neutral vs. no alignment

Wolfgang Schulze W.Schulze at LRZ.UNI-MUENCHEN.DE
Mon Sep 21 10:54:54 UTC 2009

Dear Martin and others,
starting from Peter's original question, it seems important to me to
strictly distinguish alignment from agreement. Alignment would denote
the phonological, or morphological, or syntactic, or pragmatic
behavioral pattern of referential units as pronounced in the schematic
representation of so-called intransitive and transitive structures (note
that in the following, I neglect split schemas, diathesis etc.). In
other words: alignment simply means how so-called intransitive schemas
are related to transitive ones (e.g. S=A;O, S=O;A, S;A;O, S=A=O etc.).
Alignment thus does not predict which properties of the given
referential units decide upon the position of the two co-schemas
(intransitive / transitive) on this scale (the Accustive Ergative
Continuum, AEC in my terms). I interpret the functional value of
referential units with respect to the AEC in terms of a vector-like
structure that encompasses semantic values (semantics roles), syntactic
values (such as foreground (> Subject) and Background (>Object), and
pragmatic values (topic/comment, hierarchies etc.). An alignment schema
is thus spelled out on the basis of the grammatical relations (or:
relational primitives), the vector value of which {sem;syn;pra} imay
vary from language to langue. E.g., a highly 'semantic' type of
alignment would set sem = 0.8 (in my terms '0' means not avctivated, '1'
= fully activated), but would still allow to set the pra and syn values
(e.g. A = {0.8;0.2;0.4} which would read: A is highly semantic (e.g.
Agent), but weakly syntactic (e.g. Subject) and less weakly pragmatic
(e.g. topic)). I think that there is no language that shows an absolute
value of the vector, e.g. *{1;0;0} or *{0;1;0} or *{0;0;1} (just as
there are no languages which purely ergative or accusative. Also note
that the sub-schema A=O can only be present, if A and O are
distinguished anyhow, e.g. lexically). Instead we can only describe
preferences, e.g. {0.8;02;02.} or {0.2;0.7;05}. The relational
primitives as such do not have a universal preference. Prototypically,
six primitives seem to be derivable from cognitive schemas: The four
basic primitives S , LOC (conceptual locative), A, O and the two
'variants' IO (indirect objective, the highly animate variant of O) and
IA (indirect agentive = (semantically) Instrumental), the non-animate
variant of A). The AEC suggests that A->O and S->[LOC] are intimately
related (they must have something in common, else the AEC schemas
wouldn't be motivated). In a paper entitled 'Cognitive Transitivity'
(, I have
argued that the A(~IA)->O(~IO) schema is nothing but a
metaphoric/metonymic extension of the S->[LOC] schema yielding either
[S>A -> LOC>O] or [S>O <- LOC>A]. In this sense, Martin is fully correct
in saying
> a language cannot lack alignment,
Personally, I would prefer saying that the schematic layer of human
utterances in which language so ever is always grounded in alignment.
But agreement is different: In my terms, agreement means the
articulatory 'overt' copying (or: echoing) of properties of lexically
expressed referential units onto the relational center (verb or verboid
structures). This type of 'referential echoes' primarily refers to the
relational primitives as such (preferably S, A, and O), but usually
subcategorize them according to one or more vector values (e.g. more
semantic, more, syntactic, more pragmatic). The question of which
primitive is copied onto the relational center is language-dependent,
even though there seems to be a strong preference to first copy
properties of the more central primitives (S=A or S=O etc.) before
adding echoes of the periphery ({O in S=A;O} and A in {S=O;A}). In
addition, subcategorization may export features of more specific
subcategories of a referential unit into the verbal domain, not
necessarily or fully linked to the alignment schema (such as number,
gender/class, person etc.). Now, if we have pure A-agreement or
O-agreement, this would mean nothing but a further subcategorization of
the given S/A/O alignment schema. One may hypothesize that here, we have
to deal with periphery agreement, linked an else cognitively 'distant'
referential unit to the verbal center. If we start from the schema:
[S=A][center];O[periphery] or [S=O][center];A[periphery], we may claim
that pure A-agreement is ergative in nature, and pure O-agreement is
accusative in nature (in both cases, the peripheral relational primitive
is the starting point of the echoing process [= agreement]).

For more see (among others) W. Schulze 2000. The accusative ergative
continuum. General Linguistics 37:71-155. Those of you who read German
may want to have a look at some arguments that I have put together in, etc.

Best wishes,



*Prof. Dr. Wolfgang Schulze*


/Primary contact: /

Institut fu"r Allgemeine & Typologische Sprachwissenschaft

Dept. II / F 13

Ludwig-Maximilians-Universita"t Mu"nchen

Ludwigstraße 25

Postanschrift / Postal address: Geschwister-Scholl-Platz 1

D-80539 Mu"nchen

Tel.: 0049-(0)89-2180-2486 (Secretary)

0049-(0)89-2180-5343 (Office)

Fax: 0049-(0)89-2180-16567// 0049-(0)89-2180-5345

Email: W.Schulze at
<mailto:W.Schulze at>/// Wolfgang.Schulze at
<mailto:Wolfgang.Schulze at>



/Second contact: /

Katedra Germanistiky'

Fakulta humanitny'ch vied

Univerzita Mateja Be'la / Banska' Bystrica

Tajovske'ho 40

SK-97401 Banska' Bystrica

Tel: (00421)-(0)48-4465108

Fax: (00421)-(0)48-4465512

Email: Schulze at <mailto:Schulze at>



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Lingtyp mailing list