When such factors influence grammatical choice. Psycholinguistic research

When
making grammatical choices, speakers who wish to convey an idea have many
alternative utterance forms available to accomplish their articulation goals (Quirk,
2010). For instance, in English a speaker could describe a single scene in many
different ways: The Cat is chasing the mouse, the mouse is being chased by the cat,
the mouse is fleeing the cat etc. This availability of syntactic alternatives
gives the speaker flexibility in developing an utterance plan, but also a
challenge in that in order to successfully communicate any message, they must
converge on one utterance plan (Bock, 1987). The mechanisms by which one
chooses a grammatical structure may be influenced by many factors. This essay
will describe and discuss factors that are believed to influence the final
structure of our utterances including how exposure to recent primed structures
and words may influence subsequent structures produced and how direction of
visual attention within our environment and the animacy of an entity can alter grammatical
choices. It will also briefly discuss the reasons such factors influence grammatical
choice.

Psycholinguistic
research has suggested that the structure of an utterance an individual has
previously produced or comprehended, may influence the grammatical choice they
subsequently make (Branigan & Pickering, 2017). More specifically, that
individuals are more likely to repeat the structure of a recently experienced (prime)
sentence when forming a subsequent sentence (Bock, 1986). This is referred to
as structural/syntactic priming effects or alignment.  Seminally, Bock (1986) reported such a
tendency, when participants were asked to repeat passive or active sentences
and then describe pictures that depicted transitive events. She found an
alignment effect in which participants were more likely to use a passive target
sentences (e.g. The Church is being
struck by lightning) after repeating a passive prime sentence (The referee was punched by one of the fans)
than after repeating an active (One of
the fans punched the referee). This alignment tendency result has since
been reliably replicated across psycholinguistic literature (Pickering &
Ferreira, 2008). For example, alignment effects have been found in natural
conversations (Gries, 2005; Rietter & Moore, 2006) laboratory tasks (Bock, 1986),
in spoken language (Bock, 1986) and written language (Pickering & Branigan,
1998; Gries , 2005). Moreover, structural alignment seems to transfer across
languages (Loebell & Bock, 2003; Salamoura & Williams, 2007) and even
tasks (Kaschak, 2007). Thus, given the array of contexts in which the effect
has been observed, including social and non-interactive tasks, it appears that
structural alignment is a robust factor that influences grammatical structure.

There
are several reasons why we can conclude that it is specifically the primed
syntax of a previous sentence that causes an alignment effect.  First, these effects cannot be explained in
terms of repetition of particular words in prime sentences (Pickering &
Ferreira, 2008). Bock (1989) demonstrated that participants exhibited a
tendency to use prepositional object (PO) dative sentences (The girl is handing a paintbrush to the man)
rather than double object (DO) sentences (The
girl is handing the man a paintbrush) after hearing a PO dative sentence
that did not even include the word ‘to’ (The
secretary baked the cake for her boss). Thus, word repetition was not the
reason for the prime effect as both the DO and PO target sentences share every
word bar ‘to’. Second, priming effect explanations couched in terms of meaning
can also be ruled out. For example, in both experiments outlined, the
alternative responses (the DO and PO or passive and active) reflect the same
events, as they can both describe the same picture accurately. Additionally,
research has found a priming effect between sentences that describe different
event types (e.g. Agent-Patient: The
doctor gets licked by the cow, and Experiencer-Theme: The king is being ignored by the bear), meaning that some form of ‘conceptual
prime’ cannot explain alignment (Messenger, Brangian, McLean & Sorace,
2012b).

Furthermore,
in a Dutch study, participants repeated the order of auxiliary and main verb
(geblokkeerd was ‘blocked was’ Vs. was geblokkeerd ‘was blocked’) despite
the fact that there is no difference in meaning between the two (Hartsuiker
& Westenberg, 2000). Taken together, these results suggest that there is a
form of priming which specifically primes the representations of syntactic
information but not semantic, phonological or lexical information. This
conclusion is further supported by the findings of a multitude of studies that
many syntactical structures are susceptible to priming effects, including the order of subject and locative (Hartsuiker,
1999) and the form of complex noun phrases (Cleland &
Pickering, 2003). Thus, we can
conclude relatively confidently that syntactic priming can influence one’s
subsequent grammatical choices.

Why does syntactic priming elicit such an effect on
subsequent utterances? One view posits that recently producing or comprehending
a linguistic structure causes an increase in the activation level of a
corresponding representation in memory relative to the representation of
alternative structures (e.g. the PO Vs DO dative), which increases the
probability that an individual will reuse the same sentence structure when
selecting between structural alternatives (Dell,
1986; Pickering & Branigan, 1998). Another view suggests that implicit
procedural learning mechanisms, that attend to the operations involved in sentence
production, cause this syntactic alignment tendency (Chang, Dell & Bock,
2006; Jaeger & Snider 2013). In one of the strongest formulations of this notion,
syntactic alignment the result of an ‘automatic perception-behaviour link’
(Pickering & Garrod, 2004:188), in which the mere perception of linguistic
behaviour (e.g. PO or DO use) increases automatically the probability that one
will engage in the same behaviour subsequently. 
This idea is inherited from observations that people are inherently
sensitive to their social environment and will therefore unintentionally mimic
any mannerisms and behaviours, (Bradac, Mulac & House, 1988; Chartrand
& Bargh, 1999; Maurer & Tindall, 1983). It is important to note that
the accounts outlined are not necessarily mutually exclusive and may interact
with one another, causing an alignment effect (Branigan, Pickering, Pearson
& McLean, 2010). Furthermore, the distinct components of said accounts may
vary according to the communicative context (Reitter and Moore, 2014) or may
work simultaneously at different levels of processing, causing an interaction
of effects (Bergmann and Kopp, 2012).

Interestingly,
recent research has suggested that structural priming effects can occur across
cognitive domains. Scheepers et al (2011) found that when participants were
primed with mathematical equations that either did/did not contain
parenthetical groupings e.g. 90 – (8+1) x 4 or 90–8 + 1 and then were presented
with a sentence fragment such as “The
tourist guide mentioned the bells of the church that … ,”; that the
mathematical equations structure influenced the structure of the noun phrase
used to complete the sentence- either ‘the
bells of the church’ or ‘the church’. This suggests that cross-domain
structural priming from mathematics to language could alter the structure of
people’s utterances. Therefore, the previous comprehension/production of
abstract non-linguistic structures may also influence subsequent grammatical
choices.

It is
also postulated that priming of specific verbs may affect structural choice. It
is thought that verbs carry both grammatical and lexical-semantic properties
(Myachykov, Garrod & Peepers, 2012). Therefore, it is assumed that for
example, priming a verb ‘punch’ before the presentation of a picture portraying
a transitive event (See figure 1) will not only lead to an increased
probability that the verb punch will be used to describe said event, but it
will also lead to the pre-activation of a typical associated map (e.g. the
presence of an agent acting upon a patient) and a pre-activation of structural
alternatives will ensue (e.g. a passive and an active structural frame). These
assumptions are supported by research by Melinger and Dobel (2005) who
demonstrated that isolated verb exposure can potentiate both structural frames
associated with subsequent sentence production. In this study, German
Participants were asked to read a dis-transitive verb that either permitted
solely a DO object frame (i.e. entziehen/to
deprive) or a PO frame only (i.e. adressieren/to
address). Researchers then described a visual event that was semantically
unrelated to either of the verb primes. Crucially, the verbs necessary for the
description of the target event permitted both structural frames.

(Myachykov,
Garrod & Peepers, 2012).

They
found that the frame that was associated with the isolated prime verb had an
influence over speakers’ structural choices within the target descriptions.
More specifically, Participants produced more DO alternatives after a DO
biasing verb prime and PO alternatives after a PO biasing prime. This suggests
that individual structural cues in the form of isolated verbs can implicitly
project their associated configurations onto newly generated utterances, independent
to the explicit structural frame of the sentence they appear in. Although
intriguing, there is little further empirical evidence for this effect, although
the subject of whether specific syntactic structures can be retrieved on the
basis on individual words is a current focus of psycholinguistic literature
(Salamoura & Williams, 2006).

Psycholinguistic
research has suggested that there is a link between specific syntactic
structure preferences and animacy (how alive or
sentient the referent of a noun is) (Gaméz & Vasilyeva, 2015). Many
researchers have reported that the default preference of active sentences over
passive in the English language can be reduced or reversed when the patient of
the action is animate (Branigan, Pickering & Tanaka, 2008). For example, Ferreira
(1994) found that in a paradigm in which participants were given 3 words
(consisting of 2 nouns and a past tense verb), when asked to make a sentence
with said triad; in a condition in which one noun was animate and the other was
inanimate, that participants would preferentially place nouns that denote
animates at the beginning of an utterance, even if it meant using passive
structures. Similarly, McDonald (1993) and Bock et al (1992) found in
methodologically distinct studies, that the presence of an animate word altered
participants’ structural preferences, so that the animate entity appeared
earlier in an utterance. This suggests that the presence of an animate entity can
influence grammatical choices by overriding inherent structural preferences.

Other
research has shown that word order preference can be influenced by animacy. In
a study with Japanese speakers, Tanaka et al (2005) found that participants
showed a greater tendency to misremember object-subject-verb sentences as subject-object-verb
sentences when the subject was animate compared to when it was presented as
inanimate. Similarly, Kempen and Harbusch (2004) found that altering animacy
affected participants’ choice of prepositional vs double object structures in
German, whilst in an English study, animacy was shown to affect the choice of
of-genitive versus s-genitive structures (Rosenbach,2005). Therefore, it
appears that animacy affects speakers’ structural preferences and word order
preferences, causing a tendency to place animate entities into positions that
are syntactically prominent.

However,
some of the supposed effects of animacy may perhaps be due to other important
influences on syntactic structure (Branigan, Pickering & Tanaka, 2008). For
example, in the English language, there is a general preference for short
elements to precede long elements (Hawkins, 1994; Wasow, 2002), for given
elements to precede new elements (Quirk, 2010) and for animate elements to
precede inanimate elements (Thompson, 1990). Importantly, these factors
correlate highly with one another, meaning that in general animate entities
tend to also be given and short. Thus, there is an issue whereby it is
difficult to disentangle each element’s relative influence on syntactic
preference. This can be visualised using the example below:

1.)
Helen gave him a textbook on German grammar.

2.)
Helen gave a textbook on German grammar to him.

Example
1 represents the most common structure used by English speakers (Rosenbach,
2005). It is impossible to know whether the direct object (him) was placed in a
syntactically prominent location (1) due it being highly given, short in
length, a pronoun or human (therefore animate). Arguably, any one of these
factors could therefore account for the preferential location of animate
entities in utterances (Harris, 1994).

Despite
the plausibility that some research may be explained by these ‘other
influences’, more recent research has demonstrated the independent effects of animacy.
Rosenbach (2005) found that even when the possessee and possessor were
controlled for in terms of their syntactic complexity and new/given information
structure, that participants exhibited a preference towards possessives in
which the animate entity preceded the inanimate entity (boy’s eyes) as opposed to vice versa (eyes of the boy). Likewise, Kempen and Harbusch (2004) demonstrated
in a corpus of German complement and adverbial clauses, that this tendency for
animate entities to precede inanimate entities was independent from all other
influences. This suggests that animacy is an independent processing factor that
influences grammatical choice variation, although discerning its relative
influence when all factors are involved has yet to be established (de Swart,
Lamers & Lestrade, 2007).

It is
apparent that animacy influences grammatical choice at least to some extent,
but how? Bock and Warren (1985) argued that animate entities are inherently more
‘conceptually accessible’ than inanimate entities. According to this concept of
‘conceptual accessibility’, the process by which one assigns grammatical
functions (e.g. direct object, subject etc) to words in a sentence is at least
partly determined by the ease with which lexical entries for various concepts
within the intended message are accessed. More specifically, the lemma (word
representations that encode their syntactic and semantic features) associated
with the most accessible concept will be assigned the greatest grammatical
function and as they are more quickly retrieved (Bock & Warren, 1985;
Christianson & Ferreira, 2005; Levelt, Roelofs & Meyer, 1999). This
notion can be used as an explanation as to why the presence of animate entities
can alter syntactic structure. As language production is incremental,
information that is easily accessible is processed first, meaning that animate
entities are privileged during the syntactic process of production and
therefore tend to appear as the subject or early in a sentence structure
(Branigan, Pickering & Tanaka, 2008). 

Research
has also suggested that one’s visual focus can affect subsequent structural
choice (Myachakov, 2007). For instance, Bock, Irwin, Davidson and Levelt (2003)
found that English and Dutch speakers’ preference for producing reports of time
varied as a function of their visual environment. When participants were given
a digital display, they showed a preference for using absolute reports of time
(e.g. Its nine fifteen), whereas when participants were given an analogue
display demonstrated a preference for relative reports (It’s a quarter after
nine). This suggests that grammatical choices can be influenced purely by
differences in what we visualise.

Additionally,
increasing an entity’s visual salience (by directing focus) can alter
subsequent grammatical structure. Gleitman, January, Nappa and Trueswell (2007)
found that visual cues that implicitly drew speakers attention to a specific
element of a picture (increasing its salience), lead to that entity being
mentioned earlier in the utterance, therefore affecting the structure of said
utterance. They presented participants with a picture depicting a dog chasing a
mailman and found that when they used a subliminal flash of light near the
mailman, implicitly drawing participants attention to this entity, subsequent
descriptions tended to be passive with the mailman as subject ‘The mailman is getting chased by the dog’;
whereas when the participants implicitly attended to the dog due to the flash
of light, an active voice was used in which the dog was subject , ‘The dog is chasing the mailman’.
Similarly, Forrest (1996) found that when they cued speakers attentional focus
to different locations prior to target event presentation, that differences in
cueing resulted in differences in utterance structure. More specifically, that
cueing for the location of an entity made participants more likely to make said
entity the subject of their utterance and therefore appear earlier in their
produced sentence. These results suggest that the both ones’ visual focus and
the specific visual salience of entities in our environment alter our structural
choices.

Making
something more visually salient is also thought to increase its conceptual
accessibility (Vogels, Krahmer & Maes, 2013). In the case of visual focus,
mere attention increases an entity’s salience, meaning accessibility effects
emerge from the gating functions of attention, which during an utterance plan
prioritise elements from long-term memory that are easily retrieved (MacDonald,
2013; Myachakov, Garrod & Scheepers, 2017). This results in the animate
entity tending to appear in syntactically privileged positions.  

Nevertheless,
there is not a simple linkage between early sentence positions and visually salient
elements. Rather it appears the effects of visual salience are task dependent,
in which visual information may be viewed/interpreted differently depending on
individual task goal. In a study by Kuchinsky, Bock, and Irwin (2011),
participants’ eye gaze patterns were different depending on whether they were
asked to read the time ‘normally’ (first small hand for hours and then big hand
for minutes) or ‘reversed’ (big hand minutes, small hand hours). This suggests that
goal-directed factors alter gaze pattern, meaning that the effects of visual
salience on grammatical structure is modulated by speakers’ tasks and goals in
addition to cues from their visual environment.

In
conclusion, it appears that when one is combining retrieved lexical entries
into syntactic structures several factors can independently influence the final
syntactic structure chosen. This includes, whether a previous structure has
been primed, what element of one’s visual environment is salient at that
present moment and whether an animate entity is involved in an utterance plan. However,
these are by no means the only factors that influence grammatical structure
when producing language. Prat-Sala (1998) has for example listed from
psycholinguistic literature factors such as predictability, prototypicality,
concreteness, word frequency/length, as other established influencers on grammatical
choice. Therefore, it appears that the exact factors that influence grammatical
structure in language production are vast and often multiple are present when
forming language.  How these factors
combine to produce language in real-life production is yet to be established (MacDonald,
2013; O’ Seaghdha, 2006). Thus, future studies that define which factors elicit
the most influence on grammatical choices would not only help us to decipher
which factors are most salient in producing syntactic structures but could also
aid the production of a more comprehensive, accurate model of language
production.