphemic collocation: its stem is the base and the inflectional suffix is the collocate, since it is selected for the given grammemic combination as a function of this stem. Four examples follow.

Тhe perfective forms are compositional: all the prefixes express only the perfective aspect. However, their selection cannot be described in general or systematic terms; hence the claim that there are 15 perfective morphemes rather than 15 suppletive allomorphs of a single perfective morpheme. This means that each verb has to be marked in the lexicon for the particular perfective prefix morpheme it takes. The choice of the perfective prefix for each radical is thus severely constrained, but the resulting form is compositional. This is characteristic of collocations: the stem is the base of the collocation, as well as its semantic pivot.

The plural forms of Burushaski nouns are semantically compositional, but formally unpredictable: for each individual radical (which is the base of the morphemic collocation and its semantic pivot), the corresponding plural suffix has to be specified in the lexicon. The distribution of the plural morphemes does not correspond to any more general morphological or declension class of nouns in the language: therefore, it cannot be described as allomorphy, which should be sufficiently general as to be treated as rule-governed behavior.

2.3.1 Morphemic nominemes

Morphemic nominemes are the proper names of individual places, events, human groups, etc., and, therefore, like lexemic nominemes, they represent little interest for a language lexicon: their place is rather in an encyclopedia.

Diachronically derived morphemic nominemes

Diachronically compound morphemic nominemes

2.3.2 Morphemic clichés

For the time being, I have examples for three subclasses of morphemic clichés: morphemic nicknames, morphemic termemes and morphemic formulemes (but no examples for morphemic sentencemes^[7]).

■ Morphemic nicknames

A morphemic nickname is a morphemic cliché that has a specific concrete referent, that is, it refers to an individual (in the logical sense). But unlike a nomineme, a nickname has a meaning: it not only identifies its referent, but tells us something about it; its lexical components are semantically full (= meaningful) morphemes. Nevertheless, morphemic nicknames, just like their lexemic counterparts, belong in an encyclopedia.

■ Morphemic termemes

A morphemic termeme is a morphemic cliché that has a generic concrete referent, that is, it denotes a particular class of individuals (in the logical sense). This means that a termeme corresponds to a technical term.

■ Morphemic formulemes

A formuleme is a cliché that has a specific abstract referent, that is, denotes a particular situation.

Many morphemic formulemes are constrained pragmatically, that is, by the situation of their use. Such formulemes are, at the same time, pragmatemes (Mel’čuk 2020):

3.3.1 The definition of syntactic idiom

Since idioms are a type of phraseme, in order to define syntactic idioms, syntactic phrasemes must first be defined.

The syntactic phrasemes, just like lexemic and morphemic phrasemes, are a major subclass of phrasemes. However, unlike the latter, syntactic phrasemes are non-segmental signs. In prose, this means that they do not consist exclusively of lexemes or morphemes: the signifier of a syntactic phraseme includes prosody (intonations, pauses, accents) or an operation (for instance, the duplication of an actant).

Definition 5: syntactic phraseme (see Mel’čuk 1987)

Let there be a complex sign s = 〈‘σ’ ; /s/ ; Σ_s 〉 involving at last two minimal syntactic subtrees (of the form L₁–synt→L₂) which is constrained: that is, s is such that none of its components can be selected freely by the Speaker to express the meaning ‘σ’; cf. Definition 1.

A constrained complex sign s is a syntactic phraseme if and only if its signifier /s/ is non-segmental, that is, /s/ contains prosody or a bound lexemic variable, e.g., L(X), symbolizing the operation of duplication of the phraseme’s actant X.

Remember, the signifier of a lexemic phraseme is segmental: such a phraseme contains only genuine lexemes of the language, but no special prosody and no bound lexemic variables. The difference between a lexemic phraseme and a syntactic phraseme, whose signifier is non-segmental—it contains special prosody or a bound lexemic variable—is obvious.

For the time being, only one class of syntactic phrasemes is known: syntactic idioms.

Definition 6: syntactic idiom

A syntactic phraseme is a syntactic idiom if and only if it is non-compositional.

It is not known whether compositional syntactic phrasemes—that is, syntactic collocations and syntactic clichés—are possible. Therefore, in what follows I will speak exclusively of syntactic idioms.

All the syntactic phrasemes cited above are syntactic idioms, since the meaning of each of them cannot be regularly constructed by uniting the meanings of its components.

In conformity with Definitions 5 and 6, syntactic idioms come in two structural types.

• Syntactic idioms whose signifier contains prosody

  • Idioms containing no lexemes at all, as, for instance, ˹[Petja_X] [u nas_Z] [budet guljatʹ_Y po nočam!]˺ lit. ‘Pete_X at us_Z will.go.for.a.walk_Y at night!’ = ‘If Pete goes for a walk at night, he will be severely punished by us’. All lexemes of the sentence Petja u nas budet guljatʹ po nočam! are part of the actants of the idiom (that is, the actants of the predicate ‘punish’). The meaning of the idiom (a threat of punishment: ‘If X does Y, X will be punished by Z for Y(X)’) is expressed, as explained above, by the prosody imposed on the string of its actants.

  • Idioms containing just one lexeme (plus prosody), as, for instance, čtoby ‘that’ in ˹Čtoby [Petja guljal_X po nočam]?˺ lit. ‘That Pete go.for.a.walk_X at night?’ The clause Petja guljaet_X po nočam is the actant of the idiom (‘It is quite impossible [that X takes place]’), while the conjunction čtoby ‘that’ is a component of it, even if it contributes nothing to its meaning, a sarcastic negation: ‘quite impossible for Pete to go for a walk at night’. This meaning is, again, expressed by prosody.

• Syntactic idioms whose signifier contains a bound lexemic variable L, which can be one of the two types:

  • Either L is filled with any lexical expression that, however, is semantically bound to actant X: it must denote an intensifier of X (see (28a), p. 50);

  • or L is filled with a duplicate of the idiom’s actant: [Malʹčiki_X] ˹estʹ malʹčiki_L(X)˺ ‘Boys are boys’ (see all other cases above). Syntactic idioms of the last type are the most numerous among the syntactic idioms known to me (in any case, in Russian).

3.3.2 Distinguish and avoid confounding!

Since the notion of syntactic idiom is relatively new and not stable enough, it is worth insisting on the differences between syntactic idioms and the following four other types of phrase that look similar enough to, and maybe confounded with, the former.

 1) A phrase describable by a semantically loaded (= meaningful) SSynt- relation. Thus, the well-known Russian approximate-quantitative construction of the type knig pjatnadcatʹ lit. ‘books fifteen’ = ‘maybe fifteen books’ manifests the approximate-quantitative SSynt-relation:

pjatnadcatʹ←approx-quantitative–kniga_PL ⇔ knig pjatnadcatʹ ‘maybe fifteen books’

The ordinary quantitative construction is implemented by a phrase with the inverse word order and a different SSynt-relation:

pjatnadcatʹ←quantitative–kniga_PL ⇔ pjatnadcatʹ knig ‘fifteen books’

The approximate-quantitative SSyntRel is meaningful (contrary to the quantitative SSyntRel, which does not carry any meaning): it contributes to the meaning of the phrase the semantic component ‘the Speaker is not sure of the number he indicates’. In our description, this component can be encoded in the deep-syntactic structure by means of the fictitious lexeme «primerno» ‘maybe’.^[13] It is «primerno» that appears in the deep-syntactic structure to represent the approximate-quantitative phrase; the DSynt-structure of the phrase knig pjatnadcatʹ is as follows:

              «primerno»←ATTR–pjatnadcatʹ←ATTR–kniga_pl

As another example, consider phrases of the type {Ivan –} durak durakom lit. {Ivan [is]} fool by.fool’ = ‘{Ivan is} merely a fool’ = ‘Ivan is a fool times two’, {Èto} voda vodoj lit. ‘{This [is]} water by.water’ = ‘{This [is]} merely water’, etc. They are described, on the DSynt-level, by the fictitious lexeme «prosto» = «merely» and at the SSynt-level, by means of the “merely”-reduplicative SSyntRel:

«merely»←ATTR–durak_SG ⇔ durak_SG–“merely”-reduplicative→duraк_SG (for more on this construction, see Janda et al. 2020).^[14]

 2) A phrase containing an ordinary lexical unit L (a lexemic idiom or a single lexeme) and L’s actants. Thus, the sentence U Ivana ruki češutsja vzjatʹsja za kistʹ lit. ‘At Ivan hands itch to take up a.paint.brush’ = ‘Ivan has a strong wish to start painting’ contains quite an ordinary lexemic idiom [u Х-a] ˹ruki češutsja˺ [Y-itʹ]; X and Y are the idiom’s actants, and by no means “its variable lexical parts,” as one sees them called from time to time. In an analogous way, in the expression ˹k čërtu˺ [Y]! lit. ‘To devil [with.Y]!’ Y is the actant of this lexemic idiom rather than its mystical “variable lexical part.” In the sentence Doloj ètix islamofašistov! ‘Down with these Islamofascists!’ the phrase ètix islamofašistov expresses the actant of the lexeme doloj [Y]! ‘Down [with.Y]!’: no idiom at all here. Similarly, the phrase Xren tebe! lit. ‘Horseradish to.you!’ = ‘You won’t get anything!’ contains the single lexeme xren [Y-u]!) and its actant tebe ‘to.you’.

 3) An ordinary lexemic idiom that consists of “semi”-grammatical, or “light-weight,” lexemes. Such is, for instance, the Russian lexemic idiom (31), whose lexemic components are bytʹ ‘be’ and ne ‘not’:

Its SSynt-structure is as follows:

 4) An ordinary lexemic idiom that has some syntactic peculiarities fairly often is also called “syntactic phraseme/idiom.” As an example, consider Russian lexemic idioms ˹vsë ravno˺ ≈ ‘just the same’ (described in Iomdin 2010: 156–162):

Тhe four idioms have the same SSynt-structure: ˹vsë←restrictive–ravno˺. Their syntactic peculiarities:

The four ˹vsë ravno˺ expressions are ordinary lexemic idioms featuring some syntactic particularities.

3.3.3 Fictitious lexemes

Strictly speaking, the topic of fictitious lexemes lies outside of the general discussion of phraseology. However, since fictitious lexemes are necessary to represent –both in the lexicon and in deep-syntactic structures–syntactic idioms whose signifier is only prosody, a few words have to be said about them.

Definition 7: fictitious lexeme (Mеl’čuk 2013: 37–41, 2018)

A fictitious lexeme of language L is a linguistic sign of L whose signified is similar to lexical signifieds of L, but whose signifier is non-segmental, that is, it is not a string of phonemes: it can be word order, prosody, or operations (presented by lexemic variables).

A fictitious lexeme is given a conventional name shown by « » quotes; under this name it is stored in the lexicon and appears in DSynt-structures.

Fictitious lexemes are used to represent linguistic phenomena of the three following types:

• Those semantic differences between phrases consisting of the same lexemes that cannot be expressed in the DSynt-structure only by means of DSynt- relations. Such is, for instance, the case of phrases with the subject-copredicative and object-copredicative SSynt-relations. These SSyntRels both correspond to the DSynt-relation ATTR, but carry different meanings, since their dependents semantically bear on different clause elements:

  To express this semantic difference on the DSynt-level, the fictitious lexeme «be» is introduced into the DSyntS as an ATTR-dependent of the Main Verb:

  (33)	a.	[John] met–[Mary]–subj-copredicative→dressed [John was dressed]. ⇔ vs.
  	b.	[John] met–[Mary]–obj-copredicative→dressed [Mary was dressed] ⇔ ☛ A dashed double-headed arrow indicates coreference.

• Semantically loaded (= meaningful) surface-syntactic relations, which are illustrated in 3.3.2, Item 1), with the approximate-quantitative SSyntRel. Meaningful SSyntRels are not widespread in English, but there are some; here is an example:

The so-called schm-reduplication of a noun L_(N) expresses the Speaker’s derision and skepticism about L_(N’s referent: ‘SCHM-L_(N)’ ≈ ‘I signal that I dismiss L_(N) as being ludicrous and worthless’. In the DSyntS of a sentence with the lexeme schm-L_(N, this meaning can be represented by the fictitious lexeme «derision», and English has the following deep-syntactic rule:

Note that the rule (35) actually is a part of the lexical entry for the fictitious lexeme «DERISION»: it tells us how to implement this DSynt-node in the surface-syntactic structure. On the next step, the schm-reduplicative SSyntRel is expressed as SCHM-L_(N), where SCHM- is the name of the corresponding derivational means (adding to L_(N) the prefix /šm/- and deleting the initial prevocalic cluster in L_(N), if any).

Russian is rich in semantically-loaded SSyntRels; here are two more examples of such SSyntRels and of their encoding in the DSynt-structure.

(37)	«as.for» for the focalizing-reduplicative SSyntRel (Russian phrases of the corresponding type are described in detail in Iomdin 2013):[15] ‘As for réading the novel, I did not read it, but only leafed through it’.

• Syntactic idioms whose signifier is exclusively prosody (it is because of these idioms that the notion of fictitious lexeme had to be introduced in this paper).

One such idiom—«will.be.punished»—was described in Subsection 2.2, here is another one (the conversive of the former)—«will.punish»—in sentence (38):

The DSynt- and SSynt-structures of this idiom taken together with its actants are as follows:

It is on this structure that the idiom’s prosody is imposed.

As already mentioned, each syntactic idiom has its own lexical entry, just like lexemic idioms do; this entry gives, among other things, the information necessary to expand the DSynt-node of the idiom into its SSynt-subtree (and then to properly implement this subtree in the morphological string).

Here is a list of fictitious lexemes gleaned from various languages (see also Mel’čuk 2013: 37–42); to facilitate the reader’s task, they are represented here by their English equivalents. The fictitious lexemes feature rather abstract meanings that one could loosely qualify as “grammatical”; that is why Žolkovskij (1971: 10) spoke of them as “belonging to a border zone between the lexicon and the grammar.”

3.3.4 An illustrative list of Russian syntactic idioms

Syntactic idioms include lexemes mainly from closed lexical classes or are implemented only by prosody; therefore, they cannot be very numerous. However, they are of special interest from a theoretical angle, so that it is worth presenting here those Russian idioms that are known to me at present.

This list presents both structural types of syntactic idiom:

1. A syntactic idiom whose signifier contains prosody, that is, which has less than two lexemes: either one or none. The idioms of this type are Nos. 3, 5, and 11–14.

2. A syntactic idiom whose signifier contains a lexemic variable that is filled with the duplicate of one of the idiom’s actants. The idioms of this type are all the others: Nos. 1, 2, 4, 6–10 and 15–29.