Features vs. Form

Schiller & Caramazza (2003)

Utku Turk

LING 449U, How words are built

2026-10-15

Gender in the noun phrase

	Singular	Plural
German	der Stuhl, die Tür, das Buch	die Stühle, die Türen, die Bücher
Dutch	de stoel, het boek	de stoelen, de boeken

Gender is a lexical property of the noun: car is das Auto, la voiture, el coche, de auto.
Number is set by the context: one chair or two.

In both languages the article marks gender in the singular only.

The task: picture-word interference

🪑

KLEID

Say: der Stuhl ‘the chair’
Ignore: Kleid ‘dress’, neuter (das)
Congruent distractor: Keks ‘cookie’, masculine (der)
Measure: time to speech onset

Gender congruency effect: naming is slower when the distractor’s gender differs from the picture name’s.

Where the effect shows up, and where it does not

Found: Dutch (Schriefers 1993; Van Berkum 1997; La Heij et al. 1998), German (Schriefers & Teruel 2000).
Absent with bare nouns: naming the picture without an article (La Heij et al. 1998).
Not found: Italian, Catalan, Spanish, French (Miozzo & Caramazza 1999; Costa et al. 1999; Alario & Caramazza 2002).

Why would Dutch and German show the effect when Romance languages do not?

Early and late article choice

Italian: the article depends on gender and the sound of the next word.
- il tavolo ‘the table’, lo strano tavolo ‘the strange table’
- lo scienziato ‘the scientist’, il grande scienziato ‘the great scientist’
German, Dutch: gender, number, and case alone fix the article.

Dutch and German speakers can choose the article early; Italian speakers must wait for the next word’s sounds.

Two hypotheses

Gender selection interference (GSIH): the distractor’s gender feature competes with the target’s gender feature (Schriefers 1993).
Determiner selection interference (DSIH): the distractor’s article competes with the target’s article (Miozzo & Caramazza 1999).

Both predict a singular effect. What would tell them apart?

Predictions: the plural

Response	Incongruent distractor	Gender differs?	Article differs?
der Stuhl (one chair)	Kleid	yes	yes: der vs. das
die Stühle (two chairs)	Kleid	yes	no: die for every gender

GSIH: effect in the singular and the plural.
DSIH: effect in the singular only.

Eight experiments

Exp.	Language	Response	What varies
1a	German	Det + N	one or two objects
1b	German	Adj + N	one or two objects
1c	German	Det + Adj + N	one or two objects
2a	Dutch	Det + N	one or two objects; three SOAs
2b	Dutch	Det + N	number cued by a tone
3	Dutch	Det + N	singular and plural distractors
4a	Dutch	Adj + N	singular only; extra related distractors
4b	Dutch	Det + Adj + N	same pictures as 4a

Experiment 1: conditions

	Congruent: Keks (der)	Incongruent: Kleid (das)	Response: 1a, 1b, 1c
One object	🪑KEKS	🪑KLEID	der Stuhl roter Stuhl der rote Stuhl
Two objects	🪑🪑KEKS	🪑🪑KLEID	die Stühle rote Stühle die roten Stühle

Same 60 pictures in 1a, 1b, 1c. In 1b and 1c the pictures are red or green; the responses above are for red.

Experiment 1: results

Exp.	Response	Singular effect	Plural effect
1a	Det + N	+20 ms	−9 ms
1b	Adj + N	−3 ms	−6 ms
1c	Det + Adj + N	+16 ms	−7 ms

Effect = incongruent minus congruent. Bold = reliable by participants and by items.

Articles in the singular: effect. One article in the plural: no effect.
Adjective suffix only (1b): no effect in either number.

Why Experiment 2?

Language: the effect was first found in Dutch (Schriefers 1993). Does the plural null hold there too?
Timing: the effect was largest at SOA 0 in Schriefers (1993). Experiment 1 tested only SOA 0, so 2a adds a distractor 100 ms before and 100 ms after the picture.

Experiment 2a: conditions

	Congruent: glas (het)	Incongruent: klap (de)	Response
One object	📖GLAS	📖KLAP	het boek
Two objects	📖📖GLAS	📖📖KLAP	de boeken

Every cell at three SOAs: distractor 100 ms before, with, or 100 ms after the picture.

Experiment 2a: results

Distractor onset	Singular effect	Plural effect
100 ms before the picture	+16 ms	+2 ms
with the picture	+31 ms	+11 ms
100 ms after the picture	+36 ms	+6 ms

Singular effect at every SOA; no reliable plural effect at any SOA.

Experiment 2b: conditions

Concern: on seeing two objects, speakers could pick de at once without checking gender.

	Congruent: glas (het)	Incongruent: klap (de)	Response
Low tone	📖GLAS	📖KLAP	het boek
High tone	📖GLAS	📖KLAP	de boeken

Always one object. The tone comes 300 ms after picture onset.

Experiment 2b: results

Singular effect	Plural effect
+20 ms	−2 ms

The plural null holds even when the article cannot be chosen from the picture alone.

Experiment 3: conditions

So far distractors were singular, so a neuter distractor could activate het.
Plural distractors were added: their article is de for every gender.

	Congruent, sg	Congruent, pl	Incongruent, sg	Incongruent, pl
het boek	📖GLAS	📖GLAZEN	📖KLAP	📖KLAPPEN
de boeken	📖📖GLAS	📖📖GLAZEN	📖📖KLAP	📖📖KLAPPEN

Experiment 3: results

Target	Singular distractor	Plural distractor
Singular	+21 ms	+22 ms
Plural	0 ms	−4 ms

Only the distractor’s gender matters. Number comes from the picture.

Experiment 4: conditions

Congruent	Incongruent	Semantic	Phonological
🐑NEST	🐑KERS	🐑HERT	🐑SCHAAK
nest ‘nest’ (het)	kers ‘cherry’ (de)	hert ‘deer’ (het)	schaak ‘chess’ (het)

Response to a red sheep: rood schaap (4a), het rode schaap (4b).

One object only; pictures red, green, blue, or purple.

Experiment 4: results

Exp.	Response	Congruent	Incongruent	Effect
4a	Adj + N	738 ms	734 ms	−4 ms
4b	Det + Adj + N	750 ms	766 ms	+16 ms

Phonologically related distractors sped naming in both: the distractors were processed.
4a fails to replicate Schriefers (1993, Exp. 2).
4b: reliable by participants, marginal by items (p = .056).

With the same materials, the effect appears only when the response contains an article.

What the results mean

What the eight experiments show

Response	Article differs by gender?	Congruency effect?	Experiments
Singular + article	yes	yes	1a, 1c, 2a, 2b, 3, 4b
Plural + article	no	no	1a, 1c, 2a, 2b, 3
Adjective + noun	no article	no	1b, 4a

The “gender congruency effect” is a determiner congruency effect.

Where is the competition?

Level	Competitive?	Evidence
Gender feature	no	gender still differs in the plural, but no effect
Article form	yes	effect only where the articles differ

Selecting a noun makes its gender available automatically; only forms compete.

Why no effect with adjective suffixes?

Option	Idea
Free vs. bound	free-standing der competes; the suffix -er in rot-er does not
Timing	suffixes attach late, after the competition is resolved

The paper cannot choose between them.

Article competition and serial models

	Word	Gender	Article
Picture, said	Stuhl	masculine	der
Distractor, never said	Kleid	neuter	das

das is not on the screen, so only Kleid’s gender can activate it. Discrete serial models (Levelt et al. 1999) do not allow this; cascading models do.

Is gender skipped in the plural?

If it were, gender competition (GSIH) could survive the plural null. Naming without distractors:

	Masculine	Feminine	Neuter
Singular	der 653 ms	die 655 ms	das 647 ms
Plural	die 673 ms	die 649 ms	die 681 ms
Plural cost	+20 ms	−6 ms	+34 ms

The singular article competes with die: gender is still active in the plural.

Agreement needs plural gender

Phrase	Pronoun	Gender from
eine der Katzen miaute	feminine	Katzen, plural
einer der Hunde bellte	masculine	Hunde, plural

‘one of the cats meowed’, ‘one of the dogs barked’

A Distributed Morphology reading

Not in the paper.

Vocabulary Item	Inserted in
D[definite, neuter, singular] ↔︎ /het/	het boek
D[definite] ↔︎ /de/	de tafel, de tafels, de boeken

Impoverishment: delete [gender] on plural D. The noun keeps its gender, so gender competition on D is possible only in the singular.

Testing the Impoverishment reading

Finding	Impoverishment	Article competition (DSIH)
Singular effect only	yes	yes
Plural cost: +20, −6, +34 ms	no	yes
No adjective effect	only if suffixes are late	yes