Components of linguistic know-how, levels of …ling.uni-konstanz.de/.../Syntax-I-SoSe-14/Syntax_basics_1.pdf · Components of linguistic know-how, levels of grammatical constructions:

F. Plank, SS 14 Syntax 1: English Syntax 1

Components of linguistic know-how, levels of grammatical constructions:

Spielraum for crosslinguistic variation? And whereabouts is English?


Is this what is to be expected for ALL languages, including English? 1. LEXICON = basic building blocks, grouped in paradigmatic systems (stored in long-term memory)

2. GRAMMAR = rules for and constraints on constructions (constructed, not stored)

a. Phonology grammar of sound b. grammar of meaning Morphology morphemes least complex units parts

words

phrases

Syntax clauses consisting of

sentences

paragraphs/turns Text Grammar ... texts/discourse most complex units wholes


Questions about universality

A. Do all languages (= all particular manifestations of the general human capacity for Language) have a Lexicon as well as a Grammar?

(That is, is the linguistic know-how of all speakers of Language twofold, comprising stored items on the one hand and rules/constraints for constructions on the other?)

Or could a language conceivably (or as actually attested) have only a Lexicon or only a Grammar?


Having only a Grammar and no Lexicon – constructions, but no parts from which to construct them – would be pointless. Having only a Lexicon and no Grammar is conceivable: but that would not be a very economic form of linguistic know-how, because the Lexicon required to deal with the sort of complex expressive and communicative requirements that humans are facing would have to be enormous.

No such natural languages are attested. (Perhaps it was from a “lexicon language” that Language first originated, some 200–100,000 years ago.) Dividing up the expressive labour between (i) basic building blocks and (ii) rules/constraints for constructions consisting of these basic building blocks is vastly more economical. (It enables one “von endlichen Mitteln einen unendlichen Gebrauch [zu] machen”, as Wilhelm von Humboldt liked to put it.)

And this is what all known human natural languages, as well as many animal communication systems, are doing.


B. Do all languages equally differentiate all levels of meaningful grammatical units distinguished in the diagram above:

morphemes and words, words and phrases, phrases and clauses, clauses and sentences, sentences and paragraphs/turns, paragraphs/turns and texts/ discourse?

• If all words in a language consisted of just one single morpheme, there

would be little point in distinguishing such kinds of units. (See ISOLATING/ANALYTIC languages, e.g. Vietnamese, vis-à-vis SYNTHETIC

languages, e.g. Latin and, more modestly, also English.) • If all phrases in a language consisted of just one single word, and if no two

words in any clause were closer partners in constructions than any two other words, on which grounds and for which purposes would one want to distinguish words and phrases? The single kind of unit occurring in such a language could arbitrarily be either called “word” or “phrase”.

(See NON-CONFIGURATIONAL/FLAT languages; e.g. Kalkatungu, an Australian language; perhaps Latin, too.)


• If all clauses in a language consisted of just one single (poly-morphemic) word, distinguishing Syntax (clause structure) and Morphology (word structure) would seem superfluous; there would only be a single component of grammar, to be arbitrarily called “word syntax” or “clause morphology”.

(See POLYSYNTHETIC/INCORPORATING languages with their SATZWÖRTER; e.g. Greenlandic Eskimo.)

• If all sentences in a language consisted of just one (main) clause, why

would one want to distinguish such levels of construction? (Study SPONTANEOUS/SPOKEN/CONVERSATIONAL speech, of English or any

other language whose planned, written form clearly distinguishes these levels.) • If all paragraphs/turns in a language consisted of just one single sentence,

and all texts or all kinds of discourse (conversational or other) of just one paragraph/ turn, there would be no need to assume a text/discourse grammar as distinct from sentence grammar.

(But there probably are no such languages, even among the most taciturn and terse peoples.)


Before we illustrate which kinds of languages do exist, differing in the kinds of level distinctions in constructing complex expressions their grammars recognise, a few introductory words are in order concerning the scholarly and every-day vocabulary for dealing with such matters.


C. Do the grammar-writing traditions of all languages (which have such a tradition) recognise all these levels of meaningful units through special technical terms?

Are some levels perhapsconceptually/terminologically more salient/basic than others?

English MORPHEME (ultimately borrowed from Greek morphḗ 'form, Gestalt');

WORD (basic word of everyday language, of Old English and Germanic origin, though originally of more general meaning, 'speech, talk, utterance, statement, report, sentence, word', probably from a Proto-Indo-European verb root 'to speak, say', same root as the word verb);

PHRASE (borrowed from Latin and ultimately Greek phrasis, 'speech, way of speaking, enunciation', from verb 'to tell', itself derived from 'to think');

CLAUSE (borrowed from French, ultimately Latin legal language, 'end of an argument, conclusion, stipulation, brief statement'; cf. German Klausel);

SENTENCE (borrowed from French and ultimately Latin sententia 'thought, way of thinking, judgment, authoritative pronouncement', from verb 'to feel, perceive, be of opinion'; cf. sentence in legal language)


German MORPHEM, WORT, (SATZGLIED), (GLIEDSATZ), SATZ (loan translation of Latin term from philosophical logic, propositio 'proposal, submission, fundamental assumption' = Setzung, from verb ponere 'stellen, setzen, legen'; cf. Satz in sports or musical language, corresponding to English set and movement, or as designation of a unit of movement, 'a jump', or in printing, etc.) French MORPHÈME, MOT (originally 'remark, short speech', from Latin mutum 'grunt, murmur'; cf. motto, mutter), (?), (?), PHRASE

...


Basic everyday words for the unit WORD are very common: Italian parola, Spanish palabra, Russian slovo, Maltese kelma, Turkish söz(-cük), ...

The other levels fare less well as regards everyday vocabulary.

But note the semantic vagueness/ambiguity and historical changeability of this entire vocabulary domain (English phrase ≠ French phrase, feeling ≈ thinking ≈ expressive-noise-producing ≈ speaking).


******************* [Supplementary discussion of the unit SENTENCE, from my (German) lecture notes to Einführung in die Linguistik, a long time ago]

“Was ist ein Satz?” (beliebter Titel von Aufsätzen und Büchern, in denen Hunderte von Satz-Definitionen gesammelt und gegeneinander abgewogen werden, und vom jeweiligen Autor jeweils einer anderen die Krone aufgesetzt wird)

Versuche zu antworten aus den Erkenntnissen heraus, dass Sätze (i) was bedeuten, (ii) was bezwecken, und (iii) von eigenartiger Bauform sind

• “Der Satz ist ein Bild der Wirklichkeit.” (Ludwig W., ein Österreicher)

• “Der Satz ist das Mittel dazu, die nämliche Verbindung der nämlichen Vorstellungen [die sich in der Seele des Sprechenden vollzogen hat] in der Seele des Hörenden zu erzeugen.” (Hermann Paul)

• “The sentence is the largest unit of grammatical description.” (apokryph, oft Leonard Bloomfield zugeschrieben, für den ein Satz genaugenommen das war: “an independent linguistic form, not included by virtue of any grammatical construction in any larger linguistic form”.)


Der Satz im deutschen, englischen, ... Alltagsverständnis Umfrageergebnisse: • Sätze sind (schriftsprachliche) Einheiten, an deren Anfang ein Großbuchstabe stehen muss und an deren Ende bestimmte Satzzeichen stehen (Punkt, Fragezeichen, Ausrufezeichen, evtl. Doppelpunkt, Strichpunkt). • Sätze sind (lautsprachliche) Einheiten, vor und nach denen ausgedehntere Sprechpausen gemacht werden können, ohne dass dieses durch Pausieren als besonders auffällig empfunden würde, und die durch bestimmte einheitliche Tonverlaufsmuster gekennzeichnet sind. Das heisst: Sätze haben einen Anfang und ein Ende, die in Laut und Schrift markiert werden – anders, vielleicht deutlicher als die Grenzen von anderen sprachlichen Einheiten. (Mit möglicher Ausnahme des Wortes: Wie sind Grenzen von Wörtern markiert?)


Der Satz im deutschen Wörterbuch (habe vergessen, welches) Satz, der 1. in sich abgeschlossene, nach bestimmten Regeln der Intonation (bzw. Orthographie), Grammatik, Stilistik und Logik im allgemeinen aus mehreren Wörtern zusammengefügte sinnvolle sprachliche Einheit 2. wissenschaftlicher Lehrsatz 3. das Setzen eines zum Druck bestimmten Manuskripts in Lettern und Zeilen durch den Setzer oder auf fotomechanischem Wege 4. Niederschlag, Rückstand von festen Bestandteilen, der sich aus einer Flüssigkeit abgesetzt hat 5. festgesetzter Geldbetrag, Tarif 6. in sich geschlossener Teil eines mehrteiligen Musikwerkes 7. in sich geschlossener Teil eines mehrteiligen sportlichen Wettkampfes, bes. beim Tennis [nicht z.B. beim Fußball: eine Halbzeit ist kein Satz. Warum nicht? Nicht in sich geschlossen? Nicht selbst intern wohl-strukturiert?] 8. mehrere zusammengehörige Gegenstände derselben Art (bei Maßangaben) 9. Sprung Gemeinsamer Nenner zumindest von Nr. 1, 2, 6, 7, 8 und eventuell 9:

Ein Satz ist ein in sich relativ abgeschlossener und in seinem Aufbau wohl-geregelter Teil eines (in seinem Aufbau weniger wohl-geregelten) größeren Ganzen (Paragraph, Text), der aus Teilen besteht, die ihrerseits in sich nicht bzw. weniger abgeschlossen sind (obwohl selbst auch wohl-geregelt: Teilsätze/clauses, Phrasen, Wörter).


Aus welchen Teilen (Sprach-)Sätze aufgebaut sind und welchen Regeln dieser formale Aufbau folgt, sind empirische und in manchen Hinsichten offene Fragen, deren methodische Beantwortung ein zentrale Aufgabe der Sprachwissenschaft ist. (Eine Einführung wie diese kann diese Antworten deshalb auch nicht geben: sie führt eigentlich nur ein ins erste Fragen.) Und damit das nie vergessen wird: Das empirische Betätigungsfeld der Sprachwissenschaft sind die menschlichen Sprachen in ihrer Gesamtheit, in denen sich die Einheit der menschlichen Sprachfähigkeit in ihrer Vielfalt äussert. Derzeit sind noch an die 6-7000 als Muttersprachen in Gebrauch; über nicht wenige davon wissen wir nur schlecht Bescheid, wenn überhaupt. Wieviele es davon in der 100 bis 200.000 Jahre langen Menschheitsgeschichte insgesamt gegeben hat, wissen wir nicht, denn viele haben kein Zeugnis hinterlassen. Wie wenige von den jetzt noch gesprochenen Sprachen in näherer Zukunft noch gesprochen werden, können wir in etwa sagen: ziemlich wenige, denn so an die zwei Drittel werden in hundert Jahren nicht mehr (muttersprachlich) gesprochen werden. Wollte man erst dann anfangen, Syntax zu treiben, wenn geklärt ist, einfürallemal und überallezweifelerhaben, was ein Satz ist (in jeder der menschlichen Sprachen – denn möglicherweise gibt’s da kleinere oder auch größere diesbezügliche Unterschiede), müsste man damit lang warten, und käme vielleicht nie dazu. (Wie eben diese Autoren von Büchern, die “Satz” definieren wollen.)


Der groben formalen Charakterisierung von Sätzen sollte man noch eine funktionale Charakterisierung an die Seite geben (auch in der Annahme, dass sich bei Menschenwerk sowie bei den Werken der Evolution Formen und Funktionen irgendwie entsprechen, ungefähr wie bei Schraubenziehern und anderen Werkzeugen oder Giraffen und ihren langen Hälsen): Ein Satz ist etwas, womit man etwas tun kann (was man anderweitig nicht oder nicht so gut tun kann). Und was? Antwort: Einen Sprechakt ausführen. Vielleicht nicht jeden x-beliebigen Sprechakt, aber beispielsweise Sprechakte der folgenden Art: • eine Person oder ein Fahrzeug taufen • jemanden zu einer Strafe verurteilen oder auch freisprechen • jemandem zustimmen oder widersprechen • das Wetter vorhersagen (zumindest pauschal) • jemandem ein Kompliment machen • jemandem die Uhrzeit sagen • jemandem ein Versprechen geben • jemandem eine Frage stellen • jemandem eine Anweisung bzw. einen Befehl geben • eine Bitte äussern • ein Verbot aussprechen • über einen Gegenstand (Person, Ding, Abstraktum) eine Aussage machen bzw. eine Feststellung treffen


• eine Lüge begehen • einem Wunsch Ausdruck geben • seiner Freude, Überraschung, seinem Ärger Ausdruck geben bitte fortsetzen

Wer beim Fortsetzen Hilfe braucht, kann etwa konsultieren: John L. Austin, How to do things with words (1962) – das posthume Buch eines Philosophen, Lehrer für Moral Philosophy in Oxford, das aber auch dem Linguisten was sagt. Auch wenn der Titel korrigiert werden sollte: How to do things with sentences.


Ist ein Satz auch gut geeignet, folgendes zu tun? • eine Geschichte, ein Märchen, einen Roman, einen Witz, eine Anekdote erzählen • ein Erlebnis erzählen • einen Vorfall berichten • ein Fußballspiel kommentieren • beweisen, dass die Winkelsumme eines Dreiecks gleich zwei rechten Winkeln ist (in der Euklidischen Geometrie) • jemanden davon überzeugen, dass der Abendstern der Morgenstern ist • eine Unterhaltung führen

• sich auf jemanden oder etwas beziehen und diesen Referenten so identifizieren, dass der Hörer weiss, wer oder was gemeint ist • einen Umstand (des Ortes, der Zeit, der Art und Weise, des Grundes oder Zweckes) einer Handlung oder eines anderen Ereignisses angeben

• um Hilfe schreien • jemandem rufen • jemanden beleidigen • seinem Schmerz Ausdruck geben, wenn man versehentlich mit dem Finger eine heisse Herdplatte berührt • dem Telefonpartner zu verstehen geben, dass man noch am Apparat ist • sich räuspern


• Schrauben eindrehen und ziehen bitte fortsetzen Für die erste Gruppe der oben erwähnten Handlungen (eine Geschichte erzählen usw.) wird man sich am besten komplexerer sprachlicher Formen bedienen, nämlich eines ganzen (aus mehreren Sätzen bestehenden) Textes oder Diskurses, als beim Äussern einer Bitte, eines Wunsches, einer Feststellung, einer Lüge usw., was man mit einem Satz gut machen kann. Für die zweite Gruppe von Handlungen (Referieren auf Personen/Dinge/Ereignisse, Ereignis-Umstände angeben) wird man gut mit weniger kompexen Einheiten als Sätzen auskommen, mit Phrasen (im grammatischen Sinn) nämlich. Für die dritte Gruppe von Handlungen gibt’s typischerweise Äusserungsformen von ganz eigener Art. (Wobei man sich streiten kann, ob diese Formen eine eigene Art von Sätzen sind bzw. eine Art von Satz-Wert haben.) Schraubenziehen und sowas erledigt man am besten nichtsprachlich.


(Deutsche) Beispiele: • eine Aussage machen bzw. eine Behauptung aufstellen bzw. eine Feststellung treffen: Scipio zerstörte Karthago. Der Abendstern ist der Morgenstern. Es regnet. Mir graut vor dir. Gestern wurde gestreikt. Trifft Scipio Hannibal. (nur am Anfang einer bestimmten Art von Text: Witz) Morgen fuhr Aschenbach nach Venedig. (nur in einer bestimmten Art von Text: Fiktion) • Vermutungen ausdrücken Scipio soll Karthago zerstört haben. Er wird krank sein. Sie haben wohl wieder verloren. Ich vermute, dass sie wieder verloren haben. • Voraussagen machen Morgen wird/soll es regnen. • Fragen stellen Ist Meier krank? Wo ist Meier? Warum denn so traurig? Meier ist krank, oder/nicht wahr/gell/hm? (eher: Bestätigung suchen?) • Befehle, Anweisungen geben, auffordern Halt! Kommen Sie morgen wieder. Lass das! Vor Gebrauch schütteln. Hilf mir mal einer! Dass du bloß pünktlich bist! Rauchen verboten • einem Wunsch Ausdruck geben Wenn es endlich regnete! Der Herr sei mit Euch. • seiner Freude, Überraschung, Ärger Ausdruck geben Oh! Wie ruhig es hier ist! Mir reicht’s! Meier (und) ein Heiratsschwindler?! Ausgerechnet “Scipio” haben sie ihn getauft!


• bestreiten, widersprechen, leugnen Scipio ist harmlos. Ich war es nicht. • jemanden/etwas taufen Ich taufe dich hiermit auf den Namen “Scipio”. • zwei Leute verheiraten Ich erkläre euch hiermit zu Mann und Frau. • jemanden rufen Se epp! (oder bairisch Se pp-e!) • jemanden beleidigen Esel! Sie Esel! Sie sind ein Esel. • sich auf etwas/jemanden beziehen und den Referenten geeignet identifizieren, worüber etwas ausgesagt werden soll er; Scipio; ein römischer Feldherr; der Mann, der zu sagen pflegte “ceterum censeo Carthaginem esse delendam”; die Zerstörung Karthagos durch Scipio (sich auf ein Ereignis beziehen; vergleiche mit Feststellung: Scipio zerstörte Karthago)


Die Idee ist natürlich, dass der Satz (bzw. Sätze aller Art) die Bauform, die er hat, letztlich deshalb hat, weil homo sapiens loquens damit auf zweckmäßige Art und Weise ganz bestimmte Zwecktätigkeiten ausüben kann.

Ein Satz ist – der Form nach – ein in sich relativ abgeschlossener und in seinem Aufbau wohl-geregelter Teil eines (in seinem Aufbau weniger wohl-geregelten) größeren Ganzen, der aus ihrerseits in sich nicht bzw. weniger abgeschlossenen Teilen besteht (also eine Einheit mittlerer, aber irgendwie besonderer Komplexität), womit – dem Zweck nach – Vorstellungen und Empfindungen ausgedrückt und dieser Ausdruck in Sprechakten bestimmter Art zur Ausführung kommt.

Zumindest hoffen wir, dass in diesem Fall die Form der Funktion angemessen ist. Es könnte natürlich auch sein, dass die sprachlichen Formen, mit denen wir Heutigen diese Funktionen zu erfüllen suchen, ein Zufallsprodukt der Entwicklung sind: vielleicht gab es ja Sprechergemein-schaften, die sich sehr viel zweckmäßigere Formen gebildet hatten, aber sie sind im Kampf mit den Elementen und mit ihren Nachbarn untergegangen. Wenn die Formen, von Sätzen und anderen sprachlichen Einheiten, die sich durchgesetzt haben, allzu unzweckmäßig wären, wären sie sicher im Lauf der zur Verfügung stehenden Zeit wenigstens ein wenig optimiert worden. END OF INTERLUDE.

*********************************


Back to the question of how languages differ in recognising, or not recognising, level distinctions for constructions. English, to anticipate one result, recognises them all. Languages differ in morphology in many ways. One very general difference is that languages can have more or less of it. This is what quantitative morphological typology is about, and the basic idea here is that languages as a whole can be ordered in this dimension:

ANALYTIC ------------- SYNTHETIC ------------- POLYSYNTHETIC (a.k.a. ISOLATING) (a.k.a. INCORPORATING)

Analytic languages have little morphology (compounding and perhaps reduplication and some further word formation seems the minimum that all languages have); synthetic languages have much morphology, and polysynthetic languages have even more morphology (typically clustering around the verb, with the verb as the core of the clause “incorporating” much that analytic languages would express as separate syntactic parts of clauses).


Such morphological differences go hand in hand with other differences in the use and organisation of grammar – and this is why they are of special interest for us, whose focus is on syntax:

• There is a division of grammatical labour between morphology and syntax, and what work is done by morphology won’t need to be done by syntax too, and vice versa. (The lexicon is also involved in this job-sharing:

when there are words to express a meaning, why also trouble grammar to construct it?)

• As to level distinctions, analytic languages, barely having any morphology, require almost no distinction of morphemes and words, a distinction which is crucial for synthetic languages.

• For polysynthetic languages a distinction of words and clauses is largely redundant, because these two units typically coincide.

• Certain synthetic languages do not distinguish words and phrases in the same way as analytic languages do, and perhaps not at all.


Distinguishing languages as having “little” or “much” morphology now seems rather vague. However, there are ways of being more precise about the extent to which languages have morphology.

One approach is to calculate the average ratio of numbers of morph(eme)s to numbers of words in representative texts of a language:

The higher the ratio, the more synthetic the language, and the lower, the more analytic.

Being synthetic or analytic thus is a continuum, not an either-or distinction, and the position of a language on this continuum is defined through its average ratio of morphemes to words. (Establishing what is the average for a language is one challenge here, though.)


Latin (Indo-European family, Italic subfamily, the ancestor/proto-language of the Romance languages) will serve as an example of a language sitting squarely on the synthetic side of the continuum:

domin-u-s am-a-t ancill-a-s pulchr-a-s master-THEME-NOM.SG love-THEME-3SG.PRES.IND.ACT maid-THEME-ACC.PL beautiful-THEME-ACC.PL

‘(the/a) master loves (the) beautiful maids.’

On this (plausible) analysis of Latin – expressed through segmentations (blanks = word boundaries, hyphens = morpheme boundaries) and the category labels in the interlinear morphemic glossing (for simplicity gender is here subsumed under THEME) – there are 12 morphemes and 4 (morphological-syntactic-lexical) words; syntheticity quotient therefore 12 : 4 = 3.0.

This is the quotient for this particular sentence, but it would seem to be a sentence that is typical of Latin as a whole, and more extensive text counts would no doubt confirm a figure in this area.


Compare this with the English translation equivalent (assuming both Latin noun phrases are intended as definite):

the master love-s the beauti-ful maid-s DEF master love-3SG.PRES.IND DEF beauty-ADJCT maid-PL

9 morphemes : 6 (morphological-syntactic-lexical) words = syntheticity quotient 1.5 – which is lower than that of Latin and rather close to the minimum value of 1.0, analyticity.

If the definite article is analysed as morphologically complex, th-e DEF-ART, then the figures are slightly different: 11 : 6 = 1.83, which is still lower than the result for Latin – which confirms, and makes more precise, the impression that English has less morphology than Latin.

Again, more extensive text counts are necessary to confirm this result.


A good text to begin with the more extensive counting, and in the process practise morphological analysis, is the Universal Declaration of Human Rights, because it is available in many languages, including two alternative Latin translations (the Lord’s Prayer used to be the favourite parallel text for comparative linguists in the past):

Omnes homines dignitate et iure liberi et pares nascuntur, rationis et conscientiae participes sunt, quibus inter se concordiae studio est agendum.

http://www.ohchr.org/EN/UDHR/Pages/Language.aspx?LangID=ltn

Omnes homines liberi aequique dignitate atque juribus nascuntur. Ratione conscientiaque praediti sunt et alii erga alios cum fraternitate se gerere debent.

http://www.ohchr.org/EN/UDHR/Pages/Language.aspx?LangID=ltn1

All human beings are born free and equal in dignity and rights. They are endowed with reason and conscience and should act towards one another in a

spirit of brotherhood.

http://www.ohchr.org/EN/UDHR/Pages/Language.aspx?LangID=eng


Turkish (Turkic subfamily of Altaic, a family widely spread over Eurasia, and reaching as far north as Lithuania and as far west as Berlin Kreuzberg)

(1) Ev-ler-i al-dı-k house-PLURAL-SPECIFIC.ACCUSATIVE buy-PAST-1PL.SUBJECT

‘We (have) bought the houses’

• 2 words in Turkish, 4 (or 5) in English; • 8 basic building blocks, to judge by the gloss:

4 combined in the first word, 4 in the second in Turkish, with some building blocks given separate expression that would be

cumulated in languages such as Latin, e.g. PLURAL number and ACCUSATIVE case;

morphemes-per-word ratio for this sentence: 6 : 2 = 3, comparable to Latin.


(2) Tebrik ve tehekkür-ler-im-i sun-ar-ım congratulation and thank-PLURAL-1PL.POSSESSOR-SPEC.ACC present-AORIST-1SG.SBJ ‘I offer my congratulation and thanks’

• The first word seems morphologically simplex, in comparison with the third, consisting of 4 morphological parts: How come? (Inflections are only expressed once, with the second conjunct.)

Morphemes-per-word ratio: 9 : 4 = 2.25

(3) daya-n-ıh-tır-ıl-amı-yabil-ecek mi-ymih-iz? prop.up-REFL-RECIP-CAUS-PASS-IMPOTENTIAL-POTENTIAL-FUTI INTERROG- INFERENTIAL-1PL.SBJ ‘Is it said that we may not be able to be made to practise mutual aid?’

• Remarkable! 2 words in Turkish, 16 in English!

Morphemes-per-word ratio: 11 : 2 = 5.5, reaching a new high


(4) Resim-ler-imiz kardeh-ler-iniz-in-ki-ler-den kıymet-li-dir picture-PL-1PL.POSS brother-PL-2PL.POSS-GEN-PRO-PL-ABL value-ADJECTIVISER-be.3 ‘Our pictures are more valuable than those of your brothers’

• 3 words in Turkish, 10 in English!

Morphemes-per-word ratio: 13 : 3 = 4.33

Overall, taking the average of our four examples sentences (3.77), Turkish comes out as more synthetic than Latin. [Source of examples and analyses: Lewis, G. L. 1967. Turkish grammar. Oxford: Clarendon Press.]


Bütün insanlar hür, haysiyet ve haklar bakımından eşit doğarlar. Akıl ve vicdana sahiptirler ve birbirlerine karşı kardeşlik zihniyeti ile hareket

etmelidirler.

All human beings are born free and equal in dignity and rights. They are endowed with reason and conscience and should act towards one another in a

spirit of brotherhood. (Article 1 of the Universal Declaration of Human Rights)

http://www.ohchr.org/EN/UDHR/Pages/Language.aspx?LangID=trk


But regardless of whether a language has as much morphology as Turkish or Latin (3.77 and 3.0 morphemes per word on average respectively) or as little as English (1.5 morphemes per word), such languages equally recognise a distinction between words and morphemes. Turkish, Latin, and English all have two distinct levels of constructions, morphological and syntactic, with morphemes and words as their respective constituent parts.

The difference between such languages consist in how much and exactly what responsibility they give to morphology and to syntax. In Turkish and Latin, since there is more morphology to be relied on than in English, it will have to do more grammatical work, work done by syntax in English.


But things can also be more radically different, as will be illustrated through two further languages:

Vietnamese comes close to being the extreme of an analytic language (syntheticity quotient close to 1.0);

West Greenlandic Eskimo is prototypically polysynthetic (with a syntheticity quotient much higher than that of Turkish and Latin, namely above 7).

Such languages get along without distinguishing levels of constructions in the same way as might be taken for granted on the evidence of the likes of Turkish, Latin, or English.


tiêńg Viêt(nam), Viêt-ngû (Vietmuong subfamily, Mon-Khmer family, Austro-Asiatic phylum)

(1) Sáng nay tôi uôńg hai tách cà.phê morning this me drink two cup coffee ‘I drank two cups of coffee this morning’

• 7 words in Vietnamese, 8 in the English translation (to go by blanks in the written form) – not a big deal.

• 3 or 4 words in English are complex (cup-s [PLURAL], drank [PAST of drink], I [SUBJECT case, SINGULAR number of me, we ...], th-is?),

none is in Vietnamese. • The complex “words” in English are instances of inflection, that is, one word-

form of the several word-forms realising a lexeme; in Vietnamese, lexemes are not realised by several word-forms, but – with a

very few exceptions (see below) – only by one.

The morphemes-per-word ratio for this sentence therefore: 7 : 7 = 1.0 – couldn’t be less synthetic/more analytic!


(2) Tôi àn lót-då o· câu.låc.bô, chú´ không phåi o· ho·p.tác.xá me eat line-stomach at club but not correct at cooperative ‘I ate breakfast at the club, and not at the cooperative’

• lót-då ‘to line [one’s] stomach’, two stems/words, a verb followed by a noun, in a morphological construction forming one complex word/lexeme, i.e., a compound.

Morphemes-per-word count for this sentence: 11 : 10 = 1.1, still very much at the analytic end of the continuum.

(3) mùa-màng REDUPL-crop ‘crops, vegetation’

• reduplication to express, among other notions, that of COLLECTIVE.

(4) canh-kiêć soup-EMOTIVE (a suffix) ‘soup and the like’

• probably the only genuine affix of the language


Summary:

Vietnamese confirms the hypothesis that if there is any morphology, it will be compounding; if a little more, perhaps reduplication for notions that this kind of exponent is well suited to express; if yet more, a suffix of none-too-specific meaning. Overall, such morphology does not raise the average morphemes-per-word count much above 1.0, and Vietnamese therefore is a paradigm case of an analytic language. (In fact, it is exceptionally radical in taking analyticity to the extreme, much further than Chinese, which is often cited as an example.)

Constructions in Vietnamese are almost exclusively syntactic; hardly any are morphological. Distinguishing morphemes, as constituent parts of morphological constructions, from words, the minimal units of syntax, is necessary even for Vietnamese, but it is a distinction of very limited usefulness for its grammar.

Vietnamese almost exclusively relies on syntax (and the lexicon) and gives far less responsibility morphology than do Turkish and Latin or also English.


[Source of examples and analyses: Nguyê͂n Ðình-Hoà. 1997. Vietnamese. (London Oriental and African Library 9.) Amsterdam: Benjamins.]

All human beings are born free and equal in dignity and rights.

They are endowed with reason and conscience and should act towards one another in a spirit of brotherhood.

(Article 1 of the Universal Declaration of Human Rights)

http://www.ohchr.org/EN/UDHR/Pages/Language.aspx?LangID=vie


Kalaallisut (a.k.a. West Greenlandic Eskimo, the language of the Kalaallit, the indigenous inhabitants of Greenland, having settled there in the 13th century, long before the Danes and earlier European whalers arrived; a member of the Inuit subfamily of the Eskimo-Aleut family, at home in the entire Arctic area)

(1) (kissartu-mik) kavvi-sur-put

(hot-INSTRUMENTAL) coffee-drink-3PL.INDICATIVE ‘They drank (hot) coffee’

morphemes-per-word: 5 : 2 = 2.5

(2) Nuum-muka-ssa-atit Nuuk-go.to-FUTURE-2SG.INDICATIVE ‘You will go to Nuuk’



(3) ikiu-palla-ssa-vakkit help-quickly-future-1SG/2SG.INTERROGATIVE

‘Shall I help you a moment?’


(4) tusaa-nngit-su-usaar-tuaannar-sinnaa-nngi-vip-putit hear-not-PARTICIPLEINTRANS-pretend-always-can-not-really-2SG.INDICATIVE ‘You simply cannot pretend not to hear all the time’


(5) aliikkus-irsu-i-llammas-sua-a-nira-ssa-gukku ... entertainment-provide.with-SEMITRANSITIVE-one.good.at-big-be- say.that-FUTURE-1SG/3SG.CONDITIONAL ‘If I should say that he is a good entertainer ...’



The average morphemes-per-word ratio for our example sentences is 7.13, way above Turkish (3.77), and Kalaallisut is rightly considered a paradigm case of polysyntheticity.

To flesh out this typological concept, which has implications for our question about the universality of the levels of units we introduced at the beginning:

• All example sentences are one word in Kalaallisut, but several words in the English translations – as many as ten in (4) and (5); only the adjective in (1), ‘hot’, an optional part, would add a second word (but there are also ways of “incorporating” adjectives).

• Many independent morphemes in English (“words”: adverbs, negation, modal auxiliaries, pronouns, ...), correspond to bound morphemes (affixes) in Kalaallisut.

• Many syntactic constructions in English (verb – verb, verb – noun) correspond to morphological constructions in Kalaallisut; English can do something that looks similar, but not as the normal way of clause construction: noun incorporation (a sort of compound), but not *English noun-incorporates.


• In fact, whole clauses and even multi-clause sentences in English (syntactic constructions) standardly correspond to single words in Kalaallisut (morphological constructions) – to the extent that the very distinction between these kinds of units, word and clause/sentence, which is so central to languages like English, begins to look doubtful for languages like Kalaallisut (hence the traditional term “sentence-words” for such constructions where the verb “incorporates” everything else – object, adverbials, subject, pronominals or also nominal).

[Source of examples and analyses: Fortescue, Michael. 1984. West Greenlandic. (Croom Helm Descriptive Grammars.) London: Croom Helm.]


Inuit tamarmik inunngorput nammineersinnaassuseqarlutik assigiimmillu ataqqinassuseqarlutillu pisinnaatitaaffeqarlutik.

Silaqassusermik tarnillu nalunngissusianik pilersugaapput, imminnullu iliorfigeqatigiittariaqaraluarput qatanngutigiittut peqatigiinnerup anersaavani.

All human beings are born free and equal in dignity and rights.

They are endowed with reason and conscience and should act towards one another in a spirit of brotherhood.

(Article 1 of the Universal Declaration of Human Rights)

http://www.ohchr.org/EN/UDHR/Pages/Language.aspx?LangID=esg


Kalkatungu (a language of West Queensland, now extinct, a member of the large Pama-Nyungan family which comprises most Australian Aboriginal languages) (1) cipa-yi thuku-yu yaun-tu yanyi icayi this-ERG dog-ERG big-ERG white.man bite (2) yaun-tu cipa-yi thuku-yu icayi yanyi (3) yanyi icayi cipa-yi yaun-tu thuku-yu (4) cipa-yi thuku-yu yanyi icayi yaun-tu (5) thuku-yu cipa-yi icayi yanyi yaun-tu (6) cipa-yi icayi yanyi thuku-yu yaun-tu ...

‘This big dog bit/bites the/a white man’ The morphemes-per-word ratio for such sentences in Kalkatungu is moderately synthetic: 8 : 5 = 1.6.


This is rather similar to English. But there is something else about Kalkatungu that couldn’t be more different from English (and vice versa: no judgment is intended that either one or the other is deviant): the ordering of the words of a clause: (i) Just about all permutations of the words of a clause are equally grammatical

(and more would be possible in Kalkatungu than have been given above in (1)-(6)); none of them is grammatically more basic or normal than the other, but the choice between them is a question of the pragmatic organisation of discourse (emphasis, contrast, given–new, topic–comment).

(ii) Words which would seem to closely belong together on account of their

meaning – the three which together pick out the active participant in the event described in (1)-(6): ‘this’, ‘big’, ‘dog’ (bold in the examples) – can easily be split up and be distributed all over the clause, in all chunkings and orders conceivable (and again, not all possibilities are exhausted in (1)-(6)).

When the ordering of words is “free”, not reined in by syntax but pragmatically driven, and when no linear continuity is required for words in semantic association, one begins to wonder whether there exists a level of constructions in such a language which gives groups of words cohesion: the phrase.


If there were no noun phrases in Kalkatungu clauses like those above, this would explain why those words which correspond to the members of a noun phrase in the English translation are not rigidly ordered relative to one another and can even be discontinuous. And this is indeed the conclusion that has been drawn for such languages. A structurally closer translation of (1) would therefore go like this: ‘this one, a dog, a big one, the/a white man, (it) bit/bites (him)’ The only constituents of clauses accordingly would be words, and these would be juxtaposed, being in loose apposition to one another rather than in tight syntactic construction, and arranged in any order that the context and the speaker’s priorities might dictate.


[Source of Kalkatungu examples: Blake, Barry J. 1983. Structure and word order in Kalkatungu: The anatomy of a flat language. Australian Journal of Linguistics 3. 143-166.

For another Australian language of this kind see this influential account: Hale, Kenneth. 1983. Warlpiri and the grammar of non-configurational languages. Natural Language & Linguistic Theory 1. 5-47.

Non-configurationality has been an intensely debated topic ever since. This recent survey, highlighting Australian languages, gives a useful concise summary: Nordlinger, Rachel. 2014. Constituency and grammatical relations. In Harold Koch & Rachel Nordlinger (eds.), The languages and linguistics of Australia: A comprehensive guide. Berlin: De Gruyter Mouton.]


Alongside the complexity hierarchy as set out above, there are further scales along which grammatical constructions can vary, and they are all aligned with one another: a hierarchy of cohesiveness and one of mobility: words least complex most cohesive least mobile phrases clauses sentences ...

texts/discourse most complex least cohesive most mobile

In terms of mobility, the recognition of a separate level would be justified if there were relevant differences between kinds of constructions. In Kalkatungu, while word-parts are wholly immobile (e.g., the ergative suffixes cannot be re-ordered to be prefixes), above the word level everything can be freely permuted (under pragmatic control), precluding further distinctions of levels.


In terms of cohesiveness the recognition of a separate level would only be justified if there is anything special about constructions setting them apart from other, less as well as more complex kinds of constructions.

In languages like Kalkatungu words are the most cohesive constructions, with their parts (morphemes) not only rigidly ordered, but always continuous and subject to strict construction rules; clauses are considerably less cohesive, with their parts (words) not subject to ordering and contiguity constraints and with less restrictive a grammar; and there is no recognisable further level of grouping parts into wholes in between.


In languages like English there is such an intermediate level, the phrase. A noun (or determiner) phrase in English is tightly organised, in terms of what kinds of words can form part of it and in which order they must appear; and they cannot be split up and redistributed over the entire clause: (7) QUANT DET QUANT MOD N POSTMOD [All these five ferocious dogs over there] [bit [the white men]] The internal cohesion of phrases in English and its kind is less tight than that of words, and positional licenses are more liberal: word-parts permit almost no interruptions (Is put on one word? It is interruptible: put it on), no re-orderings (affix en both suffix and prefix with adjectives, though not with one and the same: dark-en/*en-dark, en-rich/*rich-en), and no wild dispersals.

The cohesion of clauses is far less tight, even in English.

That of phrases is in between: certain additional parts can be interspersed (such as adverbs: all these five very ferocious huge dogs over there), and some re-orderings are possible, but usually require special marking (a ferocious five dogs: a pack of five dogs, not any individual five; *a too ferocious dog → too ferocious a dog).


A similar argument can be made for a verb phrase in English. Unlike apparently in Kalkatungu, the structural cohesion of the verb and its object here is tighter than that of verb and subject or that of subject and object.

The relevant evidence includes transformations as in (8), which cannot disassemble verb and object and have the pro-verb do “stand in for” just the verb alone: (8) a. What these ferocious dogs did was [bite the white men] b. *What these ferocious dogs did the white men was [bite] b'. What these ferocious dogs did (to) the white men was [bite them] Further, in nominalisations a transitive verb can “incorporate” its object, but not the subject: (9) ornithologists watch birds – the birdwatching of ornithologists *the ornithologist-watching of birds Further, verbs and objects are close enough to frequently form idiomatic units, while idioms consisting of just a subject and a transitive verb, to the exclusion of the object, which is to be interpreted literally, are quite rare:


(10) a. kick the bucket, break the ice, take the cake, split hairs, call the tune, pass the buck, smell a rat, twist someone’s arm, ... b. (German) ich glaub’ mich tritt ein Pferd; mich laust der Affe – any English counterparts? If a clause contains auxiliaries in addition to a main verb, there is clear evidence that they structurally belong with the verb: their order is strictly regulated and each auxiliary is influencing the morphological marking of the immediately following verb – which would be odd if they weren’t part of a cohesive construction. (11) MOD PERF PASS PROG V The white men must have been being bitten by these ferocious dogs inf partcp ing partcp Not only is cohesion tighter among words inside phrases than with words outside them: if there are phrases, then things grammatical must be expected to be going on inside them – special things, differing from what is going on inside other kinds of constructions. If nothing (special) were to go on, this might be because there are no phrases in the first place.


We could conclude that grammars can differ as to whether a level of phrases is recognised in between those of words and clauses, and if there is, which particular kinds of phrases are being distinguished (noun/determiner phrases, verb phrases, adpositional phrases, adjective phrases ...).

Unlike Kalkatungu, English is among those languages which do recognise phrases, and quite a variety of them are being distinguished. However, ...


... attention needs to be paid to certain manifestations of cohesion that are observed in Kalkatungu, too. While phrasehood in English is essentially manifested positionally – the members of a phrase occupy particular positions relative to one another, to a designated phrase member (such as the head of the phrase), and/or to a phrase edge – there is also overt grammar that links certain words in Kalkatungu, namely inflectional morphology. In Kalkatungu the agent of an action, the grammatical subject, is identified through a case suffix labelled “ergative”; the patient (grammatical object), by contrast, remains without overt case marking. This is how we can tell what (1)-(6) mean: ‘this big dog bit/bites the/a white man’, not ‘the/a white man bit/bites the big dog’. Now, this ergative case marker (which comes with several allomorphs) does not only occur once, but is suffixed to three words, regardless where they are placed: ‘this‘, ‘dog’, ‘big’. It is their counterparts which form a noun (a.k.a. determiner) phrase in English. That they receive (allomorphs of) the same case marker in Kalkatungu might be interpreted as another manifestation of phrasal cohesion. English largely lacks comparable inflectional morphology, but if it is available, a language may exploit it to signal cohesion through shared morphological marking – that is, through agreement.


Latin is more richly inflecting than English, as we saw earlier. And it does bind certain members of phrases together through agreement, too, including adjective and noun, agreeing in number, case, and gender (here subsumed under THEME for simplicity).

To judge by textual evidence, Latin word order was much freer than that of English, and probably almost as free as that of Kalkatungu. If permutations of the earlier (constructed) example like the following were permissible, and would not have been felt by the Romans to be strikingly exceptional, then Latin would be another candidate that might apply for membership in the club of languages lacking phrases, or at any rate noun phrases – unless phrasehood were to rest on agreement rather than position. ancill-a-s domin-u-s pulchr-a-s am-a-t maid-THEME-ACC.PL master-THEME-NOM.SG beautiful-THEME-ACC.PL love-THEME-3SG.PRES.IND.ACT ‘The master loves the maids (when they are) beautiful.’

pulchr-a-s domin-u-s ancill-a-s am-a-t beautiful-THEME-ACC.PL master-THEME-NOM.SG maid-THEME-ACC.PL love-THEME-3SG.PRES.IND.ACT ‘Of the beautiful ones the master (especially) loves the maids.’


For our purposes we will have to leave it at that. To summarise: Languages can either differ in recognising or not recognising (particular kinds of) phrases, as a type of constructions intermediate between words and clauses in terms of both complexity and cohesion. Or it may be more appropriate to attribute phrases to all languages (of the kinds considered here), and to assume that differences between languages are a matter of how phrases manifest themselves: through positionally-defined configurations of words (as in English) or through morphologically-marked agreement between words, which are thereby freed of positional restrictions (as in Kalkatungu and possibly Latin).


Postscript: Words, words, ... This class is about syntax, not morphology (see you again in Morphology I, coming winter semester) – other than insofar as it impacts syntax. But can we take it for granted that we know whether we are dealing with morphology or syntax?

As syntacticians, we would want to ignore here the internal structure of constructions of lowest complexity and highest cohesion, words. But precisely what about English utterances like the following, presented in informal spelling (transcribe phonologically), can be safely neglected by the syntactician, then?

He’ll hafta gimme Gus’s fish’n chips and a cuppa tea. He’s’nt hungry. Can’tcha eat’m? How many words? How many morphemes? Morphemes is easy: 27, here indicated through spacing. he ll haf ta gim me Gus s fish n chip s and a cup a tea he s n’t hungr y can n’t cha eat m?


Words are the crux. Of words in the sense of lexemes there are 25 (or only 24, depending on whether ’s is taken to represent the genitive or the possessive pronoun (h)is). In formal English, as preferred by dictionaries, these lexemes sometimes have other, prosodically stronger and segmentally fuller shapes than in the informally spelled example: He will have to give me Gus his (or Gus’s) fish and chips and a cup of tea. He is not hungry. Can you not eat them? (Note the order reversal: *Can not you eat them?) Syntheticity quotient therefore 1.08 (or 1.13).

If we had calculated this quotient on the basis of the informal rendering above, it would almost be double: 27 : 14 = 1.93. But that would have been wrong. The reason is that there is also a phonological sense of “word”. These weak forms written continuously with their hosts or separated from them by a hyphen (but without space) are phonologically bound. In a phonological sense they form single words with their hosts; hence only a single word stress and hence the phonological effects at the word-internal junctures (prosodic weakening, segmental reductions, devoicing of /hœv/ to /h´f/ etc.).


But they are not morphologically bound to a base word. Hence such forms, which are usually atonic versions of words which also have a full form, must be participating in syntactic rather than morphological constructing – and they therefore shouldn’t be counted in morphology and be neglected in syntax. This borderland of syntax and morphology, owed to the workings of phonology, is treacherous territory. Its name is cliticisation, and its inhabitants are clitics (which is Greek for ‘leaner’), and it should probably be proscribed for newcomers to syntax – if with a bad conscience. After all, all (spoken) languages appear to cliticise, in one way or another. And it is in this twilight zone that morphology is conceived in unions of lexicon and syntax, with phonology standing sponsor: what is now morphologically bound (i.e., an affix) is very likely to have previously been phonologically bound (i.e., a clitic). (Keywords: grammaticalisation, univerbation.) To be honest, it is sometimes very difficult to rationally decide – even for experienced grammarians and sophisticated theoretical linguists dealing with an almost overdescribed and overanalysed language such as English – whether a morpheme is morphologically bound to a base (an affix) or only phonologically leaning on a host (a clitic).


An instructive example is n’t, the English negation marker whose fuller form is not and which, in reduced form, only combines with auxiliaries (can’t, mustn’t, won’t, hasn’t, isn’t, ain’t etc.; but They don’t can/*cann’t sardines in Cannery Row now; I think not/*thinkn’t; To be or not to be/*orn’t to be).

English grammars standardly pronounce n’t an enclitic (a clitic leaning on its host’s back; front-leaners are “proclitics”); but there are rather convincing reasons to analyse -n’t as an inflectional suffix of auxiliaries – that is, as morphology rather than phonologically-touched syntax. (And English wouldn’t be the only language under the sun to have negative inflection.) Consult this paper for the argument, as well as for a useful checklist of individual criteria for the affix-clitic distinction:

Zwicky, Arnold M. & Geoffrey K. Pullum. 1983. Cliticization vs. inflection: English n’t. Language 59. 502-513.


The other way round, for all English grammars (that I am aware of) ’s as in Gus’s fish’n chips is a suffix, marking the only surviving case on nouns, the genitive. (Nominative, accusative, and dative were lost as Old English turned Middle and Modern.)

But if you compare the “genitive” with the plural marker -s of nouns (as in chip-s) or with the 3rd person singular marker -s of verbs on the one hand, both of which couldn’t be better suffixes, and with the reduced variant of the 3rd person singular indicative present tense form is of the copula verb on the other (as in He’s hungry), which fulfills all criteria for an enclitic, you are in for a surprise. Notice that these four morphemes show exactly the same phonologically conditioned allomorphy: /Iz, z, s/, depending on the preceding segment being a sibilant or otherwise voiced or voiceless. (So does the 3rd person singular masculine possessive pronoun his, with initial aitch dropped, hence virtually indistinguishable from genitive ’s.)

But they differ in many ways, including the following. (And we ignore the 3rd singular verb inflection here.)


spelling of stem ending in <y>: <ie> or <y> The cities were evacuated. Plural The city’s mayor disappeared. Genitive Singular The city’s empty. reduced copula is allomorphy of stems ending in voiced fricative: to voice or not to voice The knives are in the drawer. [naIvz] Plural The knife’s blade is blunt. [naIfs] Genitive Singular The knife’s in the drawer. [naIfs] reduced copula is location of /s/: on head noun or at the end of the whole phrase The dukes of Ruritania were all bald. Plural The duke of Ruritania’s brother is bald. Genitive Singular The duke of Ruritania’s bald. reduced copula is location of /s/: on head noun or at the end of the whole phrase Our mothers-in-law are both senile. Plural My mother-in-law’s hat is funny. Genitive Singular My mother-in-law’s senile. reduced copula is


location of /s/: on head noun or at the end of the whole phrase, regardless of the word class of the word ending a phrase (could be an adverb or a verb) word class of the carrier of /s/: noun only or no word class restrictions The cats over there are pretty. Plural The cat over there’s tail is bushy. Genitive Singular The cat over there’s pretty. reduced copula is

The cats we bought are pretty. Plural The cat we bought’s tail is bushy. Genitive Singular The cat we bought’s pretty. reduced copula is


Thus, despite identical allomorphy, there are differences between the /Iz, z, s/’s in the three constructions pointing in this direction:

• the regular exponent of plural is an affix, forming a morphological word with its base (on its own expressing singular meaning);

only nouns can be carriers of the plural suffix in English (not even adjectives can carry plural suffixes: The poor(*-s) envy the very rich(*-es)).

• The reduced form of the copula is is an (en-)clitic, forming a phonological word with its host;

words of any word class can host such an enclitic (as long as the words occur at the end of the phrase which precedes the copula verb).

Clitic and host do not form a morphological word.

• Genitive singular behaves like an enclitic rather than like a suffix on all these criteria distinguishing affixes and clitics.


There are a few complications, though. For instance, there are certain restrictions on genitive ’s which wouldn’t really be expected for a clitic, because clitics should not be choosy as to which hosts they lean on. But genitive ’s can be a bit choosy:

*the queen and I/me’s last meeting (cf. the last meeting of the queen and I/me);

*us/*we two’s meeting (cf. the meeting of us two);

the *boys/children of this country’s parents (cf. the parents of the boys/children of this country)

Perhaps it is most appropriate, therefore, to recognise genitive ’s as sharing properties of affixes and clitics; such mongrels have sometimes been called “phrasal affixes” (as opposed to “word affixes” and “phrasal clitics”) in the theoretical literature. And they are not a unique possession of English. But, as I warned above, this is treacherous ground we tread. It is safer to defer its exploration for a later occasion (Syntax II, III ...).

Documents

Components of linguistic know-how, levels of …ling.uni-konstanz.de/.../Syntax-I-SoSe-14/Syntax_basics_1.pdf · Components of linguistic know-how, levels of grammatical constructions: