Girls know almost 20 per cent more words than boys by the age of 2, a new study has found.
The Growing Up in New Zealand study, which is following almost 7000 children born in Auckland and the Waikato in 2009-10, has also found that 2-year-olds know 30 per cent more words in rich neighbourhoods than in poor ones.
Lead author Professor Elaine Reese says the gender and the socio-economic gaps persist throughout childhood, and girls and richer students do better than boys and poorer students in reading and writing in national standards in primary schools and in the Pisa (Programme in International Student Achievement) tests at age 15.
"The fact that we are seeing these gender differences and other differences so early is a concern," she said.
"The good news is we know what we can do to change it - finding books that boys will be interested in looking at, and having conversations with boys about things they want to talk about."
The study gave parents lists of 100 common English words and asked which words their children had spoken by age 2. They were allowed to count words the children pronounced incorrectly, such as "sketti" for "spaghetti".
On average, girls had spoken 51 of the words by age 2, but the boys had spoken only 43 of the words.
Children had spoken 52 of the words in the richest third of areas, 49 in the middle third and 40 in the poorest third.
There were similar differences between children of mothers with degrees (52 words) compared with mothers whose highest education was intermediate school (40 words), and between Europeans (52 words) and non-Europeans (34 words).
There was also a slight advantage for first-born children (49 words) over later children (45 words).
Reese said the gap between girls and boys showed up in many studies around the world, but the causes were still being debated.
"When people interpret the reading gap [at older ages], there are those who point to boys spending more time playing video games and girls spending more time reading for pleasure," she said.
"I would say that the main cause is how we are talking to boys and girls, that's my hunch because we know what a huge difference the language environment makes for children's language development.
"There are several theories about that. One is that it's a stereotyping behaviour where we end up doing quiet activities more often with girls, including conversation and reading books.
"There are other theories that boys are less interested in those quiet activities and are doing more rough and tumble play."
A test on a mother-child interaction task similar to a looking at a picture book found that mothers living in poorer areas asked their children fewer open-ended questions than mothers in richer areas.
"Mothers from medium-deprivation areas were also using fewer emotion words compared to mothers in less deprived [richer] areas," Reese said.
"Mothers were also using more disciplinary strategies with boys than girls, a pattern that probably speaks to the difficulty at times of engaging young boys in quiet activities like book-reading."
The study also found that 12 per cent of the children, including 40 per cent of Maori children, were described by their mothers as able to understand te reo Maori.
However only three of the 584 children whose mothers completed the questionnaire in Maori spoke only Maori words at age 2, and the other 581 spoke far more English than Maori words - 41 English words and four Maori words for girls, and 34 English words and five Maori words for boys.
"For the bilingual children, English may be already dominating their Maori language acquisition," the authors wrote.
"The future of te reo Maori depends upon this generation of NZ children and their parents, grandparents and teachers acquiring and using Maori in everyday speech at home and in educational settings."
Funding for the Growing Up in NZ study has been cut, forcing the researchers to survey only 2000 of their original 7000 children in the next data collection wave. Almost 4000 people have signed a petition asking Prime Minister Bill English to restore its funding.